Aquifer Systems Management Darcy's Legacy in A World of Impending Water Shortage

AQUIFER SYSTEMS MANAGEMENT
DARCYS LEGACY IN A WORLD OF IMPENDING WATER SHORTAGE

SELECTED PAPERS ON HYDROGEOLOGY
10
Series Editor: Dr. Nick S. Robins
Editor-in-Chief IAH Book Series
British Geological Survey
Wallingford, UK
INTERNATIONAL ASSOCIATION OF HYDROGEOLOGISTS
Aquifer Systems
Management
Darcys Legacy in a World of
Impending Water Shortage
Edited by
Laurence Chery
BRGM, The French Geological Survey, Orleans, France
Ghislain de Marsily
University Paris VI and French Academy of Sciences, Paris, France
LONDON / LEIDEN / NEW YORK / PHILADELPHIA / SINGAPORE
Taylor & Francis is an imprint of the Taylor & Francis Group, an informa business
2007 Taylor & Francis Group, London, UK
All rights reserved. No part of this publication or the information contained herein may be
reproduced, stored in a retrieval system, or transmitted in any form or by any means,
electronic, mechanical, by photocopying, recording or otherwise, without written prior
permission from the publishers.
Although all care is taken to ensure integrity and the quality of this publication and
the information herein, no responsibility is assumed by the publishers nor the author
for any damage to the property or persons as a result of operation or use of this
publication and/or the information contained herein.
Published by: Taylor & Francis/Balkema
P.O. Box 447, 2300 AK Leiden, The Netherlands
e-mail: Pub.NL@tandf.co.uk
www.balkema.nl, www.taylorandfrancis.co.uk, www.crcpress.com
Library of Congress Cataloging-in-Publication Data
Aquifer systems management : Darcys legacy in a world of impending water shortage :
selected papers from the International Association of Hydrogeologists (IAH) Dijon
symposium, Dijon, France, 30 May-1 June 2006/edited by Laurence Chery.
p. cm. (Selected papers on hydrogeology ; 10)
Includes bibliographical references and index.
ISBN 978-0-415-44355-5 (hardcover : alk. paper) 1. AquifersCongresses. 2. Groundwater
ManagementCongresses. 3. Darcy, Henry, 18031858Congresses. I. Chery, Laurence.
II. Darcy, Henry, 18031858. III. International Association of Hydrogeologists.
GB1199.A68 2007
551.19dc22 2007013656
ISBN13: 978-0-415-44355-5 (Hbk)
ISBN for e-book: 978-0-203-93459-3
Cover photograph: Rosoir Spring as the source of Dijons water supply designed by Henry Darcy
Copyright cover photograph: Gilles Creuzot DIREN Bourgogne
Sponsors:
Schlumberger
WAT E R S E R V I C E S
This edition published in the Taylor & Francis e-Library, 2007.
To purchase your own copy of this or any of Taylor & Francis or Routledges
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.
ISBN 0-203-93459-8 Master e-book ISBN
Contents
Preface xi
About the editors xix
PART I: A TRIBUTE TO HENRY DARCYS LEGACY
Chapter 1. Henry Darcy (18031858): Immortalised by his scientific legacy 3
Craig Simmons
Chapter 2. Henry Darcys Public Fountains of the City of Dijon 25
Patricia Bobeck
PART II: THE MEASURE OF PERMEABILITY
Chapter 3. What grains can tell us about Darcy velocity 35
Maria Schafmeister
Chapter 4. Measuring porosity and permeability on drill cuttings 47
Roland Lenormand & Patrick Egermann
Chapter 5. The quest for permeability evaluation in wireline logging 55
Jean-Pierre Delhomme
Chapter 6. Aquifers imagery and hydrodynamic parameters estimation using proton
Magnetic Resonance Soundings 71
Jean-Franois Girard, Jean-Michel Baltassat, Marie Boucher,
Anatoly Legchenko, Jean-Michel Vouillamoz, Alexis Gutierrez,
Marie-Luce Noyer & Patrick Lachassagne
Chapter 7. Permeability measurements in argillaceous rocks at the
Meuse/Haute-Marne Underground Research Laboratory, France 87
Jacques Delay
PART III: COMPLEX HYDROGEOLOGICAL SYSTEMS
A. Large aquifers, assessment of the resource
Chapter 8. Great aquifer systems of the World 105
Jean Margat
Chapter 9. Alluvial aquifers in the North China plain 117
Zaisheng Han
Chapter 10. Groundwater and the challenges for the future water supply of
the Republic of Djibouti in an arid climate (Horn of Africa) 127
Mohamed Jalludin
Chapter 11. Hydrogeological structure of the aquifer system in the Taoudenni
sedimentary basin, Burkina Faso 137
Johan Derouane & Denis Dakour
Chapter 12. An assessment of the groundwater resources in the western margin
of the Taoudenni basin, Mauritania 149
Franois Bertone, Philippe Renard, Jaouher Kerrou, Patrice Moix &
Pierre Perrochet
Chapter 13. Groundwater modelling for conjunctive use patterns investigation
in the upper Central Plain of Thailand 161
Werapol Bejranonda, Sucharit Koontanakulvong,
Manfred Koch & Chokchai Suthidhummajit
Chapter 14. Modelling of the Saq aquifer system (Saudi Arabia) 175
Yves Barthlemy, Olivier Bon, Yves-Michel le Nindre,
Syed Munaf, Dominique Poitrinal, Alexis Gutierrez,
Michel Vandenbeusch, Abdullah Al Shoaibi & Marcus Wijnen
Chapter 15. Holocene and present recharge of the Saharan aquifers: A numerical
modelling study 191
Mohamedou Ould Baba Sy & Mustapha Besbes
Chapter 16. Vadose zone characterisation of a hydrogeologic system in a mountain
region: Serra da Estrela case study (Central Portugal) 207
Jorge Espinha Marques, Joo M. Duarte, Alfredo T. Constantino,
Armindo Afonso Martins, Carlos Aguiar, Fernando T. Rocha, Manuela Incio,
Jos M. Marques, Helder I. Chamin, Jos Teixeira, Javier Samper,
Frederico S. Borges & Jos M. Carvalho
B. Large aquifers, water salinity and its evolution with exploitation
Chapter 17. Reconstructing natural groundwater flow and geochemical
processes in a perturbed multi-layer aquifer system 225
Dallas Baird, Corinne Le Gal La Salle, Andrew Love &
Craig Simmons
Chapter 18. The Guaran Aquifer System: state-of-the-art in Argentina 239
Ofelia Tujchneider, Marcela Perez, Marta Paris &
Mnica DElia
Chapter 19. Defining groundwater flow and mass vertical exchanges in a
regional aquifer system through geochemical modelling Aquitaine
Basin, France 253
Laurent Andr, Michel Franceschi & Olivier Atteia
vi Contents
Chapter 20. Hydrogeology of Bni Abbs: potential, hydrodynamics and
influence on the palm field (Valley of Saoura, Algerian South-West) 269
Touhami Merzougui , Abderrahmane Mekkaoui, Hamidi Mansour &
Khadidja Graine-Tazrout
Chapter 21. Groundwater system compartmentalized by a tectonic zone in the
Kanto plain, central Japan 281
Masaya Yasuhara, Akihiko Inamura, Masaaki Takahashi, Takeshi Hayashi,
Hiroshi Takahashi, Masahiko Makino, Hiroko Handa & Toshio Nakamura
C. Karstic and carbonate aquifer systems
Chapter 22. Hydrogeological consequences of the Messinian Salinity Crisis in
the Rhone River basin, France 291
Rmi de La Vaissire, Frdric Lalbat & Bernard Blavoux
Chapter 23. Hydrodynamic behaviour during pumping test and modelling
of the Cent Fonts karst system 303
Bernard Ladouche, Nathalie Drfliger, Jean-Christophe Marchal,
Patrick Lachassagne, Michel Bakalowicz, Irinia Valarie &
Philippe Lenoir
Chapter 24. Multidisciplinary approach to a karstic region for the use and protection
of the water resource. Application to the Causse de Sauveterre (South of
France) 317
Michel Lepiller, Bernard Blavoux, Stphane Brusset, Laurent Bruxelles,
Laurent Danneville, Alain Mangin & Pierre Marchet
Chapter 25. Hydrogeological characterisation of the Oxfordian limestone at the
Bure URL 333
Martin Cruchaudet, Jacques Delay & Marc Distinguin
D. Geothermal aquifer systems
Chapter 26. Tracing high pH mineral waters ascribed to ultramafic rocks
(Central Portugal): Conceptual vs numerical modelling 351
Jos M. Marques, Maria J. Matias, Paula M. Carreira,
M. Rosrio Carvalho, Fraser Goff, Maria J. Basto, Rui C. Grac
,
a,
Mrio Andrade, Lus Rocha & Lus Aires-Barros
Chapter 27. Fluid/mineral equilibrium calculations, isotopes and geophysics as
a multidisciplinary approach to the characterization of Mono
hydrothermal system (NW-Portugal) 365
Paula M. Carreira, Jos M. Marques, M. Rosrio Carvalho,
Fernando A. Monteiro Santos, Hugo Matias, Rafael Luzio & Dina Nunes
Chapter 28. Predictive model for the management of thermal resources 379
Alain Dupuy, Ronan le Fanic & Pierre Pouchan
Contents vii
E. Aquifer contamination studies
Chapter 29. Urban hydrogeology of the Paranhos sector, Porto city (NW Portugal):
A geoenvironmental perspective 391
Maria Jos Afonso, Jos M. Marques, Laura Guimares, Isilda Costa,
Jos Teixeira, Clara Seabra, Fernando T. Rocha, Lcia Guilhermino &
Helder I. Chamin
Chapter 30. Modelling of chlorinated solvent transport and natural attenuation in
groundwater 407
Fabrice Quiot, Claire Rollin, Olivier Bour, Salvador Jordana,
Eduardo Ruiz, Jordi Guimer, Jean Schwartz & Patrick Goblet
Chapter 31. Variable water saturation impact on sorption/desorption characteristics
of multiple species heavy metals from gold mining 419
Aksara Putthividhya & Srilert Chotpantarat
Chapter 32. Aquifer protection: A combined study of microstructure and transport
properties in clay-sand mixtures 435
Paolo Macini, Ezio Mesini & Rossella Viola
Chapter 33. Modelling aquifer salinity in the Potash Basin (Alsace) 445
Marie-Luce Noyer & Philippe Elsass
Chapter 34. A multidisciplinary approach for assessing the risk of seawater intrusion
in coastal aquifers: The case of the Roussillon Basin (France) 459
Bertrand Aunay, Nathalie Drfliger, Cdric Duvail, Frdric Grelot,
Paul Le Strat, Marielle Montginoul & Jean-Daniel Rinaudo
Chapter 35. Cost-Benefit analysis of the alternative scenarios for reducing
diffuse nitrogen pollution in the Seine River basin 471
Christophe Viavattene, Jean-Marie Monget, Emmanuel Ledoux &
Bernard Barraqu
Chapter 36. Integrating economic and groundwater models for developing long-term
nitrate concentration scenarios in a large aquifer 483
Nina Graveline, Jean-Daniel Rinaudo, Volker Segger, Hendrik Lambrecht,
Markus Casper, Philippe Elsass, Jost Grimm-Strele, Thomas Gudera, Remi
Koller & Paul Van Dijk
Chapter 37. Simplified vulnerability mapping of groundwater in the Seine-Normandy
Basin. Application to the BASOL sites 497
Vincent Mardhel, Annabel Gravier, Jean-Yorick Koch-Mathian,
Cline Nowak, Jean-Luc Terreyre, Mireille Raguet & Ccile Garnier
F. Aquifer monitoring systems and management rules
Chapter 38. Characterization of pesticide concentrations in the Craie du Nord aquifer
system 513
Didier Renard & Chantal de Fouquet
viii Contents
Chapter 39. Optimisation of groundwater monitoring networks: application of
geostatistics with case studies from a granitic aquifer in a semi-arid region 527
Shakeel Ahmed, Aadil Nabi, Shazrah Owais & Dewashish Kumar
Chapter 40. Use of piezometric networks for evaluating the quantitative status of
groundwater bodies: Example of the Seine-Normandy Basin in France 541
Jean-Francois Vernoux
Chapter 41. Taking groundwater into account in a drought plan 553
Philippe Verjus
Chapter 42. Risk map of high natural background of trace elements in the Water
Authority Rhne-Mditerrane & Corse basin 563
Agns Brenot, Ariane Blum, Laurence Chery, Romain Sonney,
Laurent Cadilhac & Thomas Pelte
Chapter 43. Groundwater and Soil Water System for Norway based on daily
simulations and real-time observations 569
Herv Colleuille, Stein Beldring, Zelalem Mengistu, Wai Kwok Wong &
Lars-Egil Haugen
Index 583
Author index 587
Series IAH-Selected Papers 589
Contents ix
Preface
This book contains a selection of 43 papers presented at the International Association of
Hydrogeologists (IAH) Dijon Symposium, which was held from May 30th, to June 1st, 2006,
by IAH with the Bureau de Recherches Gologiques et Minires (BRGM), the French
Geological Survey, in association with the International Association of Hydrological Sciences
(IAHS), the International Association of Engineering and Environmental Geology (IAEG),
and the Public Works Hydrogeologists Association (PWHA). The Symposium was held to
celebrate the 50th anniversary of the creation of IAH, which took place at the International
Geological Congress of Mexico City in June 1956, and the 150th anniversary of the pub-
lication of Darcys law in 1856. Darcy made his discovery while working in Dijon, where he
had been in charge of resolving the citys water supply shortage. The combined 50th anniver-
sary of the IAH and 150th of Darcys Law provided a unique opportunity for the community
of hydrogeologists to examine aquifer systems management globally and to explore how
Darcys legacy plays a critical role in the science and management of aquifers in todays
world of impending water shortage. This book was published in colour thanks to the sponsor-
ing of this edition by Andra, the French National Agency for Radioactive Waste Disposal,
BRGM, the French Geological Survey, the Norwegian Water Resources and Energy
Directorate, Schlumberger Water Services, IAH and individual contributing authors. Their
support is gratefully acknowledged.
Among the 130 papers presented at the Symposium, 43 have been selected in which their
aim is to find solutions, or develop new solutions, to solve todays water shortage prob-
lems, not just in Dijon, as Darcy did, but the whole world over Let us start with some
basic facts: today there are 6.2 billion inhabitants on planet Earth, 14% of whom are
already suffering from hunger, a number that has been increasing for the last 5 years. In
2050, the global population is predicted to rise to about 9 billion. To supply food for that
many people, will require roughly 75% more water than we use today, mostly for agricul-
ture, while maintaining natural ecosystems that are healthy and sustain biodiversity; in
comparison, the other uses of water are relatively minor. This is a tremendous task that will
require all of our energy and commitment. It will mostly be achieved by expanding rain-
fed agriculture, on todays forested or uncultivated land
1
, while trying to protect natural
ecosystems. But rainfed agriculture will not be enough, more water will also have to be
provided for irrigation, increasing its efficiency and controlling soil salinisation; for indus-
trial and domestic use, water quality will be the main issue. Natural climate variability,
possibly enhanced by forthcoming climate changes, will require that water be made avail-
able even when droughts occur, droughts that may occasionally affect several continents at
1
See e.g. Les Eaux Contnentales, Coordinated by G. de Marsily, EDP Sciences, Paris, 2006, 390 p. or
M.Griffon (2006) Nourrir la Plante, Odile Jacob, Paris, or International Water Management Institute (2007)
Water for food, water for life : the Comprehensive assessment of water management in agriculture, Report to be
published, Colombo, Sri Lanka, early 2007.
the same time, as has already happened historically e.g. in 18761878, or 18981900, in
Brazil, China, India and Ethiopia, causing dramatic famines
2
.
Groundwater will be a major player for irrigation and domestic use. Surface water stor-
age capacity will have to be increased globally, but aquifers are the major natural means to
store water during wet years and make it available during droughts. We need to deplete the
water table, so that water from wet years can be stored, but we also need to prevent the water
table from getting too low, with totally depleted reserves, where only the yearly recharge
can be exploited. We also need to control the salinisation, from seawater intrusion, from
deep saline waters, or from evaporated superficial water. All this we can do, with proper man-
agement of the aquifer. But in many cases, aquifers are in fact not managed with all these
objectives in mind, or in fact not managed at all, and only exploited at the highest possible
level. As hydrogeologists, we need to explain, demonstrate and convince ourselves, water
managers and decision makers, politicians and the community at large that aquifers can be
managed optimally and sustainably, for the benefit of all.
To be able to do that, very little time is left to learn how our aquifer systems operate,
what their reserves are, and how to manage them in order to be able to meet tomorrows
challenges. The 43 selected papers of the Dijon Symposium all deal with one (or several)
aspects of the problem, in different parts of the world and under different climates. They
more often than not use Darcys law, the simple relation between hydraulic head and flow
velocity in porous media, which allows us to make predictions of both current and future
aquifer behaviour. In the same way, the papers follow in Darcys footsteps, on a quest for
solutions of practical water supply problems and to better understand how hydrogeologic
systems work. It is thus most fitting that the symposium and this book focus on aquifer
systems management and Darcys legacy in a world of impending water shortage.
The papers have been grouped into three parts
3
.
Part I is a tribute to Henry Darcys life and scientific legacy
Two non-French hydrogeologists, Craig Simmons and Patricia Bobeck, tell us about Darcy.
Craig provides a superb overview of Darcys life and his contribution to science, including,
but not limited to, his linear flux-gradient law. Patricia, who recently completed an English
translation of the entire Darcy book Les Fontaines Publiques de la Ville de Dijon
4
, summa-
rizes the major sections of the book and presents some of Darcys engineering achievements
5
.
Part II is a short collection of invited papers on the measurement of permeability
In 1856, Darcy developed the first permeameter, an instrument which is still in use today.
But today in this modern age, do we have other means of measuring permeability, with
other types of equipment?
xii Preface
2
See e.g. A. Sen & J. Drze (1999) Omnibus, Oxford University Press, New Delhi, or M. Davis (2001) Late
Victorian Holocausts, El Nio Famines and the Making of the Third World, Verso, London (also available in
French, Gnocides Tropicaux, La Dcouverte, Paris, 2003 and 2006).
3
The Editors wish to warmly thank Dr. Nick S. Robins, from BGS, UK, who, as Book Editor in Chief of IAH,
revised and edited all the papers of these proceedings.
4
Bobeck, P. (2004) The Public Fountains of the City of Dijon by Henry Darcy (1856). Dubuque, Iowa: Kendall/Hunt
Publishing Company.
5
See also: Henry Darcy, le bicentenaire 18032003. Edition spciale Colloque International Gestion des
Grands Aquifres. 150
me
anniversaire de la loi de Darcy, 50
me
anniversaire de lAIH. Dijon. Coordination
Eliane Lochot. ISBN 2-915-128-28-6, available as a CD in English at IAH.
Maria Schafmeister presents an overview of the indirect estimate of permeability from
grain-size analysis on unconsolidated clastic sediments, a very simple method that is widely
used in Germany but often ignored elsewhere;
Roland Lenormand and Patrick Egermann show how permeability can be ingeniously
measured on very small borehole cuttings;
Jean Pierre Delhomme shows how, for the last 50 years, borehole geophysical logging
has tried to indirectly measure permeability in situ with logging tools, and has almost suc-
ceeded;
Jean Franois Girard et al. present a new surface geophysics tool, the MRS (Magnetic
Resonance Sounding method), which is able to firstly detect the presence of mobile water
at depth, and secondly, to some degree, its mobility i.e. the rock permeability;
Jacques Delay finally presents a very comprehensive series of in situ experiments to
measure the very low permeability of a clay formation, using advanced testing equipment
and methods.
Part III assembles 36 papers that explore complex hydrogeological systems
These papers address the following questions: what is the resource? How can it be stud-
ied? What is the water quality and what is its foreseeable evolution? How can the aquifers
be managed? We grouped the papers into six major topics:
Large aquifers, assessment of the resource: 9 papers
Large aquifers, water salinity and its evolution: 5 papers
Karstic and carbonate aquifer systems: 4 papers
Geothermal aquifer systems: 3 papers
Aquifer contamination studies: 9 papers
Aquifer monitoring systems and management rules: 6 papers
1 LARGE AQUIFERS, RESOURCE ASSESSMENT
J. Margat, former President of the French Chapter of IAH, first presents a study including
maps of the Worlds major aquifer systems, which occupy an area of 35 million km
2
, i.e.
one fourth of the continental surfaces, and hold the large majority of the freshwater
resources;
Z. Han reports on one of East Asias major aquifers, the alluvial plain of North China,
where the sediment thickness can reach 3,000m and the permeable aquifers 500m. This
huge reservoir is heavily exploited in this semiarid region, and its exploitation requires
understanding of current recharge and reserves, using in particular geochemical and iso-
topic tools, and the study of the consequences of subsidence;
M. Jalludin describes the difficult problem of Djiboutis (in East Africa) water resources,
with predictions made up to the year 2015, and shows the dominant role that groundwater
will play, mostly from complex basaltic aquifer systems;
J. Derouane and D. Dakour study the major sedimentary Taoudenni basin in Burkina-
Faso and Mali, extending over 40,000km
2
with a thickness of 2,000m. This system consti-
tutes the major water resource of the country, increasingly exploited, both for drinking water
and agriculture. Geological and hydrogeological models are used to assess current recharge,
Preface xiii
and the vertical conductivity between aquifers, which is of utmost importance for their
management.
F. Bertone et al. provide an extensive study of the hydrogeology of the Zouerate region
in North-Western Mauritania, containing the Archaean basement and the margin of the
same Taoudenni basin. Geological mapping, remote sensing, aerial magnetometric data,
and a survey of 624 groundwater occurrences were used. A 2-D regional model was cali-
brated to compare different recharge scenarios and to run long-term simulation scenarios.
W. Bejranonda et al. describe the water supply problem in the Central Plain of Thailand
for irrigation, domestic and industrial uses. Conjunctive use of surface water and ground-
water is analysed with a flow model of the regional aquifer, taking into account exchanges
with the rivers; optimised extraction scenarios for wet, normal, dry or drought situations
are proposed.
Y. Barthlemy et al. discuss the water resources of the Saq aquifer system in Saudi
Arabia. Twenty years ago, a first groundwater model of the Saq was built. A huge agricul-
tural development took place, generating important drawdowns; two new models are now
constructed based on 20 years of data to improve the understanding and exploitation of the
aquifer.
M. Ould Baba Sy and M. Besbes try to estimate the current recharge in one of the
largest aquifer in the World, the Northern Sahara desert system. By running an existing
groundwater model of the system over 10,000 years, they show that it is impossible to
reproduce the long-term evolution and the present status of the aquifer if no recharge is
assumed. They then quantify the necessary long-term recharge on the outcrops of the
Saharan Atlas.
J. Espinha-Marques et al. are interested in estimating recharge and recharge water qual-
ity in the mountains of Central Portugal by in situ measurements. Their aim is to estimate
the recharge to the regional aquifers in these areas where rainfall is higher than in the
plain.
2 LARGE AQUIFERS, WATER SALINITY AND ITS EVOLUTION WITH
EXPLOITATION
D. Baird et al. are interested in the multilayer aquifer system of the Northern Adelaide
Plains in Australia, where the deep freshwater aquifer is now at risk of being polluted by
salty superficial water, due to the inversion of the hydraulic gradient by heavy exploitation
and vertical leakage. They use hydraulics and hydrochemistry to assess the problem.
O. Tujchneider et al. present one of the worlds most important aquifers, the Guarani
system, which extends over four countries in South America and is increasingly exploited.
Made of aeolian and fluvial sandstone, it is covered by Cretaceous basalts providing strong
confinement which makes hydrogeological studies difficult. Analysis of the geological
and hydrochemical data in Argentina outlines the high salinity of some portions of the
aquifer and the occurrence of deep thermal groundwater.
L. Andr et al. use geochemical data and modelling to understand the aquifer/aquitard
interactions in a deep sedimentary basin, the Aquitaine basin in France, to understand the
flow direction and the origin of water mineralisation.
T. Merzougui et al. study the water resources of the Bni-Abbs oasis in Algeria. Four
aquifers are exploited, the major one being the deep great eastern desert aquifer with good
xiv Preface
quality water, but with salinisation problems in the upper terrace aquifers as a consequence
of irrigation and evaporation, which presents a threat to the main freshwater resource.
M. Yasuhara et al. describe the Kanto plain, the largest Quaternary groundwater basin in
Japan. This basin is subdivided into three distinct hydrologic areas by a central tectonic
zone, which might be the cause of major salinity differences, characterised by chloride and
isotopic studies. The water in the central tectonic zone was recharged in colder climatic
conditions and has had a long interaction with the formation, giving it its salinity.
3 KARSTIC AND CARBONATE AQUIFER SYSTEMS
R. de La Vaissire et al. present three examples of karstic aquifers in the Rhone valley
and show the strong influence on these systems of the Messinian salinity crisis in the
Mediterranean, which was responsible for the deepening of the valleys, the development
of the karst at great depth, and the filling-up of the valleys by Pliocene low-permeability
deposits; these features have a great impact on the present functioning of the aquifers.
B. Ladouche et al. report on the study of the Cent Fonts karstic aquifer in Southern France,
which is to be exploited seasonally, and therefore requires an estimate of its exploitable
reserves. A long duration pumping test was run and is interpreted with a special model com-
bining a reservoir and an analytical solution for a linear drain, to estimate these reserves.
M. Lepiller et al. present an integrated study of the large karstic system of the Causse
de Sauveterre in Southern France. The feeding zone of each spring, the discharge history,
the water chemistry, the geology and geomorphology are used together to analyse this sys-
tem, and to determine its vulnerability in order to define regulated protection zones.
M. Cruchaudet et al. use interference and local permeability tests in boreholes to esti-
mate the permeability and storativity of a deep low-permeability limestone formation,
located above a thick clay layer in the Paris basin where deep storage of nuclear waste is
envisaged.
4 GEOTHERMAL AQUIFER SYSTEMS
J.M. Marquez et al. are interested in the origin of high pH mineral waters in ultramafic
rocks in Central Portugal, and use geochemical and isotopic tools to understand the reac-
tion path in the waterrock interaction, to investigate this unique spa resource.
P.M. Carreira et al. also investigate deep thermal mineral waters in Mono, N-W
Portugal, to reconstruct with geothermometers the temperature at a depth of 3.4km in the
reservoir. They use geophysical, geochemical and isotopic tools to develop a conceptual
flow model.
A. Dupuy et al. use 3-D modelling to study the geothermal resources of the Dax spa in
the Aquitaine basin in France, in order to understand the flow path, and manage both the
quantity and the quality of the resource.
5 AQUIFER CONTAMINATION STUDIES
M.J. Afonso et al. are interested in an ancient water supply system in Porto City, Portugal,
where horizontal drainage galleries have been used for more than 500 years. But urban
Preface xv
contamination sources make this water vulnerable, and its use for irrigation therefore
requires the identification of and protection from contaminant sources.
F. Quiot et al. describe the natural attenuation of chlorinated solvents by biodegradation
in aquifers, with a real case from a contaminated alluvial aquifer in northern Spain, where
four different teams have compared their model results with the aim of improving the
modelling.
A. Putthividhya and S. Chotpantarat study groundwater pollution by gold mining activ-
ity. They focus on the impact of variable water saturation on the sorption of heavy metals
(Cd, Ni, Zn, Pb) from mining leachates on soils, illustrated by the example of the Akara
mine in Thailand.
P. Macini et al. study the functioning of sand-clay mixtures for their swell-shrinkage
properties and their permeability with variable water contents, for their role as barriers for
the protection of aquifers beneath waste landfills.
M.L. Noyer and Ph. Elsass describe the pollution of the Rhine alluvial aquifer by salt
dumps from potash mining in Alsace, Eastern France. The long-term evolution of the
salinity of the aquifers, in different remediation scenarios, is analysed through modelling.
B. Aunay et al. focus on a highly developed coastal aquifer in the Rousillon area, in
Southern France, to model the system and design effective water management strategies to
protect it from saltwater intrusion; they then present a socio-economic study of the inter-
ests at stake, of the diversity of viewpoints expressed by stakeholders, and of the political
dimension of the issue, all of which need to be taken into account for decision making.
C. Viavattene et al. examine two scenarios of water quality evolution regarding nitrates
in the Paris basin, business as usual and agro-environmental measures. A cost-benefit
analysis is used to compare the scenarios. The farmers costs are determined from their
direct actions and crop yield variations, while the benefits evaluation is the avoided costs
for the drinking water consumers.
N. Graveline et al. study the nitrate contamination of the Upper Rhine alluvial aquifer,
between France and Germany. Different models were built of economical, physico-chemical
and hydrogeological processes and their interactions. The economic model represents the
farmers production choices and practices. Several farming scenarios are tested up to the
year 2015.
V. Mardhel et al. deal with the mapping of aquifer vulnerability. They propose a new
index of persistence and development of networks (IDPR), based on an analysis of the
drainage network, which measures the capacity of the geological formations to generate
runoff or let water infiltrate. The Paris basin example shows the value of this new method.
6 AQUIFER MONITORING SYSTEMS AND MANAGEMENT RULES
D. Renard and Ch. de Fouquet give an interesting example of the use of geostatistics to
estimate the spatial and temporal distribution of pesticides in the Chalk aquifer in
Northern France. They analyse the data of three networks, which describe the water qual-
ity for different regulatory purposes, and each one gives a different perspective on the spa-
tial distribution.
S. Ahmed proposes another example of the use of geostatistics for optimising the loca-
tion of piezometers in a granitic aquifer in a semi-arid zone of Central India, both for mon-
itoring the piezometric head, and to study the fluoride distribution in the aquifer.
xvi Preface
J.F. Vernoux reflects on the design of a piezometric network to monitor the groundwater
resource in terms of quantity over a large basin, the Paris basin which includes 350 obser-
vation points. The various criteria for selecting and giving a hierarchy to these piezometers
are presented and discussed within the European Water Framework Directive.
Ph. Verjus examines the role that can be assigned to aquifers, in case of drought, when
decisions are needed to restrict the water supply and to allocate groundwater resources to
priority users, with an example in the Paris basin, France, during the 2003 drought.
A. Brenot et al. determine the risk of high concentrations in water bodies for 17 trace
elements by identifying the natural geochemical background, as required by the European
Water Framework Directive and its daughter Directive on the protection of groundwater
against pollution. The French Alps are taken as an example for the proposed methodology,
including the use of a GIS and of geochemical characterization.
H. Colleuille et al. describe a new model to generate daily soil moisture and groundwa-
ter maps based on real-time observations for the entire country of Norway, with a 1km
2
resolution. This tool is intended to provide national and local stakeholders with an innova-
tive way of assessing groundwater-related risks.
Orlans and Paris, January 15th, 2007 Laurence Chery and Ghislain de Marsily
Editors
Preface xvii
About the editors
Laurence Chery was awarded a Ph.D. in Hydrogeology at the Paris XI University
(Laboratory of Hydrology and Isotope Geochemistry). She started her professional career in
1988 as hydrogeologist in a private consulting firm in France. In 1990, she joined the Water
division at BRGM (the French Geological Survey) as a research engineer, in charge of
methodological studies as part of Public Service activities in the field of Water Quality. She
is presently in charge of a national groundwater database project, ADES, the quality and
quantity national database for groundwater. Her work includes the design of new projects
for solving specific problems in water quality, database, monitoring networks, and fieldwork.
The main subjects include hydrogeochemistry, water circulation, natural groundwater geo-
chemical background, water origin and transfer time, using isotopic techniques. She has
published several public reports, 30 articles and a book on Groundwater natural background
in France. She has been the secretary of the French Chapter of IAH since 2001.
Ghislain de Marsily is Emeritus Professor at the University Pierre et Marie Curie, Paris, VI.
He was trained as a mining engineer, and initially worked in oil well drilling and dam con-
struction, before going into teaching and research at the Paris School of Mines in
Fontainebleau (19671987) and later at the University Paris VI, on quantitative hydrogeology,
geostatistics, aquifer modelling, waste disposal and water management at the large catchment
scale. He has published several books, including the text book Quantitative Hydrogeology
at Academic Press, NY, and about 150 articles. He was an Editor of the J. of Contaminant
Hydrology for 7 years. He is a member of the French Academy of Sciences and Foreign
Associate of the US National Academy of Engineering.
I A tribute to Henry Darcys legacy
CHAPTER 1
Henry Darcy (18031858): Immortalised by his
scientific legacy
Craig Simmons
Flinders University, Adelaide, SA, Australia
ABSTRACT: This paper discusses Henry Darcys distinguished contributions to science and engi-
neering. In addition to Darcys Law, Darcy made major contributions to pipe hydraulics and pipe
friction coefficient analyses that are acknowledged in the joint naming of the Darcy-Weisbach pipe
head loss equation. He also furnished the very first evidence of the fluid boundary layer. Darcy com-
bined his law with continuity to develop the first falling head permeameter solution that we use today
and he applied it in the analysis of spring discharge. His work on open channel flow with Bazin and
his breakthroughs in pipe friction research were made possible through improvements he made to the
Pitot tube used for measuring point water velocity. This paper will demonstrate that Darcy made
numerous contributions to the study of hydraulics. Whilst Darcy is immortalised by Darcys Law, it
is immediately evident that his scientific legacy extends beyond it.
1 INTRODUCTION
Darcys Law is the fundamental equation describing the flow of fluid through porous media
including groundwater. It forms the quantitative basis of many science and engineering
disciplines including hydrology, hydrogeology, soil science, civil engineering, petroleum
engineering and chemical engineering. The year 2006 marks the 150th anniversary of the
publication of Henry Darcys most famous text Les Fontaines Publiques de la Ville de
Dijon (The Public Fountains of the City of Dijon) (Darcy, 1856). Buried in its depths was
Note D, an appendix that contained the famous sand column experiments and the discovery
of Darcys Law a discovery that marked the birth of quantitative hydrogeology.
This paper describes the many contributions Darcy made to hydraulics, including Darcys
Law. But what many hydrogeologists may not realise is that Darcy made other contribu-
tions to science and engineering that we are possibly less familiar with. He was the first to
describe aquifer resistance, he furnished the very first evidence of the fluid boundary
layer, he made major contributions to pipe hydraulics as evidenced by the joint naming of
the commonly used Darcy-Weisbach pipe friction equation, he clearly understood the
nature of laminar/turbulent flow regimes and recognised the similarity of his law to Poiseuille
flow. Many of these experimental observations were facilitated by improvements Darcy made
to the Pitot tube that both yielded its modern design and allowed for more accurate meas-
urements of the pipe fluid flow velocity distribution. Finally, not only did Darcy discover
Darcys Law, he was the first to combine it with continuity to develop the falling head per-
meameter solution that we still use today. He also applied that unsteady solution to the
analysis of spring discharge. Whilst Darcy is immortalised by Darcys Law, it is clear that
his scientific legacy extends beyond it.
This paper provides an account of Darcys contributions to engineering science. It does
so by beginning with a brief historical account of Darcys life (Section 2) in order to place
them within the necessary critical historical context and to provide some accompanying
insights on Darcys life, personality and motivations. A detailed description of Darcys con-
tributions to science and engineering is then presented (Section 3). A number of excellent
papers written recently by Brown (2002a, 2002b, 2003) form the basis for this analysis.
Finally, a brief discussion of hydrogeology in the immediate post-Darcy years (Section 4)
shows that Darcys Law was applied almost immediately after its discovery to the problem
of radial flow to a well, first treated by Dupuit (1863). It is here that we first see Darcys
Law applied to a hydrogeologic problem that resembles a modern day aquifer analysis.
2 DARCYS LIFE: A BRIEF HISTORICAL PERSPECTIVE
Henry Philibert Gaspard Darcy was born on 10th June 1803 in Dijon, France, and died in
Paris on 3rd January, 1858. He spent most of his life stationed in his native town of Dijon
working as an engineer. A large body of available literature provides compelling evidence
in support of the claim that Darcy was a great scientist, engineer and a selfless citizen.
There have been a number of historical analyses that lend insight into Darcys work and
times (e.g., Caudemberg 1858; Marsaines 1858; his great-nephew, Paul Darcy 1957; Hubbert
1969; Freeze and Back 1983; Freeze 1994; Philip 1995; Brown 2002a; Simmons 2003;
and Bobeck 2003) and some recent reviews/commentaries of Bobecks recently released
complete English translation of Les Fontaines (Bobeck, 2004) by Simmons (2004) and
Sharp and Simmons (2004). Copies of Darcys original 1856 monograph are very rare, and
few scientists have ever seen it but the new translation fills that void. From all of these
accounts, we have been able to learn some interesting things about Henry Darcy and to
answer even the most basic questions such as what did Darcy look like? Two reproductions
of Darcy one of the young Darcy at age 18 at the LEcole Polytechnique in 1821 and the
other of the mature Darcy are shown in Figure 1. And just to give a little Darcy trivia for a
moment, we also know that Darcy was 1.69m tall, had light brown hair, blue eyes and a
cleft chin! (Brown, 2002a). And what of Darcys name? As Philip (1995) points out, every-
thing he uncovered in his visit to Dijon, Darcys native town, clearly used the English spelling
Henry and not Henri, and Darcy not dArcy. It appears that Darcys name was always Henry
and never Henri but that he was born with the surname dArcy but it changed in his teen
years (see section 2.1). The subject of Darcys name has been dealt with by Brown and Hager
(2003) and provides conclusive evidence to support this claim. Indeed, it is this anglicized
form that appears on the title pages of the famous Fontaines Publiques report (see
Figure 2), on Darcys tombstone and his great-nephew Paul Darcy uses it in the title of his
Darcy biography and throughout that text (Freeze, 1994).
It is useful to highlight some of the key points in Darcys life and the timelines associ-
ated with both his major engineering projects and scientific discoveries. These important
previous accounts provide strong evidence that Darcys somewhat short life of 54 years
may be characterised by at least three distinct periods: (i) the early educative years (early
4 Aquifer Systems Management: Darcys legacy in a world of impending water shortage
1810s to mid 1820s) that establish Darcys strong technical background in engineering,
mathematics and physics, followed by, (ii) a longer period (mid 1820s to late 1840s) of
engineering service where Darcy carried out major engineering projects, including the
design and construct of the towns water supply in Dijon. This is the period in which Darcy
clearly rose to prominence and finally, (iii) the final years of Darcys life (early 1850s to
his death in 1858) where Darcys failing health sees a clear shift towards research and to
completing the writing of much of his lifes work.
2.1 The early educative years (mid 1810s1826)
Darcys father, Jacques Lazare Gaspard was a tax collector, who died in 1817 when Darcy
was only 14 (Darcy, 1957). Darcys mother, Agathe, did not have the means to finance her two
sons studies but she clearly valued it deeply. According to Henry Darcy V (2003), she
obtained a scholarship from the city of Dijon and a loan from her brother-in-law who was
also her childrens tutor. Henry Darcy V (2003) described this man as a republican brute
who advised the children to give up the particle and to transform dArcy into Darcy which
they did. But why a surname change? It is possible that the surname change was just like that
of many other people of the day who changed their surname a result of the French revo-
lution and the increasing challenges faced by the nobility. Indeed, a good number of noble
men were hanged or guillotined. It is possible that, like many others at that time, a surname
change removed associations with the old regime, made life easier and afforded oppor-
tunities that would otherwise be forsaken. The matter of Darcys name has been dealt with
extensively in other literature. Brown and Hager (2003) noted that Henry Darcys first
Henry Darcy (18031858): Immortalised by his scientific legacy 5
Figure 1. (a) Henry Darcy in 1821. (P. Darcy, 1957); (b) Henry Darcy in the later years of life.
Portrait by F. Perrodin from the collection of the Bibliothque Municipale de Dijon (from Philip,
1995; Brown, 2002a).
name is commonly spelled Henri, while his last name sometimes appears as dArcy. They
conclude that original source material shows that the correct spelling is Henry Darcy and
that the Henry spelling was his from birth, but that the Darcy spelling was adopted in his
youth and kept throughout his life.
In 1821, Darcy entered LEcole Polytechnique, Paris, and commenced science and engin-
eering studies that would set the stage for his distinguished career. Jean Baptiste Joseph
Fourier (17681830) held a Chair at the LEcole Polytechnique and in 1822 published his
Thorie analytique de la chaleur (The Analytic Theory of Heat) while based in Paris. It is
therefore possible that Fourier taught Darcy his heat law and that the earliest seeds of
Darcys Law may have been planted at this point. In 1823, at the age of 20, he was admitted
to LEcole des Ponts et Chausses (School of Bridges and Roads), Paris. This was the aca-
demic schooling arm of the Le Corps des Ponts et Chausses an elite fraternity of engin-
eers that had influential status in mid-nineteenth century France (Freeze, 1994), that was
first created in 1716 with a mission to support the construction of infrastructure through-
out France. The school was created by decree of the Royal Council in 1747 to train students
and practicing engineers for careers in the Corps. It both supported and expected excel-
lence and Darcys progression was usual for the better students at the time and would shape
the course of the rest of Darcys life (Brown, 2002a). A list of the schools graduates and
teaching staff reads like a cast of science, mathematics and engineering stars and includes
Figure 2. Darcys famous 1856 Fontaines Publiques report (from Hubbert, 1969).
Antoine Chzy (17181798), Louis Marie Henri Navier (17851836), Gaspard Gustave
de Coriolis (17921843), Arsene Jules Emile Juvenal Dupuit (18041866) and Henri
Emile Bazin (18291917), to name just a few. Coriolis also taught at the Polytechnique
during Darcys residence (Brown, 2002a). In these early educative years, we can be sure
that Darcy learned the state of the art in fluid flow, mathematics and physics. We also know
from Brown (2002a) that his class rank of 12 out of 64 at the Polytechnique, and 8 out of
15 who proceeded to LEcole des Ponts et Chausses suggests that Darcy was a good, but not
the best student.
2.2 Darcys engineering years and his rise to prominence (18261848)
Darcy joined the Corps as an engineer upon graduating in 1826 and spent most of his
working life with them stationed in Dijon. According to Freeze (1994), Darcy and other
prominent scientists and engineers attained public recognition and status in their tenure
working there. Initially, Darcy was assigned by the Corps to a position in the Department
of Jura but shortly thereafter, at the specific request of the Prefect of Cte dOr, was trans-
ferred to Dijon in 1827. He was assigned to perform a preliminary feasibility study of the
Dijon public water supply first proposed by Hugues Sambin, the 16th century architect of
Dijon. Darcy substantially completed this task in the period 18281834 and in 1834 published
Rapport M. le Maire et au Conseil Municipal, de Dijon, sur les Moyens de Fournir
lEau Ncessaire cette Ville (Report to the Mayor and the Town Council of Dijon on the
Means of Providing Necessary Water to the City). On March 5, 1835, the Municipal
Council approved his plans with no revision, and on March 31, 1837, the Dijon water pro-
ject was declared a public utility by a royal ordinance. On March 21, 1839, work began on
the Dijon water project and on 6 September 1840, water was delivered to the reservoir at
Porte Guillaume, just some 535 days later (Brown, 2002a). Darcy had transformed a
provincial capital rid with filth and squalor into a city with one of Europes best water supply
systems by about 1840. It was purported to be second only to Rome at the time and
occurred well in advance of even water supply development in Paris that was achieved by
the mid 1860s. Work on the delivery and distribution system continued until 1844 when
the Dijon water supply was largely completed. In May 1840, Darcy was appointed Chief
Engineer for the Department of Cte dOr at the young age of 37. Darcys rise to prom-
inence had begun.
At around this time, Darcy was also involved in the construction of a number of road pro-
jects, navigation works and bridges. These included two major structures over the Sane
River (Marsaines, 1858), his project to cover a 1.3km stretch of the Suzon, a small stream
that acted as an open sewer through the centre of Dijon (Caudemberg, 1858) and his
important work on the design and initiation of the component of the Paris-Lyon railroad
that passed through the Cte dOr (Darcy, 1957). This involved the construction of the four
kilometre tunnel at Blaisy which began in January of 1845 and of which Darcy completed
about one third of the tunnel before a private corporation took over the project in April of
1846 (Brown, 2002a). The Blaisy Tunnel is still used today by the TGV, the high-speed train
that connects Paris and Dijon. As noted by Brown (2002a), the tunnel equalled the longest
existing tunnel at the time.
Brown (2002a) describes the awards that followed and Darcys rise to prominence in the
period 18341848. They are also described by Philip (1995). These included a letter from
the Under Secretary of State and Director of Public Works (Dumay, 1845) that praised his
work. Darcy was awarded the Legion of Honor by King Louis Philippe on 31 August 1842.
He accepted a gold medal from the Municipal Council and a laurel wreath from the workmen
when the project was completed in 1844 but he waived all fees. It is believed that a bronze
version of the original medal was awarded to Darcy at that time. It is an heirloom of the
living Darcy descendents. The whereabouts of the original gold medal are, however, unknown
(Pierre Darcy de Moltke Huitfeldt, pers. comm.). To this authors knowledge, the original
medals have not previously been reproduced in the literature. Recent photographs taken by
this author of the front and reverse sides of the original bronze medal are shown in Figure 3.
As described by Philip (1995), Darcy, with great vision and skill, designed and built a
pure water supply system for Dijon, in place of previous squalor and filth. Dijon became
a model for the rest of Europe. Darcy selflessly waved fees due to him from the town, cor-
responding to about $1.5 million today. Medals were struck recognizing his skill and self-
lessness; and a monument celebrates his great work. The translated inscription on
Darcys tomb expresses the strong sentiment felt in Darcys time (Philip, 1995), He con-
ceived the project, made all the studies, pursued to the end the execution of the works to
which Dijon owes the creation and the abundance of its public waters. Doubly benefactor
of his native town through his talent and his selflessness. The translation to selflessness
here is arrived at from the French word dsintressement, and Philip notes that the word
means the total putting aside of ones own selfish interests. Philip also notes that the
word dsintressement appears many times in Darcy documents and that the literal trans-
lation disinterestedness would be too weak a translation. Darcy did, however, accept one
final reward for his work. In 1846, the Municipal Council resolved The town will provide
free to M. Darcy, during his life, in the house which he occupies, the quantity of water from
the public supply required for all the needs of his family and household (Philip, 1995).
Figure 3. Photographs (front and reverse) of the original medal awarded to Darcy by the municipal
council. The tower at Porte Guillaume reservoir is evident on the front side design. Photograph taken
by the author in 2006 and reproduced courtesy of Pierre Darcy de Moltke Huitfeldt.
2.3 Research excellence in Darcys final years (18481858)
It was not all fun and joy for Darcy. He suffered political persecution and in the later years
of his life his health deteriorated. In 1848, a revolution brought on by an economic depres-
sion saw the French constitutional monarchy ruled by King Louis Philippe replaced by a
provisional republican government. At only 45 years of age, Darcy was suspended from
duties since he was considered dangerous for the new state of things (Darcy, 1957) and
apparently had too much influence in Dijon for the new Commissioners liking (Brown,
2002a). Darcy was at the height of his career, and was deemed the hero of his fellow citizens
(Philip, 1995). According to Philip (1995), in Darcys very success lay his downfall. Philip
(1995) notes that despite the fact that Darcy was totally apolitical and had over the years
given generously of his own money to set up workers cooperatives, the Second Republic
saw him as dangerous and a reactionary collaborator with the ancient regime. Darcy lost
his offices and was banished from Dijon in 1848. In that period, Darcy was appointed to
Bourges to work on the Berry canal project and prepared plans for a new project to provide
drainage and irrigation over the Sologne region. Soon after the formation of the Second
Republic, however, and the election of Louis Napoleon on December 20 1848, Darcy was
transferred to Paris and appointed as Chief Director for Water and Pavements. On
December 2, 1852, the Second Republic was officially ended and the Second Empire formed.
President Louis Napoleon Bonaparte became Emperor Napoleon III. It appears that Darcy
was now politically rehabilitated but his days were numbered as Philip (1995) puts it.
Darcys health was failing. A nervous disorder accompanied by symptoms of meningitis
had been noticed as early as 1842, and he suffered a very bad period of health while direct-
ing the works at Blaisy (Darcy, 1957; Brown, 2002a) that Caudemberg (1858) attributed to
a railcar accident during the construction of the Blaisy tunnel. Darcy lost consciousness
during a conference in Paris in 1853.
In April 1850 Darcy travelled to England to collect data and information on the practice
of English road construction (including the paving of streets with layers of crushed rock
called macadam) that was published quickly upon his return to Paris (Darcy, 1850). The
report was highly regarded and Darcy was promptly promoted to the rank of Inspector
General, 2nd Class, in April of 1850 (Brown, 2002a). At around this time Darcy also con-
sulted on the City of Brussels municipal water system, for which he received the Order of
Leopold. This most significant new appointment as Inspector General provided Darcy
with major research opportunities, particularly as his new position brought with it com-
mand of the large hydraulic installation at Chaillot (Brown, 2002a).
The shift to research in last few years of Darcys life would see Darcy make some major
scientific discoveries what we might now call the Darcy scientific legacy. Importantly,
Darcys research efforts had been inspired by many years of engineering service and indeed,
it is clear that Darcys research was directly developed for engineering purposes. In the
period 18501854, Darcy designed and implemented an experimental program intended to
improve the estimation of the Prony pipe friction coefficients (Darcy, 1857). Darcys work
on pipe friction was substantially completed in the period 18501854. In the period between
its submission to the French Academy of Science in 1854, and its ultimate publication in
1857, Darcys health was failing. In 1855 he returned to Dijon and requested release from
all active duties except research. His wish was granted. In his final two years, Darcy gave
full attention to his experimentation. In Dijon, he worked on two sets of experiments, those
with Bazin on the Bourgogne Canal and the famous column experiments with Ritter in the
unnamed hospital laboratory. In this period he wrote Les Fontaines arguably his
swansong thesis completed just two years before he died. In 1857, Darcy was unani-
mously elected to hold the prestigious Chair of the French Academy of Sciences, a pos-
ition held previously by the great mathematician Cauchy but the position was not long
lived. Darcy died on January 3 1858, at the age of 54. He had apparently fallen ill with
pneumonia on a trip to Paris, no doubt brought on by the lingering effects of many years
of poor health (Freeze, 1994). Darcy (1957) notes he was carried off by pleurisy aggra-
vated by angina. His body was taken by rail to Dijon where he was given a state funeral.
The day immediately after his death square Chteau dEau, the location where the waters
of the Rosoir spring enter Dijon, was officially renamed Place Darcy a decision arrived
at unanimously and immediately by the Dijon Municipal Council.
Darcys work on improvements to the Pitot tube that yielded its modern design (Darcy,
1858; Brown, 2003) were published posthumously in 1858. His protg at the Corps,
Henri-Emile Bazin (18291917), an engineer some 26 years Darcys junior, published the
results of open channel flow experiments originally designed with Darcy in their report
titled Recherches Hydrauliques (Darcy and Bazin, 1865). Also published posthu-
mously, this publication would be Darcys last.
3 DARCYS EVERLASTING CONTRIBUTIONS TO SCIENCE AND
ENGINEERING
While the previous section sets out the major path of Darcys life within a historical con-
text and the important points in the Darcy timeline, it is important to examine Darcys sci-
entific and engineering legacy more fully. As scientists and engineers, we are the beneficiaries
of Darcys scientific legacy a legacy that not only included his well known law of water
flow through sand, but also included many other important contributions to hydraulics that
are outlined below. As an engineer, Darcys research contributions were clearly inspired by
a life of engineering excellence and driven by a deep desire to solve practical and useful
engineering problems that he had encountered along the way.
3.1 Observations of aquifer resistance (Darcy, 1834)
In 1834, Darcy published his report Rapport M. le Maire et au Conseil Municipal, de
Dijon, sur les Moyens de Fournir lEau Ncessaire cette Ville (Report to the Mayor and
the Town Council of Dijon on the Means of Providing Necessary Water to the City). In it,
Darcy described tests conducted in the groundwater system at Place Saint-Michel on
August 6, 1830. Darcy noted that the groundwater supply would not be sufficient to meet
Dijons needs and recognised that a clean water supply for Dijon would necessarily involve
more conventional surface water methods (Dumay, 1845; Brown, 2002a). Caudemberg
(1858) describes the efforts made by a society of subscribers and the Municipal Council in
hopes of repeating Moluts successful artesian well in Paris (Brown, 2002a). It is likely that
this outcome would probably have been seen as a major disappointment. However, it was
within this failed pump test that Darcy made an important new observation that the
aquifer being pumped provided significant resistance to flow, an apparently new discovery
(Brown, 2002a). Darcy noted that the amount of water yielded by the well was less than
would be expected even when friction losses within the pumping well were accounted for.
According to Brown (2002a) Darcy correctly concluded, The comparison of these figures
shows that the source did not provide to the pump what the head and the diameter of pipe
made it possible to provide, or in the least, the difference was absorbed by filtration i.e.,
aquifer losses. It appears that Darcy may have been making a connection here between real
aquifer processes and the filtration mechanics in a filter bed since he used the term filtra-
tion explicitly here and again later in Note D of the famous Les Fontaines text (Darcy,
1856) in which Darcys Law was discovered. In numerous places throughout the 1856 text,
it is clear that Darcy understood that the aquifer could provide significant resistance to
flow.
3.2 The Darcy-Weisbach equation, boundary layers, laminar/turbulent flow
(Darcy, 1857)
Pressure drop during internal pipe flow is one of the most important considerations in
designing a fluid flow system. Building upon his interest in pipe flow that had grown
whilst working on the Dijon water system throughout the 1840s, Darcy initiated, designed
and completed a comprehensive experimental program intended to improve the estimation
of the Prony pipe friction coefficients (Darcy, 1857; Brown, 2002a; Brown 2002b) that
was largely conducted in the period 18501854, although his report Recherches
Exprimentales Relatives au Mouvement de leau dans les Tuyaux, (Experimental Research
Relating to the Movement of Water in Pipes) was published later in 1857 (Darcy, 1857).
At the time, the Prony equation (Eqn. 1) was the widely accepted pipe flow resistance
equation used to calculate head losses in pipes (and open channels using different empir-
ical coefficients) but was one that was prone to error since the empirical and recommended
pipe friction coefficients did not account for pipe roughness.
(1)
where h
L
is the head loss due to friction calculated from the ratio of the length to internal
diameter of the pipe L/D, the velocity of the flow V, and a and b are two empirical friction
coefficients that account for friction. The Prony friction coefficient values were debated,
but they were believed not to be a function of pipe roughness (Brown, 2002b).
Darcys new results showed that pipe friction factor (and hence head loss) was a func-
tion of both pipe roughness and pipe diameter. Indeed, his new formulation provided a
much better estimation of losses. Darcy proposed an equation (Eqn. 2) that was similar to
the Prony equation with friction coefficients that were a function of pipe diameter D, and
which reduced to the version now known as the Darcy-Weisbach equation (Eqn. 3) at high
velocities (Brown, 2002a; Brown 2002b). As noted by Brown (2002b) the pipe friction
equation proposed by Darcy took the form:
(2)
where , , , are friction coefficients. He noted that the first term could be dropped
for old pipes and at higher velocities to yield an equation that looks similar to the Darcy-
Weisbach equation (Eqn. 3) that is commonly used today.
h
L
D
D
V
D
V
L

1
)
1
)
22
l
l
l
l
l
The Darcy Pipe Friction Equationn
h
L
D
aV bV
L
( )
2
The Prony Equation
(3)
where f is usually called the Darcy friction factor and is a complex function of the relative
roughness and Reynolds number and g is acceleration due to gravity. It may be evaluated
for a given set of hydraulic conditions by the use of various empirical or theoretical correl-
ations, or it may be obtained from published charts referred to Moody diagrams, after
Lewis F. Moody (18801953). A detailed historical account of the Darcy-Weisbach equa-
tion has been given by Brown (2003) and the reader is referred to that for further details.
It is interesting to note from that account, however, that it was actually Julius Weisbach
(18061871) who first proposed the current form of the Darcy-Weisbach equation in 1845
(Rouse and Ince, 1957) but it was Darcys work that identified surface roughness as an
important parameter in fluid flow and introduced that concept to the science of fluid
dynamics. The friction factor term f is therefore often called the Darcy f factor, although
Darcy did not propose it in that form. It was actually J.T. Fanning (18371911) who first
combined Weisbachs equation with Darcys improved estimates of the friction factor
(Brown, 2002b). Since Fanning worked in terms of radius instead of diameter in his fric-
tion analyses, the Fanning f values are
1
4 of the Darcy f values. Darcys contribution to
understanding of pipe flow friction losses and the improved Prony pipe friction coeffi-
cients is acknowledged in the joint naming of the Darcy-Weisbach equation.
In his 1857 report, Darcy also made the first accurate measurements of turbulent pipe
velocity distributions and provided the very first evidence of the existence of the fluid bound-
ary layer (Darcy, 1857) which were made possible using his improved Pitot tube designs.
Whilst limitations in technique inhibited details measurements of the boundary layer in quan-
titative terms, Darcy began to suspect the existence of the boundary layer when he compared
results in both smooth and rough pipes. In a translation by Rouse and Ince (1957, page 170) it
is immediately clear that Darcy correctly suspected that the fluid boundary layer was the cause
of the variation between smooth pipe and fully rough flows If one uses very smooth pipes, of
lead, recovered with glazed bitumen, etc, the coefficient of V
2
decreases continuously as the
degree of polish increases. But the reduction nevertheless is far from appearing proportional
to the degree of polish obtained. In vain one would say that the influence of asperities inap-
preciable to the eye persists for the fluid molecules; that explanation would not seem at all sat-
isfactory. In effect, the term in V
2
does not appear to correspond only to the resistance caused
by the asperities, but also to that produced by the fluid layer next to the boundary.
Darcy also recognised the similarity of his pipe friction formula with Poiseuilles Law
(1841) developed by Jean Louis Marie Poiseuille (17971869), an experimentally derived
physical law concerning the voluminal laminar stationary flow of incompressible uniform
viscous liquid through cylindrical capillary tubes with constant circular cross-section. Darcy
later showed that his newly proposed pipe friction formula reduced to the Poiseuilles linear
equation (Eqn. 4) at low flow and small diameters, namely,
(4)
where Q is the volumetric flow rate of the liquid and k is an empirical coefficient that
lumps constants with a second order equation for the temperature dependent viscosity
Q kD
h
L
L
4
Poiseuille's Law
h f
L
D
V
g
L

2
2
The Darcy-Weisbach Equation
(Poiseuille, 1841). Here we see the clear recognition that at low flows (i.e., laminar) flows,
a linear relationship holds between flow and head loss. Indeed, Darcy clearly understood
that such a linear relationship held in slow flow, small diameter pipes. According to Brown
(2002a) Darcy wrote, Before seeking the law for pipes that relates the gradient to the
velocity, we will make an observation: it appears that at very-low velocity, in pipes of
small diameter that the velocity increases proportionally to the gradient. He later showed
explicitly that his newly proposed pipe friction formula would reduce to equation (4) at
low flow and small diameters. Darcy noted that this was a rather remarkable result, since
we arrived, Mr. Poiseuille and I, with this expression, by means of experiments made under
completely different circumstances. Darcy had made the important connection between
real pipes and capillary tubes, My formula seems to contain the link that unites the laws
of water flow in a pipe of any diameter and in a capillary pipe (Darcy 1856, Note G). He
had probably already made a connection, based upon the expected slow speed of water
flow through sand, between his work on pipes and his work in sand columns. Indeed, a
footnote in his 1857 report notes the similarity to his 1856 results for flow in sand columns.
Similarly, Darcys 1856 report noted the similarity of his sand column results with his
(laminar flow) pipe results. Whilst workers such as Poiseuille and Hagen (17971884) had
begun to understand the differences between low and high velocity flows in capillary tubes
(what we would now call laminar and turbulent flows), Darcy had extended those insights
into real pipes and to pipes of larger (general) diameters. All available documentation clearly
shows that Darcy understood the differences in the flow regimes and the subsequent limi-
tations and applicability of his findings. There can be no doubt that Darcy clearly under-
stood how pipe diameter and flow velocity affected his results. Whilst, according to Brown
(2002a), it appears that Darcy had discovered the kernel of the truth by 1854, it was not
until the work of Osborne Reynolds (18421912) in 1883 that the differences between
laminar and turbulent flow were truly quantified.
3.3 Les Fontaines and Darcys Law (Darcy, 1856)
A city that cares for the interest of the poor class should not limit their water, just as day-
time and light are not limited(Darcy, 1856).
3.3.1 An overview of Les Fontaines
Although work on the Dijon water supply was largely conducted in the period 18341844,
it was not published until 1856. It is likely that Darcys failing health prompted him to
complete the writeup of what is now considered by many to be his most famous text on the
construction of the municipal water supply of Dijon, Les Fontaines Publiques de la Ville
de Dijon(The Public Fountains of the City of Dijon) (Darcy, 1856). This is Darcys swan-
song, Darcys attempt to write a thesis in the style of the day at a time of deteriorating
health. In it, Darcy noted that various books available at the time debated issues relating to
water supply systems but that they did so theoretically and that a publication that reports
on the construction of a large distribution system would be of interest to engineers. Full
details of this monograph are now readily accessible worldwide thanks to Patricia Bobecks
faithful English translation (Bobeck, 2004) and for which Bobeck was awarded the presti-
gious 2004 S. Edmund Berger Prize for Excellence in Scientific and Technical Translation,
which is presented by the American Foundation for Translation and Interpretation
(www.afti.org). Patricia Bobecks amazing translation of Les Fontaines Publiques de la
Ville de Dijon opens a window into the world of engineering science in the early 19th cen-
tury, as well as its challenges and implications for the present. There are many other fascin-
ating pieces of scientific, social, and historical information throughout the monograph and
the illustrative plates are amazing pieces of engineering artwork.
The original Darcy monograph was some 680 pages long and contained 28 plates of
figures in a separate atlas. While much of the material in it addresses the Dijon water sup-
ply, Darcy also discussed several other topics including groundwater, sand filters and pipe
manufacture. Darcys monograph shows how he approached the design and construction
of the Dijon water supply system by choosing the water source, building an aqueduct and
designing the water distribution system. Darcys design collected about 8m
3
/min at the
Rosoir Spring, which was dug out to improve its flow. The system did not rely on pumps
as it was gravity driven. From the original Rosoir spring source, the water was carried some
12.7km in a covered aqueduct to an enclosed reservoir located near the Porte Guillaume
(holding capacity 2,313m
3
) and another reservoir at Montmusard (holding capacity
3,177m
3
). The entire engineering design contained some 13.5km of distribution lines. It
supplied 141 public street fountains spaced 100m apart throughout Dijon that would sup-
ply abundant free water for domestic purposes (one fountain for every 200 people), for
washing streets and sewers and in fire fighting. One of the most elegant reservoir entrances
is shown in Figure 4, at Chateau dEau at La Porte Guillaume (Darcy 1856, Plate 9).
Figure 4. Darcys design of the tower at Porte Guillaume reservoir (Darcy 1856, Plate 9). The tower
and reservoir are still standing today.
In this text, Darcy also clearly emphasised the importance of science in providing and
understanding our water resources. In Darcys time, hydrogeology was still arguing about
the Greek water cycle which moved water from the sea to the continents and Father
Paramelles famous book The art of discovering springs (1856, 1859) was the best
seller, not Darcys (de Marsily, 2003). Unlike Darcys engineering views, Paramelles work
is a much more descriptive and naturalistic contribution that is quite different to that of
Darcy (1856). Darcys discussions of Father Paramelles exploration for springs and the
ancient Greek hydrologic cycle are well written and perceptive. After what initially appears
to be a significant number of pages dedicated to a gentlemanly debunking of Paramelles
methods and Darcys dismissal of dowsing as a cult (Sharp and Simmons, 2004), Darcy
eventually saw some usefulness in Paramelles observations and recognised him as a good
geologist concerned by underground hydrography. In the end, it is clear that whilst Darcy
does not agree with all the rules and methods provided by Paramelle to discover springs,
he no longer dismisses him as a water dowser.
Les Fontaines Publiques de la Ville de Dijon begins with Darcys introduction on the need
for a good watersupply, the requirements for achieving this supply, and the organization of
the book into four parts. These are outlined as follows. Part 1: History of Dijons water
supply and research conducted from the 15th to the 19th centuries, a discourse on springs
and the rationale for choosing Rosoir Spring as the source of Dijons water supply. Part 2:
The design of the Rosoir aqueduct and water distribution systems, including pipes and pipe
design, street fountains both for supply and public display, valves, and the two reservoirs,
plus a cost analysis. (Darcy intended his monograph to be a manual for future water supply
projects.) Part 3: Experiments on flow of water in the aqueduct and conduit systems. Part 4:
Administrative and judicial issues. These are followed by eight appendices: A. Springs in
the Dijon area; B. A 15th century contract for Dijon water; C. Water supply systems for
various cities, including London and Paris; D. Filtration, which includes the famous Darcy
column experiments; E. Weir gauging; F. Extracting constant volumes of water from a
varying-level stream channel; G. Pipe fabrication methods; and H. Flow in the Rosoir
aqueduct.
3.3.2 The discovery of Darcys Law
Darcy left his greatest gift buried in the depths of the report. Part 2 of Note D in a subsec-
tion titled Determination of the Laws of water flow through sand contains the results of
his famous column experiments. Freeze (1994) described their appearance as hardly front
and center. Here Darcys motivations are clear. In presenting data concerning the dis-
charge of filters in England, Scotland and France, Darcys principal motivation for the col-
umn experiments is clarified when he writes no general law can be deduced from this
data, given that the nature and the thickness of the filtration sands are not comparable,
that the heads are variable, and the water enters the equipment with different degrees of
clarity. I have tried to use precise experiments to determine the laws of water flow through
filters.Water filtration methods and galleries were becoming an increasingly common
practice at the time to improve water clarity (see Guillerme 1988 for a review) and as a
result engineers were starting to think about the behaviour of filters (e.g., Gnieys, 1835).
However, no general law governing their hydraulic behaviour had yet been discovered a
critical observation that helps understand the scientific landscape and historical context in
which Darcys Law was found. Darcy remarks on the need to decrease significantly the
surface area of artificial filters and the section of Note D on modifications to apply to
filters begins with the statement Now I would like to discuss a method of significantly
increasing the discharge of filters per given surface area and as a result, facilitating the
construction of this equipment that until now has required sites so large that the very
choice of them was one of the major difficulties of large-scale filtration. But one thing was
still missing a physical law that would express the relationship between filter volumetric
capacity, filter dimensions (area and thickness), filter bed properties, and the hydraulic
conditions under which the filter should be operated. With that motivation in mind, Darcy
set out to unravel the universal porous media flow law a flow law that he had suspected
based on his earlier work in pipes.
It is interesting to provide some details on Darcys column experiments, although full
details are now readily accessible in Bobeck (2004). Brown (2002a) also provides a compre-
hensive analysis of the experimentation. Two sets of column experiments were performed in
total. Set 1 (23 experiments) were conducted with the assistance of engineer Mr Charles
Ritter (October 29/30, November 6, 1855) and Chief Engineer Mr Baumgarten repeated
those experiments but the repeat tests are not reported. Set 2 contained an additional 12
experiments that were conducted by Mr Ritter alone (February 17 and 18, 1856). The major
difference between the experiments rested in the pressure conditions applied to the column.
The first set was undertaken with the outlet at the bottom held at atmospheric pressure, and
the second set was conducted with variable inlet and outlet pressures by methods that are not
reported. A total of 35 experiments were reported. It is said that Darcys experiments were
conducted in an unnamed hospital courtyard. The apparatus used is shown in Figure 5
(plate 24, Figure 3 of the original monograph) and consisted of a vertical column 2.50m high
(note here that the text suggests this dimension but that the original figure notes a vertical
height of 3.5m perhaps this is an error or were there two column designs?) and with an
internal diameter of 0.35m. The experiments were performed using siliceous sand from the
Sane River, and each experimental series had a different sand packing. Packing height var-
ied from 0.58m (Series 1) to 1.70m (Series 4). The column was filled with water first and
then sand was poured and packed into it. Brown (2002a) notes that the packing method used
would have resulted in coarsest particles settling at the bottom of each lift but that since the
experiments were run to equilibrium and the height of the sand was measured only at the end
of each series of experiments (after the passage of water had suitably packed the sand), that
the packing method would not have altered any of Darcys conclusions.
The column was set up so that water flowed into the sand column from above through a
pipe connected to the hospital water supply and vertically downward through the column
before exiting from the lower outlet. The pressure at the two ends of the column was meas-
ured by a U-shaped mercury manometer which, under weak heads, resulted in almost com-
plete quiescence of the mercury in the manometer and allowed measurement to the nearest
millimetre, representing 26.2mm of water. Darcy observed that when operating under
higher pressures that large (but random) fluctuations allowed the average height of mer-
cury to be measured to the nearest 5mm, and thus allowed the water pressure to be deter-
mined to an accuracy of about 13cm. Here, Darcy observes that the fluctuations were due
to water hammers produced by the operation of the numerous street fountains at the hos-
pital where the experimental apparatus was located an effect brought about by Darcys
own water supply that he had constructed some 15 years before the experiments were con-
ducted! In each experiment, the extent of oscillations was noted. When the inlet and outlet
pressure observations assured that the flow had become uniform, the discharge of the fil-
ter was noted for a certain time, and the average discharge per minute was determined.
The duration of the experiments varied between 10 and 30 minutes, and within each series,
the mean discharge per minute was both varied and measured. The smallest value of volu-
metric discharge rate used was Qlower 2.13l/min (in Set 1, Series 3, Experiment
Number 1) and the highest value of Qupper 29.40l/min (in Set 1, Series 1, Experiment
Number 10). Darcy noted that the results demonstrate that the discharge from each filter
increased proportionally with the head. Darcy denoted Q as the discharge per second
per square meter, and I as the head per meter of filter thickness and noted that for each
series, a straight line relationship existed between Q and I. However, between experiments
slightly different values of the coefficient Q/I (what we now call hydraulic conductivity)
were observed. Here Darcy noted that the sand used was not consistently homogeneous.
For the second series it was not washed; for the third series it was washed; for the 4th series,
it was very well washed and had a slightly larger grain size. He then concluded Thus, it
appears that for an identical sand, it can be assumed that the volume discharged is (directly)
proportional to the head and inversely proportional to the thickness of the sand layer that
the water passes through. And in those few words and only a few days in the hospital
courtyard, quantitative hydrogeology as we know it today was born. Darcy had provided
Figure 5. Darcys original sand column apparatus (Darcy 1856, Plate 24, Figure 3).
conclusive evidence that the water flow rate was a linear function of the total head loss across
the filter bed and not just the difference in water pressure. The subsequent experiments in
February were undertaken to ensure that the law could be generalised, and that the experi-
mental conditions employed to develop the law covered the necessary and different pres-
sure conditions that might be expected in an operational filter plant. Darcy had an extremely
good understanding of hydraulics, and he would have known that the pressure would not
have impacted his new discovery. He therefore let Mr Ritter conduct the second set of
experiments alone in February 1856, who successfully confirmed that this was indeed the
case. Darcy then stated his law (exactly as it is written in Eqn. 5) for the very first time,
noting that the pressure on the top of the layer was P h (where P atmospheric pres-
sure and h is the height of water on the sand layer), and on the bottom of the layer was
P h
o
to yield, in general terms:
(5)
where q is the volume of water discharged (per unit time), k is a coefficient that depends on
the permeability of the layer, e is the thickness of the sand layer and s is its surface area.
Eqn. 4 can easily be generalised in terms of general pressure heads and elevation heads at
the inlet and outlet accordingly to yield the more familiar version we use routinely today.
Furthermore, the Darcy unit of permeability (D) that is widely used in geology and petroleum
engineering recognises that Darcy was the first to note that flow depended upon a permea-
bility coefficient, a direct consequence of his experiments and the discovery of his law.
A number of interesting points follow from the column experiments that help to context-
ualise Darcys Law and the process of his discovery:
1. Darcy did not stumble on to his law, he probably suspected it: His column experiments
were carefully planned and meticulously executed. Darcy had a very strong under-
standing of the underlying fluid mechanics, informed by both his background educa-
tion and the great experience he had already amassed in his pipe flow research. He had
already made the connection between flow in real pipes and flow in smaller diameter
capillary tubes at low flows and knew that his pipe formulae would reduce to
Poiseuilles Law under the limiting (small pipe diameter, low flow i.e., laminar) condi-
tions. Now all that remained was for the connection to be made with sand and Darcy
did not leave that stone unturned. Indeed, when discussing his new law, Darcy notes
clearly in footnote 4 of Note D, I had already foreseen this curious result in my research
on water flow in conduit pipes of very small diameters, when the water velocity did not
exceed 10 to 11 centimetres per second. Darcy made the first clear connection between
flow in sand and flow in small pipes at low velocities. He knew that his law and
Poiseuilles Law were linear laws and most importantly, he understood why.
2. Darcy knew his discovery was new and significant: This is noted by Darcy himself
when he writes in his preface I have not seen the documents that are included in Note D
collected in any special book. In particular, to my knowledge at least, no one has
experimentally demonstrated the laws of water flow through sand. Darcys personal
view on the significance of Note D is also enforced by the fact that he dedicates almost
half the length of his preface to his entire monograph to a discussion on it.
q k
s
e
h e h
o
( ) Darcy's Law from Darcy (1856)
3. Capillary tube models of porous media and the REV: Whilst Darcy made the connec-
tion between capillary tubes and porous media, he did so primarily on the basis of flow
speed and his expectation that flow in porous media would be slow (i.e., they would be
laminar like that in small pipes with small flow speeds). He did not treat the porous
medium formally or theoretically as a bundle of capillary tubes. This would follow very
shortly after a work by Dupuit (1857) who, according to Narasimhan (2005), idealized
a permeable medium to be a collection of small diameter tubes, and showed that Darcys
Law was a special case of Pronys equation, with inertial effects neglected. Interestingly,
it can also be seen in Darcys text that he assumed proportionality of flow with surface
area, and was therefore applying the principles of continuum mechanics. For the con-
ditions under which Darcys Law was developed, this may have been entirely reasonable
but we know that such approaches are at the heart of current challenges faced by hydro-
geologists in difficult concepts such as the REV, matters of hydrogeologic scaling and
dealing with heterogeneity in the subsurface.
4. The rise of the linear gradient laws: Interestingly, the early to mid 19th century saw the
birth of the entire suite of linear gradient laws including Fouriers heat conduction law
(1822), Ohms law for electricity (1827), Poiseuilles Law (1841), and Ficks Law (1855)
for molecular diffusion. Darcys Law was the last of the great linear law discoveries.
Darcy only makes mention of Poiseuilles Law (which was obviously the most relevant
one to him) but he likely knew of the others and indeed may have been taught by
Fourier (17681830), a French Professor and academic. According to Groenevelt (2003),
it is likely that Darcy was aware of Fouriers work soon after it was published and cer-
tainly well before he conducted his famous laboratory experiments in 1856.
5. Darcy understood the practical significance of his law and he applied it: Darcy developed
the first falling head permeameter solution in Note D by combing his law with
continuity, and then applied it to determine the law of progressive decreases of a
spring from its maximum flow and for increasing their product by artificially lower-
ing their level. His work on spring discharge and artesian wells (and the discovery of
a linear relationship between discharge and spring discharge height) as shown in Figure 6,
Figure 6. Darcys measurements of artesian spring flow and his linear observations (Darcy 1856,
Plate 22).
combined with his previous pipe research and the sand column experiments, lead Darcy
to believe that the linear relationship was reasonable for laminar flow conditions i.e.,
that the wells were either supplied by very small diameter open conduits, or by con-
duits that were filled with sand. However, because observation wells were expensive,
only drawdown in the extraction well was observed, and radial flow was ignored. Darcy
continued to think of groundwater flow in terms of linear conduit flow. However, what
is critical here is that Darcy was now applying his theoretical concepts developed in
both his pipe research and sand column experiments to practical field applications in
natural geologic media and was using real field data.
3.4 Improvements to the Pitot Tube (Darcy, 1858)
In 1732, Henri Pitot (16951771) created a simple instrument to measure fluid velocity
that is called the Pitot tube. This device is lowered into a flow field and contains two tubes.
A static tube that points straight down into the field (to measure static pressure) and a
second tube that has a 90 degree bend at the bottom that faces directly into the flow (that
measures total pressure static pressure dynamic pressure). When the device is lowered
into the flow the pressure differential is recorded by observing the difference in the liquid
level in the two tubes. The difference is the dynamic pressure component that relates to the
speed of the flow. The Pitot tube is commonly used in aircraft speed determination and
other pneumatic devices. The original Pitot design had several problems as outlined in
Brown (2003) who provides an excellent account of the major developments Darcy made
to the Pitot tube and notes that Darcys contribution to the development of the device
equalled or exceeded Pitots initial work. He also notes that Darcys final design for the
instrument tip is reflected today in modern instrumentation and that it is appropriate to call
the modern design the Pitot-Darcy tube. Darcy used evolving designs to make accurate
measurements of point velocity within pipes (Darcy, 1857) and in mapping isovels (lines
of equal velocity) in open channels (Darcy and Bazin, 1865). The Pitot tube also made an
appearance in Darcy (1856) as is shown in Figure 7. Darcys 1858 publication Relative
quelques modifications introduire dans le tube de Pitot (Some modifications introduced
to the Pitot tube) was published posthumously shortly after his death and reflected several
years of work gradually perfecting its design over the period 18501857.
4 HYDROGEOLOGY IN THE IMMEDIATE POST-DARCY YEARS
It would be just seven years before Darcys Law was applied in what we might now call the
first modern-day aquifer analysis. Arsene Jules Emile Juvenal Dupuit (18041866) sub-
mitted a ground breaking report in 1863 (Dupuit, 1863) that solved the radial flow equa-
tion for steady flow to a well with a free surface. Dupuit was Darcys associate and
successor as Chief Director for Water and Pavements for Paris and Darcys contribution
was noted clearly by both Dupuit and the reviewers at the French Academy of Science
(Brown, 2002a). Both Darcys work and that of Dupuit inspired other investigations in
water supplies, soils and engineering geology. As early as 1870, the German Adolf Thiem
had modified Dupuits formula to allow for the calculation of aquifer hydraulic properties
using a pumping well and observing the resulting decline in water table in adjacent wells
(Thiem, 1887). The Austrian Philipp Forchheimer (18521933) applied Laplaces equation
and potential theory to groundwater problems by recognising the similarity between
groundwater flow and heat flow (Forchheimer, 1886). A well known American geologist,
T. C. Chamberlin further developed the relationships between groundwater and its host
geologic formations in his pioneering report The Requisite and Qualifying Conditions of
Artesian Wells (Chamberlin, 1885) the first groundwater report published by the United
States Geological Survey. In it, Chamberlin provided a theoretical basis for the scientific
study of groundwater which prompted a boom in groundwater exploration in the United
States. It is interesting to note that Darcys work on artesian wells and springs (Darcy,
1856) predated Chamberlins by some thirty years.
5 EPILOGUE
Freeze (1994) reflected upon Darcys life I can see his path through life in its various roles:
as a successful young student; as a fraternal brother in the Corps des Ponts et Chausses;
Figure 7. The Pitot tube design used by Darcy in his 1856 report (Darcy 1856, Plate 23, Figure 14).
as a young engineer of such renown that he is asked to design the water supply for the city
of Dijon; as the administrator of a large regional engineering office; as a respected leader
of the community; as a victim of political pressure in a time of tumult; and as a research
scientist who made lasting contributions to mankind. We should remember Darcy as a
man who gave selflessly to his native people of Dijon to give them free and abundant clean
water, which Darcy himself valued just as much as daytime and light. His work on the Dijon
water supply would shape the rest of his life and see him rise to prominence in the Corps.
Darcys distinguished engineering years inspired his final research years. His research was
aimed at solving practical and useful engineering problems. In the last few years of his life
and despite his rapidly deteriorating health, Darcy unrelentingly pursued his research
interests. He worked feverishly on several major research projects that were no doubt inspired
by unresolved questions brought about by his engineering projects his sand column
experiments, his improvements to Pronys pipe friction equation, his improvements to the
Pitot tube for measuring point water velocity and his work with Bazin on the open channel
hydraulic experiments. We as scientists and engineers are the beneficiaries of a scientific
legacy that includes Darcys Law but that is not limited to it. It is a legacy created by a dis-
tinguished engineer and research scientist who in his short life of 54 years achieved many
great things. But Darcy lives on forever and his contributions are everlasting. Indeed,
Darcy is immortalised by Darcys Law and his scientific legacy.
ACKNOWLEDGMENTS
I would like to express my gratitude to Professor Glenn O. Brown for many very informative
exchanges over the years about Darcy and the history of hydrogeology. Glenns Darcy
research has been a critical source of inspiration and information to me. Thank you to
Patricia Bobeck for her amazing translation that has afforded us all a wealth of accessible
new insights about Darcy and for generously sharing her Darcy knowledge with me. This
work has been influenced by my earlier work with fellow Darcy collaborator and co-author
Professor Jack Sharp and I thank Jack for sharing the journey. I am very thankful to
Professor Ghislain de Marsily for his critical reading of this manuscript and for his very
useful suggestions. Thank you to Philippe Renard for assisting me with my interpretation
on the work of Father Paramelle. I must also acknowledge J.R Philip and R.A. Freeze for
sharing their Darcy stories with the scientific community in the 1990s and for teaching me
why our scientific roots are important. Their Darcy accounts were critical in spurring me
on to learn more about Darcy and the history of hydrogeology. I have since benefited from
personal communication with Allan Freeze that has been most helpful. Thank you to Allan
Freeze and Sabine Barles for their helpful reviews of this manuscript. Finally, I must
express deep gratitude to the living Darcy descendents in France, especially Henry Darcy V
and his brother Pierre Darcy de Moltke Huitfeldt and their families. The opportunity to
know you and to learn personally from you about your famous ancestor continues to be a
life changing experience and one that I will cherish for ever. My personal contact with you
has influenced my writings on Darcy and brought about new insights and information that
were not contained in previous written accounts. These can now be shared with the world.
I thank you on behalf of the scientific community for your generosity and for allowing us
all to learn more about Henry Darcy through his descendents.
REFERENCES
Bobeck P, trans. (2004) The Public Fountains of the City of Dijon by Henry Darcy (1856). Dubuque,
Iowa: Kendall/Hunt Publishing Company.
Bobeck P (2003) Henry Darcy and the public fountains of the city of Dijon. In Proceedings of the
Darcy Memorial Symposium, World Water and Environmental Resources Congress, ed. G. Brown,
10. Philadelphia, Pennsylvania: Water Resources Institute, American Society of Civil Engineers.
Brown GO (2002a) Henry Darcy and the making of a law. Water Resour Res 38(7):
doi:10.1029/2001WR000727
Brown GO (2002b) The history of the Darcy-Weisbach equation for pipe flow resistance, in
Environmental and Water Resources History, A. Fredrich, and J. Rogers eds., pg. 3443. Reston,
Virginia: American Society of Civil Engineers.
Brown GO and Hager WH (2003) Note: Spell it Henry Darcy, in Henry P. G. Darcy and Other
Pioneers in Hydraulics: Contributions in Celebration of the 200th Birthday of Henry Philibert
Gaspard Darcy, G.O. Brown, J.D. Garbrecht, and W.H. Hager (eds), pg. 13. Reston, Virginia:
American Society of Civil Engineers.
Brown GO (2003) Henry Darcys perfection of the Pitot tube, in Henry P. G. Darcy and Other
Pioneers in Hydraulics: Contributions in Celebration of the 200th Birthday of Henry Philibert
Gaspard Darcy, G.O. Brown, J.D. Garbrecht, and W.H. Hager (eds), pg. 1423. Reston, Virginia:
American Society of Civil Engineers.
Caudemberg G (1858) Notice sur M. Henri Darcy, extract from Mmoires de lAcadmie de Dijon,
Loireau-Feuchot, Dijon, France.
Chamberlin TC (1885) Requisite and qualifying conditions of artesian wells. U.S. Geological Survey
Annual Report 5, 131173.
Darcy H (1834) Rapport le Maire et au Conseil Municipal, de Dijon, sur les Moyens de Fournir
lEau Ncessaire cette Ville, Douillier, Dijon, France.
Darcy H (1850) Rapport M. le Ministre des Travaux Publics, sur le pavage et le macadamisage des
chausses de Londres et de Paris, Ann. Ponts Chausses, Ser. 2, 10, 1264.
Darcy H (1856) Les Fontaines Publiques de la Ville de Dijon, Dalmont, Paris.
Darcy H (1857) Recherches Exprimentales Relatives au Mouvement de lEau dans les Tuyaux,
Mallet-Bachelier, Paris, 1857.
Darcy H (1858) Relative quelques modifications introduire dans le tube de Pitot, Ann. Ponts
Chausses, Ser. 3, 15, 351359, 1858.
Darcy H Bazin H (1865) Recherches Hydrauliques, Enterprises par M. H. Darcy, Imprimerie
Nationale, Paris.
Darcy H (2003) Henry Darcy: Inspecteur gnral des ponts et chausses, by Henry Darcy (V), in
Henry P. G. Darcy and Other Pioneers in Hydraulics: Contributions in Celebration of the 200th
Birthday of Henry Philibert Gaspard Darcy, G.O. Brown, J.D. Garbrecht, and W.H. Hager (eds),
pg. 413. Reston, Virginia: American Society of Civil Engineers.
Darcy P (1957) Henry Darcy: Inspecteur Gnral des Ponts et Chausses, 18031858, Imprimerie
Darantiere, Dijon, France.
de Marsily G (2003) About Darcys law. Geological Society of America. Abstracts with Programs
(Annual Meeting.), v. 35, p. 448.
Dumay V (1845) Notice Historique sur Ltablissement des Fontaines Publique de Dijon, Frantin,
Dijon, France.
Dupuit AJEJ (1857) Mmoir sur le mouvement de leau travers les terrains permables. Comptes
Rendus Hebdomadaires des Sances de lAcadmie des Sciences (Paris) 45: 9296
Dupuit AJEJ (1863) tudes Thoriques et Pratiques sur le Mouvement des Eaux dans les Canaux
Dcouverts et Travers les Terrains Permables, 2nd ed., Dunod, Paris.
Forchheimer P (1886) Uber die Ergebigkeit von Brunnen Anlagen und Sickerschlitzen. Zeitschrift
des Architekten- und Ingenieur Vereins zu Hannover, 32: 539564.
Freeze RA (1994) Henry Darcy and the fountains of Dijon. Ground Water, 32: 2330.
Freeze RA and Back W (eds) (1983) Determination of the laws of flow of water through sand. In Freeze
RA (transl) Physical hydrogeology. Benchmark Papers in Geology, vol 72, Wiley, New York,
pp 1420.
Gnieys R (1835) Clarification et dpuration des eaux, (publication posthume), Annales des ponts et
chausses, 1er sem. 1835, pp. 5676.
Groenevelt PH (2003) The place of Darcys law in the framework of non-equilibrium thermodynam-
ics. In Henry P.G. Darcy and Other Pioneers in Hydraulics, ed. G.O. Brown, J.D. Garbrecht, and
W.H. Hager, 310. Reston, Virginia: American Society of Civil Engineers.
Guillerme A (1988) The Genesis of Water Supply. Distribution, and Sewerage Systems in France,
18001850, pp. 91115, in: Tarr, J. et Dupuy, G. (ed.) Technology and the Rise of the Networked
City in Europe and America. Philadelphie, Temple University Press, 1988.
Hubbert MK (1969) The Theory of Ground-Water Motion and Related Papers, Hafner, New York.
Marsaines C (1858) Notice ncrologique sur M. Darcy, Inspecteur Gnral des Ponts et Chausses,
Ann. Ponts Chausses, Ser. 3, 15, 90109.
Narasimhan TN (2005) Hydrogeology in North America: past and future. Hydrogeology Journal,
13:724.
Philip JR (1995) Desperately seeking Darcy in Dijon. Soil Sci. Soc. Am. J., 59, 319324.
Poiseuille JL (1841) Recherches exprimentales sur le mouvement des liquides dans les tubes de
trspetits diamtres, comptes rendus, Acad. des Sci., Paris.
Reynolds O (1883) An experimental investigation of the circumstances which determine whether the
motion of water shall be direct or sinuous and of the law of resistance in parallel channel, Philos.
Trans. R. Soc., 174, 935982.
Rouse H and Ince S (1957) History of Hydraulics, Iowa Inst. of Hydraul. Res., Univ. of Iowa, Iowa City.
Sharp J and Simmons CT (2004) The Compleat Darcy: New Lessons Learned from the First English
Translation of Les Fontaines Publiques de la Ville de Dijon. Ground Water, 43, 3, 457460.
Simmons CT (2003) Happy 200th birthday Mr. Darcy and our thanks for your law! A tribute editor-
ial celebrating the life and times of the father of our science, Henry Darcy (18031858).
Hydrogeology Journal, 11, 6, 611614.
Simmons CT (2004) Book review of The Public Fountains of the City of Dijon Henry Darcy,
1856, English translation by Patricia Bobeck. Journal of International Geoscience (Episodes) 27,
4, 307309.
Theim A (1887) Verfahress fur Naturlicher Grundwassergeschwindegkiten: Polyt. Notizblatt,
42:229.
CHAPTER 2
Henry Darcys Public Fountains of the City of Dijon
Patricia Bobeck
Texas Department of State Health Services, Austin, Texas, USA
ABSTRACT: An unabridged English translation (Bobeck, 2004) of Darcys 1856 book, now a rare
book available in few libraries, makes it possible for hydrogeologists to experience Henry Darcys activ-
ities and insights first-hand. Darcys book describes the construction of Dijons water supply system in
1840 and the sand experiments that led him to formulate Darcys Law, among numerous other topics.
As an engineer in the Corps of Bridges and Roads, Darcy fulfilled his lifes dream of providing
abundant clean water for his native city of Dijon. The book describes his research into Dijons 400-
year history of surface and ground water projects and his planning and execution of all aspects of
Dijons water supply system. Darcy calculated the populations daily water needs and selected a
spring to supply the necessary amount of water to Dijon via a 12-km aqueduct. He built two reser-
voirs, 13km of pipes, and 120 street fountains in the city. The fountains supplied free water for
domestic purposes, street flushing, and fire fighting.
1 INTRODUCTION
Henry Darcy (18031858) wrote The Public Fountains of the City of Dijon as a guide for
engineers involved in the construction of water supply systems. He wrote the book in 1856
to describe the water distribution system he built in 18391844 in Dijon. As a graduate of
the Ecole Nationale des Ponts et Chausses in 1826, Darcy was familiar with the water
issues and challenges of the early nineteenth century: the choice between river water and
spring water, filtration mechanics, types of distribution systems, financing, pipe materials,
and social equality in water distribution. Darcy had studied Girards Ourcq canal that
brought water to Paris in 1813 and had seen Emmery complete the distribution of the
Ourcq water and recognize the importance of sewers in the 1830s (Barles, 1999). In the
book, Darcy quotes Emmery on the social aspects of water distribution, and the immense
public health benefit from abundant water for the poorer classes. Darcy was also familiar
with Mallet and Genieys who worked on water distribution in Paris in 18201830 (Picon,
1992). Mallet favoured an interconnected network system that operated under pressure,
and Genieys was involved in distributing water from the Seine, building storage reservoirs
and using steam pumps to raise water to heights from which it could be distributed by
gravity. Darcy used the ideas of these and other predecessors in planning and constructing
Dijons water supply system (Guillerme, 1984, 1988).
During the1830s, Darcy investigated the sources of water available to the city, estimated
the citys water needs, and chose an abundant spring located in a nearby village. Darcy
designed and built a 12.7km aqueduct to Dijon and two reservoirs within the city. He
designed and built a network-type internal distribution system to deliver water to 120
street fountains. At the completion of the project, Dijon ranked second only to Rome in
terms of water quality and quantity. Within the walls of Dijon, street fountains were no far-
ther than 100m apart, meaning that no one had to walk more than 50m to obtain water. At
the time, water fountains in Paris were spaced at an average interval of 270m.
Because Darcys book is a how-to manual for engineers, it contains discussions of a
number of topics unrelated to the Dijon project. Darcy discusses rivers, ponds and lakes as
water supply sources; his understanding of artesian wells; pipes and pipe making; and nat-
ural and artificial filtration of river water, among other topics. In Darcys day, sand filters
were used to filter river water. Because the filters were so large, it was difficult for a city
to find space to build them. Darcys research into a way to make filters smaller led to the
sand experiments described in Darcys Law.
Darcys concern for the poor shows in his discussions of the importance of numerous
street fountains located so that poorer citizens would not put be off by the length of the
journey to obtain water.
Darcys father died when he was about 14, leaving the family in difficult straits. Henry
and his brother were excellent students, and their mother put forth great effort with the
city of Dijon to obtain money to educate them. During his childhood Darcy had been sick-
ened by the only water available and had promised himself to put an end to this situation if
ever he were in a position to do so (Darcy, 1957). After Henry completed his education and
entered the Corps of Bridges and Roads, the city of Dijon requested that he be assigned to
his native city. Shortly after returning to Dijon in 1827, Darcy began working on his plan,
and by 1832 he was gauging the Rosoir Spring, which he would later divert to Dijon.
The book is divided into four parts and an appendix. The four parts are further divided
into chapters. The English translation contains the 28-plate atlas of engineering drawings
that was originally published as a separate volume.
2 PART ONE
Part One is a description of the historical water situation of Dijon. Chapter 1 is an account of
the Darcys research on old fountains in the city. His friend the city archivist helped him
determine which springs Dijon had used as water sources during the previous four centuries.
In Chapter 2, Darcy examines the Suzon, an intermittent stream that flowed through
Dijon. Darcy investigated the widespread belief that prior to 1830 it had been a perennial
stream. Based on his review of the archives, Darcy concluded that the Suzon had been an
ephemeral stream for at least 450 years.
The Suzon also presented a more difficult problem. Its streambed, which passed through
the city, was a convenient refuse dump for many city dwellers. One of the goals of Darcys
water distribution system was to construct a cover over the sewer and flush it out. Darcy
accomplished the sanitization of the Suzon sewer in 1847, seven years after the completion
of the aqueduct. Darcy discusses this project in more detail in Part Four.
Part One Chapter 3 is entitled The Rosoir Spring, the name of the spring Darcy diverted to
Dijon. This long chapter contains numerous topics. Darcy describes the source of Dijons
drinking water as private wells and wells along the city streets that tapped into alluvium sat-
urated by fluid from adjacent permeable-wall cesspools. Darcy discusses the mid 19th
century view of the relative purity of well water, cistern water, pond (stagnant) water, river
water and spring water, concluding that spring water was preferable. He provides a chemical
analysis of water collected from a well in a private house and a water analysis of the Rosoir
water from 1850. Darcy enumerates the potential sources of water for Dijon, which included
several springs, a nearby river, and an artesian well. He calculates the amount of water
needed for a city water supply, including domestic needs, manufacturing, public buildings,
fire suppression, street cleaning, public fountains and gardens. Darcy determines the per
capita water requirement to be 150l/day, 90l for domestic and industrial purposes, and 60l
for street flushing. He then eliminates various sources that do not provide enough water, or
provide water that is too hot in summer, or water that is too expensive because it must be
raised from a riverbed. Darcy shows that the Rosoir Spring provides abundant pure cool
water that will remain cool until it reaches the street fountains because of the insulation pro-
vided by the underground aqueduct. Darcy discusses water quality in terms of what water
should contain: atmospheric air, carbonic acid, sodium chloride and calcium carbonate, and
states that it should also dissolve soap well. He discusses iodine, goitre and cretinism. He
describes the gauging of the Rosoir Spring in 183233. The Rosoir Spring flows from
Jurassic limestone and discharges 4000 to 12000l per minute depending on the season.
In a section on the origin of springs Darcy notes that by 1850 most people believed springs
to be fed by infiltrated rain water, and discusses historical views on the origin of springs,
including Descartes idea of underground fires and huge stills beneath the earths surface.
Darcy provides a classification of springs and discusses the history of springs-seekers, from
the Greeks to his French contemporaries. He describes methods of creating artificial springs.
A significant portion of Chapter 3 is dedicated to artesian wells. An artesian well had
been dug in Dijon in the early 1830s, but it did not flow above the ground surface and
could not provide enough water to supply the city, so Darcy rejected it as a water supply
source. During Darcys time, it was evidently believed that water circulated in pipe-like
voids under the earths surface. Darcy also recognized that many artesian wells flow from
sandy layers under an impermeable layer. Darcy understood that friction resulting from
water movement consumes hydrostatic head, or pressure, as he called it. Darcy understood
that it was possible to increase the discharge of an artesian well by lowering its discharge
point, and calculated that it was also possible to increase its discharge by increasing the
diameter of the well. He also understood that as the number of artesian wells from a single
source increases, the discharge of the wells decrease.
3 PART TWO
Part Two describes the construction of the aqueduct and the internal distribution system. In
Chapter 1, Darcy describes the masonry aqueduct he built between the Rosoir Spring and
Dijon. Construction of the aqueduct began in March 1839 and was completed in
September 1840. The aqueduct is 12.7km long between the pavilion that covers the spring
and the Porte Guillaume reservoir in Dijon (Fig. 1). For the most part, the aqueduct is
0.60m wide and 0.90m high, and is covered by one meter of soil. Manholes are located
every 100m. The spring is located on the bank of the Suzon stream upstream of Dijon, and
the aqueduct crosses the Suzon stream three times on the way to Dijon. The aqueduct also
passes through three villages where water is diverted from the Dijon aqueduct to provide for
the villages. Just before reaching Dijon, the aqueduct becomes a viaduct to cross an area of
low topography. In the text, Darcy provides details about the slope and cross section of the
aqueduct and details of all the work involved and all costs for material and labour.
Henry Darcys Public Fountains of the City of Dijon 27
In Chapter 2, Darcy discusses the internal distribution system, which consists of two
reservoirs and a system of cast iron pipes that branch out into all neighbourhoods of the
city to provide water to street fountains and private concessions. The main artery connects
the two reservoirs, distributor pipes branch off the main artery, and service pipes branch
perpendicularly off the distributors to serve the street fountains. Darcy used masonry tun-
nels and trenches for the pipes, depending on the importance of the street above the pipe.
The pipes connect at distribution tanks to form a grid system.
The Porte-Guillaume reservoir (Figure 2) is located at the end of the aqueduct and the
beginning of the main artery of the internal distribution system. Darcy brought water into
the Porte Guillaume reservoir through a vertical pipe in the central well of the reservoir. At
a certain height, the water flowed through some openings and down a stairwell to fill the
reservoir. This height was the head that controlled water flow and pressure and maintained
maximum water pressure throughout the city. Darcy also discusses the problem of air
expansion in the reservoir, and the means he devised to minimize this problem.
The Porte-Guillaume reservoir is circular in shape, and is covered by one meter of soil.
The reservoir capacity is 2313m
3
. Reducing the water supply to the amount strictly neces-
sary for the inhabitants, which Darcy estimated at 20 litres per person, or 540m
3
per day,
the reservoir could supply enough water for four or five days. Darcy constructed an
entrance structure on top of the reservoir, which is still there today.
Darcy knew that more storage capacity was required, primarily because an aqueduct
repair could easily last more than four or five days. He built a second reservoir, rather than
one larger reservoir, to prepare for the possibility that the main artery would require repair.
By building the Montmusard Reservoir on a hill at the other end of the main artery, Darcy
Figure 1. Profile of the Aqueduct that conveys the Rosoir Spring water to Dijon, Plate 2.
ensured that all points in the city could be supplied by interconnecting pipes. In addition,
Darcy saw that supplying the main artery from both ends made it possible to furnish a
much larger amount of water than if it were served by only one end. Another reason Darcy
gives for constructing the second reservoir is the possibility that the Porte Guillaume reser-
voir may need repair. In that case, the springs themselves could not supply enough water
for street watering during the hot summer when domestic water use would be high, but
with the water from the Montmusard Reservoir, it would be possible to do so.
The Montmusard Reservoir is rectangular and is located underground, covered by one
meter of soil. Its capacity is 3177m
3
. All flow in the city pipes was suspended every night
while the Montmusard Reservoir was being filled.
Chapter 2 also contains a detailed discussion of all components of the distribution sys-
tem. He provides the calculations for the jets of the fountain he built at the Place Saint-
Pierre, which still flows today. He also provides a detailed description of the street
fountains, of which only two still exist.
4 PART THREE
Part Three presents experiments that Darcy conducted on the aqueduct and distribution
system. Chapter 1 discusses experiments on water flow in the aqueduct that conveys water
from the Rosoir spring to Dijon. Chapter 2 deals with experiments on water flow in the
conduit system. The first three sections of Chapter 2 summarize Darcys memoir
Recherches exprimentales relatives au mouvement de leau dans les tuyaux [Experiments
on Water Movement in Pipes] that was published in 1857.
Figure 2. Cross section of the Porte Guillaume Reservoir, Plate 11.
5 PART FOUR
Part Four contains a discussion of the administrative and legal questions involved in the
construction of the water supply system. These included the expropriation of the spring,
the opposition mounted by the owners of mills located downstream of the springs, the pur-
chase of land the aqueduct would cross, the sanitizing of the Suzon sewer within the city,
and private water concessions.
Because the Rosoir spring was located within the jurisdiction of a nearby village, the
question of water rights had to be resolved. The question was settled in favour of the City
of Dijon by edict of the King on 31 December 1837. The city compensated the state and
the village for the spring. Darcy proposed a water distribution formula that gave the inhab-
itants of the villages between the springs and Dijon 150% of the per capita water alloca-
tion of a Dijon resident. The claims of the mill owners were finally resolved by the
payment of simple damages rather than expropriation costs.
Darcy was in charge of the land purchases for the aqueduct. He hired an expert to han-
dle the negotiations. He says of this process, the discussions were conducted with so
much calm and sincerity on the part of the expert that the landowners exhibited such mod-
eration that I was obliged to increase the amount of the indemnities claimed by one of the
communes. From that time forward, so many landowners felt confident of us that they sent
us the acts of sale, signed blank. Not a single opposition was raised, and there were five
hundred fifty-six parcels.
6 APPENDICES
The Appendix contains eight notes, designated by the letters A through H.
Appendix A is a list of springs located near Dijon. Darcy probably made this inventory
as a part of the process of selecting a water supply source.
Appendix B is a contract dated 6 December 1445 between the City of Dijon and Pierre
Belle, a carpenter from the neighbouring village of Talant. In this contract, the carpenter
agrees to bring the waters of the Montmusard Spring to the Porte Saint-Nicolas, one of the
old city gates of Dijon, through an aqueduct constructed of hollowed-out logs.
Appendix C is a discussion of the water supply systems of London, Paris, Brussels,
Lyon, Bordeaux, Nantes, Besanon and Nmes in the 1800s. In this note, Darcy also dis-
cusses his adaptation of the Pitot tube to make it easier to use in gauging water flow in
rivers and streams.
Appendix D, entitled Filtration contains an account of the experiments that led Darcy to
formulate Darcys Law. Darcy cites London and Glasgow as cities that practiced artificial fil-
tration. The disadvantage of artificial filtration is the large surface area required for the fil-
tration beds, 4,000 square meters in the case of one filtration bed in London. Darcy proposes
a modification to decrease the size of filtration beds by increasing the discharge of the filter,
using a taller column of water or negative pressure under the filter. This leads Darcy into the
discussion entitled, Determination of the Laws of Water Flow through Sand and his
description of the experiments he conducted in Dijon in 1855 with Engineer Ritter.
In Appendix E, Darcy discusses the methods he used to gauge the Rosoir Spring.
In Appendix F, Darcy discusses methods for drawing a constant volume of water from a
variable level channel.
Appendix G is a discussion of pipe strength and fabrication of cast iron, lead, sheet
metal and bitumen pipes. Sheet metal pipes covered with bitumen were a new invention in
the 1850s.
Appendix H contains additional information on water flow in the Rosoir Aqueduct.
7 CONCLUSION
In this brief overview of Darcys book, I have had to leave out numerous topics that Darcy
discusses. The reader of Darcys entire book will encounter many treasures not mentioned
in this article, including a discussion of the cisterns of Constantinople and an ingenious
fire-suppression plan for Dijons Theatre.
Darcy is also responsible for Dijon being located on the main rail line between Paris,
Lyon and Marseille. In addition, he was a city councillor and a founder and administrator
of social service organizations that sought to lend a hand to those like him who had had
difficult beginnings (Darcy, 1957). After his unexpected death in Paris, his body was
brought back to Dijon by train, and the entire city gathered at the train station to show their
respects. The City of Dijon renamed the location of the Porte Guillaume Reservoir to Place
Darcy in his honour.
REFERENCES
Barles, S (1999) La Ville Dltre: Mdecins et ingnieurs dans lespace urbain XVIIIeXIXe
sicle. ditions Champ Vallon, Seyssel. 373 p.
Bobeck, P. (2004) The Public Fountains of the City of Dijon by Henry Darcy (1856). Dubuque, Iowa:
Kendall/Hunt Publishing Company.
Darcy, H. (1856) Les Fontaines Publiques de la Ville de Dijon. Dalmont, Paris.
Darcy, P (1957) Henry Darcy: Inspecteur gnral des ponts et chausses, 18031858. Imprimerie
Darantire, Dijon. 63 p. Unpublished English translation by Patricia Bobeck.
Guillerme, A. (1988) The Genesis of Water Supply, Distribution and Sewerage Systems in France,
18001850, pages 91115 in Tarr, J et Dupuy, G (ed) Technology and the Rise of the Networked
City in Europe and America. Temple University Press, Philadelphia.
Guillerme, A (1984) Capter, clarifier transporter leau, pages 3146 in Les Annales de la Recherche
Urbaine 2324.
Picon, A (1992) Linvention de lingnieur moderne: LEcole des Ponts et Chausses 17471851.
Presses de lEcole Nationale des Ponts et Chausses, Paris. 767 p.
II The Measure of Permeability
CHAPTER 3
What grains can tell us about Darcy velocity
Maria Schafmeister
Applied Geology, Greifswald University, Greifswald, Germany
ABSTRACT: In the pursuit of methods to estimate hydraulic conductivity the key parameter in the
description and prediction of flow in porous media grain-size based methods have the appeal of sim-
plicity, but their reliability is questionable. A number of empirical methods were surveyed and sum-
marized. Their validity was assessed by comparing the predictions to actual measurements on several
representative samples (taken from previous studies). Although the limited number of the samples pre-
cludes any sweeping statistical conclusions and does not necessarily describe all modelling situations,
the results nevertheless show that grain-size based methods, particularly those that take into account
grain-size dispersion and properly account for the controlling influence of the smaller fractions, can be
an acceptable substitute for more expensive laboratory measurements and field tests when accuracy is
not of the essence and small-volume support is desired. Among the methods described, Beyers and the
US soil classification formula are favoured.
1 INTRODUCTION
In the hydrogeological description of flow systems, the hydraulic conductivity as introduced
by Darcy in 1856 is the key parameter (1).
(1)
where K Hydraulic conductivity [L/T]
Q/A Darcy velocity [L/T]
l/h hydraulic gradient [-].
Most techniques developed to measure K are modifications of Darcys original experiment.
All these experimental methods field tests as well as laboratory measurements derive
the hydraulic conductivity by performing a flow experiment under controlled boundary
conditions. Either flow is created through columns of aquifer material or radial flow is induced
towards a pumping well. The discharge is measured and subsequently related to the geo-
metry of the experiment and to other information on the aquifer conditions.
Whereas the physical process of flow is independent of the spatial scale of the experiment,
the results of K are not. Field pumping tests always result in effective K-values, which might
be understood as an integral value over the aquifer volume affected by the experiment. The
inherent heterogeneous structure of the aquifer cannot be ascertained at a scale smaller than
K
Q
A
l
h

the affected aquifer volume. The problem of up- and down-scaling has been addressed by
many authors (Gomez-Hernandez, 1998; Renard and de Marsily, 1997; Bierkens et al., 2000).
However, for most applications dealing with groundwater resources and their accessibility, the
small-scale spatial heterogeneity of aquifers does not play an important role. However, in
solute transport, the small-scale variability of K becomes important, since the hydrodynamic
dispersion is understood as a function of aquifer heterogeneity and transport distance (Beims,
1983; Kinzelbach, 1992; Schafmeister, 1990).
K-values derived from laboratory experiments (e.g. permeameter tests) are based on
aquifer material from small-sample volumes (1000cm
3
) and thus comprise a much smaller
scale of heterogeneity than field experiments. In addition, the anisotropic characteristics of K
can be estimated by varying the direction of flow with respect to the orientation of the sample
and thus to its inherent texture.
Laboratory experiment results from many samples distributed in space assumed to
represent point-support data may thus give better information on the spatial structure of
an aquifer by means of appropriate regionalization techniques e.g. geostatistical simulation
(de Marsily et al., 1998) than one or two expensive pumping tests. Of course, the latter are
to be preferred when bulk aquifer properties are required. However permeameter experiments
are very time-consuming especially with low-permeability sediments. For many practical
purposes permeameter tests are not suitable.
Hydraulic conductivity Kis related to the intrinsic permeability k by including the specific
conditions of the fluid water, e.g. dynamic viscosity
w
, density
w
and gravitational
acceleration g (2):
(2)
where K Hydraulic conductivity [m/s]
k intrinsic permeability [m
2
]
w
g weight density of water [N/m
3
]
w
dynamic viscosity [N/(sm
2
)].
Intrinsic permeability is strongly related to the pore space, i.e. to its volume and shape.
Ultimately, detailed microscopic knowledge of grain shape and pore-size geometry can lead
to permeability prediction from image analysis and basic fluid-mechanic principles (Doyen,
1988; Jin et al., 2004; Lebron et al., 1999). While these methods are important for research,
practical applications require methods to determine hydraulic conductivity that are fast and
easy to apply. Bulk statistical measures of grain-size distribution reflect the microscopic prop-
erties while simultaneously being easy to handle. Therefore, this paper examines how the abil-
ity of materials to conduct fluids can be deduced from the grain-size spectrum, which can be
obtained easily and cheaply from small samples (1kg) of unconsolidated material.
Empirical formulae were developed which relate the hydraulic conductivity K to specific
parameters, which can be read from cumulative curves of grain-size distribution, some of
which are discussed later in this paper. The resulting values are quasi-point supported given
the small sample size compared to that of field pumping tests.
Since the appearance of these empirical formulae, many studies have investigated how the
K-values derived from grain-size analysis compare with values from other laboratory or field
methods, e.g. Pekdeger and Schulz (1975). Results from these studies are discussed below.
K k
g
w
w


However, the question concerning which of all the methods provides the true K value can
never be answered. The true, or rather, effective hydraulic conductivity depends on the
considered volume since K varies in space. In addition, hydraulic conductivity is a tensor,
whose tensorial properties can theoretically be assumed but not easily proven (de Marsily,
1986; Matheron, 1967). However, as mentioned before, K-values derived from the grain-
size distribution in small samples can be understood as point supported. Moreover, these
methods do not allow consideration of the tensorial aspect of K.
It should be noted that the empirical grain-size methods apply only to hydraulic conduc-
tivity of unconsolidated aquifers. However, since most of the socio-economically relevant
aquifers in Germany and in many other countries are situated in young unconsolidated
Quaternary and Tertiary deposits, these methods have become widely accepted.
2 METHODS
The methods discussed below are all based on grain-size curves. Grain-size curves are
derived from sieve-analyses of loose sediments with grain diameters greater than 0.063mm,
which marks the limit between coarse silt and fine sand. Below this, laser particle-size detec-
tors (Leschonski, 1987; Wachernigg, 1987) or settling tests (hydrometer analyses) are used.
Results are depicted as cumulative curves of weight percent of material against the logarith-
mic axis of grain-diameter in mm (Fig. 1). The number and opening-sizes used in practice
differ from country to country. According to the German Industrial Norm (DIN 18123 1996)
six sieves with diameters of 2000, 1000, 500, 250, 125 and 63m are used for grain diame-
ters above 0.063mm.
Most empirical formulae for hydraulic conductivity have in common that a specific grain-
size, the effective grain-size d
e
, is squared and then multiplied by a coefficient C. The latter
is either a constant or varies with the grain sorting or other grain-shape parameters.
The following parameters can be read directly from the grain-size distribution curve:
coefficient of uniformity Uand effective grain-size d
e
. Additional information on the shape
of the grains is required for the method of Kozeny-Khler (Kozeny, 1927; Khler, 1960).
What grains can tell us about Darcy velocity 37
0
10
20
30
40
50
60
70
80
90
100
0.001 0.01 0.1 1 10 100
Grain size [mm]
C
u
m
.

w
e
i
g
h
t

%
clay silt sand gravel
f m c f m c f m c
Figure 1. Example of a cumulative grain-size distribution curve for fine to medium sand.
The coefficient of uniformity Uis a dispersion measure of the grain-size distribution and
is defined as the relation of d
60
to d
10
, i.e. the 60 percentile over the 10 percentile grain frac-
tion. As U approaches 1, the better the sediment is sorted. However values of U 20 are
reported for soils dominated by the silt and clay fraction.
The effective grain-size d
e
is defined as the equivalent grain-diameter of a strictly uniform
sand, which has the same intrinsic permeability as the actual sediment. It can be related to the
specific surface O of the material (3):
(3)
where O
i
surface of grain i [L
2
]
O specific surface [L
2
/L
3
]
V
tot
total Volume of solids
m 1 for spheres and cubes
The value 1 for mcan be assumed for most fluvial and washed sediments (Beyer, 1964). It
is not always practical to measure the specific surface O for a given sample. Therefore it
becomes convenient to introduce a surrogate value for d
e
. In the usual formulas (Eqs. 4 to
7, below), this surrogate value is approached by using a low-percentile grain-diameter
because the smaller grains determine the pore space. Many empirical formulae use d
10
, but d
20
or d
50
are common as well. Beyer (1964) researched how the relation between d
10
and d
e
varies according to the coefficient of uniformity U that is as sorting measure (Fig. 2).
O
O
V
d m
O
i
tot
e

6
Figure 2. Relation between d
e
and d
10
for different ranges of U (modified after Beyer, 1964).
A simple and well acknowledged empirical formula was developed by Hazen (Hazen,
1893; Zieschang, 1961) for medium sands. It relates K to d
10
as follows (4):
(4)
d
10
10 percent grain fraction [mm]
T temperature of weakly mineralized water.
At temperatures of 10C the factor in parentheses becomes 1. The formula applies only
for well-sorted sands, i.e. d
10
greater than 0.063mm and U lower than 5.
Beyer (1964) extended Hazens formula for less well-sorted sands. He found that the rela-
tion between d
e
and d
10
depends on the coefficient of uniformity U (Fig. 2) and, therefore,
introduced the factor C that incorporates the grain sorting (5). Factor C increases with
decreasing U, i.e. with increasing grade of sorting (Fig. 3). The formula applies for Ulower
than 30 and d
10
ranging from 0.063 to 0.63mm. It can easily be seen from (5) that Beyers
method always results in smaller K-values than Hazens method.
(5)
d
10
10 percent grain fraction [mm]
C factor (120 10
4
C 60 10
4
).
It should be mentioned here that already in 1880 Seelheim developed an empirical for-
mula, which relates K to the median grain-size d
50
as follows (6):
(6) K d 0 00357
50
2
.
K C d
10
2
K d T 0 0116 0 70 0 03
10
2
. ( . . )
Figure 3. Factor C as a function of U (modified after Beyer 1964).
d
50
50 percent grain fraction [mm].
The US Bureau of Soil Classification recommends (after Mallet and Pacquant, 1954):
(7)
d
20
20 percent grain fraction [mm].
Kozenys method (Khler, 1960) relates K to porosity, grain-shape, kinematic viscosity
of water and d
e
, the latter estimated as the harmonic mean of the grain-size distribution.
Today, Kozenys method is known as the simpler Kozeny-Carman equation (modified
from Bear, 1972):
(8)
n porosity
d
50
50 percent grain fraction [m]
w
g weight density of water [N/m
3
]
w
dynamic viscosity [N/(s m
2
)].
The term (
w
g/
w
), which depends on the fluid, appears in Eq. 8 because it was devel-
oped with intrinsic permeability k in mind (see Eq. 2). Equations 3 to 7 contain the term
assumed to be constant for water and standard gravity within their numerical constants.
Mash and Denny (1966) developed an empirical formula relating K directly to d
50
and
reciprocally to the dispersion range
I,
both given in phi-grades, where phi is the negative
logarithm base 2 of the grain-size in mm. The dispersion range I, is calculated from the
marginal reaches of the cumulative grain-size curve (d
5
, d
16
, d
84
and d
95
).
3 RESULTS
From several hydrogeological research projects, eleven grain-size distribution curves were
selected and analysed and K-values determined using the empirical formulae given above.
Where K-values derived from flow tests were available for the same samples, i.e. borehole
dilution, permeameter and infiltration tests, they were compared with the results of the
empirical methods (Table 1).
Samples a and b are fine-grained poorly sorted sands from top soils on Pleistocene glacial
tills (Darsow, 2003). The d
10
values are 0.05mm and thus slightly below the lower validity
limit for Hazens and Beyers methods. Samples c and e originate from Pleistocene, well
sorted fluvial, partly eolian sands deposited on Precambrian rocks at Chalk River, Ontario
(Hoffmann, 1997). Samples d, f, g, h, i, j are medium and medium-coarse Pleistocene
glacio-fluvial sands from a sand pit in North Germany (Auer, 1990; Karnani, 1990).
Sample k is a Tertiary fluvial medium-coarse sand whose grain-size distribution was pub-
lished by Langguth and Voigt (2004).
K
g
n
n
w
w
1
)
1
)

3
2
1 ( )
dd
50
2
180
1
)
K d 0 0036
20
2 3
.
.

In Fig. 4a the data are sorted by lithology (symbols), and by method (vertical axis). It can
be seen that the infiltration method (2) results in the smallest K-values. Comparing the per-
meameter (1) and borehole dilution (3) results to those from grain-size empirical relations
it can be concluded that the latter are more sensitive to the lithology, i.e. the fine-, medium-,
and medium-coarse-grained sands are better separated from one another, except in the
method by Mash and Denny (9). For the analysed samples the methods of Hazen (4) and
Beyer (5) provide almost the same results, K-values after Hazen being slightly larger. This
is explained by the fact that the grain-size distribution curves are quite steep, i.e. the U val-
ues range between 2 and 3.5. Like those of Hazen and Beyer, the US soil classification
method (6) uses a small effective grain-diameter and thus provides 2 to 3 times lower val-
ues than Hazen and Beyer for the given samples.
Table 1. K-values in 10
5
m/s derived from flow tests (1 to 3) and grain-size distribution curves (4 to 9).
ID Classification 1 2 3 4 5 6 7 8 9
of sand Permea- Infiltration Borehole Hazen Beyer US Soil Seelheim Kozeny- Mash-
meter dilution Class. Carman Denny
a very fine 3.8 0.1 n.d. 2.5 1.9 1.1 9.6 3.9 5.5
b very fine 7.7 0.2 n.d. 2.8 2.0 1.4 10.8 4.4 5.2
c fine n.d. n.d. 6.8 5.7 4.9 n.d. 6.4 2.6 6.1
d medium 12.0 n.d. n.d. 33.5 29.0 13.5 45.9 18.5 16.7
e medium n.d. n.d. 24.0 37.6 33.7 n.d. 35.1 14.2 15.0
f medium 20.0 n.d. n.d. 61.4 52.9 22.6 75.9 30.7 17.5
g medium 13.0 n.d. n.d. 61.4 53.9 19.3 75.9 30.7 25.0
h coarse 29.0 n.d. n.d. 78.4 62.5 36.6 205.4 82.9 33.3
i medium/coarse 34.0 n.d. n.d. 78.4 66.9 32.2 113.4 45.8 15.8
j medium 34.0 n.d. n.d. 84.6 74.0 30.1 93.8 37.8 18.3
k medium n.d. n.d. n.d. 84.6 75.1 32.2 93.8 37.8 25.5
n.d. not determined
1E-006 1E-005 0.0001 0.001 0.01
K [m/s]
1
2
3
4
5
6
7
8
9
ab
a
c
ab c
ab
a b
ab c
ab c
a bc
dg
e
de f g
de f g
d g
d e g
d e g
d ef g
hij
hij k
hij k
h i jk
h i j k
h i j k
h ij k
Very fine/fine sand
Medium sand
Medium - coarse sand
0.0001
K permeameter [m/s]
1E-005
0.0001
0.001
K

[
m
/
s
]
Hazen
Beyer
US soil class.
Seelheim
Kozeny-Carman
Mash and Denny
0.0004 3E-005
0.003
a) b)
f
f
f
f
c
b
Figure 4. (a) K-values for 11 samples from permeameter (1), infiltration (2), borehole dilution
(3) tests and according to empirical methods Hazen (4), Beyer (5), US Soil Classification
(6), Seelheim (7), Kozeny-Carman (8) and Mash-Denny (9); (b) All methods compared to perme-
ameter results (1:1 line).
The methods of Seelheim (7), Kozeny-Carman (8) and Mash and Denny (9) use the
median grain- diameter as d
e
. However, the last two methods include other parameters
relating to grain shape and packing. Kozeny-Carman uses the porosity. The K-spectrum is
reduced by a factor of 2.5 compared to Seelheims results. Both the Seelheim and Kozeny-
Carman methods provide more clearly separated results for the coarser sand samples.
Mash and Dennys method includes the spreading range I which lowers the K-values
according to the degree of sorting. Seelheims method (7) tends to overestimate K com-
pared with all other methods.
On the whole, it can be seen that for the medium-grained samples, all methods provide
comparable results.
Assuming that the permeameter test reproduces Darcys original experiment best, it can
serve as a validation for the empirical grain-size methods. The results listed above were
obtained after coring the samples with minimum disturbance. The grain-size distribution was
assessed from the disturbed sample after the permeameter test. Fig. 4b demonstrates that all
grain-size methods tend to underestimate K for fine sand and overestimate K for medium and
medium-coarse sand fractions with respect to the permeameter results. Seelheims method
always overestimates K by a factor of 1.4 for fine sands up to 7 for coarse-grained sands.
Again, d
50
as the effective grain-diameter appears to be a poor choice since it puts too much
weight on the coarser grain fractions and suppresses the influence of the finer grains on the
pore space. However if d
50
is supplemented by other factors, e.g. porosity, grain shape and
spread, as in the formulae by Kozeny-Carman and Mash and Denny, the resulting K values
approach the range of the permeameter results. Nevertheless, the correlation between these
methods and the permeameter test (Table 2) is less good than the correlation with the meth-
ods of Hazen, Beyer and the US soil classification, which indicates that for the given samples,
the latter provide results that reliably reflect the process of flow through porous media.
4 DISCUSSION
The results presented above are based on a very limited number of samples from arbitrar-
ily selected sites and thus cannot be taken as a robust statistical proof. However, they com-
pare well with results from other studies which are discussed below.
Musolff et al. (2004) analysed 42 samples of fine-grained and silty soils on glacial tills,
of which samples a and b are representatives. Since d
10
was always smaller than 0.06mm,
Hazens and Beyers methods were not valid. It was shown, however, that the infiltration
method provided reliable results for these soils. This field test was designed to determine
the hydraulic conductivity of unsaturated soils in the field as a function of the tension head
of the undisturbed soil (Bohne, 1998; Wooding, 1968). K-values derived by the US soil clas-
sification method for the same samples correlate weakly with the infiltration but not with
the permeameter results (Darsow, 2003).
Hoffmann (1997) calculated K values from 145 grain-size curves from core samples of
11 wells in fluvial, partly eolian sands deposited on Precambrian bedrock in Chalk River,
Table 2. Correlation coefficient between permeameter results and empirical grain-size analyses.
n 8 Hazen Beyer US Soil Cl. Seelheim Kozeny-Carman Mash-Denny
Correlation coefficient 0.91 0.90 0.93 0.77 0.77 0.55
Ontario. Borehole dilution tests were made on 15 selected segments within the wells and
correlated with the respective samples. The d
50
values varied between 0.07 and 0.3mm,
representing fine- and medium-grained sands. K-values derived according to Hazen and
Beyer correlate with the borehole dilution results at 0.66 and 0.68, respectively and cover
the same range of hydraulic conductivity of between 0.5 and 4
.
10
4
m/s. Mash and
Dennys method shows an even higher correlation of 0.72. However, only the smaller val-
ues compare well with the borehole dilution results; K-values from samples in higher per-
meability segments are significantly lower, by a factor of 3 according to Mash and Danny.
Schafmeister and Pekdeger (1993) investigated spatial heterogeneity of K and its effect on
hydrodynamic dispersion. 219 undisturbed samples were taken from a 2m by 2m cross-
section in a sand pit in northern Germany (Auer, 1990; Karnani, 1990). With a mean d
50
of
0.5mm, the samples were classified as medium to coarse sands. Permeameter tests were per-
formed on 48 undisturbed samples in the lab and the results were compared to K-values
derived according to Beyer. The average K for Beyer and the permeameter test is 6 10
4
and
2.3 10
4
m/s, respectively. Here, the method by Beyer provides 2.5 times higher values than
the permeameter test, which is due to the fact that the dense packing of the undisturbed sam-
ples is not considered in Beyers simple formula. It can be expected that Kozeny-Carmans
method might give better results.
Pekdeger and Schulz (1975) compared K-values derived from permeameter tests, grain-
size analyses (Beyer, Hazen) and pumping tests which were evaluated by different methods.
The test sites are situated in glacio-fluvial sands in northern Germany. The d
50
grain-diameter
was 0.2mm, indicating medium-grained sands. Again the permeameter tests yielded the low-
est results (Fig. 5). For all test sites, Beyers method gave the smallest variation within the sam-
ples and slightly lower values than Hazens method, which results from the fact that
Beyers formula includes the coefficient of uniformity U. The large range of variation of the
pumping test results must be linked to the variety of different interpretation methods applied
to the same pumping test data. This example illustrates that although pumping tests may
reflect groundwater flow conditions more reliably than any laboratory test, whether or not
realistic results are obtained still depends on the level of information concerning the hydrody-
namic boundary conditions.
The governing parameter in groundwater hydrology, the hydraulic conductivity, is one
of the first parameters to be measured in any investigation. Before deciding which method
should be applied, the purpose of the investigation must be clear. Questions involving
Figure 5. Comparison of permeameter tests, downward flow (2) and upward flow (3), Hazen (4)
and Beyer (5) and pumping test results (6) (modified after Pekdeger and Schulz, 1975).
groundwater resources and production require information on the hydrodynamic flow sys-
tem on a much broader scale than most applications that deal with the question of solute
transport and hydrodynamic dispersion, the latter being influenced by small-scale differ-
ences in the hydraulic conductivity. Methods involving a larger aquifer volume, e.g. pump-
ing tests, will not reveal the small-scale heterogeneity of the aquifer. Here, any method based
on a large quantity of small samples taken uniformly within the area of interest will better
serve to assess and reproduce aquifer heterogeneity (Schafmeister and de Marsily, 1994).
There is not much doubt that laboratory permeameter tests come closest to the physical
process of fluid flow through porous media as originally described by Darcy. Thus these tests
are appropriate when small-scale flow is investigated. Even the tensorial aspect of K can
be approached by such tests. However, these tests are quite time-consuming and care must be
taken with the experimental design, e.g. samples should be undisturbed and there should
be no air inclusions during the test.
5 CONCLUSIONS
Within the framework of geological and hydrogeological studies in unconsolidated aquifers,
grain-size analysis is one of the standard investigation methods to classify the sediment. Thus,
the empirical determination of hydraulic conductivity from grain-size distribution curves
is certainly the cheapest method and fastest approach for the least effort. The use of a com-
puter is barely necessary. However special attention should be given to the validity of the cho-
sen method.
Among the methods discussed above, the one by Hazen is the easiest to use since it requires
one parameter (d
10
) only, however the validity range is restricted to uniform sands, i.e. U 5.
The method by Mash and Denny is well known in the US, but barely used in Europe. It uses the
older phi-grades rather than the metric mm. The spread measure is calculated from the extreme
reaches of the cumulative grain-size curve and may thus be subject to measurement errors.
Although Seelheims formula has not been much used recently, it was included in this
study for reasons of completeness. Based on d
50
it tends to overestimate K and thus turns out
to be the weakest of all described methods. The methods by Kozeny-Carman and Mash and
Denny, which in addition to d
50
use porosity and grain shape or a spread measure, yield bet-
ter results because they adapt better to a broader range of influencing physical properties.
However, because porosity is often not measured and the grain shape is only qualitatively
assessed, Kozeny-Carmans formula contains many sources of misinterpretation as well.
The disadvantage of Hazens formula, i.e., that of not being valid for poorly sorted sands,
has been amended by Beyer to include the coefficient of uniformity U as a spread meas-
ure. This easy-to-use and cheap method has proved to be very effective especially for glacial
and fluvial aquifers in Germany (Pekdeger and Schulz, 1975) and is widely used there but is
not well known elsewhere.
The US soil classification method, which uses d
20
as the effective grain-diameter, is equally
easy to use. For the selected samples, it yielded good results compared to the permeameter
tests. However, it seems to work best for very fine-grained soils, i.e. with a high percentage of
silt and clay. It may thus serve as an appropriate alternative for sediments that are beyond the
range of validity for Hazen and Beyer, i.e. d
10
0.06mm.
In conclusion empirical formulas for the determination of K from grain-size distribution
curves of small samples of unconsolidated aquifer material provide reasonable results with
a minimum of technical effort and are thus a cheap and fast alternative to flow tests, i.e.
pumping, permeameter or infiltration tests as used in hydrogeological investigations.
ACKNOWLEDGEMENT
The author would like to thank the numerous former students who have performed the
sieve and grain- size analyses.
REFERENCES
Auer M (1990) Der Einfluß kleinrumiger Strukturen auf die Dispersivitt in quartren Sande.
Unpublished Diploma thesis FU Berlin, 77pp
Bear J (1972) Dynamics of fluids in porous media. Amer. Elsevier, New York
Beims U (1983) Planung, Gte und Ausfhrung von Gtepumpversuchen. Zeitschrift fr
Angewandte Geologie 29, 10, 484492
Beyer W (1964) Zur Bestimmung der Wasserdurchlssigkeit von Kiesen und Sanden aus der
Kornverteilung. Wasserwirtschaft, Wassertechnik (WWT) 14, 165168, Berlin-Ost
Bierkens MFP, Finke PA, De Willigen P (2000) Upscaling and downscaling methods for environmen-
tal research. Developments in plant and soil sciences, Kluwer Academic Publishers 88, 190 pp
Bohne K (1998) Wasserbewegung und Wasserleitfhigkeit des Bodens. Book chapter Blume HP,
Felix-Henningsen P, Fischer WR, Frede HG, Horn R, Stahr K: Handbuch der Bodenkunde.
ecomed, Landsberg/Lech
Darcy H (1856) Les fontaines publiques de la ville de Dijon. Dalmont, Paris
Darsow A (2003) Aufbau der oberflchennahen Sedimente eine GIS gesttzte Kartierung der
Infiltrationskapazitt im Bereich der Ortslage Schmarsow, Landkreis Demmin. Unpublished
Diploma thesis University Greifswald, 66p
de Marsily G (1986) Quantitative Hydrogeology Groundwater hydrology for Engineers. Academic
Press, 440 pp
de Marsily G, Delay F, Teles V Schafmeister MT (1998) Some current methods to represent the het-
erogeneity of natural media in hydrogeology. Hydrogeology Journal 6, 115130
DIN 18123 (1996) Baugrund; Untersuchung von Bodenproben Bestimmung der Korngrßen-
verteilung. Beuth Berlin.
Doyen PM (1988) Permeability, conductivity, and pore geometry of sandstone. Journal of
Geophysical Research, 93, B7, 77297740
Gomez-Hernandez JJ (1998) Upscaling and downscaling: travelling up and down the scales ladder
(Solicited Paper). HSA1 Hydrology and the Earths crust.2 Identification of model parameters in
groundwater hydrology EGS meeting Nice
Hazen A (1893) Some physical properties of sands and gravels with special reference to their use in
filtration. Ann. Rep. Mass. State Bd. Health, 24, 541556, Boston
Hoffmann H (1997) Vergleichende Betrachtung verschiedener Labormethoden zur Bestimmung von
k
f
-Werten am Beispiel eines pleistoznen Grundwasserleiters in Ontario/Kanada. Unpublished
Diploma thesis FU Berlin, 77 pp
Jin G, Patzek TW, Silin DB (2004) Direct Prediction of the Absolute Permeability of Unconsolidated
and Consolidated Reservoir Rock. SPE Paper 90084, 15 pp
Karnani F (1990) Der Einfluß der Anisotropie quartrer Sande auf die Dispersivitt.
Unpublished Diploma thesis FU Berlin, 75 pp
Kinzelbach W (1992) Numerische Methoden zur Modellierung des Transports von Schadstoffen im
Grundwasser. 2nd edition, Schriftenreihe Wasser-Abwasser, Oldenbourg, 343 pp
Khler W (1960) Erweiterte Anwendung der Mittelwertkurve auf nichtlineare Teilungen
Graphische Ermittlung der spezifischen Oberflche von Krnungen. Bergakademie 12, 376382
Kozeny J (1927) ber Grundwasserbewegung. ZdDWWV 7 / Die Wasserwirtschaft 22
Langguth HR, Voigt R (2004) Hydrogeologische Methoden. 2nd edition, Springer, 1005 pp
Lebron I, Schaap MG, Suarez DL (1999) Saturated hydraulic conductivity prediction from micro-
scopic pore geometry measurements and neural network analysis. WRR 35, 10, 31493158
Leschonski K (1987) Partikelmeßtechnik, gegenwrtige und zuknftige Entwicklungen.
Erzmetall, 40, 8390
Mallet C, Pacquant J (1954) Erdstaudmme. 341 pp, Berlin
Masch FD, Denny KJ (1966) Grain Size Distribution and Its Effect on the Permeability of
Unconsolidated Sands. Water Resources Research, 2, 665677
Mathron G (1967) Elments pour une thorie des milieux poreux. Masson, Paris
Musolff A, Meyer T, Darsow A, Schafmeister MT (2004) Grundwasserneubildung und
Direktabfluß II In-situ Messung der hydraulischen Leitfhigkeit von Bden.
Schriftenreihe der Deutschen Geologischen Gesellschaft, 32, 78
Pekdeger A, Schulz HD (1975) Ein Methodenvergleich des k
f
-Wertes von Sanden. Meyniana 27,
3540
Renard P, de Marsily G (1997) Calculating the equivalent permeability: a review. Advances in Water
Resources, 20, 56, 253278
Schafmeister MT (1990) Geostatistische Simulationstechniken als Grundlage der Modellierung von
Grundwasserstrmung und Stofftransport in heterogenen Aquifersystemen. PhD thesis, Verlag
Schelzky & Jeep, 143 pp, Berlin
Schafmeister MT, de Marsily G (1994) The influence of correlation length of highly conductive
zones in alluvial media on the transport behaviour. in: Dracos & Stauffer (eds.) Transport and
Reactive Processes in Aquifers, 171176, Balkema
Schafmeister MT, Pekdeger A (1993) Spatial structure of hydraulic conductivity in various porous
media problems and experiences. In: Soares A (ed.) Geostatistics Tria 92, Quantitative
Geology and Geostatistics 5, 733744, Kluwer Academic Press
Seelheim F (1880) Methode zur Bestimmung der Durchlssigkeit des Bodens. Z. anal. Chemie 19,
387418
Wachernigg H (1987) Ein neues Laser-Partikelanalysensystem fr Materialien von 0,1600 m.
Verfahrenstechnik, 21, 1822
Wooding RA (1968) Steady infiltration from a shallow circular pond. Water Resources Research 4,
12591273
Zieschang J (1961) Zur zulssigen Hchstbelastung eines Brunnens. Z. Angew. Geol. 7: 580582
CHAPTER 4
Measuring porosity and permeability on drill cuttings
Roland Lenormand
1
& Patrick Egermann
1,2
1
Institut Franais du Ptrole (IFP), France
2
now with Gaz de France (GDF), Av du Prsident Wilson, France
ABSTRACT: The early determination of rock petrophysical properties is one of the main concerns
during the petroleum reservoir characterization process because it impacts directly the reserves (poros-
ity, saturation) and also the well deliverability (permeability). Logs generally provide a good estimate
of porosity and saturation along the well, but permeability is more difficult to measure especially in
carbonates. Drill cuttings can provide additional information on the petrophysical properties of the reser-
voir. In this paper, we have described a method for measuring permeability on drill cuttings with the
following capabilities: size of cuttings down to 1mm diameter, volume of cuttings around 1cm
3
(23
grams of dry cuttings), permeability from the microDarcy to around 100 miliDarcy. The main con-
straint of the method is the need for consolidated cuttings with a porosity value larger than 4% in order
that a minimum volume of air needs to be compressed during the cutting pressure increase. Porosity
is difficult to measure on small cuttings. The more accurate method is based on NMR relaxometry with
an apparatus dedicated to the measurement of small volumes.
1 INTRODUCTION
Cuttings are routinely used by mudloggers to build the masterlog, in which the geological
description of the drilled formation is reported. Hydrocarbon indices are also determined
from cuttings to identify the reservoir levels. Although the rock cuttings material is coming
directly from the reservoir, few applications of permeability characterization are reported
in the literature. The published works can be divided into two categories: direct and indirect
evaluations.
1.1 Indirect permeability evaluation methods from cuttings
The permeability is derived using empirical correlations with properties related to pore size
distribution, pore connectivity, or the spatial correlation between the pores. These parameters
are evaluated from capillary properties (mercury porosimetry), by NMR (Nuclear Magnetic
Resonance) or image analysis.
Several approaches have been proposed to derive permeability from mercury porosimetry
curves (Swanson, 1981). Kamath (1992) performed a comparison between these methods
and concluded that the best result is obtained with new correlations based on the Swanson
(1981) characteristic length. Swanson (1981) and Kamath (1992) also reported a possible
use of mercury porosimetry curves obtained from cuttings.
Many papers have been published on the permeability evaluation from NMR measure-
ments, and recently, some refer to applications on cuttings (Mirotchnick et al., 2004).
A thin section can also be obtained from cuttings to evaluate the porosity and the per-
meability using image analysis. The porosity corresponds to the fraction of voids whereas
the permeability is derived from an empirical law or from Carman-Kozeny type laws.
1.2 Direct permeability evaluation methods from cuttings
Few methods exist to measure permeability directly from cuttings (i.e. from a flow measure-
ment). Santarelli et al. (1998) proposed imbedding cuttings into thin disks of acrylic resin and
then measuring permeability by flowing a gas through this disk, and interpreting the meas-
urement as if it were a real core sample. However, the thickness of the disk must be smaller
than the size of the cuttings to have a permeable medium. Another method, which is based on
gas pressure diffusion, was proposed by Luffel (1993). The principle is similar to well testing,
using gas instead of liquids. However, due to the low viscosity of gases, this method can only
be used on weakly permeable rocks (permeability below the microDarcy).
2 PERMEABILITY FROM DRILL CUTTINGS
Recently, Institut Franais du Ptrole (IFP) presented an original method to measure the
permeability directly from drill cuttings (Egermann et al., 2005). The proposed method
does not require specific conditioning, is easy to apply and is suitable in typical ranges of
reservoir permeability. The main issue in determining permeability from cuttings is to
establish flow into the rock itself rather than in inter-cuttings space. The originality of the
method proposed by IFP is to achieve an effective flow inside the cuttings by compression
of the residual gas that they contain. Viscous oil is used as displacing fluid in order to
decrease the pressure diffusion kinetics into the rock. It enables measurement of permea-
bility values corresponding to reservoir rocks, which are not accessible from gas pressure
test (Luffel, 1993).
The experimental apparatus, shown in Figure 1, consists of a cuttings cell containing a
few cm
3
of cuttings and a pressure cell containing bellows that are coupled to a spring and
a pressure sensor. All the apparatus is filled with oil that can be injected under pressure with
a pump.
Initially, the dried cuttings are saturated with oil by spontaneous imbibition in a beaker.
At the end of the spontaneous imbibition, around 10% of air remains trapped inside the
cuttings as residual, disconnected gas. The cuttings are then poured into the cutting cell
that is then quickly connected to the pressure cell (around 10 bars). Oil enters into the
cuttings and the trapped air is compressed. During the displacement, the pressure is
recorded (Fig. 2) and the volume of oil entering into the cuttings is derived from the cali-
bration of the bellows (Fig. 3). The rate of invasion depends on the fluid viscosity and the
rock permeability.
The permeability is calculated by using a numerical model based on the equations describ-
ing the flow of a viscous fluid into a compressible medium of spherical geometry. The model
is similar to a well test, except that the oil compressibility is replaced by the trapped gas
compressibility, with spherical instead of cylindrical geometry (see Egermann et al., 2006).
The relaxation time depends on both oil and cuttings properties:
with the following notations: , oil viscosity (dynamic); K, cuttings permeability; , porosity;
S
g0
, initial gas saturation; P
0
, compression pressure (absolute); r, mean cuttings radius.
K
S
P
r
g
0
0
2
Measuring porosity and permeability on drill cuttings 49
Entrance valve V1
Bellow and spring
Pressure
sensor
Pressure valve V2
Cuttings cell

Outlet valve V3
Figure 1. Experimental apparatus.
9
9.1
9.2
9.3
9.4
9.5
0 20 40 60 80 100 120
Time (s)
P
r
e
s
s
u
r
e

(
b
a
r
)
Figure 2. Pressure drop in the liquid when the cuttings cell is connected. The relaxation corres-
ponds to the oil entering into the cuttings.
The determination of the permeability needs the adjustment of three parameters:
The volume of initial gas V

ini
, which corresponds to the instantaneous relaxation due to
the cell relaxation, and the gas trapped outside the cuttings. The value of V
ini
is around
20mm
3
and is adjusted using the beginning of the volume curve
The gas saturation inside the cuttings S

g0
, value generally between 0.1 and 0.2. This
parameter is adjusted from the amplitude of the relaxation of the volume curve,
The permeability K, adjusted using the curvature of the relaxation curve.

The three above-described parameters can be determined either by manual or automatic opti-
mization. The automatic optimization is based on a Levenberg-Marquardt procedure. The
three parameters are well decoupled if the experiment is long enough to reach the equilib-
rium, since the initial volume is determined at initial time and the amount of gas inside the
cuttings from the volume at end of experiment. Errors due to non-uniqueness can be con-
sidered as negligible. The numerical calculation takes also into account two corrections for
capillary and relative permeability effects.
With this apparatus, it is possible to make a measurement with a volume of trapped gas of
the order of 20mm
3
, which corresponds to a mass of about 2 grams of dry cuttings. In Figure
the solid line represents the result of the numerical simulation after optimization of the three
unknown parameters.
Generally, a first measurement is performed with an average oil viscosity around 200 cp.
Then, depending on the observed relaxation time, a second measurement is performed with
a more viscous oil (around 4000 cp) if relaxation is too fast, or a less viscous oil (10 or 1 cp)
for slow relaxation. Relaxation times can be interpreted from 1 to 10 seconds. Combining
relaxation time interpretation and oil viscosity allows a range of permeability around
6 decades.
0
10
20
30
40
0 1 2 3
Time (s)
V
o
l
u
m
e

(
m
m
3
)
Figure 3. Volume of oil injected in the cutting cell as function of time: experiments (symbols x)
and numerical simulation after optimization (solid line).
Fig. 4 shows the good agreement between the core permeability, measured by liquid dis-
placement, and the cuttings permeability obtained after crushing the core samples (size
12mm and 23mm), in a very wide range of permeability, from the microDarcy to the Darcy.
3 POROSITY MEASUREMENT TECHNIQUE
Porosity is more difficult to determine on small cuttings than permeability. For cuttings larger
than 3mm, porosity can be determined by using the standard methods used for cores. For
smaller cuttings, the liquid films on surface lead to large errors and the best method is to
determine the pore volume using NMR relaxometry. See for instance Fleury (1998) for the
description of NMR relaxometry.
Volume of solid is derived from the mass M
dry
and grain density. If the lithology of the cut-
tings is not known, an average value of 2.68 g/cm
3
is taken since the grain density of most
rocks is in the range 2.64 (quartz)-2.72 (carbonate) g/cm
3
.
Using the definition of porosity, the error introduced by this assumption is given by:

( ) 1
d
d
Vs
M
d
dry

0.0001
0.001
0.01
0.1
1
10
100
1000
0.0001 0.001 0.01 0.1 1 10 100 1000
K cuttings (mDarcy)
K

c
o
r
e

(
m
D
a
r
c
y
)
Figure 4. Comparison between core and cuttings permeability.
For a porosity 0.1, 0.001. The error on porosity due to this assumption can be
considered as negligible. The measurement using helium expansion could be done, but it
is time-consuming and does not improve the result on porosity, compared to experimental
errors.
As with core samples, the cuttings are saturated under vacuum followed by pressuriza-
tion at 150 bars for at least one hour, to dissolve the remaining air that may be trapped.
The total intensity of the NMR signal is proportional to the number of protons and con-
sequently, a mass of liquid can be determined after calibration with a known mass of
liquid. The NMR apparatus used is Maran spectrometer with a frequency of 23MHz and
a probe of 18mm. With this apparatus, the accuracy is of the order of 1mg.
For small cuttings, the water that remains between the cuttings presents a longer relax-
ation time and can be quantified like the liquid in the vugs of a vuggy carbonate (see for
instance Fig. 2 in Fleury, 1998). In Fig. 5 showing the T2 distribution, the first peak corre-
sponds to water into the cuttings and the second to water between the cuttings. On the T2
distribution (Fig. 5), the area under the two peaks is proportional to the volume (or mass)
of water inside and outside the cuttings.
This approach was tested on crushed core. In Fig. 6, the agreement between cores and
cuttings is not always very good, but for most of samples, the porosity was measured on com-
panion cores only. What is more important is the good agreement between the cuttings of dif-
ferent sizes, not obtained with other methods.
4 CONCLUSIONS
In this paper, we have described a method for measuring permeability on drill cuttings
with the following capabilities:
size of cuttings down to 1mm diameter,
volume of cuttings around 1cm

3
(23 grams of dry cuttings),
permeability from the microDarcy to around 100mDarcy,

0
5
10
15
20
25
30
1 10 100 1000 10000
Relaxation time (in ms)
A
m
p
l
i
t
u
d
e
Figure 5. Principle of NMR porosity measurement on cuttings. Like in vuggy carbonates, the T2
relaxation time of water between the cuttings is larger than the relaxation inside the pores.
The main constraint of the method is the need for consolidated cuttings with a porosity value
larger than 4% in order to have a minimum volume of air trapped to be compressed during
the cutting pressure testing to derive permeability.
Porosity is difficult to measure on small cuttings. The more accurate method is based on
NMR relaxometry with an apparatus dedicated to the measurement of small volumes.
REFERENCES
Egermann, P., Lenormand, R., Longeron, D., and Zarcone, C. (2005) A fast and direct method of per-
meability measurements from drill cuttings, SPEREE (August 2005) 4, 269275.
Egermann, P., Doerler, N., Fleury, M., Behot, J., Deflandre, F. and Lenormand, R. (2006) Petrophysical
measurements from drill cuttings: an added value for the reservoir characterization processes,
paper SPE 88684, published in SPE Reservoir Evaluation and Engineering, August 2006, pp.
303307.
Fleury, M. (1998) Caractrisation de structures poreuses par relaxomtrie RMN, Revue de lInstitut
Franais du Ptrole, Vol. 53, 4, 489493
Kamath, J. (1992) Evaluation of accuracy of estimating air permeability from mercury injection
data, SPEFE (December 1992) 4, 304310.
Luffel, D. L. (1993) Devonian shale matrix permeability successfully measured on cores and drill
cuttings, Gas Shales Technology Review, 8, 2, 4655.
Mirotchnik K., Ktyuchkov S. and Strack K. (2004) A novel method to determine NMR petrophysical
parameters from drill cuttings, SPWLA n 45th Annual Logging Symposium, The Netherlands.
Santarelli, F. J., A. F. Marsala, M. Brignoli et al. (1998) Formation evaluation from logging on cut-
tings, SPEREE (June 1998), 238244.
Swanson, B. F. (1981) A simple correlation between permeability and mercury capillary pressures,
J. of Petrol. Technology (December 1981), 24982504.
0.0
0.1
0.2
0.3
0 0.1 0.2 0.3
Core porosity
C
u
t
t
i
n
g
s

p
o
r
o
s
i
t
y
Size A (12 mm)
Size B (23 mm)
Size C (35 mm)
Median
Figure 6. Comparison between core and cuttings porosity using the NMR method.
CHAPTER 5
The quest for permeability evaluation in wireline logging
Jean-Pierre Delhomme
Schlumberger Water Services, cours du Triangle, La Dfense Cedex, France
ABSTRACT: For decades, a constant objective of wireline logging has been to obtain a continu-
ous permeability log. Except for a few attempts such as the search for an acoustic log response that
could directly yield a permeability indicator, most of the initial efforts have been directed towards
deriving permeability from the combination of porosity with some other log-derived property related
to the type of pore geometry. In sandstones, excellent results have recently been obtained with nuclear
magnetic resonance (NMR) logging that, by itself, provides information on both porosity and pore
size distribution. In carbonates, the NMR approach sometimes breaks down, but the information about
carbonate rock facies carried by continuous electrical images of the borehole walls has, when coupled
with conventional porosity logs, generated continuous permeability indicators in complex carbonate
formations.
1 THE CHALLENGE
Since 1856 when Henri-Philibert-Gaspard Darcy first defined fluid conductivity of a porous
material in his famous technical report known as the Mmoire sur les fontaines publiques
de la ville de Dijon, permeability has become one of the most studied, yet stubbornly elu-
sive, properties of rocks. For decades, hydrogeologists have been using pumping tests to meas-
ure permeability in aquifers, or rather to access an average permeability-thickness value,
called transmissivity, masking permeability differences in different layers. Similarly, many
well testing techniques were developed by the petroleum industry. Well testing rapidly became
an oilfield standard because it was investigating the rock and fluid in situ, under actual reservoir
flow conditions. However, none of the well testing methods, except a rather cumbersome and
lengthy one called layer reservoir testing, are providing information about the variations of
permeability versus depth.
To achieve this goal, cores are often taken at different depths when drilling, and core
samples are analyzed under controlled laboratory conditions to measure permeability.
Coring and laboratory analyses are quite expensive procedures. Wells are, therefore, rarely
cored continuously but, even when they are, core permeability data can be of questionable
value when only 6-inch spaced core plugs are analyzed in heterogeneous rocks where per-
meability over just a few inches can vary by five orders of magnitude. The idea of a continu-
ous permeability profile is attractive, using the same approach that had been successful in
providing continuous profiles of porosity and fluid saturations in the formations crossed
by oil and gas, and sometimes water, wells: wireline logging.
2 FIRST METHODS FOR DERIVING PERMEABILITY FROM CONVENTIONAL
WIRELINE LOGS
The first suggestion was to link conventional wireline log data, or log-derived rock prop-
erties such as porosity, with permeability. This idea is almost as old as wireline logging
itself. Over the years, many transforms were proposed and, under certain conditions, they
have been providing acceptable approximations of hydraulic conductivity, i.e. of single-
phase intrinsic permeability. In the multiphase situation encountered by the oil industry,
dimensionless terms called relative permeabilities were added to adapt Darcys equation
in order to describe the ability of a rock to conduct one fluid in the presence of one or more
other fluids, but no wireline logging solution has yet been found to estimate these relative
permeability values in situ, and they have always been so far measured in laboratories on
core samples.
The first formula relating intrinsic permeability with other measurable rock properties
was proposed by Kozeny (1927) and later modified by Carman (1937). This formula is
commonly written as: k
3
/S
2
, in which S is the grain surface area per bulk volume,
is the porosity and an empirical constant. It describes permeability in packs of spheres
of uniform size but unfortunately breaks down in any real world formation other than uncon-
solidated well-sorted sands with almost spherical grains, since not only grain size but also
sorting, compaction, and cementation affect permeability in sandstones (e.g., see Beard
and Weyl, 1973). However, for log analysts, its greatest drawback was that the grain (or
pore) surface area could only be determined on core samples.
To alleviate this problem, Wyllie and Rose (1957) conjectured that grain surface area
can be, in water-wet formations, approximately related to the irreducible water saturation
Sw
irr
(i.e. the amount of water in the pore space that cannot be displaced by oil), because
they had noted that both grain surface area and Sw
irr
increase when grain size decreases
and when sorting becomes poorer. The advantage was that Sw
irr
could be obtained from
logs, although sometimes with difficulty. A consistent minimum of the bulk volume water
over an oil- or gas-bearing sandstone interval usually provides a good Sw
irr
estimate, but
Sw
irr
cannot be easily determined from resistivity logs when the reservoir is not at irre-
ducible conditions i.e. when the hydrocarbon-bearing zone also produces water.
Timur (1968a), based on laboratory studies of 155 sandstone cores from different US oil
fields, then proposed a slightly different relationship that was adopted by the entire oil
industry: k
1
2.25
/Sw
irr
(Fig. 1).
By the same time, in clay-rich formations, the Archie equation, used for computing water
saturation in hydrocarbon-bearing formations from resistivity logs, started to be replaced
by the so-called shaly sand models, and some log analysts derived Sw
irr
, and thereby k,
from the new expressions proposed for shaly sands by Coates and Dumanoir (1973). None
of these interpretation models, however, was realistically accounting for the effects of clay
type and morphology on permeability, which sometimes was leading to poor k estimates.
Neasham (1977) studied the impact of clay on the porosity-permeability relationship in
sandstones. From a survey of 14 very well sorted sandstones from North Sea reservoirs,
with similar textures but different types of clay morphology in the pore space, he showed
that throat-bridging clay connected across the pore space was causing major reduction in
permeability, while porosity was much less affected. In other words, all the empirical cor-
relations based on Sw
irr
were likely to be working well in clean mature sandstones but mar-
ginally elsewhere.
3 BETTER PERMEABILITY TRANSFORMS BASED ON MORE RECENT
WIRELINE LOGGING TOOLS
In the 1980s, the information brought by the conventional logs was complemented by the
first geochemical logging tools that were using neutron-induced gamma-ray spectroscopy
to measure the presence of elements in a formation, then transformed into mineral concen-
trations. The basis for obtaining permeability from these data was that changes in miner-
alogy are normally accompanied by changes in the size, shape, and morphology of rock
grains; these changes affect the pore system geometry, which directly influences permea-
bility. A linear combination of mineral abundances was substituted for the surface area
term in the Kozeny-Carman relationship by Herron (1987). This approach has been suc-
cessfully used in the US Gulf Coast. More than anything else, it seems that the technique
was deriving a textural maturity term from feldspar content computed from the geochem-
ical tool readings.
In carbonates, the traditional permeability transforms, based on Sw
irr
and , soon appeared
to be of limited use. The reason is that, as shown by Nurmi (1986), porosity in carbonates
is often not intergranular as in sandstones, and quite different pore types may result from
the various diagenetic effects, such as dolomitization, leaching, and fracturation. For a given
carbonate pore type, permeability generally increases with porosity along a fairly consistent
The quest for permeability evaluation in wireline logging 57
Figure 1. Charts based on permeability transforms proposed by Timur (top) and by Wyllie and
Rose (bottom).
trend, but pore connectivity is critical (Fig.2): for instance, non-connected vugs contribute to
porosity but very little to permeability. Conversely, the presence of fractures significantly
increases permeability, but creates little additional porosity if fractures have not been
enlarged by dissolution.
Guided by the intuition that the Archie exponent, m, is correlated with the pore tortuos-
ity that also affects permeability for a given porosity value, Watfa (1987) observed that the
presence of vugs that reduces permeability was typically leading to high values of m (2),
whereas the presence of fractures that increases permeability was leading to low values of
m (close to 1). He thus assumed that permeability could be taken proportional to
m
. This
relationship at least well agrees with the observation: vugs increase m, which lowers
m
and thereby the k estimate; fractures decrease m, which increases the k estimate. The pro-
portionality constant that Watfa said to be related to an equivalent pore radius was fitted
using core permeability data, for a given carbonate formation, which then permitted a
continuous derivation of k, provided that m could be continuously estimated versus
depth. A method was devised that permitted estimating m continuously. It made use of a
logging tool that was developed in the 1980s: the Electromagnetic Propagation Tool (EPT)
records a high-frequency electromagnetic propagation travel time that responds to water-
filled porosity but, contrary to resistivity measurements, does it without an exponent. As
a consequence, combining this log with a resistivity log allows a continuous evaluation
of m, after eliminating porosity. The method has been successfully used in the Middle-
East (Fig. 3).
Figure 2. Porosity, pore type, and permeability in carbonates.
4 CORRELATION OF PERMEABILITY WITH ACOUSTIC LOGGING
MEASUREMENTS
During the late 1970s, Lebreton has advocated for some years that a permeability index may
be derived from a ratio of the absolute peak values of the three first half-cycles of the acoustic
waveform recorded by a sonic logging tool. There was no convincing explanation why this ratio
and permeability should be related. Improved acoustic coupling into fractures may have been
causing the observation reported by Lebreton et al. (1978) since, right at a fracture, there is
far better coupling between the borehole and the formation than elsewhere. In the 1980s and
early 1990s this triggered several attempts to correlate permeability with the Stoneley wave
data recorded by sonic logging tools, such as the DSI (Dipole Shear Sonic Imager) tool.
The DSI tool generates low-frequency tube waves called Stoneley waves that propa-
gate up and down the borehole with a special monopole transmitter operating at frequen-
cies of 600Hz to 5KHz. While these waves preserve most of their energy in the borehole,
some energy is attenuated in front of permeable formations as the wave pressure pushes
fluid from the borehole into the formation, similar to a quick small-scale pressure test. In
so doing, this technique gains direct entry to permeability by physically moving fluid through
the formation. The velocity of the wave is slowed down as a function of the ratio of forma-
tion permeability to fluid velocity (Winkler et al., 1989). In the absence of mudcake, and
knowing the acoustic velocity of the borehole fluid, the permeability can be estimated.
A preferred method, based on energy and not velocity attenuation, establishes permea-
bility from Stoneley waves without needing any further information (Cassel et al., 1994).
Figure 3. Comparison of variable m permeability and core permeability, in a Middle-East
carbonate.
Furthermore, rather than measuring the Stoneley energy transmission between sonic tool
transmitter and receiver, it measures the attenuation seen between two adjacent receivers,
thus narrowing the field of investigation to the distance between these near and far receivers,
i.e. about 15cm, which provides higher resolution. Excellent agreement has been observed
in Middle-East carbonates between core measurements and such permeability indicators
(Fig.4). However, it may be difficult to get quantitative permeability estimates in the pres-
ence of mudcake that interferes with the waves ability to move fluid into the formation.
It remains that Stoneley wave interpretation has been instrumental in fractured forma-
tions. The way fractures affect Stoneley waves is different than for compressional and shear
waves: acoustic energy is not lost through mode conversions but as a result of moving the
fluid into the fracture, and Stoneley attenuation is, therefore, quite independent of the frac-
ture dipping angle and mostly a function of fracture permeability. Stoneley waves have
thus proven to be an excellent fracture indicator (e.g., see Hornby et al., 1987) in tight for-
mations where finding open fractures is equivalent to finding permeable zones.
Whereas all approaches based on Sw
irr
were geared to oilfield conditions and hydrogeo-
logists could not utilize them, the techniques based on geochemical, electromagnetic propa-
gation, and sonic logging could very well be used in water wells. The same holds true for
the techniques described below.
5 NUCLEAR MAGNETIC RESONANCE LOGGING: A NEW WAY TO
ESTIMATE PERMEABILITY
Magnetic resonance imaging instruments are commonly used as diagnostic tools in medi-
cine, but nuclear magnetic resonance (NMR) is also extensively used by the oil industry in
wireline logging, as part of its quest for permeability. The physics and interpretation of NMR
logs is reviewed below, starting from the earlier NMR tools, so as to provide some infor-
mation on a technique that is not widely understood. NMR logging gives unprecedented
Figure 4. Comparison of core measurements and permeability indices from Stoneley waves, in a
Middle-East carbonate.
information about both porosity and pore size distribution, which successfully provides
continuous permeability logs, notably in siliciclastic formations.
5.1 A brief history of early NMR logging techniques
The physical principle called nuclear magnetic resonance refers to the response of atom
nuclei to externally applied magnetic fields. Many atom nuclei have a magnetic moment,
i.e. behave like tiny spinning magnets. These spinning nuclei can interact with a magnetic
field, producing detectable signals. For most elements, nevertheless, the measured signals
are weak, but hydrogen, which is abundant in both water and hydrocarbons contained in
the pore space of rocks, has a relatively large magnetic moment. As far back as 1946, NMR
signals from hydrogen atom nuclei (i.e. protons) were observed by Purcell and Bloch. Oil
industry interest followed right away, with several patents for NMR logging tools filed in
the 1950s. The first NMR logging tool was developed by Brown and Gamson (1960) of
Chevron Research and the first log was run in 1960. Schlumberger ran two versions of this
tool, under license from Chevron, and later developed a tool commercialized at the end of
the 1970s.
The principle of the early NMR tools was the following: the protons spinning in the for-
mation are initially aligned to the Earths magnetic field; the logging tool has a horizontally-
mounted coil that transmits an oscillating magnetic field perpendicular, or transverse, to the
direction of the Earths magnetic field which tips the protons 90, and then turns it off; the
tipped protons immediately start to wobble or precess about the Earths magnetic field
( just as a childs spinning top precesses in the Earths gravitational field, its spinning axis
describing a cone) at a frequency called the Larmor frequency, and they gradually relax
back towards the Earths magnetic field; the precessing protons create a small magnetic
field, oscillating at the Larmor frequency, which is detected by the same tool coil. At first
all the protons precess in unison but, as the protons precess about the static field, they
gradually lose synchronization, mainly due to irreversible molecular interactions. This
causes the magnetic field in the transverse plane, and hence the detected signal, to decay.
The quantities measured were NMR signal amplitude and decay. Because the voltage
level in the tool coil was reduced by several orders of magnitude in going from transmit-
ting to receiving mode, there was a delay before the induced signal could be measured, and
amplitude had to be extrapolated back to time zero. However, continuing research into the
interpretation of these measurements has produced some outstanding contributions. The
amplitude of the returned signal was found to indicate free-fluid porosity. Timur (1968b)
developed the concept of free-fluid index (FFI) that he related to Sw
irr
(Sw
irr
1 FFI/)
and he proposed a method to estimate permeability using NMR in 1968. However, the decay
of the NMR signal during each measurement cycle, called the (transverse) relaxation time
or T
2
, generated the most excitement among the petrophysical community. Relaxation time
was found to depend on pore size, larger pores that contain the most readily producible fluids
allowing longer relaxation times. Seevers (1965) developed a first relationship between
relaxation time and permeability of sandstones. A relationship between pore size, fluid and
matrix properties was then presented by Loren and Robinson (1969).
However, with these early NMR logging tools, the volume of investigation could not be
controlled and, to prevent the tool from reading borehole fluid, drilling mud had to be treated
with a magnetite sIurry before logging, in order to reduce the borehole signal below the
measurement threshold. This time consuming treatment was not very popular with drillers
and hindered the acceptance of NMR logging. The 1970s and 1980s saw continuation of
this work on NMR logging by many oil companies or oilfied service companies (e.g., see
Kenyon et al., 1986), in parallel with laboratory NMR techniques developed to characterize
core samples. To make the logging technique more widely acceptable meant a radical design
change to use permanent magnets instead of the Earths magnetic field for aligning protons,
and to profit from advances in pulsed NMR technology commonly used in the laboratory.
5.2 The more recent generations of NMR logging tools
The use of powerful permanent magnets, applied to the formation as the logging tool moves
up the borehole, allows the position of the measurement volume to be controlled by tool
design, thus eliminating the need for borehole mud doping. The use of a pulse sequence helps
compensate for some reversible dephasing effects caused by the inhomogeneity of the static
magnetic field. When this field is not perfectly homogeneous, the protons precess at slightly
different Larmor frequencies, causing a decay that is not a property of the formation. The
protons can be rephased when pulses that tip them 180 are transmitted. Pulses are applied
repeatedly in an evenly spaced train. Each time the protons rephase, they generate a signal,
called a spin echo.
This configuration was proposed by Jackson (1980 & 1984) who filed his patent in
1978 and the first experimental pulsed logging tools were eventually developed in the late
1980s. The MRIL (Magnetic Resonance Imager) tool built in 1990 by NUMAR, now a
subsidiary of Halliburton, was the first commercial pulsed NMR tool. It incorporates a long
permanent magnet to create a static lateral field in the formation. The tool is run central-
ized in the borehole, and the measurement volume consists of a thin concentric cylindrical
shell with a length of 61cm along hole and a depth of investigation varying with the bore-
hole diameter (about 7.5cm for a 10in. or 25cm hole).
A side-looking configuration invented by Schlumberger (Kleinberg et al., 1992) was the
basis for the CMR (Combinable Magnetic Resonance) tool commercialized in 1995. It is
run pressed against the borehole wall by a bowspring. A short directional antenna sandwiched
between a pair of permanent magnets focuses the measurement on a zone located 2.8cm inside
the formation, with a length along hole of only 15cm providing high vertical resolution.
By design, the area between the CMR tool skid and the measurement volume does not
contribute to the NMR signal. This provides immunity to the effects of mudcake and hole
rugosity. The two permanent magnets generate a focused static magnetic field, which is
about 1000 times stronger than the Earths magnetic field, i.e. of about 550 gauss in the meas-
urement region. The measurement sequence starts with a wait time of about 1.3sec to allow
for complete polarization of the hydrogen protons in the formation along the length of the
skid. Then the antenna typically transmits a train of several hundred magnetic pulses into
the formation. The entire pulse sequence, a 90 pulse of 4 gauss switched on for 16 sec
oscillating at the Larmor frequency followed by a long series of 180 pulses, is called a CPMG
sequence after its inventors: Carr, and Purcell (1954) and Melboom and Gill (1958). The
antenna also acts as a receiver and records each spin echo amplitude. The Larmor fre-
quency for hydrogen nuclei in a field of 550 gauss is about 2.3MHz. The echo spacing is
320 sec for the CMR tool. T
2
distribution is derived from the decaying spin echo curve.
In the latest Schlumberger tool, the MRX (Magnetic Resonance eXpert) tool, the num-
ber of echoes and their spacing are programmable, among other novel features, so as to adapt
to conditions where it is needed to measure long T
2
values (e.g., see Freedman, 2006).
5.3 A deeper insight into the interpretation of NMR logs
Molecules in fluids are in constant Brownian motion. Besides the relaxation by molecular
diffusion in magnetic field gradients that the CPMG pulse sequence is compensating for,
there exist two main NMR relaxation mechanisms, i.e. bulk fluid relaxation and grain sur-
face relaxation. Both mechanisms result from molecular interactions and create the irre-
versible dephasing that can be observed by means of the decaying amplitude of spin echoes.
The bulk relaxation is caused by the magnetic interactions between neighbouring precess-
ing protons in the fluid itself, while the grain surface relaxation is caused by the probabil-
ity for a precessing proton moving about the pore space to collide with a grain surface.
Bulk fluid relaxation can often be neglected but can be important when water is in very
large pores, which may be the case in vuggy carbonates, and when, therefore, hydrogen pro-
tons rarely contact a surface. Water in a test tube has a long T
2
relaxation time of 3700 msec
at 40C, a value that may be approached in a rock with very large vugs. Bulk relaxation
also matters when non-movable hydrocarbon is present in the measurement region: the non-
wetting phase does not contact the pore surface, and so it cannot be relaxed by the surface
relaxation mechanism; in addition, increasing fluid viscosity shortens bulk relaxation times.
Grain surface relaxation is, by far, the most important process affecting relaxation times.
Because of complex atomic-level electromagnetic field interactions at the grain surface,
there is a high probability, characterized by a parameter called the surface relaxivity,
2
,
that the proton in the fluid will relax when it encounters a grain surface. For a given grain
type, e.g. in sandstones, the speed of relaxation depends on how frequently protons can
collide with the surface, and this depends on the surface-to-volume ratio (s/v) and thereby
on pore size. For example, relaxation times for a sandstone typically range from 10msec
for small pores to 500msec for large pores. Collisions are less frequent in large pores that
have a small s/v and relaxation times are, therefore, relatively long. Conversely, small
pores have a large s/v and short relaxation times.
For a single pore, nuclear spin magnetization decays exponentially, and the signal ampli-
tude decays with time constant T
2
(
2
(s/v))
1
. Rocks have a distribution of pore sizes,
each with its own value of s/v. The NMR signal is the sum of the signals coming from all the
pores located in the measurement volume. The initial NMR signal amplitude is thus propor-
tional to porosity; its overall decay is the sum of the individual decays, which reflects pore
size distribution. Separating out ranges of T
2
values by a mathematical inversion process pro-
duces the T
2
distribution curve. The area under the curve represents the porosity and the
curve shape the distribution of pore sizes. This inversion process normally requires stacking,
in order to improve the signal-to-noise ratio, which slightly degrades the vertical resolution.
5.4 From NMR-derived porosity and pore size distribution to permeability
Traditionally, the total porosity seen in formations is subdivided into three components:
free-fluid porosity, capillary-bound water, and clay-bound water. NMR free-fluid porosity
is determined by applying a cut-off, of generally 33msec for sandstones, to the T
2
distri-
bution curve. The area underneath the curve above the cut-off gives free-fluid porosity
(Fig.5). As NMR tool technology has improved over the last decade with shorter echo spa-
cing (today, for example, the CMR-200 and CMR-Plus tools can measure T
2
down to the
0.3msec range), the fast decaying clay-bound water signal with T
2
values below 3msec
can also now be measured and distinguished from capillary-bound water.
NMR permeability is derived from empirical relationships that were developed from
brine permeability measurements and NMR measurements concurrently made in the labora-
tory on hundreds of different core samples. The two widely applied permeability trans-
forms are the Timur-Coates and the Schlumberger-Doll Research (SDR) equations. While
the Timur-Coates transform contains the total porosity and the ratio of the free-fluid vol-
ume to the bound-fluid volume, the SDR transform is based on the NMR porosity and the
logarithmic mean of T
2
: k
NMR
C(
NMR
)
4
(T
2, log
)
2
where k
NMR
is the estimated permeabil-
ity,
NMR
is CMR total porosity, T
2,log
is the logarithmic mean of the T
2
distribution, and C
is a constant depending upon the formation, e.g. 4 for sandstones and 0.1 for carbonates.
In Fig. 6, CMR porosity shows a good match with core porosity measurements and, after
fine-tuning of constant C, CMR permeability overlays core permeability points over the
whole interval. Notably, over the zone with little porosity variation and where permeability
varies from 0.07md to 10md, CMR permeability values compares well to that of core meas-
urements. The value of C used for this well was applied to subsequent CMR logs in the
same formation, enabling the oil company to reduce coring costs.
It has also been observed that the sum of all spin-echo amplitudes is proportional to the
product of porosity and average T
2
, and correlates well with permeability. This alternative
yields better results in high noise environments and can be interpreted without stacking,
which leads to a new NMR permeability indicator (Sezginer, 1999) with higher vertical
resolution (typically 20cm).
5.5 Some specific interpretation issues related to NMR logs in carbonates
The interpretation model assuming that, in water-saturated reservoir rocks, the T
2
and
pore-size distributions are directly related well explains why NMR T
2
curves are success-
fully used to characterize sandstones containing mixed pore-size distributions. However,
there is some concern within the oil industry that NMR does not work as well in carbon-
ate reservoirs. First, NMR responses in carbonates differ from those in sandstones: all pore
surfaces are not equally effective in relaxing hydrogen nuclei and carbonates are about
Figure 5. T
2
distributions for two sandstones with same porosity but different permeabilities and
pore sizes (the yellow area corresponds to free-fluid porosity).
three times less efficient in relaxing the nuclear magnetism than sandstones. For carbon-
ates, relaxation times therefore tend to be three times longer and a 100msec cut-off was
proposed for free-fluid porosity. This cut-off value has often to be locally adapted. For instance,
in the Thamama formations of Abu Dhabi, permeable grainstones could be distinguished
from lower permeability packstones and mudstones with a 225msec cut-off. But, while
carbonate formations contain mixed pore-size distributions, e.g., intergranular porosity and
vugs, NMR logging data in these formations nevertheless frequently yield unimodal T
2
distributions, which often results in inconsistent T
2
cut-off values to distinguish bound and
free fluids, and leads to unreliable permeability predictions.
Developments in NMR research (Ramakhrisna, 1999) have now explained why the con-
ventional approach breaks down in grain-supported carbonates which have dual pore systems.
The breakdown is due to diffusion of spinning protons between the micro- and macrop-
ores. If the surface relaxivity is small enough, protons originally in the micropores can dif-
fuse into the macropores before their nuclear spins relax; the decay of these spins then
proceeds much more slowly. Conversely, spinning protons originally in the macropores
can penetrate into the micropores where they encounter more surface interactions, speed-
ing up their decay. Diffusion, therefore, causes the area under the short T
2
peak, the porosity
fraction associated with micropores, to decrease; at the same time, the position of the higher
T
2
peak shifts towards shorter times. Acting together, these two effects tend to merge the
two peaks and produce a unimodal T
2
distribution that bears little resemblance to the bimodal
distribution one would expect from a dual-porosity system.
In chalk formations with a single pore system, NMR logging performs very well, as demon-
strated by an example from the Ekofisk formation in the North Sea (Fig.7). While it is
widely believed that chalk formations are homogeneous, borehole electrical images have
revealed thin laminations. In the image, light yellow indicates electrically resistive low-
porosity chalk and dark brown more conductive higher porosity chalk. While the standard
CMR permeability transform shows little evidence of these laminations, the high-resolution
permeability indicator log shows permeability variations that are consistent with the lam-
inations seen in the images.
Figure 6. Comparison of CMR porosity and CMR permeability with core measurements.
6 BOREHOLE IMAGE ANALYSIS: A WAY TO ACCESS PERMEABILITY
THROUGH ROCK FACIES TYPING
In carbonates with complex pore structure and sometimes difficult NMR interpretation,
a saving grace for permeability logging (Akbar et al., 1995 & 2000) has been the develop-
ment, in the late 1980s, of high-resolution borehole imaging tools, such as the FMI (Fullbore
formation Micro Imager) tool which provides a picture of most of the borehole wall with
192 small current-emitting electrodes mounted on four pads and four flaps pressed against
the formation. As the tool is pulled up the hole, a measurement is made every 2.5mm and
the small electrodes also have an effective horizontal spacing of 2.5mm. Borehole orien-
tation, tool azimuthal orientation, and borehole diameter are all recorded, allowing the
3-dimensional positioning of every measurement.
From small-scale conductivity variations in the electrical images (Fig.8) the presence of
macro or vuggy porosity in carbonates can be identified and the facies recognized. The
permeability in carbonates is predominantly a function of the facies (or rock type).
While pores in clastic rocks are located between grains and uniformly distributed through-
out the rock, in carbonates the diagenesis can significantly modify pore space and permea-
bility because those rocks are highly susceptible to dissolution: grains can be dissolved to
form new pore space, shells can be dissolved creating moldic porosity, dissolution along
fractures or cracks can create large vugs or even caves; depositional bedding is rarely pre-
served; also, whereas clastic diagenesis normally does not involve a change in mineralogy,
in carbonates a diagenetic process, i.e. the replacement of calcium carbonate by magne-
sium carbonate, called dolomitization, can significantly improve the permeability.
Figure 7. Comparison of CMRPlus high-resolution permeability with FMI borehole electrical
images.
A trend in FMI interpretation in the early 1990s has been towards automated quantita-
tive image analysis (Delhomme, 1992), and an innovative method for characterizing rock
type and permeability was later developed. First the textural variations in the borehole
electrical images are captured: the types, sizes, and densities of both conductive and resist-
ive features are determined, conductive paths between large conductive features (usually
cracks or fractures connecting large vugs) are identified.
Information about the internal organization of the rock is summarized as textural logs that
are then combined with conventional logs such as gamma ray, neutron, and density providing
information about porosity and lithology. This is achieved by means of artificial neural network
(ANN) software that produces a continuous identification of the rock types (carbonate facies).
Once the rock type is identified, a porosity-permeability transform could be specified,
at each depth, to estimate permeability, as suggested in Fig.2. However, it has been found
simpler, and more efficient, to use the ANN software for producing a continuous quantita-
tive permeability estimate directly.
The ANNs for both rock type and permeability determination are trained on cored inter-
vals, from the same well or from a nearby well.
This approach has proved to be so powerful that it has been successfully retrofitted to
old wells where only high-resolution dipmeter (e.g. SHDT) data, and not images, had been
acquired. Fig. 9 displays results obtained in that way from an Abu Dhabi well. Photographs
in the composite plot show blown-up pictures of 3 distinct rock types. Note the more pre-
cise log-derived rock type zonation, and the good agreement of log-derived permeability
estimates with core permeability data.
7 WHAT ABOUT PERMEABILITY ANISOTROPY?
In the past years, reservoir engineers have increasingly paid attention to permeability
anisotropy. With more and more highly deviated and horizontal wells in the oil and gas
Figure 8. Mottled fabric of a Middle-East carbonate rock shown by a FMI image (dark pores,
light grains and matrix).
fields, vertical permeability may be the most important reservoir parameter because it affects
production the larger the vertical anisotropy, the higher the productivity index injection
performance, or gas and water coning. Vertical permeability is routinely determined from
cores, but the problem with anisotropy is that it varies with scale: permeability barriers
anticipated from core plug data may have, or lack, lateral extension and influence, or not, the
flow patterns at a larger scale. Vertical interference testing with the Modular formation
Dynamics Tester (MDT) tool (Pop, 1993) is more a wireline-conveyed technique than
a true logging one, but it provides this type of information.
Figure 9. Rock type zonation and continuous permeability indicator derives from high-resolution
dipmeter data combined with conventional porosity and lithology logs.
Horizontal anisotropy is also a major concern in oil and gas fields. A horizontal well
drilled normal to the direction of larger horizontal permeability will be a much better pro-
ducer, or injector, than one drilled parallel to it. Wireline logging measurements in a pilot
vertical well provides valuable information for horizontal well design. Shear sonic logging
may, for instance, be used to identify the maximum and minimum stress directions that usu-
ally coincide with the maximum and minimum horizontal permeability directions: natural
(micro) fractures aligned with the maximum stress direction open up in the direction normal
to it, but stress anisotropy may also cause minor permeability anisotropies in the absence
of fractures, by distorting the pore space.
Hydrogeologists may soon be facing the same situation than reservoir engineers if horizon-
tal wells start to be drilled for aquifer storage and recovery or to mitigate saltwater intrusion
in coastal aquifers.
ACKNOWLEDGEMENT
The author is indebted to the numerous Schlumberger geologists, petrophysicists and reser-
voir engineers who co-authored the articles published by Schlumberger on permeability
since 1980. The present paper borrows ideas and even sentences, sometimes cited verbatim,
from the 9 papers that are listed at the beginning of the bibliographical section. Nevertheless,
rather than corporate views, this paper mainly reflects the authors opinion.
REFERENCES
Articles published in the Oilfield Review and other Schlumberger reviews since 1980:
Akbar M et al. (1995) Classic interpretation problems: Evaluating carbonates. Oilfield Review 7, 1,
3857
Akbar M et al. (2000) A snapshot of carbonate reservoir evaluation. Oilfield Review 12, 4, 2041
Allen D et al. (1988) Probing for permeability: an introduction to measurements. The Technical
Review 36, 1, 620
Allen D et al. (2000) Trends in NMR logging. Oilfield Review 12, 3, 219
Ayan C et al. (1994) Measuring permeability anisotropy: The latest approach. Oilfield Review 6,
4, 2435
Kenyon W et al. (1994) Nuclear magnetic resonance imaging: Technology for the 21st century.
Oilfield Review 7, 3, 1933
Mathieu G et al. (1985) Applying sonic data: Fracture detection. The Technical Review 33, 1, 6979
Nurmi R (1984) Permeability in sandstones. The Technical Review 32, 1, 410
Nurmi R (1986) Carbonate close up. Middle East Well Evaluation Review 1, 2235
Other articles on permeability and logging:
Beard DC, Weyl PK (1973) Influence of texture on porosity and permeability of unconsolidated
sands. AAPG Bulletin 57, 349369
Brown RJS, Gamson BW (1960) Nuclear magnetism logging. Journal of Petroleum Technology 12,
199207
Carman PC (1937) Fluid flow through granular beds. Transactions of the Institution of Chemical
Engineers 15, 150166
Carr HY, Purcell EM (1954) Effects of diffusion on free precession in nuclear magnetic resonance
experiments. Physical Review 94, 3, 630638
Cassel B et al. (1994) Permeability prediction based on anelastic attenuation using dipole shear and
low frequency monopole sources in a carbonate reservoir in Saudi Arabia. Proceedings of the
GEO94 Middle-East geosciences conference, Bahrain.
Coates GR, Dumanoir JL (1973) A new approach to improved log-derived permeability. Transactions
of the 14th annual logging symposium, SPWLA, paper R
Delhomme JP (1992) A quantitative characterization of formation heterogeneities based on borehole
image analysis. Transactions of the 33rd annual logging symposium, SPWLA, paper T
Freedman R (2006) Advances in NMR logging. Society of Petroleum Engineers, paper 89177
Herron MM (1987) Estimating the intrinsic permeability of clastic sediments from geochemical
data. Transactions of the 28th annual logging symposium, SPWLA, paper HH
Hornby BE et al. (1987) Fracture evaluation from the borehole Stoneley wave. Expanded abstracts
of the 57th annual international meeting, SEG, pp 688691
Jackson JA (1984) Nuclear magnetic resonance well logging. The Log Analyst 25:1630
Jackson JA, Cooper RK (1980) Magnetic resonance apparatus. US patent #4, 350, 995
Kenyon WE et al. (1986) Compact and consistent representation of rock NMR data for permeability
estimation. Society of Petroleum Engineers, paper 15643
Kleinberg RL et al. (1992) Novel NMR apparatus for investigating an external sample. Journal of
Magnetic Resonance 97, 3, 466485
Lebreton F et al. (1978) Logging tests in porous medium to evaluate the influence of their permea-
bility on acoustic waveforms. Transactions of the 19th Annual Logging Symposium, SPWLA,
paper Q
Loren JD, Robinson JD (1969) Relations between pore size, fluid and matrix properties, and NML
measurements. Society of Petroleum Engineers, paper 2529
Meiboom S, Gill D (1958) Modified spin-echo method for measuring nuclear relaxation times. The
Review of Physical Instruments 29, 8, 688691
Neasham JW (1977) The morphology of dispersed clay in sandstone reservoirs and its effects on sand-
stone shaliness, pore space, and fluid flow properties. Society of Petroleum Engineers, paper 6858
Pop JJ et al. (1993) Vertical interference testing with a wireline-conveyed straddle-packer tool.
Society of Petroleum Engineers, paper 26481
Ramakhrisna TS et al. (1999) Forward models for nuclear magnetic resonance in carbonate rocks.
The Log Analyst 40, 4, 260270
Seevers DO (1966) A Nuclear magnetic method for determining the permeability of sandstones.
Transactions of the 7th Annual Logging Symposium, SPWLA, paper X
Sezginer A et al. (1999) An NMR high-resolution permeability indicator. Transactions of the 40th
Annual Logging Symposium, SPWLA, paper NNN
Timur A (1968a) An investigation of permeability, porosity, and residual water saturation relation-
ships. Transactions of the 9th Annual Logging Symposium, SPWLA, paper J
Timur A (1968b) Effective porosity and permeability of sandstones investigated through nuclear mag-
netic resonance principles. Transactions of the 9th Annual Logging Symposium, SPWLA, paper K
Watfa M, Youssef FZ (1987) An improved technique for estimating permeability in carbonates.
Society of Petroleum Engineers, paper 15732
Wyllie MRJ, Rose WD (1950) Some theoretical considerations related to the quantitative evaluation
of the physical characteristics of reservoir rocks from electrical log data. Petroleum Transactions
of AIME 189, 105118
Winkler KW et al. (1989) Permeability and borehole Stoneley waves: Comparison between experiment
and theory. Geophysics 54, 6675
CHAPTER 6
Aquifers imagery and hydrodynamic parameters estimation
using proton Magnetic Resonance Soundings
Jean-Franois Girard
1
, Jean-Michel Baltassat
1
, Marie Boucher
1,3,4
,
Anatoly Legchenko
2
, Jean-Michel Vouillamoz
2
, Alexis Gutierrez
5
,
Marie-Luce Noyer
5
& Patrick Lachassagne
6
1
BRGM, ARN, Orlans, France
2
IRD-LTHE, LTHE, Grenoble Cedex, France
3
IRIS-Instruments, Orlans, France
4
ISTO, Orlans Cedex, France
5
BRGM, EAU, Orlans, France
6
BRGM, EAU/RMD, Montpellier, France
ABSTRACT: In this paper, we discuss about hydrogeological parameters that can be measured
with the magnetic resonance sounding method (MRS). This non-invasive surface geophysical
method is directly sensitive to groundwater and meets a growing interest in the scientific and engin-
eering communities of the earth sciences. It has been used for years now, both for water resources
prospecting all around the world and to obtain additional information to the classical method for
improving aquifer management. The main results to be obtained from the method and its limits are
presented and illustrated from numerical models and case histories.
1 INTRODUCTION
The proton Magnetic Resonance Soundings (MRS) method is a non-invasive geophysical tech-
nique specially adapted to hydrogeological investigations. Indeed, this method has the decisive
advantage of measuring a signal with the amplitude directly related to groundwater content
and the duration (characterized by relaxation time) linked to the size of water filled pores.
In this paper, basic principles of the method are briefly presented. Practical use for
hydrological parameter estimation (specific yield, piezometric level, transmissivity) is
illustrated with examples. This technique has been in use for years, both for water
resources survey all around the world and to obtain additional information to the classical
method for improving the management of an aquifer. A non-exhaustive review of several
field case studies illustrate various uses of MRS, such as prospecting water resource, map-
ping the limits of a basin, detecting change of a hydrogeological context, estimating
groundwater reserves or detecting and locating karstic conduits.
Resolution of the method is a limiting factor for characterizing hydrogeological vari-
ations. Because hydraulic conductivity estimation relies on both MRS water content and
relaxation times, several factors that influence the accuracy of magnetic resonance meas-
urements should be considered. MRS parameters variation with geology was studied in the
field and at the laboratory using Nuclear Magnetic Resonance (NMR) measurements.
Results in sandy-clayey formation of Sologne and in chalk in the Hallue catchment (both
in France) are illustrated.
2 BASIC PRINCIPLES
Magnetic Resonance Sounding (MRS) is based on the measurement of the nuclear mag-
netic resonance signal which is produced by the hydrogen proton of water molecules after
they have been stimulated by an alternative electromagnetic field at a specific frequency
(the Larmor frequency). It uses the same basic principles as Magnetic Resonance Imaging
(MRI) and Nuclear Magnetic Resonance logging (NMR) widely used for medical applica-
tions and in petroleum well-logging respectively. It differs according to the investigated
volume (thousands of cubic meters instead of cubic centimetres or millimetres) and the
use of the natural geomagnetic field as a static field (instead of artificially created fields).
The MRS signal (called free induction decay or FID) oscillates at the Larmor frequency
after an excitation pulse has been generated. The MRS response decreases exponentially
with the decay time T
2
* (Figure 1). Its initial amplitude, e
0
and the relaxation time T
2
* are
related to the water content W (Legchenko et al., 2004) and the mean pore size in the sat-
urated aquifer respectively: the smaller the pore size, the shorter is T
2
*. But T
2
* is also
influenced by the local inhomogeneities of the magnetic field and a more reliable param-
eter, T
1
can be obtained using a sequence composed of two electromagnetic stimulation
pulses (Figure 1). Relaxation time T
1
is linked to the mean pore size of saturated aquifer
as follows: 1/T
1
1
S/ V (Kenyon, 1997), where V is the volume of water saturated pores,
S the pore surface and the surface relaxivity depending on rock mineralogy. In
the non-saturated zone, the MRS response could be intuitively predicted considering
that water volume, V and total water content W increase with saturation while pore sur-
face, S remains constant. Thus, initial amplitude e
0
and decay times should increase with
saturation.
e(t) = e
0
. exp(t/ T
2
*
) . sin(w
0
t +
0
)
Pulse P1
Ambiant
noise
Signal FID2
Pulse P2
Amplitude e
0FID1
e
0FID2 Amplitude
D
ecay tim
e D
ecay tim
e
T
2
*
T
2
*
Signal FID1
Delay t
e
0FID2
e
0FID1
= 1 exp(t/T
1
)
t

Figure 1. MRS signal recording sequence.
3 POROSITY ESTIMATION
The instrumental dead time of 40ms between the end of excitation pulse and the beginning
of recording does not allow measurement of the early time part of the signal (Figure 2a). For
example, in the case of a 20ms decay time signal the recorded initial amplitude is only 13%
of the actual magnetic resonance initial amplitude, whereas it is more than 50% for a 60ms
decay time signal. Thus, the part of the water characterized by a short relaxation time T
2
*
(i.e. bound and part of capillary water) is not detected (Figure 2b). MRS water content
would then be a good estimation of the effective porosity if unconnected and dead-end
water can be neglected (Lubczynski et al., 2003; Lachassagne et al., 2005). It is also a good
estimation of the specific yield if capillary water can be neglected. However, in fine grain
sediments, because of capillary water, MRS water content overestimates the specific yield.
Moreover, the relaxation time T
2
* depends also on the magnetic properties of the rock
the more magnetic the rock, the shorter the relaxation time T
2
* and the larger the part of the
water that is invisible to MRS. Consequently, the MRS water content can be smaller than
the specific yield. Finally, depending on the geological context the specific yield can be
over or under estimated. Using measurements from other hydrogeological methods, a cal-
ibration can be established and applied to a whole MRS survey in the same context.
In the laboratory NMR measures the water content in samples using the same physical
principle as MRS. Because the instrumental condition in the laboratory are better controlled,
the recorded signal is not affected by a dead time as for MRS. It allows the measurement of
all the water in the samples. Figure 3 shows an example of comparison between the water
content estimated with lab NMR and by weighting and there is a good correlation.
4 AQUIFER DETECTION
Because the MRS signal is generated by water hydrogen protons, it is specifically linked
to the presence of groundwater. It is the only geophysical method which allows an unam-
biguous detection of underground water. Water content estimation from electrical resistiv-
ity through formulas like Archies Law needs to know the electrical conductivity of water
and empirical parameters to characterize the rock. When MRS is used in a combined
Aquifers imagery and hydrodynamic parameters estimation 73
S
W
W
visible for MRS
Noise Pulse Signal
Short T2*
Long T2*
(a)
DEAD
TIME
DEAD
TIME
Solid grain
Water with short T2*
invisible for MRS
Water with long T2*
DEAD
TIME
DEAD
TIME
DEAD
TIME
DEAD
TIME
(b)
Figure 2. Amount of water detectable with MRS (MRS water content) generates a signal longer
than instrumental dead time (a) and corresponds to water molecules far from pore surface (b).
methodology with the electric-electromagnetic mapping method (Yaramanci et al., 2002,
Vouillamoz et al., 2007) for locating and delineating subsurface structure, MRS provides
significant advantages as it helps to distinguish electrically conductive anomalies due to
high clay content from those due to conductive water.
5 AQUIFER GEOMETRY
MRS data are inverted using a 1D (layered earth) model to provide a vertical log of water
content. As a result, the geometry of the saturated aquifer can usually be defined accur-
ately. In Figure 4, borehole information and MRS results in a sandy aquifer (near
Chuisnes, Beauce, France) are compared. Saturated zones are distinguished from unsat-
urated zones where the total water content increases in the saturated condition. In this case
the water table in an unconfined aquifer can be estimated accurately from the MRS data.
For example, the top and bottom of the saturated alterite reservoirs estimated by MRS
were found to be comparable to borehole information in Burkina Faso (Vouillamoz, 2005)
within a relative average difference of 20%.
In the case of a wide capillary fringe, the difference between MRS and piezometric level
may be significant but it can be corrected after calibration at monitoring boreholes.
Because MRS is not sensitive to pressure changes inside the saturated zone, then only the
top of the confined aquifer is detected while the hydraulic head cannot be estimated.
As with other surface geophysical methods, MRS is non-invasive and it is possible to
investigate large areas at a reasonable cost. Though measurements are time consuming, the
number of measurements is not constrained. MRS was used to map the water table of an
unconfined sand aquifer over the Montreuil-sur-Epte experimental catchment (Paris
basin). Because the borehole network is often coarse and irregular, interpolation of MRS
results over a more regular and dense grid of soundings is relevant because it can detect
unexpected lateral variations due to faulting, lenses and lateral change of geology, which
10% 30 % 20 % 40 %
Water content measured by weighting
10%
20%
30%
40%
L
a
b

N
M
R

w
a
t
e
r

c
o
n
t
e
n
t
Figure 3. Comparison of water content measured with laboratory NMR (at 2MHz) and by weight-
ing on sandy-clayey samples from Marcilly site.
can otherwise be missed because of the limited borehole distribution. Moreover, MRS
information can be reliably extended outside the validation zone of the boreholes. MRS
water table determination calibrated with piezometers makes it possible to map the piezo-
metric water level over the whole of the Montreuil catchment (Figure 5) and helped to
define its boundaries.
0 10 20 30
Water content (%)
0
10
20
30
40
50
60
70
80
90
100
D
e
p
t
h

(
m
)
1
E
-
6
1
E
-
5
1
E
-
4
1
E
-
3
Permeability (m/sec)
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
clay
chalk
clay
Water static level
1
E
-
6
1
E
-
5
1
E
-
4
1
E
-
3
0 10 20 30
Aquifer
Bound Free
water water
0 100 200 300 400
Relaxation time T1 (ms)
0
10
20
30
40
50
60
70
80
90
100
0 100 200 300 400
sand
Figure 4. Borehole information and MRS results near Chuisnes (Beauce, France) site (transmissiv-
ity is 5e-3m
2
/s, with a similar estimation from pumping test and MRS).
Figure 5. Water table elevation mapping using MRS over the Cuisien aquifer in the Montreuil-
sur-Epte (Paris Basin) catchment.
6 NATURE AND HETEROGENEITY OF AN AQUIFER
The type of water bearing material is characterized not only by the water content, but also
the decay time of the MRS signal (Table 1): the smaller the pores the shorter the decay time.
In a known geological context where there are noticeable variations of the MRS parameters
then the MRS response is an indicator of the nature of the medium. It makes it possible to
separate the response of sand aquifer from chalk, for example, as it was observed and con-
firmed over the Montreuil-sur-Epte catchment (Figure 6). The thickness variation of the
sandy aquifer is, moreover, in good agreement with the borehole information.
7 TRANSMISSIVITY ESTIMATION
Because MRS data are linked to water filled pore size, hydraulic conductivity can be esti-
mated using empirical models (Legchenko et al., 2002, 2004) which combine the MRS
Table 1. MRS parameters for various geological media (
1
from Legchenko et al., 2004,
2
from Vouillamoz et al. 2005,
3
from Boucher et al. 2006).
Saturated aquifer material T
2
* Decay time (ms) T
1
Decay time (ms)
Chalk (France) 6080 6080
Clay and fine sand
1
70 310
Alterite (Burkina)
2
150 400
Medium sand (France)
1
120 420
Fractured limestone (Cyprus)
1
130 430
Highly fractured limestone (France)
1
280 800
Gravel and coarse sand (France)
1
330 600
Karstic limestone (Cyprus, France)
1,3
460 1000
Figure 6. Discrimination between unsaturated sand (area B, C, E), chalk (area D) and saturated
sand (central area) aquifers using MRS water content in Montreuil-sur-Epte site.
parameters. For example, a MRS transmissivity can be defined such as: T
RMP
C.(W.
z).T
1
2
where W is the MRS water content, z the saturated aquifer thickness, and T
1
the
decay time of MRS signal. A calibration constant C is used and varies with the medium
properties. An empirical approach is used to overcome the problems of, on the one hand, the
water which is detected by the MRS but does not contribute to the hydraulic conductivity
(dead-end and unconnected pores) and, on the other hand, the fact that MRS is not sensitive
to pore throat size. Calibration should first be performed by comparing results with those
from pumping tests (Figure 7) to allow a reliable hydraulic characterisation to be achieved
within a given hydrogeological region.
In the Beauce aquifer, such a methodology was developed and nowadays MRS is com-
monly used before drilling to identify the most favourable drilling site among a set of loca-
tions previously selected with conventional hydrogeological criteria. The transmissivity
can also be estimated by using a calibration well established earlier in the region. Over the
last years, several sites were prospected and drilled and a good correlation has been
observed between observed and measured transmisivity with MRS results (Baltassat et al.
2003).
Comparison of MRS results with pumping test characterization for an aquifer is particu-
larly relevant because the investigated volume for both methods is similar. The MRS inves-
tigated volume is roughly a vertical cylinder of the loop size diameter (usually 60100m).
There is no scaling effect between the two methods. MRS is not affected by the problem of
the quality of connection between the borehole and the water bearing structure (drilling
across a clay lens, missing a fracture network, etc.).
8 ESTIMATION OF THE SPECIFIC YIELD IN A SANDY-CLAYEY FORMATION
Laboratory NMR measurements on rock samples and magnetic resonance soundings have
been performed in the sandy-clayey formation of Sologne in the test site named Marcilly.
In an attempt to minimize water content disturbance within the samples, a direct push
1x10
-3
1x10
-2
Pumping tests Transmissivity (m
2
/s)
1x10
-3
1x10
-2
M
R
S

T
r
a
n
s
m
i
s
s
i
v
i
t
y
x 2
x1

x1/2

Figure 7. Correlation between MRS and pumping test transmissivity estimation for different sites of
the Beauce aquifer. Uncertainty on calibrated transmissivities is considered to be within a ratio of 2.
core sampler was used. The samples were analyzed with the NMR spectrometer at 2MHz
in the Department of Applied Geophysics of the Technical University of Berlin (TUB).
Results (Figure 8) show a good correlation between NMR parameters and the geological
log. Similarly, MRS (using a 25m square loop, 3 turns) inversion results over the first 20m
provide a good estimation of the saturated part of the sandy aquifer (Figure 9). However,
0
0.6
1.5
2.6
5.5
6
8
D
e
p
t
h

(
m
)
0
%
1
0
%
2
0
%
3
0
%
4
0
%
Water content
1 10 100 1000
Decay time (ms)
soil
sandy clay
clayey sand
clay
fine sand
sand
Figure 8. Comparison of lab NMR and geology in Marcilly test site.
top soil
sand and gravel
grey clay
ochre clayey sand
fine to coarse sand
compact clay
fine to coarse sand
grey clay
clayey sand
very clayey coarse sand
sandy grey clay
0
1
2.2
4.6
6
9.5
10.1
13.1
15.2
17.5
19.5
21.5
D
e
p
t
h

(
m
)
0% 10% 20% 30%
MRS water content
25
20
15
10
5
0
D
e
p
t
h

(
m
)
0 100200300400500
Decay time T1 (ms)
25
20
15
10
5
0
Figure 9. Comparison of field MRS results with geology from borehole cuttings in the Marcilly test site.
the water content measured with MRS is lower than the water content measured with
NMR. This result is in agreement with the fact that the MRS water content, due to the
instrumental dead time, is an estimation of the free water, whereas the laboratory derived
NMR spectrometer measures the total saturated porosity. However, the difference between
MRS water content (15%) and laboratory NMR water content (30%) is very import-
ant at this site. Beside, the MRS relaxation time (200400ms) is more than ten times
longer than the relaxation time from laboratory NMR measurements (20ms). In previ-
ous works (see next paragraph on chalk aquifer, and Mller et al., 2005), the relaxation
times measured both in the laboratory and in the field were comparable.
The artificial static magnetic field used in the laboratory is many times stronger than the
earth magnetic field used in MRS (47T in the field vs. 47mT in the laboratory).
Perturbation of the magnetic field inside the investigated volume is linked both to the para-
magnetic properties of the medium and to the strength of the magnetic field. This can
explain the decrease of the relaxation time in the laboratory, but it does not affect the water
content estimation (see Figure 3) because the instrument in the laboratory allows the signal
to decay down to 1ms and still be reliably measured. In this case it is assumed that the relax-
ation time that is measured is so influenced by the magnetic susceptibility of the sample that
the relationship with pore size is severely affected in laboratory measurement: T
2
10ms
in the coarse sandy part of the aquifer is normally observed in a clayey medium. In this con-
text, MRS has the advantage of performing measurements in a lower magnetic field (earth
field) than in the laboratory. Moreover, field experiments have shown that magnetic suscep-
tibility is not such a problem in many geological contexts when calibration is performed
against available pumping test data for the area.
9 CHALK UNSATURATED ZONE
Previous field surveys have shown that MRS is able to detect water above the piezometric
level, particularly in chalk aquifers. These preliminary observations lead to the develop-
ment of specific equipment and a methodology that focuses on detecting variations in the
MRS parameters in the vadose zone.
In the investigated chalk aquifer a thick (about 30m) unsaturated zone was observed.
Typically, in the chalk unsaturated zone, the fracture porosity (2%) is empty of water and
nearly all the matrix (40% porosity) is saturated (Price et al., 2000). The chalk has a very
low magnetic susceptibility. These two conditions (thick unsaturated zone with high water
content and low magnetic susceptibility) enable MRS to detect part of the capillary water
above the water table.
Field and laboratory experiments in the Hallue catchment show (Figure 10) that MRS
water content increases with depth whereas the water content measured on samples was
found to be not very variable and close to full saturation of the chalk (40%). Thus, less and
less water is detectable with MRS from the water table up towards the surface. Because of
the low magnetic susceptibility of the chalk, the relaxation time T
2
* was found to be simi-
lar in the laboratory and in the field. In Figure 10, T
2
* increases with depth as well as MRS
water content up to the water table and tends towards the water content measured in the
laboratory, thus confirming that the quantity of water undetectable with MRS is linked to
the relaxation time T
2
*. The experimental results (Figure 10) and numerical modeling
results (Figure 11) show good agreement.
10 LIMITS OF APPLICABILITY
The MRS investigation depth is linked with the loop size (maximum depth is generally
about 100m with the largest usual square loop of 100m side). The depth of investigation
50
40
30
20
10
0
D
e
p
t
h

(
m
)
0 20 40 60 80 100
Decay time T2 (ms)
0 10 20 30 40
Water content (%)
50
40
30
20
10
0
D
e
p
t
h

(
m
)
Field MRS
Lab NMR
Water
table
Part of water
not observed
with MRS
Figure 10. Comparison of laboratory NMR (diamonds) and field MRS (lines) results in the chalk
Hallue catchment.
0 10 20 30 40 50
Water content (%)
60
40
20
0
D
e
p
t
h

(
m
)
10 20 30 40 50
Decay time T2* (ms)
60
40
20
0
0 10 20 30 40 50
MRS water content (%)
MODELLING
d
e
c
a
y

t
i
m
e
w
a
t
e
r

c
o
n
t
e
n
t
Part of water
not observed
with MRS
M
R
S

w
a
t
e
r

c
o
n
t
e
n
t
Figure 11. Modeling of a typical water content profile in a chalk environment. Left: model param-
eters (black: water content, gray: decay time); right: MRS water content log obtained from corres-
ponding synthetic data.
decreases in highly electrically conductive medium. If the loop size and electrical power
are instrumental limits, electrical conductivity imposes a physical limit, which can be
roughly estimated with the electromagnetic skin depth (skin depth is around 110m for a
100 m material, but only 35 m for 10 m and 11m for 1 m).
But the main limit that is often encountered is caused by the electromagnetic noise close
to urban area and power-lines. It is generally possible to work at no less than 250m from
power lines (and 500m in case of high tension lines). The vulnerability is site specific
because the MRS signal amplitude may vary by a ratio of 100 depending on the water con-
tent. Different filtering strategies have been developed to decrease the noise: stacking, dig-
ital filtering of industrial harmonic frequency and analogic filtering with a remote
reference loop. The efficiency of the filtering varies from site to site because of the nature
of the noise sources. Generally, it is a matter of time to obtain a sufficient signal to noise
ratio, resulting in sounding duration varying from 2h up to 20 hours.
As with many geophysical methods applied at the surface, and the integrative property
of MRS, it is not always possible to characterize each layer individually in the case of a
multi-layered aquifer. Responses from aquifers located at the same depth with an equiva-
lent volume of water (product of thickness by water content) are similar. A numerical
study was performed to underline the accuracy of MRS results (Legchenko et al., 2004).
The error on the parameter P was estimated as: 100.(P
inv
-P
mod
)/P
mod
. The top of the
layer and the water volume (product W*z) were reliably estimated (Figure 12). This
explains why the aquifer transmissivity was faithfully replicated whereas the vertical
distribution of hydraulic conductivity was poorly defined (low accuracy on a single esti-
mation of aquifer thickness or water content) (Figure 12). The layer geometry input from
geology or other geophysics has to be applied to reduce the number of free parameters in
the inversion process. It increases the accuracy of the inversion results and leads to more
reliable estimation of the water content and hydraulic conductivity distribution with depth.
It is noticeable that magnetic rocks influence the magnetic resonance phenomenon and
may shorten the decay of the signal below the instrumental threshold. It is both an instru-
mental and a methodological limitation. Future devices may be available for use even in
such magnetic environments as volcanic areas.
Figure 12. Numerical modelling: resolution of a 10m thick layer (20% water content) when using
a 100m side square loop (after Legchenko et al., 2004).
11 GROUNDWATER RESERVES MAPPING IN WEATHERED HARD-ROCK
AQUIFERS IN BRITTANY, FRANCE
MRS provides a quantitative estimation of the water content. Coupled with geometrical
aquifer modelling, it has been used to create a map of groundwater reserves over a 270km
2
study area in a weathered basement setting (Figure 13 from Wyns et al. 2004). The depths
to the interfaces determined by MRS are considered and validated by comparison with the
geometrical approach. MRS water content is compared with the hydrogeological model
obtained from boreholes and surface geology in an attempt to define the average water con-
tent of the alterite and the underlying fissured zone aquifers for each formation. The results
of the study show a decrease in water content from the top downwards for the three main
aquifer layers (respectively: unconsolidated alterite, and an upper and a lower fissured
zone). The groundwater reserves (80% in the fissured zone and 20% in unconsolidated
alterite) represent approximately three years of average infiltration (Wyns et al. 2004).
12 KARST GALLERY IMAGING USING 2D INVERSION OF MRS SECTION
In most cases, interpolation of MRS 1D inversion results is sufficient to reveal the hetero-
geneity of an aquifer along a profile. However, when studying targets of limited size with
respect to the measurement loop size, such as karstic conduits, the use of a refined step
between the soundings (down to 5 or 10m step) can provide a better image of the conduit.
A true 2D inversion is then necessary to process the data. Such a methodology was applied
over the underground Ouysse River near Rocamadour (France). This karstic conduit is full
of water during the whole year and its topography was controlled by speleologists. The 2D
0 10
Km
Legend
Fault
River
Geologic boundaries
from the 1/250.000 map
Total
reserve
(m)
0,009-0,250
0,251-0,500
0,501-0,750
0,751-1,000
1,001-1,250
1,251-1,500
1,501-1,750
1,751-2,000
2,001-2,250
2,251-2,500
Avrille
granite
Biotite -
Sillimanite
granite
Mica
Schist
Schist
Schist
Granite
with 2
micas
Granite
with
2 micas
Biotite
granite
Biotite
granite
(a)
(b)
-40
-30
-20
-10
0
D
e
p
t
h
(
m
)
0 2 4 6 8 10 12 14
Water content (%)
MRS28
AvrillGranite
(V_=1.802m)
MRS14
Biotite/sillimanite
granite
(V_=1.167m)
MRS16
Micaschist
(V_=0.425m)
Figure 13. La Roche-sur-Yon region: (a) Thickness map of total reserves in the weathered aquifers.
(b) Example of MRS water content profiles measured in different geological settings.
inversion of the MRS results provided the location and size of the conduit (Figure 14)
(Boucher et al. 2006). Karstic conduit detection is an important issue for water supply in
many places. If today such a survey cannot be widely performed, it is only due to the num-
ber of soundings needed and the time to perform them. Because the MRS signal generated
by a karst conduit is very low, the survey is even more vulnerable to background noise than
MRS used above a normal aquifer.
13 FUTURE WORK
Future development will be impacted by reducing the MRS vulnerability to electromag-
netic noise. It is the main limitation to the application of the method. Because MRS can-
not be used close to urban areas or electric power-lines, the signal to noise ratio will be
enhanced and it will result in better accuracy of measurement which then will improve the
reliability of results.
There is also a need to establish hydrogeophysical models for various geological con-
texts, and to refine the calibration i.e. which part of water remains undetectable to MRS
but contributes to the hydrodynamic parameters? As in the case of karstic conduit, MRS
can provide high resolution results. A methodology for 2D3D tomography will need to be
3000
2000
1000
0
P
u
l
s
e

(
A

m
s
)
Amplitude (nV)
50 0 50100 50 100 50 100 50 100 50 100 50 100 50 100 50 100 100 0 50100
3000
2000
1000
0
P
u
l
s
e
s

(
A

m
s
)
Model
Data
NW SE
40 45 50 55 60 65 70 75 80 85 90
Abscise (m)
25
20
15
10
5
0
D
e
p
t
h

(
m
)
Model
Real section
NW SE

.
.
Figure 14. Karst conduit imaging near Pou Meyssen (France). 10 MRS were performed along a
profile crossing the conduit. Data curves (top) were simultaneously inverted to obtain a 2D model
(bottom).
developed to improve the resolution. This will become a useful tool for highly heteroge-
neous media commonly encountered at shallow depth.
A new measurement scheme will have to be developed to measure MRS in the vadose
zone for all media, not just for media such as chalk with abundant capillary water and low
magnetic susceptibility.
14 CONCLUSION
MRS provides useful information, which may not otherwise be accessible from standard
methods because the borehole network is often coarse and irregular. MRS will not replace
information from boreholes and pumping tests but it can detect water with little ambiguity
and provide the geometry and hydrodynamic characterization of an aquifer at least within the
top 100m below ground level. Calibration from borehole information improves the accuracy
of MRS results. MRS does not just interpolate between boreholes, it also enhances the spa-
tial resolution and detects any lateral variations that may be present. In addition, MRS pro-
vides an estimate of transmissivity that is not biased by the quality of the connection between
the borehole and the surrounding aquifer.
Case studies show that, within its domain of applicability, MRS is nowadays a mature
method that can be applied over a wide range of purposes, including water resource prospect-
ing, reserve estimation, water table mapping, and karstic conduit imagery.
ACKNOWLEDGEMENTS
Results on the Montreuil-sur-Epte catchment were funded by the European FP5 PEGASE
project (Contract EVK1-CT1999-00028 financed by the EU through its 5th PCRDT), the
European Union FP6 Integrated AquaTerra Project (Project no. GOCE 505428) under the
thematic priority Sustainable development, global change and ecosystems, the BRGM
POLDIF research project, the Agreement 012095 with the Seine-Normandy Water
Authority (lAgence de lEau Seine Normandie) and within the framework of the Ph.D.
thesis of X. Morvan, financed by the Centre Regional Council (Conseil Rgional Centre).
Results on the chalk in the Hallue basin, at the Warloy-Baillon monitoring test site in
north France (www.flood1.info) were funded by the INTERREG III A FLOOD1 project,
the financial partners being the EU through ERDF funds, the MEDD (Ministre de
lEcologie et du Dveloppement Durable) through the DIREN Picardie, and two local end-
users: the Conseil Rgional de Picardie and the Conseil Gnral de la Somme.
Results on the Pou Meyssen karst were funded in the framework of the French national
INSU ECCO-PNRH program in the HYKAR and WATERSCAN projects. Researches on
the topic of MRS estimation of hydrodynamic parameters are funded by the French research
network and innovative technology for water and environment RITEAU-PROTON project.
REFERENCES
Baltassat J.-M., Legchenko A., Vouillamoz J-M., Sabatier S., Chigot D., Schmidt J-C., (2003).
Calibration of MRS results using hydrodynamic characteristics from pumping tests, 2nd
International Workshop on the Magnetic Resonance Sounding method applied to non-invasive
groundwater investigations, November 1921, 2003, Orlans, France (ISBN 2-7159-0936-5),
pp. 912.
Boucher M., Girard J.F., Legchenko A., Baltassat J.M., Doerfliger N., Chalikakis K. (2006). Using
2D inversion of magnetic resonance soundings to locate a water-filled karst conduit, Journal of
Hydrology, Vol. 330, issues 34, p. 413421.
Kenyon WE (1997). Petrophysical principles of applications of NMR logging. The Log Analist,
March-April, 2143.
Lachassagne P., Baltassat J.M., Legchenko A., Machard de Grammont H. (2005). The links between
MRS and the hydrogeological parameters., Near Surface Geophysics, 3, 4, 259265.
Legchenko, A., Baltassat, J.-M., Beauce, A., Bernard, J. (2002). Nuclear resonance as a geophysical
tool for hydrogeologists. Journal of Applied Geophysics. 50: pp. 2146.
Legchenko A, Baltassat JM, Bobachev A, Martin C, Robain H, Vouillamoz JM (2004). Magnetic
resonance sounding applied to aquifer characterization. Ground Water 42, 3, 363373.
Lubczynsky M, Roy J (2003). Hydrogeological interpretation and potential of the new magnetic res-
onance sounding (MRS) method. Journal of Hydrology 283, 1940.
Mller M., Kooman S., Yaramanci U. (2005). Nuclear magnetic resonance (NMR) properties of
unconsolidated sediments in field and laboratory, , Near Surface Geophysics, 3, 275285.
Price M., Low RG., McCann C. (2000). Mechanisms of water storage and flow in unsaturated zone
of Chalk aquifer. Journal of Hydrology, 233, 4771.
Vouillamoz J.M., Descloitres M., Toe G., Legchenko A., (2005). Characterization of crystalline base-
ment aquifers with MRS: a case study in Burkina Faso, Near Surface Geophysics, 3, 205213.
Vouillamoz J-M., Chatenoux B., Mathieu F., Baltassat J-M., Legchenko A. (2007). Efficiency of
joint use of MRS and VES to characterize coastal aquifer in Myanmar, Journal of Applied
Geophysics 61, 142154.
Wyns R., Baltassat J-M., Lachassagne P., Legchenko A.V., Vairon J., Mathieu F. (2004). Application
of proton magnetic resonance soudings to groundwater reserve mapping in weathered basement
rocks (Brittany, France)., Bulletin de la Socit Gologique de France, 175, 1, 2134.
Yaramanci U., Lange G., Hertrich M. (2002). Aquifer characterisation using Surface NMR jointly
with other geophysical techniques at the Nauen/Berlin test site, Journal of Applied Geophysics
50, 4765.
CHAPTER 7
Permeability measurements in argillaceous rocks
at the Meuse/Haute-Marne Underground Research
Laboratory, France
Jacques Delay
Andra, Route Dpartementale, Bure, France
ABSTRACT: In November 1999 Andra began building an Underground Research Laboratory
(URL) on the border of the Meuse and Haute-Marne administrative regions in eastern France. The
research activities of the URL are dedicated to reversible, deep geological disposal of high-activity,
long-lived radioactive wastes in an argillaceous host rock. This paper presents an overview of con-
tainment studies, carried out from surface boreholes and underground drifts, and a comprehensive
site characterisation strategy for studying containment properties of the Callovo-Oxfordian argilla-
ceous rocks. Results of hydro-chemical and diffusion studies are used to understand the mechanisms
of flow in a very low porosity formation and help to understand phenomena such as overpressures or
osmotic effects. Use of complementary hydraulic methods for testing hydraulic properties at differ-
ent scales provides a robust estimate of hydraulic parameters. Factors affecting test interpretations
are also presented. When conducting permeability tests, it would be ideal to initiate them in the most
stable pressure conditions possible. However, in very low-permeability clay formations (i.e.
10
13
10
14
m/s), the pressure disturbances induced by drilling and testing take months or even
years to dissipate. The main disturbances affecting transient pressure responses include techno-
logical effects due to the test equipment, drilling-history effects, thermal effects, physico-chemical
and hydromechanical effects due to interactions between the test fluid and the formation, and mechan-
ical effects due to borehole-wall creep.
1 INTRODUCTION
Constituted as an autonomous public industrial authority, by the Law of 30 December
1991, the French National Radioactive Waste Management Agency (Agence nationale
pour la gestion des dchets radioactifs Andra) is responsible for the long-term manage-
ment of radioactive waste in France. Andra is also responsible for providing the French
Parliament with sound scientific arguments as the basis for the debate to take place in
2006, concerning potential options for a HLW-LL waste disposal.
On 3 August 1999, The French government authorised Andra to implement and operate
the first French Underground Research Laboratory (URL) on a site straddling the Meuse and
Haute-Marne administrative regions, near the village of Bure (Meuse, France, Figure 1).
The geological formation selected for this laboratory is a 130m thick argillaceous rock
layer, about 155 million years in age, the Callovo-Oxfordian argillaceous rocks. This
Callovo-Oxfordian clay rich layer is located at a depth of between 400 and 600m (Figure 2)
(Vigneron et al., 2004).
In order to select the URL construction site and verify the existence and the physical
characteristics of the selected host formation, investigations were carried out from 1994 to
1996.
The work performed during this period was relatively conventional, using sedimentary
investigation tools (surface mapping, seismic geophysics, borehole drilling and drill cores
analyses) and relied as much as possible on oil exploration techniques. Nevertheless, from
the outset, specific techniques for geochemical borehole monitoring and hydrogeological
testing in very low permeability formations were implemented.
The objectives of the URL for the years 19992005 were mainly the in situ character-
isation of the physical and chemical properties of the rock. This involved achieving a level
of knowledge that could support the development of disposal designs and safety studies.
The work was carried out mainly from the shafts and experimental drifts, but also from
deep boreholes drilled from the surface in the vicinity of the URL.
Studies and experimental work from deep boreholes and drifts cover three major aspects
(Andra, 2005a):
Containment capability of the host formation

This containment capability depends on the specific physical characteristics of the
rock, the physical-chemical characteristics of the interstitial fluids and their interaction
with the rock. A fundamental physical characteristic is permeability, which can be
investigated with specific tests (Distinguin et al., 2006). The chemical characteristics
Figure 1. Location of the URL Meuse/Haute-Marne.
of the interstitial fluids condition the mobility of the various radionuclides likely to be
found in the natural environment (Pearson et al., 2003). The studies focus on know-
ledge of the geochemistry of the interstitial fluids at equilibrium with the minerals in
the rock, and on the diffusion and retention of the radionuclides.
Geomechanical properties Creation of damaged and disturbed zones associated with

drift excavation assessment of the sealing zone concept.
The main purpose of the studies is to investigate how the rock reacts to the excavation
of shafts and drifts, and the associated development of the damaged and disturbed zone
(Tsang et al., 2004). The various geomechanical measurement campaigns conducted in
the Callovo-Oxfordian formation from deep boreholes provided essential information
on the natural stress conditions: confirmed the amplitude of the minor horizontal com-
ponent (
h
), yielded the anisotropy ratio of the horizontal stresses and showed that the
principle stress (
1
) corresponds to the major horizontal stress (
H
). The sealing of a
drift is a major issue when considering the disposal construction and safety options
(Andra, 2005b). It involves designing systems to re-establish the original low permea-
bility of the formation by overcoming potentially negative effects from the damaged
zone surrounding the drifts and shafts.
Regional knowledge of geological and hydrogeological properties of the host rocks

and the surrounding aquifers.
Survey work focused on studying the deep hydrogeology of the area. The aim of these
studies is to evaluate the vertical and horizontal variability of the geological formation
in order to identify an area where a deep disposal could be implemented.
Permeability measurements in argillaceous rocks at the Meuse/Haute-Marne 89
Figure 2. Simplified stratigraphic column.
2 OVERVIEW OF THE EXPERIMENTS RELATED TO THE CONTAINMENT
PROPERTIES CARRIED OUT FROM DRIFTS
The initial hydrogeological, geochemical and diffusion values used in the safety models
and design files presented by Andra (Andra, 2005c) were obtained through sample meas-
urements or deep borehole measurements carried out during the period 19942004. Drift
experiments made at 445m and 490m depths (Figure 3) provided new sets of values, usu-
ally more reliable, since they were obtained under much more controlled conditions and,
therefore, assumed to be more representative of the real values of the rock parameters
taken into account in the files.
For the hydrogeological characterisation, the permeability programme in drifts relied on
measurements made in boreholes equipped with permanent completions with five test
intervals. These 200 to 400cm
3
test intervals were filled with water of a composition close
to that of the interstitial water in order to limit any chemical disturbance. More than fifty
tests were carried out from the boreholes drilled in the experimental drifts. Due to the spe-
cific testing conditions of the URL, a single test could last up to three months.
For the geochemical characterisation of the interstitial water, gas and water samples are
taken from two dedicated boreholes from the drifts. The aim of this experiment is to deter-
mine the chemical and isotopic composition of interstitial fluids in their natural initial
state. The design of this experiment relied on measurements made on solid core samples
taken for pore water extraction (Gaucher et al., 2004), and analyses and tests required for
modelling water rock equilibria (Parkhurst et al., 1999). The results were used to define
the major ion composition and the water/rock interaction mechanisms, which govern the
composition of the interstitial water. Figure 4 shows the results obtained after 113 days of
synthetic water circulation and the evolution of its composition in major ions.
Diffusion models determined from samples required confirmation of the retention and
diffusion properties. Tracer operations in six short boreholes were carried out from the
Figure 3. General layouts of the underground drifts of the Bure URL.
drifts at 445m and 490m depths. The tests were designed to investigate the behaviour of
(i) inert HTO, (ii) anions:
125
I,
36
Cl,
75
Se, (iii) cations:
22
Na,
134
Cs,
85
Sr. The decrease in
tracer concentration was followed up through sampling and an online gamma spectrom-
eter. The first interpretations of the diffusion tests were based on the decrease in concen-
tration of the injected tracers and showed three expected types of behaviour (Figure 5).
This methodology had been previously tested and validated at Mont Terri Rock Laboratory
(Wersin et al., 2005; Van Loon et al., 2004).
0.0001
0.0010
0.0100
0.1000
Na K Mg Ca Sr Cl SO4 Alkalinity
e
q
/
L
Water best estimated composition
by modelling
Synthetic water
Circulated water, 113 days
Figure 4. Pore water chemical composition obtained after 113 days of water circulation.
Figure 5. Evolution of tracer concentration in a diffusion borehole showing water (HTO), cationic
retention (
134
Cs) and anionic exclusion (
36
Cl).
3 KEY FIGURES
The permeability and head values presented here rely on a large amount of data collected
from various tests in deep boreholes, dedicated boreholes drilled from the drifts and core
sample analyses. Seventeen permeability tests were performed and about 40 cores were
analysed in the deep boreholes. Sixty intervals were tested in more than 20 boreholes in the
drifts, and 20 permeability tests were carried out on cores. The permeability values range
between 1 10
14
m/s and 4 10
12
m/s with most of the values lying between
5 10
14
m/s and 10
12
m/s. Long-term natural hydraulic head measurements were made at
12 intervals in the deep boreholes.
4 SPECIFIC ASPECTS OF PERMEABILITY IN ARGILLACEOUS ROCKS
The permeability of a geological environment, a notion experimentally established by
Darcy in 1856, describes the linearity between water flow and the hydraulic head gradient.
Thus, a macroscopic parameter assumes that the porous environment cannot be deformed,
that the fluid does not interact with the solid skeleton of the environment and that flow
migration is laminar.
However, in predominantly argillaceous rock, owing to the H
2
O molecular structure
(Revil et al., 1998) which allows electrostatic links to form with the crystalline structure
of minerals, the interstitial water is found in different states: (i) adsorbed on the argilla-
ceous minerals, (ii) adsorbed at the surface of argillaceous minerals, but this water would
be little or not at all mobile through hydraulic load gradient (Horseman et al., 1996) and
(iii) free between the minerals and argillaceous aggregates.
Thus, the interaction between water molecules and argillaceous minerals and the size
of the pores prompts us to question the validity of Darcys law in such an environment. If
the size of the pores is less than 10nm, the structure of the water molecules in the double
layer zone along the argillaceous layers eliminates advection transport (Horseman et al.,
1996).
In the case of Callovo-Oxfordian argillaceous rocks, various methods for measuring
porosity on samples (mercury, petrol, helium, nitrogen adsorption) and the images
acquired through autoradiography and SEM (Scanning Electron Microscope) were used to
establish a conceptual model of the porosity of the environment. Below this 10nm thresh-
old, the porosity is about 40% (Figure 6). Therefore, the pore network allowing advective
transport is above the percolation threshold and Darcys law would be valid.
Determining the permeability of argillaceous rocks is not an easy task when no migra-
tion can be observed. Therefore, resistance to the migration of free water must be over-
come. This requires injecting a fluid under a strong pressure gradient in order to be able to
measure the flow in a steady migration state. However, the experimental protocols may
generate artefacts:
Excessive hydraulic head gradients may bring about (i) a change in porosity due to the
decrease in actual stress, (ii) turbulent flow, (iii) a flow of adsorbed water modifying
the volume of water involved (Crois et al., 2004);
Lack of equilibrium between a percolation fluid and the clay minerals in the argilla-
ceous rocks, which may bring about dissolution processes, e.g. precipitation, diffusion,
osmotic processes, or a change in thickness of the double layer.
The most commonly used method, according to different protocols, consists of inducing a
temporary and limited disturbance of the fluid pressure and measuring its return to equi-
librium (pulse test), which avoids the need to apply excessive hydraulic gradients.
Short-term hydraulic tests were made in several deep boreholes prior to carrying out
experiments in the underground laboratory. In the underground laboratory, a systematic
pulse test measurement programme was implemented by monitoring the recovery of static
pressure over several months.
In addition to these in situ measurements, a set of laboratory measurements was per-
formed on samples taken during the drilling of boreholes.
Three types of measurements were carried out including:
A method consisting of interpreting the evolution of saturation profiles in an argilla-

ceous rock based on its water sorption isotherm;
A method for measuring the flow of a Tritium tracer performed in association with dif-
fusion measurements;
Tests in steady and transient states using a rock mechanics laboratory cell.
Other indirect permeability measurements included either measuring the petrophysical
characteristics of rock formations through geophysical methods, or performing a detailed
analysis of pressure logs and stress profiles. For example, measurements carried out with
NMR (Nuclear Magnetic Resonance) probes (Klopff et al., 2004) provided a detailed
description of the porosity, not only in terms of volume but also of geometrical structure.
5 FACTORS AFFECTING HYDRAULIC TESTS
When conducting a short-term packer test, the ideal would be to carry out permeability
tests in the most stable pressure conditions possible. However, in very low-permeability
Figure 6. Pore size distribution in the Callovo-Oxfordian clay.
clay formations (i.e. 10
13
10
14
m/s), the pressure disturbances induced by drilling and
testing take months or even years to dissipate.
The main disturbances affecting transient pressure responses when testing such a for-
mation in a deep borehole include:
Technological effects due to the test equipment. These include irreversible deformation
(particularly that of packers at the contact with the formation) and reversible deform-
ation controlled by equipment elasticity (test tool compressibility varying with pressure);
Drilling-history effects, i.e. impact of hydraulic pressure conditions different from the for-
mation pressure prior to the hydraulic test (e.g. drilling phase, geophysical logging, etc.);
Thermal effects due to circulating or injecting fluid with a significantly different tem-
perature from the formation temperature;
Chemical effects due to interactions between drilling and formation fluids;
Physical-chemical and hydromechanical effects due to interactions between the test

fluid and the formation (clay hydration, swelling);
Mechanical effects due to stress release (borehole-wall creep).

All these effects need to be minimized during the operating phase, and should be taken
into account when interpreting the measurements.
It has been recognized that hydraulic heads estimated in such formations, based on
short-term packer tests, are marred by uncertainties related to these disturbance processes.
However the permeabilities estimated with this technique have been shown to be reliable
for characterizing near borehole conditions, assuming that the causes of disturbance (i.e.
borehole history, temperature changes) are measured and accounted for in the analyses.
From the drifts only long-term tests (one month or more) are carried out allowing a sig-
nificant reduction of the disturbance factors. However, the construction of the drift itself
could create a significant change in local hydraulic and geomechanical conditions.
6 IN SITU PERMEABILITY MEASUREMENTS
6.1 Testing tools and measurement strategy
Three types of equipment, based on different concepts and methodologies, have been
applied to the Bure Site for permeability and head measurements in deep boreholes (Delay
et al., 2004).
The first testing tool is the single or double packer test equipment deployed during or at
the end of the drilling phase. It is a conventional test tool used in the petroleum industry.
This equipment consists of a packer tool assembly. Inflation of the packers is controlled
from the surface (both packers can be inflated together or separately). As stainless steel is
being used for the tubing as well as for the hydraulic lines, the test zone compressibility is
low (1 10
09
1/Pa). The test interval can be connected and isolated from the surface
with a down-hole shut-in tool. This valve can be activated with minimal water displace-
ment in order to avoid pressure perturbations in the test interval. In very low permeability
environments, testing of the interval is performed with pulse tests or slug tests, thereby
modifying the water level in the 2
7
8 (7.3cm) tubing and opening up the shut-in valve.
Finally, pressure and temperature in the test interval, as well as below and above the inter-
val, is continuously monitored.
The second tool is the Electromagnetic Pressure Gauge (EPG), which is a monitoring
tool used in the petroleum industry and adapted to Andras requirements (Soulier and
Lematre, 1993). This equipment is used for long-term monitoring of the formation pres-
sure and temperature. The advantage of this system is its ability to perform measurements
in a down-hole isolated interval completely protected from external disturbances (i.e. with-
out any influence from the upper/lower parts of the borehole). The test interval, about ten
meters long, is totally isolated from upper borehole influences. A permanent packer inflated
with cement is placed above the gauge and then the upper section of the well is filled with
a very-low-permeability cement plug (8.25Darcy). Using an independent down-hole
power supply, measurements can be carried out over several years with data transmitted to
the surface by electromagnetic waves. The pressure sensor has a measurement range of
013.78MPa, 10
4
% full-scale accuracy, and 2 10
4
% full-scale resolution over three
years (full-scale resolution 3kPa) (Cecconi et al., 2004). The life-time of the latest EPG
installed by Andra in 2004 is expected to be more than 10 years. This tool measures the
early time pressure recovery from highly disturbed conditions (i.e. the influence of drilling
as well as testing and subsequent cementation activities) towards the initial formation pres-
sure reached after several months. While late time data provide a direct reading of the for-
mation pressure, early and middle-time data provide transient pressure measurements,
which can be interpreted in terms of permeability. Andra has installed five EPG gauges at
420 to 540m depths, 2 of them sealed one above the other in a single borehole.
The third tool, multipacker equipment (Eldred, 1995) was first developed for the hydro-
geological monitoring of aquifers. This equipment measures the pressure and temperature
in several packer-isolated intervals in a single borehole. Pumping tests and water sampling
can also be performed in each of the packed-off intervals. The PVC system installed at
Bure is less rigid than the stainless steel EPG system, which contributes, besides the larger
test interval volume, to a higher well-bore storage coefficient. Another constraint is that
this equipment requires the borehole to be filled with freshwater for installation, which is
critical in argillaceous formations. However, its main advantages are its active testing and
monitoring capabilities as well as the amount of intervals for investigating formation prop-
erties (i.e. head or permeability profiles along a single well). The equipment installed in
the Callovo-Oxfordian at Bure consists of 11 measuring chambers installed over 165m at
410575m depths.
For the drifts, only permanent multipacker equipment is used. It is installed in 101 or
86mm diameter boreholes. The lengths of the boreholes are generally between 15 and
20m. The boreholes are cored using air as drilling fluid, which ensures good stability. The
completion is installed immediately after drilling and a specific procedure for removing
air and filling the test intervals with synthetic water is carried out. One equipped borehole
consists of five chambers.
6.2 Measurement strategy
The measurement strategy relies on the successive use of short-term packer tests associ-
ated with in-situ long-term monitoring techniques.
Initial short-term packer tests provide a first estimate of the formation properties (per-
meability and head). More specifically, the tests are conducted under the following condi-
tions: a 24 to 72 hour test procedure designed to obtain the borehole-formation flow model
as well as the near hole and the undisturbed formation permeability.
Thus the test procedure typically includes:
An initial passive pressure recovery and;
An active pulse test, followed by the continuation of the overall pressure recovery.
This procedure is designed to provide an order of magnitude of hydraulic conductivity, but
it is in most cases unable to provide a sufficiently precise value of the initial head. Various
borehole configurations: orientation (vertical, inclined), nature of fluids (polymer-based
mud, water, oil-based mud, diesel oil) were tested and provided similar results.
Long-term monitoring equipments, such as EPGs or multipacker equipment, are then
installed in the boreholes to accurately measure the pore water pressure. Furthermore, the
transient pressure recovery from the post-drilling conditions toward equilibrium at static
conditions provide excellent long-term data to determine the borehole-formation flow
model as well as the corresponding formation permeability. As the static pressure is no
longer estimated from transient data but measured at late time, the permeability determined
from long-term data is more accurate than the permeability obtained from short-term tests.
6.3 Interpretation methods
All short-term packer tests conducted during the 20032004 campaign were analysed
according to two approaches based on the same conceptual model, an analytical approach
and a numerical approach with the nSIGHTS Code (Roberts 2002). nSIGHTS is a radial
numerical model developed for Sandia National Laboratories in support of the WIPP Site
program (Kessel, 2005). nSIGHTS is capable of taking into account large spectra of for-
mation and borehole environments. Optimisation of the parameters is done by inverse
modelling. Through the analytical approach, it was possible to derive the best preliminary
hydraulic parameters (model type and transmissivity) and to use them as the input param-
eters for the numerical analysis. Two independent numerical analyses were performed as
follows:
The first phase of numerical interpretation, i.e. the 19941996 campaign, was carried
out from early-middle time data by using the results of the analytical interpretation as
input parameter sets. Parameters were then optimised by a non-linear regression in
order to identify the best parameter set (formation transmissivity, storage and pressure)
(Figure 7);
The second phase of numerical interpretation was carried out at late time, i.e. 2004, by
using the second parameter set, and the stabilized formation pressure as measured
from the long-term monitoring system (Figure 8).
7 LABORATORY MEASUREMENTS ON SAMPLES
Permeability tests on samples were carried out as a complement to in situ tests. Indeed, it
was not always technically feasible to test the formations in situ, and experimental measure-
ments on samples can be carried out over longer periods of time. However, samples can be
disturbed by the conditions under which they have been taken and conditioned. Laboratory
tests assume a good control of the conditions under which the samples were taken, of the
mechanical confinement and saturation as well as a minimal cohesion of the samples.
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03
Eapsed Time [h]
D
e
c
o
n
v
o
l
v
e
d

p
r
e
s
s
u
r
e

&

d
e
r
i
v
a
t
i
v
e

[
1
/
h
]
Deconvolved pressure derivative; pi=cst.
Deconvolved pressure derivative; pi(t)
Deconvolved pressure; pi=cst.
Deconvolved pressure; pi(t)
Composite TC
Composite TC derivative
Homogeneous TC derivative
I
I
n
n
e
r

z
o
n
e

b
o
u
n
d
a
r
y
Inner zone IARF stabilization
Outer zone IARF stabilization
Figure 7. Interpretation of a short term pulse test.
Figure 8. Interpretation of the recovery period of an EPG.
7.1 Permeability estimates of the argillaceous rocks through the interpretation of
water sorption/desorption isotherms and moisture profiles
In the case of mass transfer in the partially saturated argillaceous rocks, permeability can
be determined according to the saturation level, assuming that the evolution of the overall
diffusivity is known.
Two different experimental devices were built on this principle.
The first one uses a conventional oedometrical frame in which there is a cell consisting
of superimposed steel rings containing the sample. Based on a given relative humidity, the
sample is re-saturated through injection at the base. The moisture profile for a determined
re-saturation period is obtained by cutting the sample with the rings. The permeability is
obtained from several profiles for different periods of time, for a given sorption-desorption
curve (Robinet and Rhattas, 1995).
In the second case, the sample is placed under controlled relative humidity in sealed con-
tainment with an isothermal lid. Pressure applied around the sample is modified, which
induces a variation in the sample mass. The evolution kinetics of the mass is monitored
through frequent weighing on microscales. The permeability to fluid is calculated by an
inverse method by working out the mass transfer equation in a non-saturated environment.
The results obtained from samples taken in deep boreholes show a very low and rela-
tively homogeneous permeability over the formation as a whole. Values range between
1 10
14
and 4 10
14
m/s.
7.2 Permeability measurements through Tritium-traced water flow
This method for assessing permeability relies on the diffusion tests carried out with
Tritium-traced water. Following the diffusion tests, after the steady state was reached, a
difference in pressure was applied between the upstream and downstream cavities of the
diffusion cells. An increase in the fluid flow was observed in the downstream cavity. This
increase was interpreted as a result of water migration through the sample.
However, although the order of magnitude of permeability of 10
14
m/s obtained
through these tests for the argillaceous rocks seem plausible, problems related to the
equipment and certain analysis hypotheses call into question the reliability of these results.
7.3 Permeability measurements with a geomechanical cell
The tests are carried out on cylindrical test tubes, cut in a piece of core and stored in a cell
immediately out of the borehole in order to maintain the mechanical confinement. Thus,
the argillaceous rock retains most of its original water (usually over 95% saturation). In
some cases, a blind hole is drilled in the axis of the test tube to obtain a radial flow.
Permeability measurements are carried out parallel or vertical to the stratification. The
geomechanical load makes it possible to recover the original in situ state of stress of the test
tube. However, even in drained conditions, it increases the pore pressure. The hydromechan-
ical coupling is higher when the permeability is low and the overpressure resulting from the
applied mechanical stress takes a long time to dissipate or become homogeneous inside the
test tube. Water at chemical equilibrium with the rock is used as the injection fluid.
Two types of tests were carried out:
Steady-state tests (stationary);
Transient-state tests of the pulse-test type.

Steady-state tests were only carried out with the radial injection device. Since the meas-
ured flows were extremely low, relatively high gradients were applied.
7.4 Results of tests on samples
These devices underwent cross-comparison tests in the framework of the ForPro Research
Group (ForPro, 2003). Interpretation of the tests was based on the Hsieh analytical solution
(Hsieh et al., 1981) either directly from graphs (in a dimensional form), or through the
inverse method. The storage coefficient can only be determined if the pressure is close to sta-
bilization (which requires carrying out the tests over a significantly longer period of time).
Permeability measurements on core samples were carried out parallel and perpendicu-
lar to the bedding planes in order to obtain estimates for the permeability anisotropy. The
permeability estimates are of the same order as those obtained through short-term tests
carried out with straddle packers and long-term pressure logs. Estimates of the vertical
permeability range between 2.4 10
16
m/s and 1.5 10
12
m/s. Horizontal permeabil-
ity estimates on samples range between 9.1 10
15
m/s and 7.6 10
13
m/s.
8 CONCLUSIONS
The means used by Andra since 1994 to determine the hydraulic properties of the Callovo-
Oxfordian argillaceous rocks have demonstrated their robustness in terms of the comple-
mentarity of methods and their implementation, as well as the coherence of results.
Four distinct methods contributed significantly to determining the permeability in
Callovo-Oxfordian argillaceous rocks.
In situ investigations in deep boreholes were tested along considerable lengths (meters
or tens of meters) in order to obtain the bulk properties of the environment along a hori-
zontal plane, mostly parallel to the bedding. Complementary in situ investigations carried
out in the underground laboratory experimental drifts provided values of the vertical and
horizontal permeability of the argillaceous rocks at a more local scale (decimetric cham-
bers). Laboratory measurements on samples, mostly carried out vertically, only tested
smaller borehole lengths (below 0.1m).
The permeability of the Callovo-Oxfordian formation obtained through in situ measure-
ments ranges between 10
14
m/s and 10
12
m/s. Most values range between 5 10
14
m/s
and 10
12
m/s. Permeability measured through in situ investigations is of the same order as
that obtained through laboratory measurements on samples, even though the values obtained
from the latter are more scattered. At this stage, no significant anisotropy has been observed.
REFERENCES
Andra (2005a) Dossier 2005 Argile Evaluation de la faisabilit du stockage gologique en forma-
tion argileuse profonde Rapport de synthse, Dcembre 2005, Andra, France (available at
www.andra.fr).
Andra (2005b) Dossier 2005 Argile Tome Evaluation de sret du stockage gologique . Andra
report n C.RP.ADSQ.05.0018, Andra, France (available at www.andra.fr).
Andra (2005c) Dossier 2005 Rfrentiel du site Meuse/Haute-Marne. Andra report n
C.RP.ADS.04.0022, Andra, France.
Cecconi F, Chatelet V and Delay J (2004) Monitoring Abnormal Pore Pressures in Low Permeability
argillite Using Wireless Transmission Gauges, 66th EAGE Conference and Exhibition 710 June
2004 Paris.
Crois J, Schlickenrieder L, Marshall P, Boisson JY, Vogel P and Yamamoto S (2004) Hydrogeological
investigations in a low permeability claystone formation: the Mont Terri Rock Laboratory, Physics
and Chemistry of the Earth, 29, 2004, 315.
Delay J and Distinguin M (2004) Hydrogeological Investigations in Deep Wells at the Meuse/Haute
Marne Underground Research Laboratory. In: R. Hack, R. Azzam, R. Charlier (Eds.), Engineering
Geology for Infrastructure Planning in Europe, Lecture Notes in Earth Sciences, 104. Springer, pp.
219225.
Distinguin M and Lavanchy JM (2006) Determination hydraulic properties of the Callovo-
Oxfordian Argillites at the Bure Site: Synthesis of the results obtained in deep boreholes using
several in-situ investigation techniques. International Meeting (Tours, 2005). Physics and
Chemistry of the Earth, special issue (Elsevier Ed.) (accepted).
Eldred CD, Scarrow JA and Smith A (1995) An integrated system for groundwater monitoring at
Sellafield PNWR, UK. High-Level Radioactive Waste Management Conference, May 15, Las
Vegas, Nevada, USA, 1995.
ForPro (2003) Benchmark Permabilits: Caractrisation des milieux de trs faible permabilit.
Rapport GdR FORPRO 2003/06.
Gaucher E, Robelin C, Matray JM, Ngrel G, Gros Y, Heitz JF, Vinsot A and Rebours H,
Cassagnabre A, Bouchet A (2004) Andra Underground research Laboratory: interpretation of the
mineralogical and geochemical data acquired in the Callovian-Oxfordian formation by investiga-
tive drilling, Physics and Chemistry of the Earth, 29, 5577.
Horseman S et al (1996) Water, gas and solute movement through argillaceaous medias Report
CC-961/1, OCDE Publications, Paris, 1996, pp 205213.
Hsieh PA, Tracy JV, Neuzil CE, Bredehoeft JD and Silliman SE (1981) A transient method for deter-
mining the hydraulic properties of tight rocks1, Theory: lnternational Journal of Rock
Mechanics and Mining Science, 18, 3, 245252.
Kessel DS (2005) Sandia National Laboratories Waste Isolation Pilot Plant (WIPP) Program
Description. Available at http://www.nwmp.sandia.gov.
Klopff W, Vinsot A and Delay J (2004) Expanded Application of NMR Well logs for Poresize
Distribution and Permeability in Low-Permeability Argillites and Carbonates, 66th EAGE
Conference and Exhibition 710 June 2004, Paris.
Parkhurst DL and Appelo CAJ (1999) Users guide to PREEQC (Vers.2) A computer program for
speciation, batch-reaction, one dimentional transport, and inverse geochemical calculations. U.S.
Geol. Surv. Water-Ressour. Invest. Rep. 99/4259, 312p.
Pearson FJ. Arcos D, Bath A, Boisson JY, Fernndez AM, Gbler HE, Gaucher E, Gautschi A,
Griffault L, Hernn P and Waber HN (2003) Geochemistry of Water in the Opalinus Clay
Formation at the Mont Terri Laboratory. Report of the Federal Office for Water and Geology
(Bern, Switzerland), Geology Series, N 5.
Revil A and Glover PWJ (1998) Nature of surface electrical conductivity in natural sands, sand-
stones and clays, Geophys. Res. Lett., 25, pp 691694.
Roberts RM, Beauheim RL and Avis JD (2002) Quantifying Parameter Uncertainty in Well-Test
Analysis. Proceedings Volume of International Groundwater Symposium Bridging the Gap
between Measurements and Modeling in Heterogeneous Media. March 2529, 2002. Berkeley,
California.
Robinet JC and Rhattas M (1995) Dtermination de la permabilit non sature des matriaux
argileux faible porosit. Can Geotech J., 32, 6, 10351043.
Soulier L and Lemaitre M (1993) E.M. MWD Data Transmission Status and Perspectives SPE/
IADC 25686 SPE/IADC Drilling Conference, Amsterdam 2325 February 1993.
Tsang CF and Bernier F (2004) Definitions of excavation disturbed zone and excavation damaged
zone, Luxembourg, 35 November 2003, European Commission report EUR 21028 EN.
Van Loon LR, Wersin P, Soler JM, Eikenberg J, Gimmi T, Hernn P, Dewonck S and Matray JM
(2004) In-situ diffusion of HTO, 22Na, Csand I- in Opalinus Clay at the Mont Terri under-
ground rock laboratory. Radiochim. Acta, 92, 757763.
Vigneron G, Delay J, Distinguin M, Lebon P and Trouiller A (2004) Apport des investigations multi
chelles pour la construction dun modle conceptuel des plateformes carbonates de lOxfordien
moyen et suprieur de lest du Bassin de Paris. Colloque: Gologie et hydrogologie du Bassin de
Paris. Avances et perspectives: hommage Claude Mgnien, Paris, 1617 novembre 2004.
Wersin P, Dewonck S, Savoye S, Hernan P, Van Loon L, Soler J, Gimmi T, Cartalade A, Yllera A,
Eikenberg J, Grolimund D, Bayens B and Van Drop F (2005) Diffusion of HTO, iodide, sodium-
22 and caesium in Opalinus Clay: A combined field, Laboratory and modelling study.
International meeting March 1414 Tours, France. Clays in natural and engineered barriers for
radioactive waste confinement.
III Complex hydrogeological systems
A. Large aquifers, assessment of the resource
CHAPTER 8
Great aquifer systems of the world
Jean Margat
Honorary member of IAH, BRGM, BP 36009, 45060 Orlans Cedex 2, France
ABSTRACT: A macroscopic vision of the hydrogeological conditions of the world cannot be
based on only one typology deduced from the predominant regional geological features. It could, and
should, also identify and outline all first-order aquifer systems on the basis of analyses of their struc-
ture. Just as hydrographic structures come in various sizes, so too do groundwater systems. Less
well-known than the giant river systems, there are nevertheless around forty very large aquifer sys-
tems with a surface area ranging from 100,000 to 1,000,000km
2
or more, distributed across the con-
tinents and all the climate zones.
These first-order aquifers are usually linked to great sedimentary basins which are generally multi-
layered and whose maximum thickness may exceed 10,000m. Altogether, they occupy 35 million
km
2
and probably contain the greater part of the continental fresh groundwater reserves. They often
have multinational drainage basins, and social and economic problems have to be overcome in order
to study them and manage their resources.
1 INTRODUCTION
The acquisition of new information on the hydrogeology in many countries has produced
a wealth of articles and national and regional scale maps making it possible and pertinent,
to take a global view of aquifer systems. The first step is to consider a typology of the salient
hydrogeological conditions based on the principles of the international hydrogeological maps
(IAH/UNESCO, 1995) i.e. the purpose of the mapping synthesis of the WHYMAP proj-
ect undertaken by the IAH, UNESCO and the Commission of the Geological Map of the
World (CGMW). The view of global hydrogeological conditions may also be structural and
will show the first order aquifer systems whose geography is less well documented rather than
that of the large drainage basins.
In the same way as surface water is organised into drainage basins of all sizes with a few
large river basins monopolising a large portion of all the flow, the identifiable aquifer systems
are also of different sizes and a few large ones contain a considerable share of the worlds
groundwater reserves and flow. However, their geography is less obvious and must be investi-
gated through hydrogeologic studies.
All the large groundwater systems are contained within sedimentary rocks and are
included in the blue class of the above-mentioned WHYMAP map.
37 first-order aquifer systems, whose extents range between 100,000 and 2,000,000km
2
or more, are distributed across the continents and through all the climate zones. Their locations
are shown in Figure 1 and their main characteristics listed in Table 1.
2 STRUCTURAL TYPES
The large aquifer systems belong to three structural types.
(1) Sedimentary, generally multilayered basins, of varying complexity depend on their
structure and their stratigraphy, from the Pre-Cambrian to the Quaternary, and which may
reach a thickness of several thousand metres (up to 20,000m on the Russian platform).
They are composed of alternating, very permeable layers with a high transmissivity
(detrital formations sand, sandstone or carbonate rocks) and layers with low permeabil-
ity (clay, marl) which do not prevent vertical communication (leakage) or impervious lay-
ers (saline rocks), complicated by lateral facies changes.
They contain one or several water-bearing layers with unconfined freshwater aquifers,
and a varying number of layers with confined, partly inter-connected horizons. These may
be predominantly fresh but with brackish or saline water at depth where the recharge is
weaker and the residence times longer. This is the most common type. The main character-
istic is the difference between the horizontal and vertical anisotropy. There are many
examples: Great Artesian Basin in Australia, Sahara Basins (SASS, Nubian Aquifer),
Northern Great Plains Aquifer System in North America (Canada, United States), and the
Western Siberia Basin. These basins may be divided according to two structural types:
In the Paris model, of which the Paris Basin is a classic prototype, outcropping
water-bearing sedimentary rocks on the periphery contain unconfined aquifers and
overflowing springs that are in continuous connection with the deeper confined, some-
times artesian, horizons towards the centre.
Figure 1. Large aquifers in the World.
T
a
b
l
e

1
.
G
r
e
a
t

a
q
u
i
f
e
r

s
y
s
t
e
m
s

o
f

t
h
e

W
o
r
l
d
.
N
c
f
.
N
a
m
e
C
o
u
n
t
r
y
S
u
r
f
a
c
e

a
r
e
a
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
T
h
e
o
r
e
t
i
c
a
l
M
e
a
n

r
e
c
h
a
r
g
e
R
e
f
e
r
e
n
c
e
s
m
a
p
1
0
0
0
k
m
2
s
t
r
u
c
t
u
r
e

(
T
:

m
a
x
i
m
u
m
r
e
s
e
r
v
e

v
o
l
u
m
e
f
l
u
x

i
n

k
m
3
/
a
n
t
h
i
c
k
n
e
s
s

i
n

m
)
1
0
0
0
k
m
3
A
F
R
I
C
A
1
N
u
b
i
a
n

a
q
u
i
f
e
r

s
y
s
t
e
m

(
N
A
S
)
:
E
g
y
p
t
2
1
9
9
M
u
l
t
i
l
a
y
e
r
5
4
2
1
M
.

B
a
k
h
b
a
k
h
i

2
0
0
2
N
u
b
i
a
n

S
a
n
d
s
t
o
n
e

a
q
u
i
f
e
r
L
i
b
y
a
(
o
f

w
h
i
c
h
C
a
m
b
r
o
-
O
r
d
o
v
i
c
i
a
n

t
o

f
r
e
s
h
w
a
t
e
r
C
E
D
A
R
E
/
I
F
A
D
s
y
s
t
e
m

(
N
S
A
S
)
S
u
d
a
n
1
8
0
0

w
i
t
h
O
l
i
g
o
c
e
n
e
(
F
A
O

2
0
0
3
)
P
o
s
t
-
N
u
b
i
a
n

a
q
u
i
f
e
r

s
y
s
t
e
m
C
h
a
d
f
r
e
s
h
w
a
t
e
r
)
P
r
e
d
o
m
i
n
a
n
t
l
y

U
N
E
S
C
O
/
O
S
S

2
0
0
5
(
P
N
A
S
)
c
o
n
t
i
n
e
n
t
a
l

s
a
n
d
s
t
o
n
e
T
3
5
0
0
m
2
N
o
r
t
h
e
r
n

S
a
h
a
r
a

a
q
u
i
f
e
r
A
l
g
e
r
i
a
1
0
1
9
M
u
l
t
i
l
a
y
e
r

6
0
1
O
S
S

2
0
0
3
,

2
0
0
4
s
y
s
t
e
m

(
S
A
S
S
)
L
i
b
y
a
C
a
m
b
r
o
-
O
r
d
o
v
i
c
i
a
n

t
o
U
N
E
S
C
O
/
O
S
S

2
0
0
5
T
u
n
i
s
i
a
M
i
o
c
e
n
e
C
o
n
t
i
n
e
n
t
a
l

i
n
t
e
r
c
a
l
a
r
y
a
n
d

t
e
r
m
i
n
a
l

c
o
m
p
l
e
x
T
1
6
0
0
m
3
B
a
s
i
n

o
f

M
u
r
z
u
k
-
D
j
a
d
o
L
i
b
y
a
4
5
0
M
u
l
t
i
l
a
y
e
r

4
.
8

i
n

L
i
b
y
a
0
.
1
5
U
N
E
S
C
O
/
O
S
S

2
0
0
5
A
l
g
e
r
i
a
C
a
m
b
r
o
-
O
r
d
o
v
i
c
i
a
n

t
o
N
i
g
e
r
C
r
e
t
a
c
e
o
u
s
T
2
5
0
0
m
4
B
a
s
i
n

o
f

T
a
o
u
d
e
n
i

T
a
n
e
z
r
o
u
f
t
A
l
g
e
r
i
a
2
0
0
0
M
u
l
t
i
l
a
y
e
r

0
.
0
1
8
0
.
3

i
n

M
a
l
i
I
.

Z
e
k
s
t
e
r

2
0
0
4
,
M
a
u
r
i
t
a
n
i
a
I
n
f
r
a
c
a
m
b
r
i
a
n

t
o
e
x
p
l
o
i
t
a
b
l
e

i
n
U
N
E
S
C
O
/
O
S
S

2
0
0
5
M
a
l
i
,

B
u
r
k
i
n
a
T
e
r
t
i
a
r
y

(
C
T
)
M
a
l
i

a
n
d
F
a
s
o
,

G
u
i
n
e
a
T
4
0
0
0
M
a
u
r
i
t
a
n
i
a
5
S
e
n
e
g
a
l
-
M
a
u
r
i
t
a
n
i
a
n

B
a
s
i
n
M
a
u
r
i
t
a
n
i
a
,
3
0
0
M
u
l
t
i
l
a
y
e
r

1
.
5
9
B
.

D
i
a
g
a
n
a

1
9
9
7
S
e
n
e
g
a
l
,
M
a
i
n

a
q
u
i
f
e
r
:
U
N
E
S
C
O
/
O
S
S

2
0
0
5
G
a
m
b
i
a
,
M
a
e
s
t
r
i
c
h
t
i
a
n
G
u
i
n
e
a

B
i
s
s
a
u
T
6
0
0

m
(
C
o
n
t
i
n
u
e
d
)
T
a
b
l
e

1
.
(
C
o
n
t
i
n
u
e
d
)
N
c
f
.
N
a
m
e
C
o
u
n
t
r
y
S
u
r
f
a
c
e

a
r
e
a
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
T
h
e
o
r
e
t
i
c
a
l
M
e
a
n

r
e
c
h
a
r
g
e
R
e
f
e
r
e
n
c
e
s
m
a
p
1
0
0
0
k
m
2
s
t
r
u
c
t
u
r
e

(
T
:

m
a
x
i
m
u
m
r
e
s
e
r
v
e

v
o
l
u
m
e
f
l
u
x

i
n

k
m
3
/
a
n
t
h
i
c
k
n
e
s
s

i
n

m
)
1
0
0
0
k
m
3
6
I
u
l
l
e
m
e
d
e
n
-
I
r
h
a
z
e
r

a
q
u
i
f
e
r
N
i
g
e
r
,
6
3
5
M
u
l
t
i
l
a
y
e
r

1
0

t
o

1
5
A
.

D
O
D
O

1
9
9
2
s
y
s
t
e
m
A
l
g
e
r
i
a
,
C
a
m
b
r
o
-
O
r
d
o
v
i
c
i
a
n

t
o

U
N
E
S
C
O
/
O
S
S

2
0
0
5
M
a
l
i
,
E
o
c
e
n
e
N
i
g
e
r
i
a
3

s
u
b
-
b
a
s
i
n
s
T
1
5
0
0
m
7
L
a
k
e

C
h
a
d

B
a
s
i
n

N
i
g
e
r
,
1
9
1
7
M
u
l
t
i
l
a
y
e
r

0
.
6

i
n

N
i
g
e
r
3
.
6

i
n

N
i
g
e
r
C
B
L
T

9
7
7
N
i
g
e
r
i
a
,
C
o
n
t
i
n
e
n
t
a
l

i
n
t
e
r
c
a
l
a
r
y
,

(
0
,
0
5
J
.
L
.

S
c
h
n
e
i
d
e
r

2
0
0
1
C
h
a
d
,
C
o
n
t
i
n
e
n
t
a
l

t
e
r
m
i
n
a
l

e
t
e
x
p
l
o
i
t
a
b
l
e
)
U
N
E
S
C
O
/
O
S
S

2
0
0
5
C
a
m
e
r
o
o
n
,
P
l
i
o
-
Q
u
a
t
e
r
n
a
r
y
0
.
4

e
x
p
l
o
i
t
a
b
l
e
C
A
R
T
7
0
0
0
m

i
n

C
h
a
d
8
S
u
d
d
-
B
a
s
i
n

U
m
m

R
u
w
a
b
a

S
u
d
a
n
,
3
6
5
M
u
l
t
i
l
a
y
e
r

0
.
1
1
0
.
3
4
I
C
I
D

1
9
8
3
A
q
u
i
f
e
r
E
t
h
i
o
p
i
a
N
e
o
g
e
n
e

Q
u
a
t
e
r
n
a
r
y

U
N

1
9
8
7
T
3
0
0
0
m
O
A
C
T

1
9
9
3
9
O
g
a
d
e
n
-
J
u
b
a

B
a
s
i
n
E
t
h
i
o
p
i
a
,
1
0
0
0
M
u
l
t
i
l
a
y
e
r
1
0
U
N

1
9
8
7
S
o
m
a
l
i
a
,
C
o
n
f
i
n
e
d

a
n
d
K
e
n
y
a
u
n
c
o
n
f
i
n
e
d

a
q
u
i
f
e
r
s
T
1
2
0
0
0
m
1
0
C
o
n
g
o

B
a
s
i
n
C
o
n
g
o

D
R
,
7
5
0
M
u
l
t
i
l
a
y
e
r

1
0
0
I
.

Z
e
k
t
s
e
r

2
0
0
4
C
o
n
g
o
,
M
e
s
o
z
o
i
c

(
K
a
r
o
o
)

t
o
A
A
C

1
9
9
3
A
n
g
o
l
a
,
Q
u
a
t
e
r
n
a
r
y

a
l
l
u
v
i
a
l
C
R

o
f

A
f
r
i
c
a
T
3
5
0
0
m
G
a
b
o
n
1
1
C
u
v
e
l
a
i

B
a
s
i
n

U
p
p
e
r
A
n
g
o
l
a
,
7
0
0
M
u
l
t
i
l
a
y
e
r

3
0

t
o

6
0
I
.

Z
e
k
t
s
e
r

2
0
0
4
Z
a
m
b
e
z
i

B
a
s
i
n

(
U
p
p
e
r
B
o
t
s
w
a
n
a
,
C
a
r
b
o
n
i
f
e
r
o
u
s
,
K
a
l
a
h
a
r
i
)
Z
a
m
b
i
a
,
C
r
e
t
a
c
e
o
u
s

(

K
a
r
o
o

)

Z
i
m
b
a
b
w
e
t
o

N
e
o
g
e
n
e
1
2
S
t
a
m
p
r
i
e
t
-
K
a
l
a
h
a
r
i

B
a
s
i
n
S
o
u
t
h

A
f
r
i
c
a
,
3
5
0
M
u
l
t
i
l
a
y
e
r

1

t
o

2
I
.

Z
e
k
t
s
e
r

2
0
0
4
(
L
o
w
e
r

K
a
l
a
h
a
r
i
)
B
o
t
s
w
a
n
a
,

K
a
r
o
o

t
o

N
e
o
g
e
n
e
N
a
m
i
b
i
a
1
3
K
a
r
o
o

B
a
s
i
n
S
o
u
t
h

A
f
r
i
c
a
6
0
0
M
u
l
t
i
l
a
y
e
r

3

t
o

5

(
d
o
l
o
m
i
t
e
)
1
6

t
o

3
7
A
A
C

1
9
9
3
C
a
m
b
r
i
a
n

t
o

J
u
r
a
s
s
i
c
I
.

Z
e
k
t
s
e
r

2
0
0
4
T
7
0
0
0
m
A
M
E
R
I
C
A
1
4
N
o
r
t
h
e
r
n

G
r
e
a
t

P
l
a
i
n
s
C
a
n
a
d
a
,
2
0
0
0
M
u
l
t
i
l
a
y
e
r

U
N

1
9
7
6
A
q
u
i
f
e
r

S
y
s
t
e
m

(
U
S
A
)
/
U
S
A
P
a
l
e
o
z
o
i
c

t
o

E
o
c
e
n
e
U
S
G
S

2
0
0
3
I
n
t
e
r
i
o
r

P
l
a
i
n
s

(
C
a
n
a
d
a
)
1
5
C
a
m
b
r
i
a
n

O
r
d
o
v
i
c
i
a
n

U
S
A
2
5
0
M
u
l
t
i
l
a
y
e
r

U
S
G
S

2
0
0
3
A
q
u
i
f
e
r

S
y
s
t
e
m

(
C
e
n
t
r
a
l
C
a
m
b
r
i
a
n
-
O
r
d
o
v
i
c
i
a
n
R
e
g
i
o
n
)
1
6
C
e
n
t
r
a
l

V
a
l
l
e
y

C
a
l
i
f
o
r
n
i
a
U
S
A
8
0
C
o
n
f
i
n
e
d

a
n
d

u
n
c
o
n
f
i
n
e
d
1
.
7
3
7
U
S
G
S

M
u
l
t
i
l
a
y
e
r

Q
u
a
t
e
r
n
a
r
y
J
.
A
.

M
i
l
l
e
r
,

2
0
0
0
a
l
l
u
v
i
a
l

a
q
u
i
f
e
r
,

P
6
0
0
m
1
7
O
g
a
l
l
a
l
a

A
q
u
i
f
e
r

H
i
g
h

P
l
a
i
n
s
U
S
A
4
5
0
T
e
r
t
i
a
r
y

a
l
l
u
v
i
a
1
5
6
t
o

8
U
S
G
S
T
1
5
0
m
J
.
A
.

M
i
l
l
e
r
,

2
0
0
0
1
8
A
t
l
a
n
t
i
c

a
n
d

G
u
l
f

C
o
a
s
t
a
l
U
S
A
,
1
1
5
0
M
u
l
t
i
l
a
y
e
r

U
S
G
S
P
l
a
i
n

A
q
u
i
f
e
r
s
M
e
x
i
c
o
P
a
l
e
o
z
o
i
c

t
o

T
e
r
t
i
a
r
y
J
.
A
.

M
i
l
l
e
r
,

2
0
0
0
G
r
o
u
p

o
f

c
o
n
t
i
g
u
o
u
s
a
q
u
i
f
e
r
s
T
1
2
0
0
0
m
1
9
A
m
a
z
o
n

B
a
s
i
n

B
r
a
z
i
l
,
1
5
0
0
M
u
l
t
i
l
a
y
e
r

3
2
.
5
A
.

R
e
b
o
u
a
s

1
9
8
8
C
o
l
o
m
b
i
a
,
P
a
l
e
o
z
o
i
c

t
o

T
e
r
t
i
a
r
y
P
e
r
u
,
T
2
0
0
0
m
B
o
l
i
v
i
a
2
0
M
a
r
a
n
h
a
o

B
a
s
i
n
B
r
a
z
i
l
7
0
0
M
u
l
t
i
l
a
y
e
r

1
7
.
5
4
A
.

R
e
b
o
u
a
s

A
I
H
O
r
d
o
v
i
c
i
a
n

t
o

C
r
e
t
a
c
e
o
u
s
1
9
7
6
1
9
8
8
T
3
0
0
0
m
(
C
o
n
t
i
n
u
e
d
)
T
a
b
l
e

1
.
(
C
o
n
t
i
n
u
e
d
)
N
c
f
.
N
a
m
e
C
o
u
n
t
r
y
S
u
r
f
a
c
e

a
r
e
a
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
T
h
e
o
r
e
t
i
c
a
l
M
e
a
n

r
e
c
h
a
r
g
e
R
e
f
e
r
e
n
c
e
s
m
a
p
1
0
0
0
k
m
2
s
t
r
u
c
t
u
r
e

(
T
:

m
a
x
i
m
u
m
r
e
s
e
r
v
e

v
o
l
u
m
e
f
l
u
x

i
n

k
m
3
/
a
n
t
h
i
c
k
n
e
s
s

i
n

m
)
1
0
0
0
k
m
3
2
1
G
u
a
r
a
n
i

A
q
u
i
f
e
r

(
P
a
r
a
n
a

&
B
r
a
z
i
l
1
1
9
5
M
u
l
t
i
l
a
y
e
r

5
7
2
3
4
A
.

R
e
b
o
u
a
s
C
h
a
c
o
)

o
r

M
e
r
c
o
s
u
l

A
q
u
i
f
e
r
A
r
g
e
n
t
i
n
a
S
i
l
u
r
i
a
n

t
o

C
r
e
t
a
c
e
o
u
s
1
9
8
8
1
9
7
6
S
y
s
t
e
m
P
a
r
a
g
u
a
y
T
8
0
0
m
L
.
M
.

A
r
a
j
o

&

a
l
.

1
9
7
9
U
r
u
g
u
a
y
T
o
t
a
l

b
a
s
i
n
:

5
0
0
0
m
A
S
I
A
2
2
A
r
a
b
i
a
n

A
q
u
i
f
e
r

S
y
s
t
e
m
S
a
u
d
i

A
r
a
b
i
a
,
1
4
8
5

i
n

M
u
l
t
i
l
a
y
e
r

2
.
1
8
5

i
n

1
.
2
W
.

A
b
d
e
r
r
h
a
m
a
n
J
o
r
d
a
n
,
S
a
u
d
i

A
r
a
b
i
a
C
a
m
b
r
i
a
n

t
o

N
e
o
g
e
n
e
S
a
u
d
i

A
r
a
b
i
a
U
N
E
S
C
O

2
0
0
2
K
u
w
a
i
t
,
T
6
5
0
0
m
U
N

1
9
8
2
B
a
h
r
a
i
n
,
Q
a
t
a
r
2
3
I
n
d
u
s

B
a
s
i
n
P
a
k
i
s
t
a
n
3
2
0
A
l
l
u
v
i
a
l

u
n
c
o
n
f
i
n
e
d
6
6

(
u
s
a
b
l
e
I
C
I
D

2
0
0
0
a
q
u
i
f
e
r
p
o
t
e
n
t
i
a
l
)
T
3
0
0
m
2
4
I
n
d
o
-
G
a
n
g
a
-
B
r
a
h
m
a
p
u
t
r
a
I
n
d
i
a
,
6
0
0
C
o
n
f
i
n
e
d

a
n
d
U
N

1
9
8
6
B
a
s
i
n
N
e
p
a
l
,
u
n
c
o
n
f
i
n
e
d

a
l
l
u
v
i
a
l
D
.
K
.

D
u
t
t

1
9
8
8
,
1
9
8
9
B
h
a
b
h
a
r

e
t

T
a
r
a
i

A
q
u
i
f
e
r
s
B
a
n
g
l
a
d
e
s
h
a
q
u
i
f
e
r

T
6
0
0
0
m
2
5
W
e
s
t

S
i
b
e
r
i
a

A
r
t
e
s
i
a
n

B
a
s
i
n
R
u
s
s
i
a
3
2
0
0
M
u
l
t
i
l
a
y
e
r

1
0
0
0
8
2

o
v
e
r

U
N

1
9
8
6
P
a
l
e
o
z
o
i
c

t
o

C
r
e
t
a
c
e
o
u
s
2
.
5
M

k
m
2
I
.

Z
e
k
t
s
e
r

2
0
0
4
I
n

p
a
r
t

s
u
b
-
p
e
r
m
a
f
r
o
s
t
(
p
r
e
d
i
c
t
i
v
e
T
6
0
0
0
m
f
r
e
s
h

g
r
o
u
n
d
-
F
r
e
s
h
w
a
t
e
r
s
a
l
i
n
e
w
a
t
e
r

f
i
e
l
d
)
w
a
t
e
r
2
6
T
u
n
g
u
s
s

B
a
s
i
n
R
u
s
s
i
a
1
0
0
0
M
u
l
t
i
l
a
y
e
r

U
N

1
9
8
6
C
a
m
b
r
i
a
n

t
o

T
r
i
a
s
s
i
c
.
I
.

Z
e
k
t
s
e
r

2
0
0
4
S
u
b
-
p
e
r
m
a
f
r
o
s
t

F
r
e
s
h

a
n
d

s
a
l
i
n
e

w
a
t
e
r
T
4
0
0
0
m
2
7
A
n
g
a
r
a

L
e
n
a

B
a
s
i
n
R
u
s
s
i
a
6
0
0
M
u
l
t
i
l
a
y
e
r

1
2
.
6
U
N

1
9
8
6
C
a
m
b
r
i
a
n

t
o

J
u
r
a
s
s
i
c

I
.

Z
e
k
t
s
e
r

2
0
0
4
F
r
e
s
h

a
n
d

s
a
l
i
n
e

w
a
t
e
r
T
3
0
0
0
m
2
8
Y
a
k
u
t

B
a
s
i
n
R
u
s
s
i
a
7
2
0
M
u
l
t
i
l
a
y
e
r

1
8
U
N

1
9
8
6
S
u
b
-
p
e
r
m
a
f
r
o
s
t
I
.

Z
e
k
t
s
e
r

2
0
0
4
F
r
e
s
h

a
n
d

s
a
l
i
n
e

w
a
t
e
r
I
n
f
r
a
-
C
a
m
b
r
i
a
n

t
o
C
r
e
t
a
c
e
o
u
s
T
1
2
,
0
0
0
m
2
9
N
o
r
t
h

C
h
i
n
a
C
h
i
n
a
3
2
0
M
u
l
t
i
l
a
y
e
r

Q
u
a
t
e
r
n
a
r
y
0
.
0
1
8
4
9
,
2
C
h
i
n
a

W
a
t
e
r

V
i
s
i
o
n
A
q
u
i
f
e
r

s
y
s
t
e
m
a
l
l
u
v
i
a
l

a
q
u
i
f
e
r
,

c
o
n
f
i
n
e
d
(
4
8
,
1
2
0
0
0
(
H
u
a
n
g
-
H
u
a
i
-
H
a
i

P
l
a
i
n
)
a
n
d

u
n
c
o
n
f
i
n
e
d
e
x
p
l
o
i
t
a
b
l
e
)
I
.

Z
e
k
t
s
e
r

2
0
0
4
T
1
0
0
0
m
A
r
t
e
s
i
a
n
3
0
S
o
n
g
-

L
i
a
o

P
l
a
i
n
C
h
i
n
a
3
1
1
M
u
l
t
i
l
a
y
e
r

a
l
l
u
v
i
a
l
3
3
.
4
I
.

Z
e
k
t
s
e
r

2
0
0
4
Q
u
a
t
e
r
n
a
r
y

a
q
u
i
f
e
r
(
2
3
.
8
T
3
0
0
m
e
x
p
l
o
i
t
a
b
l
e
)
C
o
n
f
i
n
e
d

a
n
d
u
n
c
o
n
f
i
n
e
d
A
r
t
e
s
i
a
n
3
1
T
a
r
i
m

B
a
s
i
n
C
h
i
n
a
5
2
0
M
u
l
t
i
l
a
y
e
r

a
l
l
u
v
i
a
l
a
q
u
i
f
e
r
.
U
N

1
9
8
6
C
o
n
f
i
n
e
d

a
n
d

u
n
c
o
n
f
i
n
e
d
T
1
2
0
0
m
E
U
R
O
P
E
3
2
P
a
r
i
s

B
a
s
i
n
F
r
a
n
c
e
1
9
0
M
u
l
t
i
l
a
y
e
r

0
.
5

t
o

1
2
0

t
o

3
0
U
N

1
9
9
0
T
r
i
a
s
s
i
c

t
o

N
e
o
g
e
n
e
T
3
2
0
0
m
3
3
R
u
s
s
i
a
n

P
l
a
t
f
o
r
m

B
a
s
i
n
s
R
u
s
s
i
a
3
1
0
0
G
r
o
u
p

o
f

c
o
n
t
i
g
u
o
u
s

U
N

1
9
9
0
m
u
l
t
i
l
a
y
e
r

a
q
u
i
f
e
r
s

I
n
f
r
a
-
C
a
m
b
r
i
a
n

t
o
Q
u
a
t
e
r
n
a
r
y
S
u
b
-
p
e
r
m
a
f
r
o
s
t

i
n

N
o
r
t
h
T
2
0
,
0
0
0
m
(
C
o
n
t
i
n
u
e
d
)
T
a
b
l
e

1
.
(
C
o
n
t
i
n
u
e
d
)
N
c
f
.
N
a
m
e
C
o
u
n
t
r
y
S
u
r
f
a
c
e

a
r
e
a
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
T
h
e
o
r
e
t
i
c
a
l
M
e
a
n

r
e
c
h
a
r
g
e
R
e
f
e
r
e
n
c
e
s
m
a
p
1
0
0
0
k
m
2
s
t
r
u
c
t
u
r
e

(
T
:

m
a
x
i
m
u
m
r
e
s
e
r
v
e

v
o
l
u
m
e
f
l
u
x

i
n

k
m
3
/
a
n
t
h
i
c
k
n
e
s
s

i
n

m
)
1
0
0
0
k
m
3
3
4
N
o
r
t
h

C
a
u
c
a
s
u
s

B
a
s
i
n
R
u
s
s
i
a
2
3
0
M
u
l
t
i
l
a
y
e
r

U
N

1
9
9
0
C
a
r
b
o
n
i
f
e
r
o
u
s

t
o
N
e
o
g
e
n
e
T
1
0
,
0
0
0
m
3
5
P
e
c
h
o
r
a

B
a
s
i
n
R
u
s
s
i
a
3
5
0
M
u
l
t
i
l
a
y
e
r

U
N

1
9
9
0
O
r
d
o
v
i
c
i
a
n

t
o

T
e
r
t
i
a
r
y
T
3
0
0
0
m
P
a
r
t
l
y

s
u
b
-
p
e
r
m
a
f
r
o
s
t
A
U
S
T
R
A
L
I
A
3
6
G
r
e
a
t

A
r
t
e
s
i
a
n

B
a
s
i
n
A
u
s
t
r
a
l
i
a
1
7
0
0
M
u
l
t
i
l
a
y
e
r

T
r
i
a
s
s
i
c

2
0
1
.
1
M
.
A
.

H
a
b
e
r
m
e
h
l
t
o

C
r
e
t
a
c
e
o
u
s
8
.
7
U
N
E
S
C
O

2
0
0
2
T
3
0
0
0
m
W
a
t
e
r

R
e
s
o
u
r
c
e
s

a
n
d
U
s
e
,

A
u
s
t
r
a
l
i
a
,

2
0
0
3
3
7
C
a
n
n
i
n
g

B
a
s
i
n
A
u
s
t
r
a
l
i
a
4
3
0
M
u
l
t
i
l
a
y
e
r

J
.
E
.

L
a
u

&

a
l
.

1
9
8
7
D
e
v
o
n
i
a
n

t
o

C
r
e
t
a
c
e
o
u
s
I
.

Z
e
k
t
s
e
r

2
0
0
4
T
1
0
0
0
m
In the Pannonian model, of which the Hungarian Basin is an example, the upper layers
completely cover the deeper aquifers, which are only accessible by drilling and are less
easy to identify, and where vertical transport by leakage or through faults predominates.
These basins are often also artesian where they contain zones of initial potential outflow
over large surface areas sourced by favourable elevation and head providing them with
energy.
(2) Subsidence troughs are mainly filled with alluvium, with an unconfined aquifer
which may also be multilayered and contain several confined aquifers including artesian zones.
Examples are: Central Valley California, Northern China Plains, Indo-Ganga-Brahmaputra
Basin, India.
(3) Detrital accumulation at the foot of large mountain chains: mainly unconfined
aquifers such as, for example, the High Plains (Ogallala) aquifer, United States.
However, some of the great aquifers referred to as single systems are, in fact, aggregates
of adjacent and similar, but nevertheless, independent aquifers, e.g., the Ogallala aquifer
of the High Plains in the United States and the Basins on the Russian platform.
These large aquifer systems jointly cover a total area of nearly 35 million km
2
, almost a
quarter of the entire land surface. They are likely to contain the greater part of the conti-
nental fresh groundwater reserves although reserve estimates are patchy for some aquifers
(Table 1). They contain most of the identified non-renewable groundwater resources
(reserves that can be withdrawn in economically and environmentally acceptable condi-
tions) and are partly exploited by several countries in arid zones (mainly Saudi Arabia,
Libya and Algeria). They have also been subjected to the highest depletion rates during the
twentieth century, either through over-exploitation or to groundwater mining. Overall,
this depletion amounts to at least 1000km
3
and has caused vast drawdowns, often greater
than 100m. The great aquifers are minor participants in world groundwater flow (on the
order of 12,000 to 13,000km
3
/y) since the flow is more a characteristic of unconfined
aquifers of which many are local phreatic aquifers in humid zones.
The functions of these large aquifer systems in the cycle of global continental waters and
their place in the water economy are closely linked to their prevailing climatic. In humid
zones, they contribute to the regulation of surface flow and they are also heavily exploited,
even over-exploited, for example, in China, India and the United States.
In arid or semi-arid climate zones, they contribute very little to scarce surface water
flow, and abstraction draws mostly on their non-renewable resources, particularly in
Saharan Africa, in the Middle-East and in Australia (where it is kept to a minimum).
In the sub-arctic regions of northern Europe and Asia (Russia, Siberia), the permafrost,
which reaches great depths (several hundred metres), freezes the hydrological function
and constrains the potential for exploitation.
The incomplete records of current groundwater abstraction from one country to another,
make it impossible to put an accurate figure on the role of the great aquifer systems in
global groundwater exploitation, the latter estimated to be at least 800km
3
/y. It is possible
to provide a few examples (Table 2), and they already represent some 135km
3
/y of which
around twenty are from non-renewable resources (Sahara and Arabian basins) but the
intensity varies greatly.
The aquifer systems that cross national borders are difficult to study and not easy to
manage whether the resources are renewable or not. The problems of co-operation and
sharing are similar to those arising from the management of multinational hydrographic
Great aquifer systems of the world 113
basins but with the additional possible problem of the transfer of the impact of exploita-
tion to neighbouring countries.
REFERENCES
ADES (http://www.ades.eaufrance.fr) Institut franais de lenvironnement (2004): Les prlve-
ments deau en France en 2001, Orlans
Akram Kahlown M. and Abdul Majeed (2004) Water Resources Situation in Pakistan : challenges
and future strategies (Pakistan Council of Research in Water Resources, 12p, Islamabad.
Anonyme (2003) Water Resources and Use in Australia. Site farmweb.au.com.
Arajo L.M. & al. (1999) Hydrogeology of the Mercosul aquifer system in the Parana and Chaco-
Parana Basins, South America, and comparison with the Navajo-Nugget aquifer system, USA.
Hydrogeology Journal, vol 7, N.3, June, pp 317336. IAH, Springer Ed.
Table 2. Present groundwater withdrawals in some large aquifer systems in the World.
Aquifer system Index of Country Date Withdrawals Reference
tab. 1 in km
3
/y
Nubian Sandstone 1 Egypt, Libya, 2000 2.2 CEDARE/IFAD, FAO
aquifer system (NSAS) Sudan 2003
with Post-Nubian
Northern Sahara 2 Algeria, Libya, 2000 2.5 OSS 2003
aquifer system (SASS) Tunisia
Murzuk Basin 3 Libya 2000 1.75 A. M. EL GHERIANI,
2002
UNESCO/Project
ISARM 2004
Senegal-Mauritanian 5 Mauritania, 2000 0.26 UNESCO/project
Basin Senegal, Guinea ISARM 2004
Bissau, Gambia
Iullemeden Irhazer 6 Algeria, Mali, 2000 0.22 UNESCO/project
aquifer system Niger, Nigeria ISARM 2004
Lake Chad Basin 7 Niger, Nigeria, 2000 0.25 UNESCO/project
Chad ISARM 2004
Central Valley 16 USA 1990 20 USGS
California
Ogallala Aquifer 17 USA 2000 21.5 USGS
Arabian Aquifer 22 Saudi Arabia, 1995 15.2 FAO 1997
System United Arab A.A. Ghum Alghamdi
Emirates, 2002
Bahrain, Qatar
Indus Basin 23 Pakistan 2000 46.7 IGRAC/M.A.
Kahlown 2004
Huang-Huai-Hai 29 China 1997 20 China Water Vision 2000
Paris Basin 32 France 2000 2 BRGM
IFEN 2004
Great Artesian Basin 3 Australia 2000 0.55 M.A. Habermehl,
UNESCO, 2002
Bakhbakhi M. (2002) Hydrogeological framework of the Nubian Sandstone Aquifer System. GW-
MATE, UNESCO Expert group meeting, Socially-sustainable-management of Groundwater
Mining from Aquifer storage- 6p, Paris.
CEDARE (2002) Regional Strategy for the Utilisation of the Nubian Sandstone Aquifer System.
Volume II, CEDARE, Heliopolis Barhy, Cairo, Egypt (CEDARE: Center for Environment and
Development for the Arab Region and Europe).
CGMW/UNESCO (2000) Geological Map of The World 1: 25,000,000 -2nd edition. Paris
China Water Vision, The Ecosphere of Water, Life, Environment & Development. Ed by Rusong
Wang, Hongsun Ren & Zhiyun Ouyang. China Meteorogical Press, 178p, 2000 Beijing
Diagana B. et A/PNUD/DASDG/RAF/94/01C SAT1 (1997) Gestion des eaux internationales en
Afrique Sub-Saharienne. Bilan diagnostic de la gestion intgre des eaux et des contraintes envi-
ronnementales dans la valle du fleuve Sngal
Dodo A. (1992) Etude des circulations profondes dans le grand bassin sdimentaire du Niger:
identification des aquifres et comprhension de leurs fonctionnements. Thse de Doctorat s Sc.,
Univ. Neuchtel (Suisse), 101p
Dutt D.K. (19871988) Techno-economic Considerations for Groundwater Development in India.
International Symposium on Groundwater Economics, Barcelona, Spain, Oct.1987, 20p. /
Natural Resources Forum, pp159167,UN, New-York.
Habermehl M.A. (2002) Groundwater Development in the Great Artesian Basin, Australia. GW-
MATE UNESCO Expert group meeting, Socially Sustainable Management of Groundwater
Mining from Aquifer Storage, 35p, Paris.
El Gheriani A.M. (2002) The Great Man Made River Project. Colloque Socit Hydrotechniqiue
de France, Eau et Economie, 241244, Paris.
FAO (1997) Irrigation in the NearEast Region in Figures. FAO, Water Reports 9, Saudi Arabia 205
212, Roma.
Ghurm Alghamdi A.A. (2002) Sustainable Management for Deep Aquifer Storage in the Kingdom
of Saudi Arabia. GW-MATE UNESCO Expert group meeting, Socially Sustainable Management
of Groundwater Mining from Aquifer Storage, 9p, Paris.
ICID (1983) Irrigation and Drainage in the World. Vol.III, Sudan, Third ed. 1983, 12621280, New
Delhi.
ICID (2000) Irrigation and Drainage in the World. Chap. Pakistan, 12p. New Delhi.
Lau J.E. and al. (1987) Hydrogeology of Australia. Bureau of Mineral Resources, Geology and
Geophysics, Bull. 227, 21p. Australian Government Publishing Service, Camberra.
Miller J. A. (2000) Ground Water Atlas of the United States, US geological Survey, 404p.
OACT (1993) Safar-Zitoun M./AAC 1993; Notice explicative de la carte hydrogologique interna-
tionale de lAfrique. Feuille 3. Association Africaine de Cartographie, Alger
OSS (2003 ) Systme aquifre du Sahara septentrional. Rapport de synthse. Observatoire du
Sahara et du Sahel, 2003, 129p. Tunis
0SS (2004) Besbes M.& al./(2004) Conceptual framework of the North Western Sahara Aquifer
System. UNESCO/IHP-VI, Series on Groundwater No 8, Managing Shared Aquifer Resources in
Africa, pp. 163169. Paris
Rebouas A. (1976) Le grand bassin hydrogologique du Maranhao. Brsil. Perspectives sur lex-
ploitation. Association internationale des hydrogologues, Budapest, Mmoires de lAIH XI.
pp.448458.
Rebouas A.(1988) Groundwater in Brazil. Episodes, vol.11, No 3, sept. pp.209214.
Schneider J.L.(2001) Carte de valorisation des eaux souterraines de la Rpublique du Tchad (1/1
500,000) et Gologie- Archologie- Hydrologie de la Rpublique du Tchad. 2 vol., 1100 p.
UN (19761990) Groundwater in the Western hemisphere, in the Eastern Mediterranean and
Western Asia, in the Pacific Region, in Continental Asia, in Africa, in Europa. UN-DTCD, Natural
Resources. Water series n 4, 9, 12, 15, 18, 19, New-York.
UNESCO/IAH/CGMW/BGR (2004) Groundwater Resources of the World (special edition
Aug.2004, 1/50,000,000). Groundwater Resources of the World-Transboundary Aquifer Systems
(special edition for the 4th World Water Forum, Mexico, March 2006, 1/50,000,000).
UNESCO / Foster, S.S.D. Loucks, P., Eds. (2006) Non renewable Groundwater Resources. A guide-
book on socially-sustainable management for water policy-makers. UNESCO-IHP publications,
Paris.
Great aquifer systems of the world 115
UNESCO/0SS (2005) UNESCO/055 (2006) Ressources en eau et gestion des aquifres trans-
frontaliers de lAfrique du Nord et du Sahel. ISARM-AFRICA. UNESCO/IHP. IV, Series on
Groundwater No 11. Paris.
UNESCO/Project ISARM (2004) ISARM-Africa, Managing Shared Aquifer Resources in Africa
UNESCO/IHP. VI, Series on Groundwater N.8, 2004, 238 p., Paris
USGS (2003) Principal Aquifers, in the National Atlas of the United States of America, scale1/
5,000,000. USGS, revised 2003
Zekster I.S., Everett, L.G., Eds. (2004) Groundwater resources of the world and their use.
UNESCO, IHP-VI, series on Groundwater, n 6, Paris, 346p.
CHAPTER 9
Alluvial aquifers in the North China plain
Zaisheng Han
China Geological Survey, Beijing, China
ABSTRACT: With an area of 140,000km
2
, the alluvial plain located in northern China has one of
the largest aquifers in East Asia. The recharge of this shallow groundwater system is mainly from
rainfall. There is also recharge from surface water through riverbanks and from irrigated areas.
Groundwater flows from the foot of the mountains to the coastal plain. Isotopic data from deeper
aquifers show that the water flowing from the foot of mountains needs several thousand to over 10
thousand years to reach the middle and coastal plains. Long periods of groundwater overexploita-
tion, especially in the deeper aquifers, not only cause many boreholes to dry up but also create many
eco-environmental problems. A detailed investigation of the aquifers was undertaken. Progress has
been made on sustainable use of groundwater in socio-economic development.
Keywords: Alluvial Aquifers, plain, China, East Asia.
1 GEOGRAPHY AND HYDROLOGY
The North China Plain is one of the focus regions of China. Beijing, the capital of China,
Tianjin city, most of Hebei province and part of Shandong and Henan provinces are situ-
ated in this plain. The alluvial plain also has one of the largest aquifers in East Asia. From
the Taihang Mountains in the west to the Bohai Sea in the east, from the Yan Mountain in
the north to the Yellow River in the south, the surface area of the North China Plain is
140,000 km
2
. The elevation is lower than 100m a.s.l. According to formation and physiog-
nomy, it can be divided into the foothill alluvial plain, the middle plain and the coastal
plain. There are some lakes and billabongs (blind river channels and ponds which are
seasonally filled), especially in the middle and eastern part. The plain belongs to semi-arid
and semi-humid climate regions. The average annual rainfall is 500600mm. The rainfall
and evaporation are asymmetric, and 80% of the rain falls between June and September.
The plain is part of the Yellow River, Hai River and Luan River catchments. These rivers
are almost dry during most of the year due to upstream dams (Han, 2004).
2 QUATERNARY AQUIFERS
The plain is a large Cenozoic sedimentary basin. It consists of alluvial sediments up to
10003000m thick where the Quaternary accounts for more than 600m in the depression
area and 200m at the extremity. Groundwater exists in the Quaternary and Tertiary porous
aquifers. The aquifer system is 350500m thick. The Quaternary aquifers can be divided
into four groups in the middle and the coastal plain. The upper two groups are shallow
aquifers. The lower two groups are deep confined aquifers. The base of the shallow aquifers
is generally at a depth of 4060m. In the foothill plain, with mixed extraction from multi-
layered aquifers, the shallow groundwater systems extend to depths of 120150m. The bot-
tom of the deep confined aquifers is at 350550m. In the middle and coastal plain, saline
aquifers exist between the shallow and deep freshwater aquifers (Foster, 2004).
According to the chemical characteristics of these groundwater systems, there are three
aquifer systems in the North China plain: A) the Luanhe River system, B) the Hai River
system and C) the ancient Yellow River system. They could be subdivided into 7 second-
grade subsystems, 17 third-grade subsystems and 28 fourth-grade subsystems. They form
the basis of the groundwater system analysis and assessment (see Figure 1 and Tables 1
and 2). The Quaternary aquifers of the Luan river system consist of coarse sands. They are
recharged by direct rainfall recharge and surface water. The aquifers of the Hai River sys-
tem are complex. They vary from the foothills to the coastal area.
The cross-section in Figure 2 shows the structure of the Quaternary layers. The shallow
aquifers of the ancient Yellow River system receive significant recharge, but the deep
aquifers which are interbedded with clay receive less recharge.
3 WATER FLOW AND CHARACTERISTICS
The recharge to the shallow groundwater system is mainly from rainfall. There is also
recharge from riverbank filtration and irrigation. The groundwater flows from the foot of the
mountains toward the coastal plain. The coefficients of transmissivity are 5001000m
2
/d in
the foothills, 100500m
2
/d in the middle plain, and generally less than 50m
2
/d in the coastal
plain. The shallow groundwater system discharges by exploitation and evaporation. Due to
shallow groundwater exploitation, evaporation has decreased and recharge has increased in
recent years. The TDS of the shallow groundwater is 0.21.8g/l. The freshwater is distrib-
uted in the foothills, whereas saline water can be found in the coastal area with TDS exceed-
ing 5g/l. The space between them is a transition area (Zhang, 2000).
The deep confined aquifers cannot be recharged directly by rainfall and the flank
groundwater flow may recharge these deep aquifers. Isotopic data of water in the deep
aquifers show that the water from the foothills needs several thousand to more than 10 thou-
sand years to reach the middle and coastal plains. In periods of intensive exploitation, these
aquifers are recharged by bordering shallow aquifers and aquitards. Exploitation such as
pumping constitutes the main discharge of the deep groundwater system. The pumping
mainly concerns water stored in deep confined aquifers and neighbouring aquitards, and
thus causes land subsidence. From Table 1, we can see that this groundwater is mainly of the
Na-HCO
3
.SO
4
and Na.Mg-HCO
3
chemical types. Other geochemical and isotopic parame-
ters of groundwater in deep confined aquifers are also shown in this table.
4 GROUNDWATER RESOURCES
The assessment of groundwater resources is based upon an understanding of the recharge
across a region. The recharge consists of groundwater flow from the mountain area,
infiltration from rainfall, river water and irrigation. Artificial recharge is included. The
exploitable amounts of groundwater were evaluated from the recharge and other environ-
mental factors. The evaluation of the Long Term Average groundwater resource is
described in Table 2. The average amount of annual fresh groundwater recharge in the
North China Plain is 19 10
9
m
3
per year, the recharge to the shallow aquifers is
16.8 10
9
m
3
per year and that to the deep confined aquifers is 2.4 10
9
m
3
per year. The
dissymmetry typically results in an exploitable yield lower than the recharge volume.
Alluvial aquifers in the North China plain 119
Figure 1. Groundwater systems of the North China plain (sub-system A, B, C as in Table 2).
These results are restricted by the technical, economic and eco-environmental conditions.
The deep confined aquifers are recharged by the flow from the basin margin in the west as
well as by leakage from overlying aquifers. However, the rate of horizontal flow is very
slow and vertical leakage is weak. Much of the groundwater in the deep confined aquifers
can be regarded as fossil water with low recharge.
Groundwater resources of the alluvial plain constitute the primary water supply for
urban water, industry and irrigation. In the past 30 years, the exploitation and utilization of
groundwater has been intense in the plain. For example, the amount of groundwater
Table 1. Geochemical and isotopic parameters of groundwater in deep confined aquifers.
Situation Foothill plain Middle plain Costal plain
C (pmc) 24.50111.07 6.2918.40 2.4418.52
18
O() 10.58.7 12.510.2 10.79.5
D () 7265 8774 8175
Cl (mg/l) 5.5762.30 1.64329.60 242.70524.7
18
O temperature (C) 9.4015.06 4.068.43 7.9710.31
3
H(TU) 647 13 16
13
C (PDP) 10.399.28 10.237.75 11.467.92
TDS (mg/l) 342 908 1786
pH 7.7 7.95 7.9
Chemical type HCO
3
-Na.Ca Cl.HCO
3
-Na Cl-Na
Vogel (ka B.P.) 10Modern 2112 3012
Table 2. Groundwater resources in the North China plain.
Sub systems Area Annual Recharge Exploitable
(km
2
) preci. (mm) (Mm
3
/yr) (Mm
3
/yr)
A Luan River system 7000 552 1440 1238
B Hai River system 75340 500 14400 13018
C Ancient Yellow River system 56900 584 3280 3036
Total (or Long Term Average) 139240 537 19120 17292
Figure 2. Profile of groundwater flow system.
extracted in 2000 was 21 10
9
m
3
, with 17.8 10
9
m
3
extracted from the shallow
aquifers and 3.3 10
9
m
3
from the deeper aquifers. The groundwater usage is shown in
Table 3. Since groundwater makes up about two thirds of all the water supplied to the area,
groundwater over-exploitation is appearing in some parts of the plain.
5 GROUNDWATER-RELATED ECOLOGICAL ISSUES
Groundwater has played a key role in the history and development of the plain. The huge
extent of groundwater over-exploitation in the alluvial plain has created a broad range of
largely environmental problems. Long-term groundwater over-exploitation, especially in
deep aquifers, not only causes many boreholes to dry up but also creates many eco-
environmental problems. During the past 30 years, groundwater levels in both the shallow
and deep aquifers in most parts of the plain have fallen by more than 15m. In some urban and
irrigated area, the groundwater levels have fallen even further. For example, groundwater
over-exploitation frequently occurs within Beijing city and Hebei Province where groundwa-
ter levels decline continuously. Figure 3 shows the situation in downtown Beijing. The
Table 3. Groundwater usage in the North China plain (Mm
3
/yr).
Province Total Industry Agriculture Domestic Other
Beijing 2487 436 1532 528
Tianjin 555 15 316 105 119
Hebei 12862 1476 10136 125
Shandong 2756 571 1827 358
Henan 2538 294 1999 245
Total 21198 2792 15801 2486 119
Figure 3. Groundwater level in Downtown Beijing.
North China plain has witnessed the most excessive pumping of groundwater in the world
and covers the largest subsidence area with the most numerous cones of depression. The
groundwater level in the coastal region has fallen below sea level.
Several ecological issues have appeared due to groundwater over-exploitation, includ-
ing land subsidence, sea water intrusion and water quality degradation, which have been
investigated and verified. In 2005, the area of land subsidence of more than 1m reached
8,635km
2
; the area of land subsidence of more than 0.5m reached 30,080km
2
. Most of the
land has subsided by 0.79m in Beijing city, by 3.18m in Tianjin city, by 2.45m in Hebei
Province, by 0.94m in Shandong Province. The areas of subsidence are still growing and
in several parts of the plain they have tended to join up.
Land subsidence caused by groundwater withdrawal causes very serious economic and
social problems. Subsidence in urban areas, such as Tianjin and Cangzhou, affects not only
the natural structures but also man-made ones. Damage, if it could be evaluated, would be
enormous. The main hazards are structural damages, damage to well casing, less efficiency
of storm-drainage facility, submergence of coastal lowland, etc. The disastrous land subsi-
dence leads to tide and water intrusion due to diminished land surface elevation. The rela-
tionship between land subsidence and groundwater levels in deep aquifers in the eastern
part of the North China Plain is demonstrated in Figure 4. The depth to the groundwater
level is between 50 and 70m at the key points of land subsidence. The aquifers dynamic
pressure conditions clearly changed at these water levels.
To control land subsidence, a monitoring network focusing on the North China Plain is
being established. It will monitor the rate of land subsidence as well as the groundwater
level across an area of 70 000km
2
.
The shallow aquifers are polluted by urban and industrial wastewater. The contaminated
area is about half of the plain. Saline water has moved down to the fresh water aquifers due
to excessive exploitation.
6 SUSTAINABLE GROUNDWATER USE
For the social and economical development in the North China Plain, one of Chinas most
important regions, sustainable use of groundwater resources is imperative (Liu, 2001).
Figure 4. Relationship between land subsidence and the groundwater level of deep aquifers
(Modified from He, 2005).
A detailed investigation of the aquifers has been undertaken. A numerical model of the
plain with the software FEFLOW of WASY has been established. All the groundwater
models are based on Darcys law, which was presented in 1856. The 3-D model is used to
simulate and forecast the situation of both shallow and deep aquifers (Figure 5). The data
bases of the aquifers are established. They consist of basic geographic information, geo-
logical information, hydrogeological conditions, groundwater quantity and quality infor-
mation and dynamic monitoring data, etc. The data of the years 20002003 are used to
establish and calibrate the model. The hydrogeological conceptual model covers the whole
plain with a surface area of 140, 000km
2
and a depth of 550m. Several measures for
groundwater conservation are evaluated by the model. The simulated water table contour
lines of shallow aquifers are shown in Figure 6.
Figure 5. Numerical model of the North China plain.
The natural water flow has been interrupted by surface reservoirs in the rivers upstream.
The alluvial aquifers are recharged only in the flood seasons. Groundwater extraction
exceeds recharge. Figure 7 shows that the groundwater level in the foothills of the Taihang
Mountains rose again after recharge by rainfall. Artificial recharge is the most important
means of protecting groundwater and induces aquifer recovery. Aquifer recharge enhance-
ment by excess surface runoff and urban wastewater are water resource safeguard meth-
ods. There are large areas with coarse sands and gravel outcrops only in the foothill plain
where artificial surface recharge would be effective.
Figure 6. Simulated water table contour lines of the shallow aquifers.
Reducing groundwater withdrawal, by a variety of means, will generally be the most
costeffective method of solving groundwater-related eco-environmental problems. A
long-distance water transfer project from the Yangtze River to the North China Plain will
be implemented in 2010. By a combined use of surface water and groundwater, the
exploitation of groundwater in deep confined aquifers will decrease. Some surface runoff
would be resumed and the hydrological environment should improve.
Agricultural sustainability in the North China Plain depends heavily on water resource
availability. Over twenty percent of the plain is irrigated by groundwater. Now it is impera-
tive to demonstrate that new practices aimed at conserving groundwater resources and
tested on small plots are effective commercially to ensure the agricultural sustainability
needed to feed Chinas increasing population. Reducing groundwater withdrawal for irriga-
tion, together with measures to save water, is the preferred option (Zhang and Fei, 2003).
To use local wastewater and brackish groundwater more efficiently is one way to save
freshwater. Treatment and reuse of urban and industrial wastewater, which amounts to
7 10
9
m
3
per year in this area, would also be beneficial. Moreover, new groundwater
well fields could supply water in an emergency. Utilization of other water resources, such
as surface reservoirs and water transferred from other catchments, must also be regulated.
To sum up, sustainable groundwater use in the North China Plain could be achieved by
integrating all available management options.
7 CONCLUSION
The alluvial aquifers in the North China Plain are an important source of water supply in
this intensively developed area. A detailed hydrogeological investigation of the aquifers
Figure 7. Groundwater levels at the riverside and monthly precipitation.
was undertaken. Several eco-environmental issues have arisen due to groundwater over-
exploitation. Measures to ensure sustainable groundwater use have been evaluated and
brought into effect, including aquifer recharge enhancement, saving of irrigation water and
more effective use of local wastewater and saline groundwater.
ACKNOWLEDGEMENT
Many thanks to the scientific committee of the Dijon International Symposium. The
Chairman Ghislain de Marsily and two reviewers gave us valuable comments and sugges-
tions to improve the original manuscripts. Mr. Lifeng Hu and Miss. Xu Han of Beijing
University of Posts & Telecommunications helped me revise this paper in the 2007 New
Year Holiday.
REFERENCES
Foster, S., Garduno, H., Evans, R., Olson, D., Tian, Y., Zhang, W., Han, Z. (2004) Quaternary Aquifer
of the North China Plainassessing and achieving groundwater resource sustainability,
Hydrogeology Journal, Springer-Verlag GmbH, Vol.12, No. 1, 8193 pp
Han Zaisheng (2004) Groundwater Investigation in China. Research Basin and Hydrological
Planning, A.A. Balkema Publishers, 165169 pp
He Qingcheng (2005) Land subsidence in the North China Plain, Proceeding of the 7th International
symposium on land Subsidence, Shanghai Scientific & Technical Publisher, 1829 pp
Liu Changming, Yu Jingjie (2001), Groundwater Exploitation and its Impact on the Environment in
the North China Plain, Water International, 2(26), 261272 pp
Zhang Guanghui, Fei Yuhong (2003) Groundwater evolvement and countermeasure in Hai river
plain, Academic Publishing House [in Chinese]
Zhang Zonghu, Shen Zhaoli, Xue Yuqun et al. (2000) Evolution of Groundwater Environment in the
North China Plain, Beijing Geological Publishing House [in Chinese]
CHAPTER 10
Groundwater and the challenges for the future water supply
of the Republic of Djibouti in an arid climate (Horn of
Africa)
Mohamed Jalludin
Centre dEtude et de Recherche de Djibouti (CERD), Djibouti, Rpublique de Djibouti
ABSTRACT: Located in the Horn of Africa where the separation of several tectonics plates
occurs, the Republic of Djibouti has an arid climate. The country continuously has to face difficul-
ties in its water supply which is 95% derived from groundwater. 29.4 million m
3
per year are pro-
duced from volcanic and sedimentary aquifers of which 5.7% is used by the rural population and for
cattle farming, 42.5% for irrigation and 51.7% for urban areas. Given the problems of falling water
tables, water quality degradation and high TDS contents, alternative solutions have been used, i.e.,
surface water and desalination of high TDS-content aquifers. Nevertheless, the projected economic
development programmes and population growth will continue to increase groundwater demand
over the next ten years. Future water requirements, estimated for 2010 and 2015, amount to 41 and
51 million m
3
/y. The production rate increase for 19772005 was between 0.55 and 0.7 million m
3
/y.
To fulfil the demands for water over the next ten years, the production rate increase must reach 2.17
million m
3
/y, three times more than the present production rate increase. The new phase of ground-
water resource development must inevitably be followed by management and protection of the
aquifer systems and not just the search for new groundwater resources.
1 INTRODUCTION
The Republic of Djibouti (23,000km
2
) is located in the Horn of Africa. 83% of the total
population of 60,0000 inhabitants live in urban areas. Given the arid climate conditions,
the freshwater supply is based mainly on groundwater which is available in volcanic and
sedimentary aquifers. Since the urbanisation of the 1960s, and with the development of
irrigated land and livestock, groundwater production has now reached around 29 million
m
3
/y. The intensive exploitation of some aquifer systems has led to water table decreases
and/or increased mineralization of pumped water as a result of the lack of management
and/or insufficient groundwater recharge.
Considering the population increase (national growth rate of 3%), economic develop-
ment, particularly of the Doral port, agricultural development, increase of livestock and
the cattle pen project, the freshwater supply for the next five to ten years presents a serious
challenge (MAEPH, 2000). Although solutions such as desalination and surface water use
are programmed or applied in specific situations, groundwater exploitation will continue
to satisfy most of the water needs.
In the first two parts of the paper, the aquifer systems and the means of groundwater
production are described and analysed. The third part deals with current groundwater
exploitation and the role of the aquifer systems in water supply. Finally, the last part analy-
ses different possible scenarios for water needs and the solutions that may be foreseen for
the next ten years.
2 THE VOLCANIC AND SEDIMENTARY AQUIFER SYSTEMS
The geodynamical situation in the region is related to the plate tectonic movements since
2530 My. The first volcanic outcrops, Adolei basalts, correspond to the first splitting of
the Arabian and the Nubian blocks. This phase was followed by a low expansion period
where the Mabla rhyolites (15 My) were formed. After an erosion period, the Dalha basalts
(3.49 My) were formed and outcrop in discordance over the old series. Somali basalts are
contemporaneous with Dalha basalts and occupy the eastern part of the Ali Sabieh block.
Between 3.4 and 1 My, plate expansion created the Afar Depression which is filled by stra-
toid basalts and the ridge of Tadjourah, which is characterized by Gulf basalts. In the Ali
Sabieh area, Jurassic limestone and Cretaceous sandstone form sedimentary rocks before
the tectonic blocks split. Sedimentary rocks are found on the coastal plains, in the tectonic
basins, along the wadi beds and intercalated in the volcanic series. They are represented by
coarse alluvial and marine deposits in the coastal plains. Lacustrine and fluviatile series
related to the successions of humid and arid climates form the sedimentary sequences of
the endorheic basins.
The aquifer classification is based on the geometry, the dimension, the type of aquifer
and the recharge mode (BGR, 1982; Jalludin and Razack, 2004). Recent sedimentary
rocks deposited along the wadi beds represent the alluvial aquifers. Their width varies
between some tens of meters and some hundreds of meters and their length can exceed
several tens of kilometres. Generally their thickness remains under 10m but in certain situ-
ations it can reach several tens of meters. They play an important role in the recharge of
underlying aquifer systems during run off periods. The alluvial aquifers are unconfined
and the water table is generally located some meters below ground level.
The category of local aquifers, recharged by alluvial aquifers, comprises both sediment-
ary and volcanic (Fig. 1). Local sedimentary aquifers are represented by the coastal sedi-
mentary plains and the sedimentary basins and have a total area of around several
hundreds of km
2
. The local volcanic aquifers exceed 500km
2
in area excluding the Adolei
basalt and the Southern Mabla rhyolite aquifers. The local aquifers are unconfined and the
water table varies between some meters and some tens of meters below the surface.
The stratoid basalt series cover all of the western part of the country with an area of 10,
900km
2
and extend into Ethiopia. Recharge to the regional volcanic aquifer is controlled
mainly by underground flows from the downstream Awash valley region in Ethiopia (BGR
1982). The aquifer is confined and feeds several more or less mineralised and/or hot
springs in the northern part of the Hanl and Gobaad basins.
The mean transmissivity of the Quaternary sediments is 0.004m
2
/s (N 25; STD
0.0032) while it is 0.005m
2
/s (N 1) for the Cretaceous sandstones. The volcanic aquifer
transmissivity varies according to the age of the formation with regard to hydrothermalism
and weathering effects (Jalludin and Razack, 2004). The Gulf basalt and stratoid basalt
transmissivities vary between 0.00015 and 0.314m
2
/s (N 26; STD 0.066). In Dalha
basalts, T ranges between 0.00015 and 0.01m
2
/s (N 21; STD 0.0034). T ranges
between 0.0003 and 0.061m
2
/s (N 8; STD 0.022) for the Somali basalts and between
0.00006 and 0.002m
2
/s (N 4) for the old Adolei basalts. The transmissivities of recent
rhyolites from the stratoid series and Mabla rhyolites are respect-ively from 0.015 to
0.15m
2
/s (N 3) and from 0.002 to 0.034m
2
/s (N 5).
The groundwater is characterized by relatively high TDS contents, 2g/l (Houssein
and Jalludin, 1996). Low TDS values are observed in alluvial aquifers, in mountainous
Groundwater and the challenges for the future water supply 129
KM
40 20 0
Local aquifer of the old sediments
Local sedimentary aquifers
Local rhyolitic aquifers
Local basaltic aquifers
Regional volcanic aquifer
Supposed boundary of the regional aqu
ETHIOPIA
ERYTHRE
A
SOMALIA
REP.OF
DJIBOUT
I
Hanle
Gobaad
Bissidourou
Obock
Nagad
Ali Sabieh
Djibouti
Tadjourah
Arta
Dikhil
Asal
Abhe
Figure 1. Aquifer systems of the Republic of Djibouti.
0
5
10
15
20
25
30
35
40
1
9
3
9
1
9
4
4
1
9
4
9
1
9
5
4
1
9
5
9
1
9
6
4
1
9
6
9
1
9
7
4
1
9
7
9
1
9
8
4
1
9
8
9
1
9
9
4
1
9
9
9
2
0
0
4
N
u
m
b
e
r

o
f

w
e
l
l
s
0
50
100
150
200
250
300
350
400
C
u
m
u
l
a
t
e
d

n
u
m
b
e
r

o
f

w
e
l
l
s
Number of wells
Cumulated number of wells
Figure 2. Evolution of the number of wells.
zones and in certain local aquifers. Most of the catchments show a regular increase of TDS
with time. Recent geochemical studies demonstrate that high TDS water underlies fresh
groundwater (Bouh, 2006). This is due to old marine intrusions in the coastal aquifer sys-
tems which emerged during vertical movements during the plate tectonic splitting
processes. In the continental part, the more mineralised water results from evaporation of
groundwater and lakes in the last arid climate period during the Holocene. Groundwater
generally contains sodium chloride and/or alkaline earths. There is a bicarbonate trend in
the alluvial aquifers and a sulphate one in the sedimentary aquifers of the basins.
3 GROUNDWATER PRODUCTION MEANS
The country has 371 wells of which only a hundred are effectively exploited for urban and
rural water supply (ONED, 2004; BGR, 1982; MAEPH, 1997). The remaining wells are
abandoned, unproductive or characterized by high TDS. The evolution of the number of
wells throughout the country has not been regular in the past (fig. 2). Overall, a relatively
regular increase can be seen from the 1960s until the maximum in 1989, followed by a
decrease until 1994 and again an increase from 1998 onward. One hundred wells were
drilled in 1980, 200 in 1987 and 300 in 1992. The total number of water wells is set to
increase given the current water resource projects. 66.4% of the wells are in volcanic rocks
and 33.6 % in sedimentary formations. The cumulated depths of the wells in each aquifer
are shown in Figure 3. The total cumulated depths in the Republic of Djibouti amount to
over 33km. In rural areas, the water supply is based mainly on shallow large-diameter
wells (1 000 to 1 200) and drilled wells (Labonne, 1987).
Yields in the range of 12m
3
/hour for rural needs and of 1520m
3
/hour for urban wells
are considered a success. The best yields are obtained in recent sediments, Gulf basalts and
the stratoid basalts (Fig. 4; Fig. 5). The global success rate calculated for all the wells is
69.4%. The low success rate for the basin sediments might be explained by the lacustrine
and clayey deposits. High TDS wells are encountered in practically all the aquifer systems
but seem to be more frequent in sedimentary aquifers, 21.3%, against 10.5% in volcanic
aquifers. The individual success rate of sedimentary and volcanic aquifers is respectively
59.8% and 74.9%.
0
10
20
30
40
50
60
70
80
50 100 150 200 250 300
Interval of depths
N
u
m
b
e
r

o
f

w
e
l
l
s
Sediment
Volcanic
0.72
0.91
4.40
3.60
0.38 0.35 0.33
8.51
2.74
7.47
2.27
0.53
0
2
4
6
8
10
G AR SC SB BA RM BM BD BS BG BST RST
C
u
m
u
l
a
t
e
d

d
e
p
t
h
s

k
m
G: cretaceous sandstones; AR: recent alluvium; SC: coastal sediment; SB:
basin sediment; BA: Adolei basalts; RM: Mabla rhyolite; BM: Mabla basalts;
BD: dalha basalts; BS: Somali basalts; BG: Gulf basalts; BST: stratoide
basalts; RST: stratoide rhyolite
Figure 3. Cumulated depths of wells for each aquifer system.
4 PRODUCTION OF GROUNDWATER RESOURCES
The highest groundwater withdrawal rates occur around Djibouti and for irrigation in rural
areas (Fig. 6). In third place comes production for the smaller towns. Production for the
city of Djibouti has been more or less regular since 1920 but has increased significantly
since the 1960s. This is related to the development of the city and the new groundwater
extraction in the Gulf basalts. Current production is around 14 million m
3
/y (ONED,
2004). An inventory by the agricultural department (MAEPH, 1995) revealed the exist-
ence of 1 000 ha of irrigated land in several parts of the country. Contrary to the most
intensive groundwater pumping which began in 1960, agricultural development started
just after independence in 1977 (Fig. 6). The total irrigated volume estimated for 1998
exceeded 12 million m
3
/y. The production of groundwater for irrigation is of the same
order of magnitude as the production of groundwater for urban zones.
The groundwater production for the population and livestock in rural zones is relatively
small compared to the irrigation sector, which uses 88%. The water consumption of the
rural populations and livestock (around 1 million animals) is less than 2 million m
3
/y.
0
10
20
30
40
50
60
70
80
G SR SC SB BA RM BM BD BS BG BST RST
N
u
m
b
e
r

o
f

w
e
l
l
s
Positive wells
Negative wells
High TDS waters
Figure 4. Positive, negative and high TDS wells (see legend on Fig 3).
76
24
32
27
23
164
0
20
40
60
80
100
120
140
160
180
Sedimentary Volcanic
N
u
m
b
e
r

o
f

w
e
l
l
s
Positive
Unproductive
High TDS
5
9
.
8
1
8
.
9
1
4
.
6 2
1
.
3
1
0
.
5
7
4
.
9
0
10
20
30
40
50
60
70
80
Sedimentary Volcanic
P
e
r
c
e
n
t
a
g
e
Figure 5. Success rates in sedimentary and volcanic aquifers.
Most of the 29.3 million m
3
/y pumped by the Republic of Djibouti come from the sedi-
mentary and volcanic aquifers (Fig. 7). The volumes from the shallow wells cannot be dis-
tributed according to the aquifer because of the lack of lithological data for the wells. Most
of them tap the recent alluvium in the wadi beds but a number of them reach underlying
aquifers where the recent alluvium layer is thinner.
The most exploited aquifers are those in the Gulf basalts supplying the town of Djibouti
(Gamal, 1988; Jalludin & Razack, 1997) and the alluvial aquifers exploited in rural areas
mainly for irrigation (Fig. 7). The total annual volume pumped in these aquifers is far higher
than in the other aquifers. It represents 86.9% of the total annual production. The Somali
basalt aquifer occupies the third position with 1.31 million m
3
/y mainly destined for the town
of Djibouti. The Dalha basalt and coastal sediment aquifers are next in importance with
respectively 0.7 and 0.65 million m
3
/y for the urban zones. The Dalha basalt aquifer is more
heavily exploited than the coastal sedimentary aquifers in rural zones. The high TDS aquifer
in the Cretaceous sandstone provides 0.46 million m
3
/y to the town of Ali-Sabieh, after
desalination. The other aquifer systems: Adolei basalt, Mabla basalt, Mabla rhyolite, Stratoid
basalt and the basin sediments contribute only 0.1 million m
3
/y to the water supply.
A
0
2000000
4000000
6000000
8000000
10000000
12000000
14000000
1900 1925 1950 1975 2000
A
n
n
u
a
l

v
o
l
u
m
e

m
3
B
0
20
40
60
80
100
120
140
160
180
19601965197019751980198519901995
I
r
r
i
g
a
t
e
d

a
r
e
a

h
a
0
2
4
6
8
10
12
14
C
u
m
u
l
a
t
e
d

a
n
n
u
a
l

v
o
l
u
m
e
.
M
i
l
l
i
o
n
s

m
3
Irrigated area
Cumulated annual
volume
Figure 6. Evolution of the production for: A) Djibouti town; B) irrigation.
1
3
.
6
0
.
0
1
2
0
.
0
2
5
0
.
0
1
2
0
.
0
4
9
0
.
2
5
0
.
0
1
2
0
.
0
8
6
0
.
0
7
4
0
.
0
3
7
0
.
0
3
7
0
.
0
8
6
0
.
4
6
0
.
7
0
1
.
3
1
1
1
.
8
9
0
.
2
2
0
.
6
5
0.01
0.1
1
10
100
Puits G BA BM RM BD BS BG BST AR SC SB
Rural. Millions m
3
/an
Urban. Millions m
3
/an
G: cretaceous sandstones; AR: recent alluvium; SC: coastal sediment; SB
basin sediment;BA: Adolei basalts; RM: Mabla rhyolite; BM: Mabla basalts
BD: dalha basalts; BS: Somali basalts ; BG: Gulf basalts; BST: stratoide
basalts ; RST: stratoide rhyolite
Figure 7. Aquifer systems exploitation.
5 POSSIBLE SCENARIOS FOR FUTURE WATER NEEDS
Given the difficulties of the Republic of Djibouti to ensure fresh water supply for the dif-
ferent sectors today, the volumes needed for the next ten years raise questions over which
water resources can be developed.
Concerning the water supply of the towns, the study by Lavalin-Tractebel (1993) projects
the demand for water until 2015, based mainly on the growth rate, i.e., 3%. The greatest need
concerns the town of Djibouti with 25 million m
3
/yr. The proposed solutions for this situa-
tion are seawater desalination and the construction of a dam for surface water impoundment
(BCEOM, 2005). Another option for the town of Djibouti is exploitation of the Stratoid basalt
regional aquifer but this will demand significantly more time than the previous solutions as
the aquifer potential has not yet been evaluated. Secondly, groundwater from the regional
aquifer contains an excess of fluoride according to the WHO standards (BGR, 1997). The
total needs for the urban zones in 2015 will be 28.5 million m
3
/y (Fig. 8).
Water supply in the rural areas also remains a major problem. In 2015, the livestock
water needs will have increased by around 12% without taking into account the needs aris-
ing from the cattle exportation project near Djibouti. Rural community needs and the
demand from livestock would be 3.2 million m
3
/y. Compared to the total production these
water demands remain below 6.5%. Nevertheless, these volumes are not negligible and
will require the creation of new shallow and deep wells. Surface water exploitation represents
an increasingly realistic option for the rural water supply.
Although the agricultural sector is relatively limited, it represents an important compon-
ent of the water demand with more than 12 million m
3
/y used to irrigate 1 000 ha. The
additional needs of this sector over the next ten years depend on further extensions of the
irrigated areas. The current production of 12.5 million m
3
/y might, therefore, reach 19.3
million m
3
/y by 2015 (Fig. 8). Such a large increase in the withdrawal of water for irriga-
tion would definitely intensify the exploitation of some aquifers and new aquifers would
have to be tapped, for instance in the area of Bissidourou or the Hanl plain.
The estimated total needs for 2015 is 51 million m
3
/y. Hence, the production must be
increased by 21.5 million m
3
/y compared to the production of 2005. As a consequence, the
Republic of Djibouti will have to double its total production by 2015.
The mean annual production rate increase for the town of Djibouti calculated since 1960
is 0.3 million m
3
/y (table 1). For the period 1977 to 2005 and for the whole country, the
mean annual production rate increase was estimated at between 0.55 and 0.7 million m
3
/y.
To achieve the objective of securing a water supply in 2015 without any shortage, the
0
2
4
6
8
10
12
2010. Total: 11.32 millions m
3
2015. Total: 21.55 millions m
3
1
.
7
2
.
6
3
.
2
1
5
.
2
2
2
.
9
2
8
.
5
1
2
.
5
1
5
.
2
1
9
.
3
2
9
.
4
4
0
.
8
5
1
0
10
20
30
40
50
60
2005 2010 2015
A
n
n
u
a
l

v
o
l
u
m
e
.

M
i
l
l
i
o
n
s

m
3
Rural
Urban
Irrigation
Total
Ground water Surface water Desalination
A
n
n
u
a
l

v
o
l
u
m
e
.

M
i
l
l
i
o
n
s

m
3
B
A
Figure 8. Future water needs: A) per sector; B) additional needs per source.
Republic of Djibouti will have to maintain a production rate increase of 2.17 million m
3
/y.
This objective represents a serious challenge both scientifically and financially given that,
in the next few years, the mean production index might be multiplied by a factor of
between 3.1 and 3.9.
However, when these rates are calculated separately for the town of Djibouti, the rural
areas and for irrigation, it appears that they are dominated by the town of Djibouti, 1.18
million m
3
/y, due to its size and its current water deficit which is estimated to be between
3 and 5 million m
3
/y. Irrigation takes second place with 0.68 million m
3
/y. The mean
annual production rates for the rural areas (populations and livestock) and the other towns
are similar, i.e., 0.15 million m
3
/y.
About 1.5 million m
3
/y needed by the rural communities and their livestock might be
covered by groundwater but over 70%, the remainder could be supplied by surface water.
The surface water exploitation infrastructures are already operational in the northwestern
part of the country and similar ones would be useful in other silty plains. Carrying out pre-
liminary hydrological studies is essential prior to building such infrastructures.
By 2015, the agricultural sector will require an extra 6.84 million m
3
/y for irrigation.
According to the projected programmes, these volumes would be provided by ground-
water. The additional water needs for Djibouti town will reach 11.8 million m
3
/y. The Gulf
basalt aquifer may not be exploited for this extra volume as it is already over-exploited and
the proposed options of the Nagad dam and the desalination of seawater will not be oper-
ational before 20082010 at the earliest. The additional needs until 2010, 6.8 million m
3
/y,
may still be covered by groundwater. These groundwater resources will require new hydro-
geological studies of the upper part of the aquifer and its western zones as well as of its
adjacent aquifers. If the projected dam and desalination are operational by 20082010,
they would cover the water needs of Djibouti town and contribute to reduce the pressure
on the Djibouti aquifer.
6 CONCLUSION
Water supply in the Republic of Djibouti from volcanic and sedimentary aquifers currently
represents 29.4 million m
3
/y. This intensive groundwater withdrawal that started in the 1960s
results in over- exploitation of several aquifers: water table lowering and deterioration of the
Table 1. Mean annual production rate (million m
3
/y)
Mean annual production rate increase
Period before 2005
Djibouti town. 19602005 0.3
Republic of Djibouti. 19772005 0.55 to 0.7
Period 20052015
Republic of Djibouti 2.17
Djibouti town 1.18
Other towns 0.15
Rural zones 0.15
Irrigation 0.68
water quality by seawater intrusion and upwelling of high TDS groundwater. Together with
groundwater, solutions such as desalination and surface-water impoundment are already
applied and/or projected for new projects to satisfy the increasing demands for fresh water.
51 million m
3
/y represent the predicted water needs by 2015 considering the population
growth and the programmes for economic development. The mean annual production rate
increase estimated to 0.7 million m
3
/y in the past might become 2.17 million m
3
/y for the next
ten years to fulfil the new demands for fresh water. Alternative solutions will contribute to sat-
isfy part of the water requirements and groundwater will still remain the most important
source of water, which will result in more intense aquifer exploitation.
In this situation, it is essential to apply the measures recommended by the Master Plan for
water resources for the management and protection of groundwater resources. Evaluations
of the groundwater resources would be useful to manage adequately the aquifer systems.
New hydrogeological studies must, therefore, be undertaken to pursue these objectives and
to develop new groundwater resources for urban and rural needs. At the same time, a com-
plete economic analysis is necessary to better define the new water requirements.
REFERENCES
BCEOM (2005) Plan damnagement intgr du bassin versant de loued Ambouli. Proposition pour
la construction du barrage de Nagad. MAEPH Rapport
BGR (1997) Forages de reconnaissance dans la rgion de Hanl. ONED Coopration
Hydrogologique Djibouto-Allemande Rapport
BGR (1982) Inventaire des resources en eau de la Rpublique de Djibouti. Gnie Rual. Coopration
Hydrogologique Allemande 5 vol.
Bouh Houssein (2006) Etude de laquifre des basaltes du Golfe et des aquifres volcaniques adja-
cents. Approche hydrochimique et isotopique. Thse de Doctorat. Universit de Paris Sud.
Gamal Eldin HA (1988) Etude de laquifre basaltique de la zone de captage de la ville de Djibouti.
Thse de doctorat, Universit Albert-Ludwig de Fribourg, p.200
Houssein I and Jalludin M (1996) The salinity of Djiboutis aquifer. Journal of African Earth
Sciences, 22, 4 , 409414
Jalludin M and Razack M 2004 Assessment of hydraulic properties of aquifer systems in the
Republic of Djibouti. Horn of Africa. Hydrogeology Journal, 12, 159170
Jalludin M et Razack M (1997) Modlisation de laquifre fissur basaltique. Nappe de Djibouti.
5
me
Assemble Scientifique de lAssociation Internationale des Sciences Hydrologiques. Rabat
Maroc 23/043/05
Labonne M (1987) Projet dencadrement de llevage pastoral dans le Nord de la Rpublique de
Djibouti. MAEPH, p.44
Lavalin et Tractebel (1993) Plan Directeur dalimentation en eau potable des centres urbains. ONED
2 vol.
MAEPH (2000) Schma Directeur de lEau. Ministre de lAgriculture, de lElevage et de la Pche,
charg de lHydraulique. 3 rapports
MAEPH (1997) Ministre de lAgriculture, de lElevage et de la Pche, charg de lHydraulique
(1997) Inventaire des forages deau Rapport
MAEPH (1995) Rpertoire des exploitations agricoles. Ministre de lAgriculture, de lElevage et
de la Pche, charg de lHydraulique Service de lAgriculture et des forts Rapport
ONED (2004) Rapport dactivit de lanne 2004. Office National des Eaux de Djibouti
CHAPTER 11
Hydrogeological structure of the aquifer system in the
Taoudenni sedimentary basin, Burkina Faso
Johan Derouane
1
& Denis Dakour
2
1
SOFRECO-SAWES, Programme VREO, Bobo-Dioulasso, Burkina Faso
2
MAHRH-DGAEP, Programme VREO, Bobo-Dioulasso, Burkina Faso
ABSTRACT: South-western Burkina Faso and southern Mali share an aquifer system that forms part
of the large West-African sedimentary basin of Taoudeni. With a surface area of around 40,000km
2
inside Burkina Faso and a depth that can reach over 2000m, this sedimentary aquifer system provides
most of the water for the human (drinking water) and economic (mainly agricultural) needs of the coun-
try. In the present semi-arid conditions characterised by a considerable decrease in the rainfall during the
last 40 years and a demographic explosion resulting in a greatly increased need of water for both human
consumption and economic development, the management of this important groundwater resource has
become a priority not only for Burkina Faso but also for the international community. A hydrogeologi-
cal, multi-disciplinary approach, combining, among others, stratigraphy, hydrodynamic field data, geo-
chemistry and isotopic techniques with geological and hydrogeological modelling was used to improve
the knowledge of the aquifer and the efficiency of its management. While the geochemical and isotopic
techniques employed in the study area do not provide convincing arguments in favour of the existence
of individual aquifers in the separate geological formations, the hypothesis of vertical hydraulic continu-
ity throughout the Burkina Faso part of the sedimentary basin is still questioned as well the estimated
effective recharge rate. The piezometry of the system indicates that the drainage basin locally discharges
into the main Burkina Faso rivers. However, in view of the low relief of the topography as compared to
the base level depth of the flow system, regional flow with major discharge into the Bani and Niger rivers
in Mali cannot be excluded according to Toths (1963) model. On the basis of the results and interpreta-
tions of field data, a conceptual representation of the hydrogeological situation was put together and built
into a geological (aquifer geometry) model. Further studies on both sides of the frontier with Mali
should contribute to the validation of some assumptions that are still being considered as conceptual
hypotheses as well as improvement of our knowledge of this important aquifer system.
1 MANAGEMENT AND EXPLOITATION OF GROUNDWATER IN
BURKINA FASO
Burkina Faso is situated in the centre of West Africa and has a population of around 12 mil-
lion of which 80% live in a rural environment. It is arid or semi-arid and its surface area is
274,000 km
2
. The main part of its national territory is a laterite plateau with a mean eleva-
tion of 300m sloping from the north-east to the south-west. The climate is Sudanese-
Sahelian and is characterised by considerable spatial and year-to-year rainfall variations,
from 350mm/yr in the north (Sahelian climate) to 1100mm/yr in the south-west (south
Sudanese climate). The rainy season is spread over three months in the north and over
seven months in the south-west. The hydrographic network consists of four national water-
shed basins (Como, Mouhoun, Nakamb and Niger) and has two major permanent rivers:
the Como and the Mouhoun. On average, over the last 40 years, the surface-water flow
represents around 4% of the rainfall while infiltration accounts for some 10 to 15%. The
difference, i.e., over 80% of the rainfall, is taken up by evapotranspiration (Bro, 2001). The
demand for water, estimated at 505 million m
3
/yr is primarily intended for irrigation
(64%), domestic consumption (21%) and livestock (14%). All other requirements, includ-
ing those of industry are relatively minor.
Groundwater provides 85% of the water for human consumption in Burkina Faso while
only 15% are supplied by surface water. These figures are much higher than the world
average and demonstrate the urgent need to preserve the groundwater resources for both
quantitative and qualitative reasons. They justify the concern shown by the international
community and expressed by its support for the management of the natural resources of the
country. Overall, the Burkina Faso aquifers can be divided into two large groups of forma-
tions: firstly, the crystalline bedrock, which covers 82% of the territory, but has a poor
exploitation potential (low flow rates, frequent dry boreholes), secondly, the sedimentary zone
in the extreme south-eastern part of the Taoudeni basin. Although this represents less than
20% of the surface area of the country, it contains a great multi-layered aquifer system, which
offers a large production potential: it provides considerable support for the economic
development of the country.
2 HYDROGEOLOGICAL DESCRIPTION OF THE AQUIFER SYSTEM IN THE
TAOUDENI SEDIMENTARY BASIN, BURKINA FASO
West Africa consists of a vast Precambrian craton, which has remained stable for approxi-
mately 1600 My and is surrounded by mobile zones where tectonic activity occurred during
the Upper Precambrian and the Palaeozoic. The Upper Precambrian and Palaeozoic cover of
the craton, which started to form some 1000 My ago and ceased in the Carboniferous period,
and is divided into three main basins (the basins of Tindouf, Taoudeni and Volta) whose ero-
sion limits give only a vague indication of its original extent. The Taoudeni basin is the largest
sedimentary basin of Upper Precambrian and Palaeozoic age in Africa. It stretches across
Mali, Mauritania and the two Guineas and overlaps slightly into Algeria, Senegal, Sierra
Leon and Burkina Faso. Overall, its structure is simple: the most recent formations crop
out in its sandy centre and the dip rarely exceeds 1.
The south-eastern margin of the Taoudeni basin coincides with the right bank of the river
Niger which cuts through the basin for almost 1500km along a west-east axis. The surface
area of the south-eastern margin of the sedimentary basin (Mali and Burkina Faso) is close
to 260,000km
2
of which 40,000km
2
(i.e., less than 20%) belong to Burkina Faso. It is par-
tially covered by surface formations of Tertiary and Quaternary age among which are the
so-called Continental Terminal Tertiary formation as well as the recent and modern clay-
sand alluvia found along the main rivers.
In the Burkina Faso part of the basin, nine formations have been identified from the
bottom up (Ouedraogo, 1998): Lower Sandstone (LS), Kawara-Sindou Sandstone (KSS),
Glauconitic Fine Sandstone (GFS), Sandstone with Quartz Grains (SQG), Guena-
Souroukoundinga Siltstone, Argillite and Carbonate rocks (SAC 1), Fine Pink Sandstone
(FPS), Samandeni-Kibani Sandstone, Argillite and Carbonate rocks (SAC 2 or Toun
schist), Fo Siltstone and Quartzite (SQ or Koutiala sandstone) and Fo-Bandiagara
Sandstone (FBS). These formations, mainly dominated by sandstones, are piled up in
nearly horizontal layers with a slight dip toward the north-west (around 2) and generally
thicken toward the west. The Taoudeni sedimentary basin has an elevation ranging
between 790 and 240m in Burkina Faso. The highest points are located in the extreme
south-west while the lowest zone is the Sourou plain in the north-east of the study area.
The principal river in the Burkina Faso sedimentary zone is the Mouhoun; it rises inside
the sedimentary zone and crosses it in a west-east direction over a distance of approximately
350km. The other main perennial rivers in the zone are the Como and the tributaries of the
Bani on the Mali border. The sedimentary series has a crucial supporting role during low flow
in these rivers. The mean annual rainfall rate in the Burkina Faso sedimentary zone varies
according to latitude between 600mm in the extreme northern part and 1100mm in the south.
The great spatial variation in the rainfall rate, together with that of potential evapotranspira-
tion, indicate a steep gradient in recharge to the aquifers. Moreover, the evolution in time
of mean 10-year rainfall rates reveal a current deficit of nearly 20% compared to the values
recorded in the 1960s (Dakour, 2003).
The hydrogeological characteristics of the Burkina Faso sedimentary series can be deter-
mined by interpreting, among others, the available quantitative and qualitative data collected
in the study area. There are over a thousand drilled wells for which data are available, either
for statistical analyses by formation or analysis of the whole basin. Note that the distinctions
Hydrogeological structure of the aquifer system in the Taoudenni sedimentary basin 139
Figure 1. Overall view of the south-eastern margin of the Taoudeni basin.
between the nine formations listed above are based on purely petrographic criteria: no perti-
nent characteristic makes it possible to associate one particular aquifer with any one of these
formations. However, within the RESO programme (Gombert, 1998) the hydrogeological
data were summarised for each formation. The main results, from a quantitative standpoint,
are shown in the following table.
The aquifers are assumed to be confined in about 30% of the boreholes where data
required to estimate the storage coefficient are insufficient. The effective infiltration was
estimated from water balances in the Mouhoun and the Como watershed basins, calculated
on the portions overlying the sedimentary zone and estimated from actual evapotranspiration
and runoff. Given the spatial variation of the rainfall, the estimate of the effective infiltration
yields a value of around 10 to 15% of the rainfall, which represents a mean annual depth of
infiltrated water ranging between 60mm (extreme north-east of the area) and 110mm (in the
extreme south-west).
The depth of the water levels in the aquifers was estimated throughout the sedimentary
series in Burkina Faso within the VREO programme (Derouane, 2005) from over 1000
boreholes and wells. The water levels lie at a depth of between 0 and 128m below ground
level. The interpretation of the depth to water map strengthens the hypotheses concerning the
relationship between the rivers and the aquifer. For example, the Mouhoun, which is a peren-
nial river, is at equilibrium with the aquifer over a large part of its course. However, it becomes
perched in the downstream part of the study zone below Dedougou, where the depth of the
water table close to the river, is about 20m. The same is true for the Sourou, which is proba-
bly isolated from the water table along its entire course through the study area. Overall, the
Figure 2. Simplified map of the geological formations of the sedimentary series in Burkina Faso.
depth to the water table is greater than 20m at almost all of the observation points situated in
the Sourou basin (inside the study zone): thus, the aquifer has no natural outlet in this zone.
The zone covered by the Continental Terminal also lacks a natural outlet of the aquifer, which
becomes deeper in this area, descending to depths of more than 20m. Thus, across these two
adjacent zones, covering the northern part of the studied area, the groundwater flow is no
longer drained by the surface water; this is the beginning of the Gondo piezometric depres-
sion the centre of which can be found further north, outside the study zone (Archambault,
1960). Moreover, when the maps of aquifer depth and piezometry are examined together, it
seems clear that the principal rivers in the Banifing basin, in particular those along the admin-
istrative border with Mali, have marked drainage characteristics and are probably at equilib-
rium with the aquifer. Whether or not this equilibrium is permanent still has to be ascertained,
but this will require a hydrological database that has yet to be established.
The piezometry of the aquifer was estimated across the whole sedimentary series in
Burkina Faso within the VREO programme (Derouane, 2005), on the basis of observations
made on a sample of 400 points chosen for their perceived reliability. However, for most of
these points, only one piezometric measurement is available and taken after drilling the bore-
hole. The actual piezometric network in the study zone, where data are recorded periodically,
consists of only thirty observation points.
The analysis of the spatial interpretation of the piezometry highlights:
the piezometric maps are based on existing observation points, which generally with-
draw water at a relatively shallow depth (200m) from the uppermost water-bearing
layers. An integration of all these points, regardless of the catchment layer, requires an
overall hydrostatic equilibrium across the area and must, therefore, be based on the
assumption of a single multi-layered aquifer;
the study area in Burkina Faso shows mean piezometric heights ranging between 635
and 165 metres;
various hydrogeological ridges in Burkina Faso reveal different hydrogeological recharge

areas. There is a strong similarity between the topographical ridges that define the main
watershed basins (Mouhoun, Camo, Banifing) and the piezometric highs. For the
groundwater flow in the sedimentary aquifers, there is general conformity between the
hydrogeological basins and the main surface basins. Future work to be carried out within
Table 1. Hydrodynamic characteristics of the formations in the Burkina Faso sedimentary series.
Hydrogeological Max Average Average Average Average Average Average
unit nbr of depth of groundwater flowrate specific transmissivity permability
boreholes boreholes level (m
3
/h) flowrate (10
4
m
2
/s) (10
6
m/s)
(m) depth (m) (m
3
/h.m)
LS 21 75 8,9 5,1 0,6 0,51 0,2
KSS 69 48 5,9 5,1 0,4 2,7 1,2
GFS 166 80 15,5 9,1 0,5 2,8 0,5
SQG 271 62 13,1 13 1 8,5 3
SAC1 179 66 10,8 13,4 1,4 4,9 2,2
FPS 58 62 10,8 6,8 1,3 1,7 1,7
SAC2 99 53 7,6 7,6 5,5 2,3 1,1
SQ FBS 9 56 1 13,2 18,6 2,8 2,6
Complete set 872 60 10,4 9,8 2,6 4,1 1,8
Figure 3. Piezometric map of the sedimentary series in Burkina Faso.
the VREO programme will explore this hypothesis, and may indicate the presence of large
groundwater transfers through deep layers toward a linear outlet situated outside Burkina
Faso, e.g., the Niger and Bani rivers (according to the 2-D or 3-D model by Toth, 1963);
the piezometry indicates four main hydrogeological basins:

the Mouhoun hydrogeological basin which covers almost 80% of the study zone and
drains most of the groundwater in the Burkina Faso sedimentary zone along a north-
north-east axis toward the plain of Sourou and even further downstream, outside the
zone included in the VREO programme, in the direction of the Gondo depression. The
effect of the Mouhoun River is clearly visible in the piezometry. This entire hydrogeo-
logical basin is located inside Burkina Faso. The only natural outgoing groundwater
flow occurs in the Mouhoun river where it is at equilibrium with the aquifer at the north-
north-eastern limit of the study zone (frontier with Mali and northern boundary of the
Sourou watershed basin). Furthermore, this hydrogeological recharge basin cuts across
all the geological layers present in the region;
the Banifing hydrogeological basin, which drains the groundwater along a north-
west axis in the direction of Mali. The principal rivers in this basin, including those
that flow along the frontier, have an effect on the piezometric map. The ground-
water flows leaving the study zone are, in this basin, concentrated on the border with
Mali; they then continue inside Mali, probably toward the principal outlets, i.e., the
rivers Bani and Niger;
the Com hydrogeological basin in the extreme south-west of the study area. Overall,
the groundwater flow in this basin has a westerly direction toward Mali and leaves at the
frontier, and a southerly one where it leaves the sedimentary zone inside Burkina Faso
in the direction of the crystalline bedrock;
the Bougouriba-Grand Bal hydrogeological basin in which a piezometric ridge runs
along the Bobo-Banfora cliff and its northern extension, i.e., on the inner margin of
the sedimentary zone. It is inside the study area and is quite small. Groundwater flow
is generally oriented toward the south-east and soon leaves the sedimentary zone in the
direction of the crystalline bedrock.
the administrative border between Burkina Faso and Mali largely coincides with the
boundaries of the main surface basins. The general coincidence of the boundaries of
the principal watershed basins and those of the hydrogeological basins indicates that no
groundwater exchanges occurs between the two countries along much of the adminis-
trative border. In other words, the administrative border between Mali and Burkina
Faso follows a hydrogeological ridge which acts as a groundwater divide. This is evi-
dent from the piezometric map, of the south-west extension of the Bandiagara cliff
which clearly follows the groundwater divide along the Mali frontier. Thus, the only
large groundwater exchanges between the two countries are essentially in the north
close to the Continental Terminal and to the Sourou, in the west at the piezometric
trough of Banifing and in the south-west in the Como basin. This interpretation does
not exclude, given the present state of the knowledge, the possibility of groundwater
transport at depth.
finally, on the subject of common cross-border water-resource management, note that

no groundwater flow occurs from Mali in the sedimentary formations. The flow, there-
fore, comes exclusively from Burkina Faso, i.e., all the groundwater circulating in the
sedimentary zone in Burkina Faso covered by the VREO programme derives from rain-
fall recharge within the Burkina Faso sedimentary formations.
The evolution over time of the piezometry, albeit based on a small number of observations,
shows a general lowering of the water levels over the last five decades.
The hydrochemistry of the groundwater in the Burkina Faso sedimentary series has
been described on the basis of analyses (ionic balance within 5%) of samples collected
from around 500 observation points (Dakour, 2003). A distribution of the hydrochemical
facies in the different formations in the sedimentary series was defined by statistical analy-
sis, interpretation on Piper and Schoeller diagrams and Principal Component Analyses.
The main results are shown in table 2.
The data show that the majority of the samples are of the calcium bicarbonate type
(86%). There is a high degree of pollution. The Schoeller diagram illustrates the uniformity
of the data on a regional scale and reflects a hydraulic continuum across the various basins.
Isotopic analyses (Dakour, 2003) have been gathered over three sampling phases
(19962002). The stable isotopes
18
O and
2
H help to establish the origin of the waters and
the radioactive isotopes
3
H and
14
C indicate their age. The isotopic studies show that most
of the water stored in the sedimentary reservoir is old (from 50 to over 1000 years old)
which is not incompatible with the calculations of the present aquifer recharge given the
large amount of water in storage. The currently interpretation is that of a mixing model con-
taining contemporary and recent water mixed with the old water.
3 CONCEPTUAL GROUNDWATER FLOW MODEL FOR THE TAOUDENI
SEDIMENTARY BASIN
It is assumed that there is hydraulic continuity between all the hydrogeological basins, and
that the Burkina Faso sedimentary series contains a large single multi-layered aquifer.
Although there are, locally, impervious horizons in the series, their horizontal extent is minor
compared to the surface area of the basin. The weakly permeable layers, therefore, only raise
Table 2. Hydrochemical characteristics of the Burkina Faso sedimentary series.
Hydrogeological Population Temp. pH Conductivity Hydrochemica facies (number)
unit (C) (S/cm)
Bicarbonate Chloride Chloride Bicarbonate
calcium-like and/or and/or alkaline
sulphate sulphate
calcium-like alkaline
LS 12 30,4 5,4 59 7 4 1
KSS 12 30,2 6,4 36 14 1
GFS 67 30,6 6,9 183 60 7 3
SQG 116 29,5 6,4 144 104 7 5
SAC1 104 30,8 7,3 359 91 10 3
FPS 20 31,3 7,3 294 15 5
SAC2 55 31,9 7,3 347 37 7 2 9
SQ 16 31,9 6,8 154 14 2
FBS 21 6,8 118 18 1 2
CT 82 7,5 413 78 4
Complete set 438 12 33 28
86% 3% 6% 5%
the hydraulic head in selected parts of the aquifer. Geologically, the sedimentary series com-
prises potential reservoir formations across more than 75% of its thickness; these are mainly
sandstones and carbonates. For the most part, between the bedrock and the SAC1 base, there
is a regular sandstone-on-sandstone contact over a thickness of around 1000m with the
exception of a discontinuous layer of argillite at the base of the fine glauconitic sandstone,
which locally confines the Kawara-Sindou sandstone. There are also thick layers of locally
impervious argillite in the SAC1 and SAC2 formations although the lateral and deep continu-
ity of these aquicludes is by no means certain since there are productive boreholes on both
sides in the same stratigraphic layer. Furthermore, the discontinuities observed at various
levels (fissures, fractures, faults) inside the reservoir rock, provide an interconnected network
of potentially water-filled voids. This secondary porosity is superimposed on the primary
matrix one of the aquifer and allows groundwater to circulate locally in otherwise impervious
media such as the argillite. It is difficult to argue for the existence of impervious aquifer com-
partments since deep boreholes into the same sandstone around Sikasso (Mali) show the pres-
ence of fractured and water-bearing layers down to a depth of almost 400m.
The evolution of the water chemistry in the sedimentary series does not show any signif-
icant difference between individual aquifers. There is an only calcium-magnesium bicar-
bonate type water. An inspection of the major ions reveals an enrichment in bicarbonate as
groundwater flows from the fine glauconitic sandstone or sandstone with quartz granules
toward the Toun schist (SAC2). This is in keeping with a single continuous aquifer system in
which all the groundwater types are mixed together. The available isotopic data are also con-
sistent with the concept of a single aquifer. The mean residence time of the water ranges from
less than 50 years in the fine glauconitic sandstone (Bobo-Dioulasso) to almost 500 years at
the top of the sandstone with quartz granules (Nasso) and over 4000 years in the SAC1 (Fon).
This evolution is consistent with slow and continuous flow in a general south-north direction.
Thus, all the available data reflect a single multi-layered aquifer system, locally partitioned by
a few discontinuous impervious layers of which the principal ones are situated at the base of
the fine glauconitic sandstones (between Kklesso and Pni) and at the top of the SAC1
(between Banzon and Samandni) as well as inside the SAC2 (Toun schist).
Overall discharge from the system requires consideration of the whole south-eastern
margin of the Taoudeni basin, i.e., beyond Burkina Faso towards the River Niger. Although
the piezometry of the regional drainage basin is fairly well known because of the density
of observation points, they are measured in boreholes that are rarely deeper than 100m
whereas the aquifer thickness may exceed 2000m. It is impossible that deep groundwater
flow occurs perhaps in opposing directions. The geological characteristics of this large
sedimentary basin (the Taoudeni basin) reflect local groundwater basins near the surface
(Kou, Mouhoun, Banifing, etc.), which drain the superficial groundwater, and large regional
basal outlets (e.g., the river Niger and the Bani in Mali), which may discharge deeper
groundwater flow. This is in keeping with Toth (1963), who characterised a groundwater
flow system on two or three levels (local, intermediate and regional) in a single multi-layered
aquifer.
Surface water hydrograph analysis is also instructive (Mouhoun, Como, Banifing).
The Mouhoun, whose catchment covers 80% of the sedimentary zone in Burkina Faso,
appears to be draining and at equilibrium with the aquifer over a major part of its course,
but seems to become perched below Ddougou, where the water table is about 20m below
ground level. This is the extreme south-western part of the Gondo depression which is cen-
tred on the Continental Terminal and whose origin is still a matter of debate. Locally, the
surface-water groundwater relationship can be observed in a few zones with springs,
some of which are quite large, for example, the Nasso-Guinguette springs in the Bobo-
Dioulasso region whose flow rate exceeds 10,000m
3
/h. The recharge area of these springs
probably extends beyond the Kou watershed basin into the high basins of the Orodara.
There is also the important question of the recharge to the multi-layered aquifer and esti-
mates of the renewable resource potential. Several methods have been used in the Sudan-
Sahel region to arrive at this estimate with very diverse results due to the variability of the
media and their characteristics as well as that of the techniques themselves (Bazie & Dieng,
1995). In the case of the Burkina Faso sedimentary zone, there is a strong north-south gradi-
ent in rainfall, temperature and evapotranspiration. An estimate using the Thornthwaite for-
mula for the Bobo-Diolasso region (19811990) yields a mean infiltration rate of 75 to
120mm/yr for a soil water reserve (SWR) of between 140 and 200mm (Dakour, 2003). The
spatial variability of the recharge is estimated to be 0 in the north (where the rainfall does not
exceed 600mm/yr) and almost 250mm in the extreme south (high basins). These figures,
however, are based on inadequate data (Dakour, 2003).
4 GEOLOGICAL MODELLING OF THE TAOUDENI SEDIMENTARY BASIN,
BURKINA FASO
The 3-D geometrical model of the Burkina Faso sedimentary series was developed within
the VREO programme (Derouane, 2005) for the purpose of:
providing a complete 3-dimensional representation of the extent and boundaries of the

10 geological formations across the Burkina Faso sedimentary series, and to generate
geological cross-sections;
defining hydrogeological layers which could be used in a mathematical model to sim-

ulate groundwater flow;
The basic data for the geometrical model are the topography, the geological raster map and
18 available geological cross-sections obtained in the field within the VREO programme.
The 3D geological model represents the known characteristics of the region:
the geological layers show a monoclinal structure with a dip on the order of 2% in a
general north-westerly direction. The lower layers become deeper as they approach the
Figure 4. Coding of the stratigraphic points for the development of the 3-D geometrical model
(left) and integration of the topography into the geometrical model (right).
Mali frontier, perhaps up to 2000m deep. This is based on present knowledge of the
geological structure and on air-borne geophysical investigations;
the depth to the crystalline bedrock also indicates the major structural zones;
the SQG geological layer pinches out in a north-eastern direction;
the Continental Terminal (CT) has a tabular (sub-horizontal) shape and extends in
stratigraphic discordance over the GFS, SQG, SAC1, FPS and SAC2 layers in the north
of the region.
5 CONCLUSION
The understanding of the aquifer system on the south-eastern margin of the Taoudeni basin
in Burkina Faso has gained by the interpretation of the piezometry across this 40,000km
3
zone and the development of a 3-D geological model. This work was undertaken as a
multi-disciplinary effort and combined stratigraphy, hydrodynamics, hydrogeochemistry
and isotopic studies to support the geological and hydrogeological modelling.
The observed piezometry of the system comprises a typical large drainage basin with
outflow towards the major Burkina Faso rivers. Given the low topographical relief com-
pared to the depth of the base of the aquifer system, regional flow toward the Bani and
Niger Rivers in Mali cannot currently be excluded. The interpretation of all the available
field data supports the conceptual groundwater flow model.
REFERENCES
Archambault, J. (1960) Les eaux souterraines de lAfrique occidentale. Berger-Levrault, Nancy,
France.
Figure 5. 3-D view of lithostratographic cross-sections generated within the geological model.
Bazi, P., Dieng, B., Ackerer, P. (1995) Bilan des transferts verticaux deau en zone non-sature sous
climat soudano-sahlien : aplication lestimation de la recharge des nappes. Revue des sciences
de lEau, 8, 2 237260.
Besbes, M., Dakour, D. (2002) Quantification des ressources en eaux souterraines du bassin sdi-
mentaire du Sud-Ouest. AIEA Projet BKF/8/002.
Bro, C., DHI et IWACO (2001) Etat des lieux des ressources en eau du Burkina Faso et de leur cadre
de gestion. Rapport final. Ministre de lAgriculture, de lHydraulique et des Ressources
Halieutiques du Burkina Faso.
Dakour, D. (2003) Etude hydrogologique et gochimique de la bordure Sud-Est du bassin sdi-
mentaire de Taoudeni (Burkina Faso Mali) Essai de modlisation. Thse de Doctorat,
Universit Paris VI.
Derouane, J. (2005) Rapport de modlisation des eaux souterraines de la srie sdimentaire dans la
zone dintervention du Programme VREO au Burkina Faso pour lAssistance Technique Sofreco-
Sawes partie 1. SOFRECO-SAWES, Programme VREO.
Gombert, P. (1998) Synthse sur la gologie et lhydrogologie de la srie sdimentaire du sud-ouest
du Burkina Faso. DRH-HB Programme RESO.
Ouedraogo, C. (1998) Cartographie gologique de la rgion Sud-Ouest du Burkina Faso au
1/200.000 Synthse gologique. Aquater/Bumigeb.
Toth, J. (1963) A theoretical analysis of groundwater flow in small drainage basins. J. Geophys. Res.,
68, 47954812.
Wood, W., Sanford, W. (1995) Chemical and isotopic methods for quantifying groundwater recharge
in a regional, semi-arid environment. Groundwater, 33, 458468.
CHAPTER 12
An assessment of the groundwater resources in the western
margin of the Taoudenni basin, Mauritania
Franois Bertone
1
, Philippe Renard
2
, Jaouher Kerrou
2
, Patrice Moix
2
&
Pierre Perrochet
2
1
BCEOM, Montpellier, France
2
Centre dHydrogologie, Universit de Neuchtel, Switzerland
ABSTRACT: An extensive study was made of the hydrogeology of the region around Zouerate in
the north-western part of Mauritania. This area consists of two main zones: the Archaean basement
in the west and the margin of the Taoudenni sedimentary basin in the east. The geometry of the
aquifers in the Taoudenni basin was deduced by a combination of different techniques including:
refinement of the geological map using remote sensing, interpretation of aerial magnetometric data
and construction of a hydrostratigraphic log. A survey of 624 groundwater points, including GPS
levelling, was conducted. It became apparent that a few sebkhas are the main regional discharge
zones. To estimate the fluxes, a 2D regional model was constructed and calibrated. It allowed a com-
parison of different recharge scenarios and may be used to run long-term simulations. A great deal
of uncertainty remains in the model results, especially concerning the recharge area distribution.
1 INTRODUCTION
To develop the Mauritanian mining sector, the World Bank funded the PRSIM2 project
whose aim, among others, was the evaluation and mapping of the groundwater resources
around Zouerate, in the north-western part of Mauritania (Figure 1). To sustain the mining
exploitation in this part of the Sahara a reliable water supply is required.
1.1 Climate and hydrology
The climate in this region is of Saharan type. It is characterized by high daily temperatures
and less than 70mm/year of rainfall. The rainfall is highly variable both in space and time
(Figure 2). Furthermore, since 1970, the annual rainfall has decreased by 35% (Figure 2)
as in most parts of North Africa.
The city of Atar is situated in the south of the area and Zouerate in the north. In the west,
there is a plain called Amsaga in the south and Tiris in the north; in the east there is a series of
plateaux constrained by north-south oriented cliffs. The whole structure is locally covered by
the Hammami, the Maqteir and the Oum Arouaba sand dunes.
There are no perennial rivers in the area. Except near the highest topographical points, but
the hydrographic network is otherwise not developed. It drains water for only a few days per
year and can remain dry during long periods. When rainfall does occur, the surface runoff
both on the plains and in the wadis converges towards a large number of closed endorheic
depressions. Among them are the sebkhas and the garaas. The sebkhas, or saline lakes, are
characterized by an abundance of soluble salt, especially chloride and sulphates that are pre-
cipitated at the surface. The piezometric levels below the sebkhas are shallow and there is
intense groundwater evaporation (up to 4m per year). The garaas are humid zones depend-
ing on the runoff in the ephemeral wadis. They are less brackish than the sebkhas.
M

a

l

i
M a u r i t a n i a
A

l

g

e

r

i

a
W

e

s

t

e

r

n

S

a

h

a

r

a
S e n e g a l
Atar
Zouerate
Figure 1. General map of the study area location.
1920 1930 1940 1950 1960 1970 1980 1990 2000
0
50
100
150
200
250
P
r
c
i
p
i
t
a
t
i
o
n
s

(
m
m
)
Figure 2. Annual rainfall and moving average over 20 years in Atar city.
1.2 Geological setting
The study area (80000km
2
) comprises two distinct zones: the north-western part of the
Taoudenni sedimentary basin in the east, and the Precambrian basement in the west
(Figure 3).
The basement includes mainly fractured metamorphic formations, intrusive granite and
a few allocthonous units forming small or major inselbergs like the Kdia dIdjil, the
MHaoudat, or the Guelb el Rhein. These units belong to the Idjil group, which is considered
to be an old deformed sedimentary system, metamorphosed and displaced within the base-
ment. This group is mainly comprises ferruginous quartzite, quartzite and mica schist. Other
lithologies are present including amphibolites, metagabbros, dolomite. The depth of the
basement eastward under the Taoudenni basin is not precisely known.
To the east, the sedimentary cover corresponds to the north-western extremity of the
Taoudenni sedimentary basin, which is composed of a complex series of fissured lime-
stone, sandstone, conglomerates and shale. The lithology and the stratigraphy of the basin
have been described in detail (Delpy et al., 1963; Trompette, 1973; Lahondere et al., 2003;
Pitfield et al., 2004). The oldest deposits are regrouped under the denomination of the
Hodh Supergroup (Infra-Cambrian Series). They are covered by deposits of the Adrar
Supergroup (Cambro-Ordovician Series) from which they are separated by the so-called
lower tillite (glacial deposits). These supergroups have been subdivided into groups, each
one comprising several sub-ensembles. The groups rest uncomformably on top of each
other and some groups are not present everywhere. The whole sedimentary basin is not
folded, but is affected by minor shear deformations (Villemur, 1967).
An assessment of the groundwater resources in the western margin of the Taoudenni basin 151
Figure 3. Extent of the studied area drawn on a geological sketch modified from Lahondere et al.
(2003).
1.3 Groundwater systems in the oldest sedimentary formations
In the south of the area, the outcrop limits of the oldest basin formations (Char Group and
Atar Group formations) are sites where small parallel valleys have formed, all of them
draining into the Seguelil wadi toward the Archaean plain of Amsaga in the south. In these
valleys, often partially filled by Quaternary deposits, are the oases of the sector of Atar, irri-
gated by numerous wells. Moussu and Trompette (1966a, b) describe the hydrogeological
setting of this area based on the oasis wells in the Quaternary deposits. Up to the 1970s,
these wells exploited the alluvial aquifer. With the decline in rainfall, this aquifer now dries
up after the recharge episodes. The oasis wells have very often been deepened to reach the
sedimentary formations under the Quaternary alluvium. The main outlet of the system is, at
present, still beneath the oases, where more than 6.5Mm
3
/year are extracted, but more and
more is pumped from the old sedimentary formations. In this sector, in addition to the 2500
wells, located in the 22 oases, the aquifers have been explored around Atar by drillings. An
analysis of these data shows the discontinuous nature of the aquifers. 25% of the boreholes
have a yield of less than 1m
3
/h and 20% have a yield greater than 15m
3
/h. Only the most
productive wells have been hydraulically tested and the transmissivities range between
1 10
4
and 5 10
3
m
2
/s. The boreholes intended for the water supply of Atar city pro-
duce some 0.6Mm
3
/year in addition to the water drawn from the oasis wells.
In the north of the area, the relief is dominated by the plateaux of the oldest formations of
the Hodh Super-group. The slope of these plateaux, drained by the valley of Assabet el
Hassiane, is gentle. The main relief is that of the Kdia dIdjil and the MHaoudat, on the
border of the basin, partially covered by sedimentary formations. Plote (1966) drew a hydro-
geological map based mainly on observations in the Kdia dIdjil, where iron ore is mined.
The first boreholes were drilled in the Taoudenni basin to supply water to the mining indus-
try in the early 1970s. The withdrawal for this activity represents some 27.8Mm
3
since its
inception. Pumping at present is 0.9Mm
3
/year for the mines and a total of 2.4Mm
3
/year for
the Zouerate city water supply and other uses. As for the Atar sector, the main characteristic
of the aquifers is a strong variability of productivity indicating a discontinuous aquifer. 33%
of the wells have a yield of less than 1m
3
/h and 10% have a yield greater than 30m
3
/h. The
range of transmissivities for the highly productive wells is between 1 10
4
and 4 10
3
m
2
/s. The piezometric variations of the Zouerate sectors have been well known since the start
of the exploitation of groundwater in this zone. However, the evolution of the numerous
piezometers does not show the same trend around the well fields. Even if the general trend
is toward is of decline, it does not occur everywhere at the same rate. The piezometric records
often present strong or weak changes that deviate very strongly from a clear overall trend
(Figure 4) and there are phases when the water levels rise and there are fast and erratic vari-
ations. The fast rises are sometimes limited to one or two boreholes and only sometimes vis-
ible in all the boreholes.
The aquifer system of the oldest Taoudenni basin formations (Hodh Supergroup) con-
sists of a series of three main lithologies:
1. Sandstone horizons are present over the whole stratigraphic log. Their thickness is
highly variable from small interbedded lenses to large monotonous horizons. Their pri-
mary porosity is often very low, but they have a significant secondary porosity due to
fissures and fractures.
2. Carbonate horizons are localized mainly at the base of the stratigraphic log. They have
features that indicate a karstic development.
3. Mudstone horizons, usually considered of low permeability. The hydraulic communi-
cation between these aquifer horizons occurs through the dense fracture network that
cross-cuts the whole series. The transmissivities are not correlated with the thickness
of the aquifer or with the lithology. It appears that the main controlling factors are the
orientation of the fractures with respect to the current state of stress and the relative
position of the boreholes with respect to the topography (Fnart et al., 2006). The most
productive fractures are oriented N120E in the south and between N140E and N010E
in the north. In the sectors where the relief is marked, the boreholes located at the foot
of the cuestas are less productive than the wells located at top.
The hydrogeological potential of the oldest Taoudenni basin formations was identified in
the late 1960s but not quantified. Although the existence of one or several aquifer systems
was clear to all the hydrogeologists working in the area, their spatial extent, the regional
flow directions and more generally their behaviour were unknown. The recharge processes
were unknown as well; furthermore, the vertical continuity between these horizons has
never been ascertained. The work presented in this paper aims at clarifying these points
and providing a basis for future water management in the region.
2 METHODOLOGY
Using ArcView and ArcGis (ESRI products), a spatial database was built to store all
the collected data, process them and generate map output. This GIS database contains dif-
ferent types of cartographic data such as the digital elevation model, the location of
groundwater observations and withdrawal points, and the geological map (polygons) that
constrains the extent of the groundwater model.
2.1 Measuring groundwater levels
Over the study area, 624 groundwater points were identified and levelled using high pre-
cision differential GPS. The geographical distribution of this network is irregular; the
water points are mainly located around the two cities, Atar in the south and Zouerate in the
Figure 4. Drawdown measured in boreholes of the mining company and average trend.
north. A 20-day piezometric field study allowed us to measure more than 532 water levels
including 421 static ones.
2.2 Bedrock mapping and aquifer geometry
An interpreted map of the depth of the basement under the Taoudenni basin was created
(Figure 5) by combining an interpretation of the main tectonic events, inverted geophysical
data and extrapolation of the observed thicknesses of the outcropping sedimentary layers and
their dip angles. All these data were combined by kriging. This analysis shows that it is highly
probable that the overall thickness of the sedimentary system can reach more than 4000m in
the eastern part of the area.
2.3 Groundwater modelling
The numerical groundwater modelling was carried out with Feflow (Diersch, 2005).
This regional model is conceived in 2D because of the lack of hydraulic conductivity data
Figure 5. Result of kriging on the zone of interpolation representing the depth of the basement
under the basin of Taoudenni.
as a function of depth and because of the wide lateral extent of the aquifer as compared to
its thickness (the ratio is less than 1%).
During this study, a series of pumping test data were re-interpreted. Surprisingly, all but
one data set showed typical 2D radial flow behaviour (constant log derivative). There was no
evidence of double porosity behaviour or finite fracture behaviour. This shows that even if
the medium is fractured, the fractures must be continuous enough in space to produce radi-
ally convergent 2D flow around the wells. As a result, it is reasonable to assume that an
equivalent continuous medium can represent the large-scale regional flow. A study was made
of the sensitivity of the groundwater model to the presence of preferential flow directions. It
was done by including five major faults oriented N120E in the south and N140E in the
north, and by testing different regional anisotropy factors while keeping the same principal
anisotropy directions.
The measured transmissivities do not correlate with the thickness of the aquifer. An
acceptable conceptual model for the Taoudenni basin is to assign to each geological forma-
tion a transmissivity value that is independent of its thickness because borehole measure-
ments of transmissivity depend on the number of fractures crossed by the borehole (Figure
6). However, the degree of open fractures may be different from one formation to another.
The sebkhas constitute the principal natural groundwater outlets in the area. The location
of potentially active sebkhas was determined by satellite radar images, the ENVISAT pic-
tures present a characteristic signal of the moisture presence on the earth surface. An esti-
mate of the order of magnitude of the evaporation rates beneath the sebkhas was made with
the formulas by Coudrain-Ribstein et al. (1998), based on an evaluation of the water table
depth (0.4 to 4m). The output fluxes were imposed as source terms on the surface shape of
the sebkhas, allowing the use of PEST and automatic calibration of these fluxes.
Evaluation of the recharge is always difficult in arid zones. Published values are extremely
variable, from less then 1mm/year in New Mexico to 110mm/year in the Neguev desert
(Scanlon et al., 1997). Recharge may be significant as demonstrated by Ould Baba Sy and
Besbes (2006) for the North Western Sahara Aquifer System. Recharge was calibrated in the
model, by zone, as a constant source term. Three variants for the spatial distribution of
recharge were tested (Figure 7). The choice of these possible preferential recharge areas is
based on an analysis of existing data:
1. The Kdia dIdjil, where groundwater is locally fresher than elsewhere in the area and
the piezometry reacts to the most intensive rain events.
1
2
3
Figure 6. Conceptual model of the transmissivity used in the groundwater model. The transmissiv-
ity of the first and second wells will probably be higher than that of the third well because of the
number of fractures penetrated.
2. The MHaoudat in the north of the Kdia, which presents comparable geological and
hydrogeological settings.
3. The dune massifs (Hammami, Maqteir, Oum Arouaba) where precipitation can pene-
trate quickly.
4. The most permeable formation outcrops: Azougui, Aguni, and Foum Chor, with a
high density of open fractures at the surface.
5. The concentration zones of the hydrological network and of runoff accumulation.
There are three scenarios: all of these zones are assumed to be potentially active, infiltra-
tion through the dunes is considered negligible, and only local recharge is considered.
Based on the model for transmissivity distribution and the three scenarios of recharge
distribution presented above, three steady-state flow models were successively and satis-
factorily calibrated. The calculated steady-state heads were used as initial conditions for
the transient simulations, covering the period from 1957 to 2004. In addition, different
time-dependent or constant-recharge scenarios, as well as different values of the storage
coefficient, were tested.
3 REGIONAL FLOW SYSTEMS
The piezometric map constructed from the groundwater levels measured in all the Hodh
Supergroup formations shows a remarkable coherence. It shows at least two major flow sys-
tems, separated by a piezometric crest: the Atar system in the south of the Oum Arouaba
dunes massif, and the Bou Talha systemto the north of these dunes. In the south-eastern part
of the area, there is an upper aquifer system located within the Adrar Supergroup, and over-
laying the Hodh another which is isolated by a thick aquitard.
3.1 The Atar system
Overall, water flows from the north (Oum Arouaba dunes) to the south (Amsaga plain).
The groundwater is drained by the oases in the small Adrar valleys. The share of this
Figure 7. Delimitation of the potential recharge zones (grey) considered in the three recharge
scenarios.
regional flow that reaches the southern outlet is probably very small, on the order of
0.2Mm
3
/year.
If part of the recharge of the system can occur during the floods beneath the alluvial
deposits, which fill the wadies, and may also occur during rainy episodes between valleys
beneath the sandstone and limestone outcrops, whose surface is heavily fractured, the piezo-
metric map suggests that the system is recharged from the piezometric crest in the North,
under the Oum Arouaba dunes massif. On the whole, the recharge by rainfall may be on the
order of 5.4Mm
3
/year. The piezometric variations reported by the operators of the oasis wells
show that the system refills during very rainy seasons. Depletion is then observed until the fol-
lowing heavy rains, one to four years later (Figure 2). During seasons with heavy rainfall, the
recharge is intensive and the piezometric levels in the oasis wells are often less than 1 to 2m
below the ground surface. As the withdrawal is limited by the depth of the wells, the system is
in a state of unstable equilibrium between maximum recharge during very rainy seasons and
maximum drainage after several years without recharge, when the piezometric levels reach
the limit of the well depth. In addition to rainfall recharge, the system is also recharged from
the overlying system of the Adrar Supergroup, probably by around 2.1Mm
3
/year.
"
!
!
!
!
!
!(
!(
!(
! !
!
!!
!
!
!
!
!
!
!
!
! !
!
!
!
! ! ! !
!
!
!
!
!
!
!
!
!
!
!
!
!
! !
! !! !
!
!
!(
!
!
!
! !
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
! !
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
! !
!
!!
!
!(
!
!
!
!
!
!
!
!
!
!
!(
!
!

!
!
! !
!
!
!
!
!
! !
!
!
!
!
!
!
!

!!
!
!
!
!
!
!
!
!
!
!
!
! !
! ! !
!!
!
!
!
!
! !
!
!
!
!!
!
!
!
!
!
!
!
!
! !
!
! !
!
!
!!
! !
!
!
!
!
!
!
!
!
! !
! !
! !
!
!
! !
!!
!
!(
!
!( !(
!
!(
!(
!
!(
!(
!
!
!
!
!

!

!
!
!
!
!(

!
!
!
!
!(
!
!
!
!
!
!
!
!
!
!
!
!
!
!
OUM AROUABA
dunes
Tachot
Azougui
Aguenni
Taizent
Ksar Torchane
Atar
2
9
0
2
8
0
3
0
0
2
7
0
3
3
0
3
1
0
3
2
0
2
6
0
340
2
5
0
2
4
0
2
3
0
2
1
0
2
2
0
2
0
0
1
9
0
1
7
0
1
8
0
1
6
0
1
5
0
1
90
2
2
0
170
2
3
0

2
9
0
2
8
0
700 000
2

2
5
0

0
0
0
2

2
5
0

0
0
0

2

3
0
0

0
0
0
2

3
0
0

0
0
0
13
2
1
13
700 000
2
1
M a u r i t a n i a
W

e

s

t

e

r

n
S

a

h

a

r

a
Zouerate
Atar
Figure 8. Piezometric map of the Atar system, May 2005.
3.2 The Bou Talha system
The Bou Talha aquifer system is, like the Atar system to the south, part of the aquifer forma-
tions in the Hodh Supergroup. It extends over a vast domain from the piezometric crest
beneath the Oum Arouaba dunes massif in the south up to the northern border of the sedi-
mentary basin (Figure 9). In the west, it is also limited by the border of the sedimentary
basin. In the east it extends beyond the mapped zone, certainly under the Maqteir dunes mas-
sif (the available piezometric information is too scarce to define the extent of the system
eastward). Two zones can be distinguished in this system, to the south and in the north of the
Bou Talha sebkha.
The piezometry in the south is known only at a few groundwater points (10). It seems that
it follows the topographic slope to converge towards the Bou Thala sebkha. It should be
"
"

!(
!(
!(!( !(

"
Choum
Zouerate
280
2
7
0
2
9
0
320
3
0
0
310
330
260
340
2
5
0
2
4
0
2
7
0
2
9
0
3
0
0
3
0
0
2
8
0
2
7
0
3
0
0
3
0
0
3
2
0
2
9
0
2
9
0
3
1
0
2
6
0
3
1
0
3
1
0
700 0 00 750 000
800 000
800 000
850 0 00
2

3
0
0

0
0
0
2

3
5
0

0
0
0
2

3
5
0

0
0
0
2

4
0
0

0
0
0
2

4
0
0

0
0
0
2

4
5
0

0
0
0
2

4
5
0

0
0
0
2

5
0
0

0
0
0
2

5
0
0

0
0
0
2

5
5
0

0
0
0
13
12
N
W
S
E
12
2
2
2
2
2
3
2
1
M a u r i t a n i a
W

e

s

t

e

r

n
S

a

h

a

r

a
Atar
Zouerate

Figure 9. Piezometric map of the Bou Talha system, May 2005.
noticed here that the piezometric crest beneath the Oum Arouaba dunes massif also corres-
ponds to a topographic high. The flow comes from the dunes massifs: Oum Arouaba in the
south, Maqteir in the east and Hammami in the west and north-west, and converges towards
the Bou Thala sebkha which seems to constitute the main outlet. Between Bou Talha and
Tourine, a watershed is outlined between the water from the southern sector drained by the
Bou Thala sebkha and the water from the north drained by the Meddahia sebkha.
In the northern sector of the Bou Talha system there are numerous boreholes supplying
water to the mines around Zouerate. Large drawdown cones have formed beneath the well
fields, which tap the groundwater on the border of the basin. The piezometry in the rest of the
zone is less well known with only 29 measurement points. The sector east of the Hammami
dunes massif is drained eastward by the Meddahia sebkha which, together with the boreholes
for the mining company, represents one of the major outlets of the northern sector of the Bou
Talha system. This sebkha also drains flow from the western extremity of the Hammami,
which is not diverted towards the Bou Thala sebkha. Furthermore, it seems to drain ground-
water stored beneath the Maqteir dunes massif, in the south-east.
Like the Atar system, the discontinuous aquifer layers in the oldest sedimentary forma-
tions are recharged by rainfall, probably some 6.8Mm
3
/year. Uncertainties persist, however,
concerning the exact recharge sites, i.e. beneath the dunes massif (Oum Arouaba in the
south, Maqteir in the east and Hammami in the centre) and/or beneath the most permeable
rocky outcrops or beneath the surface water accumulation sites, notably in the alluvial val-
leys of wadis. The system is also recharged by the basement formations (of the order of
1.4Mm
3
/year), in particular from the Kdia dIdjil and the MHaoudat. The current data sets
are not sufficient to clarify whether or not recharge occurs beneath the dunes and it would be
necessary to know the hydraulic gradients in the basin in the east and the south-east of the
zone, under very remote, hardly accessible, dune massifs. The interpretation here is important
because it modifies the regional balance by an amount of the order of 15Mm
3
/year, which
corresponds, for example, to five times the current withdrawal required by the mining activity.
The existence or the absence of such a recharge in the dunes is thus a major question, which
remains to be answered in order to understand the hydrogeology of the region.
The system is drained naturally at the low points where there are springs (0.1Mm
3
/year)
and oases at the border of the basin (Choum wadi and Char) and the Bou Talha sebkhas and
nearby Aouchich, Meddaiha and Erguiya where evaporation is high (11.3Mm
3
/year). It is
also drained by the numerous production boreholes (2.4Mm
3
/year), in particular those sup-
plying the mines. Between the Hammami dunes massif and the Bou Talha sebkhas, it is very
likely that the system is drained by evaporation beneath the Timnjati sebkha. This draining
does not modify the flow direction, as it only affects the flow from the north-west. The quan-
titative balance proposed by the mathematical models shows a depletion of the reserves on
the order of 2.4Mm
3
/year. It thus seems that the withdrawal through boreholes, although
localized, diverts only an insignificant part of the groundwater flow away from the sebkhas.
4 CONCLUSION
The data analysed in this work made it possible to describe the main behaviour of two vast
aquifer systems in the region. In the Atar system in the south, a long term equilibrium is estab-
lished, constrained by the limitations on the withdrawals, in particular due to the shallow depth
of the wells. In the Bou Talha system in the north, the high withdrawal rates for the mining
activities are concentrated to the border of the Taoudenni basin and do not in practice divert
any significant component of the large flow away from the sebkhas. In this system, unlike that
of the Atar, the withdrawals have led to a continuous decline of the piezometric levels.
ACKNOWLEDGEMENTS
The work presented in this paper was done under the supervision of A. Dassargues and
M.Y. Ould El Joud for the Second Project for Institutional Strengthening of the Mining Sector
(PRISM2) engaged by the Ministry of Mines and Industry and funded by the World Bank and
the Mauritanian Government. The authors of this paper would like to thank the PRIMS2 for
having given the permission to publish the work. Special thanks to G. de Marsily, M. Perez,
and M. Besbes for their constructive comments while reviewing the paper.
REFERENCES
Coudrain-Ribstein A, Pratx B, Talbi A, Jusserand C (1998) Lvaporation des nappes phratiques en
climat aride est elle indpendante de la nature des sols?, C. R. Acad. Sci. Paris, Sciences de la terre
et des plantes, 326 pp159165
Delpy J, Trompette R, Villemur J R (1963) Essai de synthse stratigraphique du Palozoque
infrieur de la bordure septentrionale de la synclise de Taoudeni (Sahara occidental) Bull. Soc.
Gol. De France (7), V, pp 10581062
Diersch HJ (2005) FEFLOW 5.2, Finite element subsurface flow and transport simulation system.
Users Manual. Rapport Technique: Wasy GMBH, Berlin
Lahondere D, Thieblemont D, Goujou J-C, Roger J, Moussine-Pouchkine A, Le Metour J, Cocherie A,
Guerrot C (2003), Notice explicative des cartes gologiques et gtologiques 1/200000 et 1/500000
du Nord de la Mauritanie. Volume 1. DMG, Ministre des Mines et de lIndustrie, Nouakchott
Moussu H, Trompette R (1966a) Notice explicative de la carte hydrogologique au 1/200.000, Atar
BRGM pour le compte de la direction des services techniques du Ministre de la construction, des
travaux publics et des transports. RIM
Moussu H, Trompette R (1966b) Notice explicative de la carte hydrogologique au 1/200.000,
Chinguetti. BRGM pour le compte de la direction des services techniques du Ministre de la con-
struction, des travaux publics et des transports. RIM
Ould Baba Sy, M, Besbes M (2007) Holocene recharge and present recharge of the Saharan aquifers;
a study by numerical modelling. This volume
Fnart P, Pistre S, Bertone F (2006) Application dune analyse multicritres pour la dtermination
des zones potentielles de productivit daquifres fissurs en pays aride (rgion dAtar,
Mauritanie). Proceedings of the Integrated Water Resources Management and Challenges of the
Sustainable Development conference, Marrakech, May 2006
Pitfield PE, Key RM, Waters CN, Hawkins MPH, Schofield DI, Loughlin S, Barnes RP (2004)
Notice explicative des cartes gologiques et gtologiques 1/200000 et 1/500000 du Sud de la
Mauritanie. Volume 1 gologie. DMG, Ministre des Mines et de lIndustrie, Nouakchott
Plote H (1966) Notice explicative de la carte hydrogologique au 1/200,000 Fort Gouraud Oum
DFerat, RIM
Scanlon BR, Tyler SW, Wierenga PJ (1997) Hydrologic issues in arid, unsaturated systems and
implications for contaminant transport. Reviews of Geophysics, 35, 4, 461490
Trompette R (1973) Le Prcambrien Suprieur et le Palozoique infrieur de lAdrar de Mauritanie
(bordure occidentale du Bassin de Taoudenni, Afrique de lOuest). Un exemple de sdimentation
de craton. Etude stratigraphique et sdimentologique. Trav. Lab. Sc. de la Terre, St Jrme,
Marseille, (B) n7, Marseille, 702 p
Villemur JR (1967) Reconnaissance gologique et structurale du nord du bassin de Taoudenni.
Mmoires du BRGM, n51, 151 p
CHAPTER 13
Groundwater modelling for conjunctive use patterns
investigation in the upper Central Plain of Thailand
Werapol Bejranonda
1
, Sucharit Koontanakulvong
1
, Manfred Koch
2
&
Chokchai Suthidhummajit
1
1
Department of Water Resources Engineering, Chulalongkorn Uni., Bangkok, Thailand
2
Department of Geohydraulics and Hydrology Engineering, Uni. of Kassel, Germany
ABSTRACT: In the upper part of the Central Plain of Thailand which covers about 38,000km
2
, due
to the government price-subsidized policy, farmers tend to grow rice more often now, resulting in a high
demand for irrigation water, with the latter being drafted increasingly from groundwater resources.
This has not only put pressure on the regional aquifers but, owing to river-groundwater interactions,
also on the surface waters in the region. As part of a major national effort, conjunctive water use pat-
terns are to be explored to optimise the water resources in the upper Central Plain for the various stake-
holders involved. In the present study, conjunctive use pattern of surface- as well as groundwater are
investigated by field surveys and groundwater flow modelling, using the MODFLOW model. The
groundwater model of the upper Central Plain is developed to simulate the groundwater movement
over the last 10 years, before an investigation of the groundwater use was conducted. The latter can be
categorized into three main types, namely, industrial, domestic and agricultural. Groundwater use pat-
terns were considered for different seasons (wet and dry) and different water availability situations.
A pilot area in the study area was selected to investigate the actual water use patterns, farmers irriga-
tion behaviour and constraints; i.e. harvest terms, groundwater pumping hours, pumping ability, etc.
The conjunctive patterns vary significantly in response to the water availability situation, such that the
portion of groundwater covering the total water-demand in years of wet, normal, dry and drought con-
ditions is 13%, 17%, 13%, and 19%, respectively. The groundwater ratio in the dry and wet season is
6% and 38%, respectively, of the demand. Finally, the future groundwater demand in the groundwater
basin has been predicted and it is found that the conjunctive use pattern is a key factor for the estima-
tion of groundwater consumption and for the assistance in the proper conjunctive planning in order to
mitigate future water shortages and to sustain the groundwater resources in the area.
Keywords: Groundwater demand; river-aquifer interaction, modelling; MODFLOW; conjunctive
use; Thailand.
1 INTRODUCTION
In spite of the tremendous steps made in recent years towards becoming an industrialised
country, Thailand still defines itself economically as an agricultural country, as the export
of agricultural products, namely rice, is still bringing in a large portion of the national
revenue. Boosting up rice production and, at the same time, the often precarious living
conditions of the rice farmer has, thus, been an active policy of the Thai government in
recent years and has lead it to develop many irrigation projects and agricultural price-sub-
sidized schemes to support local farmers. At the same time, both groundwater and surface
water resources have been developed to respond to an increased water consumption in the
private, domestic and agricultural sector.
The upper part of the Central Plain of Thailand is located in a large plain that is very suitable
for agriculture, as water resources are normally plentiful. However, with the active price poli-
cies mentioned, farmers nowadays tend to grow rice more often, which can only be achieved
through increased irrigation using both surface- and, lately, also more groundwater, putting
more pressure on the available water resources in the region. This precarious situation asks for
the use of techniques of so-called conjunctive management (Chun, 1964) which is a manage-
ment approach similar to IWRM (Integrated Water Resources Management), with the empha-
sis on the combined use of both surface- and subsurface waters to meet the total water demand
(cf. Bealaineh, 1999; Azaiez and Hariga, 2001). In the upper Central Plain many large irriga-
tion-serviced fields are scattered, making a central delivery of irrigation water through canals
difficult. This has led farmers to mostly set up their own groundwater wells on their paddies
and pump groundwater individually to compensate surface water shortages. The subsequent
and often uncontrolled heavy pumping has induced a decline of the groundwater table in parts
of the irrigated areas, causing future problems of groundwater accessibility for the farmers.
Since the groundwater level is going down mainly in the dry season when pumping for rice
paddy irrigation is at its highest, artificial aquifer recharge during the wet season has been sug-
gested to alleviate the water storage problem in this region (Chulalongkorn, 1998). However,
given its huge costs, such a recharge project has not been implemented yet up-to-date.
As it is not possible to provide sufficient surface water for irrigation, a conjunctive use
scheme should be developed (RID, 2005). Although there are many long-term hydro-
meteorology data and surface-water development projects available, no groundwater-use-
behaviour study in this region has been conducted up to now, and most of the pumping wells
are not well recorded. Moreover, groundwater levels have been monitored judiciously for only
just a few years. Hence, there is a lack of a comprehensive groundwater study, which is nec-
essary to understand the subsurface-, and, because of the intertwined interaction, also the sur-
face water resources, both of which are prerequisites for a conjunctive use analysis.
Beforehand it is necessary to understand the present conjunctive use-pattern, i.e. the propor-
tion of local agricultural, industrial and domestic water demand as a function of the prevail-
ing conditions of surface-water supply and the geographical characteristics. These use-pattern
have been established, sometimes sketchily, from field surveys and questionnaires handed out
to farmers and are to be used in the groundwater model, MODFLOW (Harbaugh et al.,
2000), to simulate the long-term behaviour of the exploited groundwater system and to come
up, eventually, with sustainable conjunctive water use patterns for the future.
2 STUDY AREA
The upper Central Plain, Thailand, covers about 38,000km
2
(180km 300km) of 8
provinces with a population of 4 million people. The main land-use is 63% agricultural,
out of which 21% is irrigated, and 24% forest. More than 90,000 groundwater wells exist in
the region. The main groundwater basin is dissected by five major rivers that flow from north
to south and which have formed the basis geologically as a depositional flood plain. The
basin is surrounded in the east and west by mountains of volcanic rocks. The average eleva-
tion of the basin is 4060m.a.s.l. The basin drains into the lower basin in the south, though
the free discharge is partially obstructed by crystalline rocks there. The 9001450mm annual
rainfall within the study region is apportioned to 81% in the wet (Apr.Sep.) and to 19% in
the dry season (Oct.Mar.).
3 GROUNDWATER MODELLING FOR CONJUNCTIVE USE PATTERN
ANALYSIS
3.1 Methodology and data collection
Groundwater levels and movements in the study area are simulated with the GMS/MOD-
FLOW groundwater flow model. The modelling approach follows the usual steps of build-
ing the conceptual model, the model design, calibration and verification/prediction (cf.
Anderson and Woessner, 1992). Groundwater use is a key input parameter in this study.
There are three main types of groundwater uses: agricultural, domestic and industrial.
Questionnaires were distributed to farmers to acquire data on the estimated groundwater
use. A further verification of the latter was gained from an analysis of several recent surface
water shortages. The reported pumping rates were grouped and classified with respect to
the surface-water availability in the year considered (wet, normal, dry, and drought), the
Groundwater modelling for conjunctive use patterns investigation 163
Southeast Asia
Thailand
Ping river
Yom river
Nan river
Chao Phraya river
Sakae Krang river
Figure 1. Map of the study area with the alluvial aquifers.
season of the year (wet and dry) and the location (inside or outside an irrigation project, sur-
face basins and aquifer characteristics). Groundwater levels were collected in the field
and/or taken from historical records. The geohydraulic properties were estimated from
pumping tests and groundwater recharge was computed from rainfall and an assumed infil-
tration rate (Koontanakulvong, 2002).
During the calibration of the model, the groundwater pumping rates were further adjusted
on the grid-cells, in order to wipe out as much as possible inconsistencies in the reported
data. A surface water balance analysis (using MIKE BASIN and WUSMO) was eventually
carried out to estimate the ratio between surface and subsurface water use and compare this
with the actual water demand. A scheme of the methodology used is depicted in Fig. 2.
3.2 Model development
The groundwater conceptual model, namely the aquifers and their confining boundaries,
were defined using the concept of the hydrostratigraphic units which is defined as geologic
units of similar hydrogeologic properties. The aquifer system in this study was defined as a
two-layer aquifer, whereby the thickness of the upper, semi-confined layer varies between 40
and 100m and that of the lower, confined layer between 100 and 300m (cf. Fig. 3). The 3-D
block-centred grid model representing the groundwater basin has a grid-size 10km10km,
resulting in 320 elements in the upper and 346 elements in the lower layer (Fig. 4).
The western, eastern and northern borders of the model where assumed as an imperme-
able body of consolidated rock and were defined as specific inflow boundaries (total 587
million m
3
/year) derived from the available head distribution along these boundaries. The
Math Model
Design
Define
Table of Outputs
Review
Conjunctive Use in
study area
Conceptual
Groundwater Model
Calibration and
Verification
Modeling
Result
GW Use Pattern
Analyze
Space Dimension
- Irrigation Area
- Surface Basins
- Aquifer Types
Time Dimension
- Water Situations
- Seasons
Hydraulics Properties
Groundwater Uses
Recharge
Surface Water Balance
Analysis Result
Surface
Water Uses
Surface Water
Shortage
Industrial/Private Use
Agricultural Use
Domestic Use
Questionnaires
Groundwater Levels
Figure 2. Overview of study scheme.
southern boundary, which is partially blocked by impermeable rocks and forms a narrow
trough between the mountains in the east and west, was set as an outflow boundary. A previ-
ous study on the lower Central Plain groundwater basin (Siriputtichaikul, 2003) provided an
outgoing flow rate between the upper and lower plain of 56 million m
3
/year and this number
was also used here. An average areal recharge of 555 million m
3
/year, derived from rainfall
Figure 3. 2-layer aquifer conceptual model.
Figure 4. Model grid design with vertical north-south cross-section.
and from a map of the soil-type and its infiltration rate (Koontanakulvong, 2002), was applied
on the top layer and on the outcropping sections of the lower layer. The river-aquifer interac-
tion of the five main rivers giving an average annual recharge of 337 million m
3
were derived
from the hydraulic properties of the river bed materials, the river cross-sections, the river
stages and the seasonally varying computed groundwater table. As for the possibility of return
flow of irrigated water into the canals, we assumed it negligible since, (1) the drainage canals
in the irrigation area are usually nearly dried out, except during the flood season and, (2) the
irrigation area covers only 13% of the entire model where the overall recharge takes place.
The hydraulic properties of the aquifer, namely hydraulic conductivity, transmissivity
and specific storage, were estimated from pumping tests. In addition, the aquifer proper-
ties as well as vertical leakage were obtained from three previous sub-regional groundwa-
ter models of the area (Jindasagnon, 1997; Chulalongkorn, 1998). The recharge, river stages,
surface and groundwater use were adapted in response to the climatic conditions, namely,
in terms of the amount of rainfall and the reservoir storage.
3.3 Groundwater use
As mentioned, groundwater-use estimation has been categorized into three main types: indus-
trial, domestic and agricultural. Industrial groundwater use has been determined from govern-
ment records that list the location of the well, its depth and the pump rate, all in all 992
registered items. The summary pumping rate is 380 million m
3
/year, but with only 26 million
m
3
/year extracted in the upper aquifer. Domestic groundwater use has been divided into two
types: village tap-water and water from private wells. The groundwater use of the village tap-
water depends on the number of families there and amounts to a total of 30 million m
3
/year.
The total number of shallow wells in the study area in 2003 has been 78,114 with a ratio of
agricultural to domestic consumption-well of 1:3 (Chulalongkorn, 2002) and an average daily
domestic consumption of 0.71m
3
/well, amounting to a total domestic-consumption from
wells of 15 million m
3
/year in 2003.
The major groundwater use in this area is by agriculture, namely, for rice and some
sugar cane in the western section of study area. Since the crop pattern is seasonally planed,
the agricultural stress-period used in the model is also based on the climatic conditions,
i.e. the wet and dry seasonal cycle. Agricultural wells are usually installed by the farmer to
supplement a shortage of surface irrigation water, therefore, records often do not exist and
the pumping behaviour is unknown. Because of this, the C-38 service unit inside the
Plychumpol irrigation project area in Phitsanulok Province has been selected as a pilot
study area to investigate the actual water use pattern, farmers behaviour and constraints,
i.e. harvest terms, groundwater pumping hours, pumping rates, maximum water drawdown,
etc. Moreover, 500 questionnaires were distributed to 30 sample sub-districts located in
five surface-basins throughout the entire study area.
The major pumping statistics retrieved from the survey is summarised in Table 1. From the
data listed there one can deduce that the average pumping capacity per well is 41m
3
/hour,
whereas the average pumping rate per well is 79m
3
/day inside the irrigation project, and
76m
3
/day outside. As for the groundwater-well database, it is based on records of the year
2003. The historical yearly record of the wells in each province during 19932003 has been
converted to a growth rate of the well concentration for the future. As mentioned, besides the
seasonally triggered agricultural water use, the latter depends also on the surface water supply
available during the time under question which, in turn, is linked to the actual storage of two
main upstream reservoirs (Koontanakulvong, 2002), the Bhumibol and Sirikit reservoirs
which provide surface-water and irrigation water to this area. The usable storage of these two
reservoirs on January 1st was used to define the situation of surface water availability, namely,
wet, normal, dry and drought, as shown in Fig. 5.
The yearly pumping rates were weighted relative to this surface water situation, using
1999 as the base year as it has been a drought year, i.e. when the pumping rate has been at
a maximum. In addition, agricultural groundwater use was rechecked by considering the
amount of compensable water to the agricultural surface-water shortage, which was calcu-
lated from (1) the water demand using the model WUSMO and, (2) a water balance using
the model MIKE BASIN.
3.4 Calibration and verification
Model calibration and verification/prediction was performed in steady state as well as in
transient state. Following the seasonal crop pattern, the seasonal stress period was used in
Table 1. Average pumping frequency from five surface-basins with 500 questionnaires.
Area Harvest Season Number of pumpings Duration of Pumping
frequency for each crop each pumping period each
(crops/year) (times) (days) day (hours)
Irrigation 2.5 dry 6.0 2.6 19.3
wet 3.8 2.3 19.3
Rainfed 2.0 dry 6.5 3.1 22.0
wet 3.1 2.1 16.0
Pilot area 2.28 dry 5.4 4.9 20.4
(irrigation) wet 3.5 4.5 23.8
Bhumibol+Sirikit Reservior Storage at J anuary 1st
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
Year
T
o
t
a
l

r
e
s
e
r
v
i
o
r

u
s
a
b
l
e

s
t
o
r
a
g
e

(

(
M
i
l
l
i
o
n

m
3
)
less 4,000 MCM : Drought Year
4,000 - 6,500 MCM : Dry Year
6,500 - 12,500 MCM : Normal Year
more 12,500 MCM : Wet Year
Figure 5. Historical storages of the Bhumiphol and Sirikit reservoirs.
the calibration of two years of recorded historical groundwater levels. The early water level
data were obtained from registered wells that recorded water levels during well construc-
tion. The last updated well records are from 2003. Since during 20012003, the groundwa-
ter use was almost stable, due to a constant situation for the surface water (see Fig. 5), the
average water level during the dry season of 2003 was selected to be the representative
steady-state water level for the calibration. 13 groups of the hydraulic conductivity were
adjusted during the steady-state calibration process. Figure 6 illustrates the observed and
simulated steady-state groundwater levels for the semi-confined layer 1. One notices a
rather good agreement between the two which is also manifested by the scatter-plot of the
observed versus modelled heads shown in the left panel of Fig. 7.
Figure 6. Observed (left panel) and simulated (right panel) steady-state groundwater levels in 2003.
15
20
25
30
35
40
45
50
55
60
65
70
75
80
15 20 25 30 35 40 45 50 55 60 65 70 75 80
Computed vs. Observed Values
Head
C
o
m
p
u
t
e
d
Observed
Computed Head Error
-20.0
-15.0
-10.0
-5.0
0.0
5.0
10.0
15.0
20.0
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Year
E
r
r
o
r

(
m
.
)
Figure 7. Computed versus observed heads for layer 1 in steady state (left panel) and transient
error as a function of time in layer 1 for the transient simulation (right panel).
Calibration in transient state has been carried out, using the 19932003 historical water
levels, whereby groups of specific storage have been calibrated. The transient simulation
is initialised from an average wet-season water level. During the transient-state calibration,
the pumping rate weights were fine-tuned, as these are often prone to errors. In summary,
the root mean square calibration error is 3.70m in steady-state mode and 5.11m in tran-
sient mode (see Fig. 7, right panel). An a posteriori transient-state verification/forecast,
using two years of groundwater level monitoring data (20042005) and water level data
from 50 extra observation wells collected during the study period (2005), has been per-
formed, resulting in a root mean square error of 5.95m.
3.5 Model results
The groundwater flow simulations show that, depending on the surface water availability,
the water levels are, on average, about 4m below ground surface in the wet season, but
drop to 69m below GS in the dry season. Significant head drops of 2.57m are observed
between the wet and the dry season in one year, especially in the dry season of a drought
year, when the head changes amount to 38m. The water balance (shown in Fig. 8 for both
the wet and dry season) illustrates that the total groundwater use was 812 million m
3
/year
in 2003, increasing to 1,068 million m
3
/year in 2005. For the year 2003, the total inflow
amounts to 1,142 million m
3
/year and the natural outflow to 489 million m
3
/year.
Furthermore Fig. 9 indicates that the aquifer contributes only an average 12% of the
annual aquifer-recharge into the rivers in the wet season, but is recharged from the rivers
in the dry season with 42% of the total recharge in dry season. Moreover, Fig. 9 shows that
over recent times, while the groundwater use has been increasing and the surface water
supply decreasing, the river-aquifer interaction has been declining.
The groundwater flow model has been used to compute historical seasonal groundwater
uses, based on the assumption that the ratio of groundwater use in the dry season is 24.3
times that in the wet season of the same year. Moreover, the results of the study show clearly
that the farmers are the major groundwater users in this region with 715 million m
3
/year,
with a ratio of groundwater use of 91%:5%:4% for the agricultural, domestic and industrial
sectors, respectively. The groundwater use patterns vary significantly with the water avail-
ability situation, as farmers are attempting to compensate the lack of surface water by
groundwater during drought years. For example, Fig. 9 illustrates that the groundwater use
runs inversely with the surface water use, and that during the drought years 1994 and 1999,
an increasing amount of groundwater had to make up for the scarcity of surface water. The
conjunctive use ratios of groundwater and surface water as a function of water-demands in
Figure 8. Average seasonal water balance in year 2003 (unit:million m
3
/day).
response to the surface water situation are listed in Table 2. One notices that the groundwa-
ter use ratio increases when less surface water can be supplied. However, inside the irrigation
area, the groundwater use ratio in a drought year is not too different from that of a normal dry
year, as the irrigation policy is to restrict the irrigation water consumption in such precarious-
time situations. The average ratios of groundwater use to water demand in a wet, normal, dry
and drought year are 13%, 17%, 13%, 19% and 6%, 7%, 9%, 10% in the irrigation and rain-
fed areas, respectively. In Table 3, the conjunctive use ratios of groundwater to surface water
use to water-demands are listed separately for the various surface basins. Obviously, almost
all of the groundwater use ratios are higher inside the irrigation project than for the rain-fed
areas. The highest groundwater ratio is observed in the Nan basin where a potential high-
yield groundwater aquifer is encountered. In fact, Table 4 indicates furthermore that the
400
600
800
1000
1200
1400
1600
1800
2000
2200
2400
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
W
a
t
e
r

u
s
e

(
m
i
l
l
i
o
n

m
3
/
y
e
a
r
)
1500
1000
500
0
500
1000
1500
S
e
a
s
o
n
a
l

r
i
v
e
r
s
-
a
q
u
i
f
e
r
s

i
n
t
e
r
a
c
t
i
o
n

(
m
i
l
l
i
o
n

m
3
)
Surface water
use
Ground water use
rivers contribute to aquifers
aquifers contribute to rivers
wet dry
wet
dry
wet
dry
Figure 9. Comparison of total groundwater and surface water use with river-aquifer interaction.
Table 2. Average conjunctive use ratio as a function of the surface water situation.
Irrigation condition Water demand:SW:GW
1
Water situation Wet season Dry season Whole year
Irrigation area
Wet 1:0.74:0.05 1:0.50:0.29 1:0.66:0.13
Normal 1:0.77:0.06 1:0.37:0.38 1:0.62:0.17
Dry 1:0.77:0.05 1:0.25:0.54 1:0.68:0.13
Drought 1:0.80:0.06 1:0.63:0.53 1:0.74:0.19
Rain-fed (no irrigation)
Wet 1:0.98:0.02 1:0.54:0.46 1:0.94:0.06
Normal 1:0.98:0.02 1:0.36:0.64 1:0.93:0.07
Dry 1:0.97:0.03 1:0.24:0.76 1:0.91:0.09
Drought 1:0.97:0.03 1:0.18:0.82 1:0.90:0.10
1
If the sum of SWGW is less than 1, this points to an unmet water demand.
highest pump yields are obtained in those aquifers whose alluvial deposits are the most con-
ducive to groundwater flow.
For the pilot study area located within the Plychumphon irrigation project, where there
is a concentration of rice farming (2.5 crops/year), the groundwater pumping behaviour
has been explored in more detail. Table 5 illustrates that, whereas the groundwater use in
the pilot study area in the wet season is much higher than the average use inside the irriga-
tion area (see Table 2), it is similar in the dry season. In the irrigation project area, water is
allocated by a rotation rule, whereby farmers obtain irrigation water for one week, but
have to wait another three weeks for the next allocation. In the pilot study area, on the other
hand, farmers tend to pump groundwater when irrigation water is rotated to other farms in
order to keep their young rice alive. Even when irrigation water was allocated to farms,
water was collected in ponds or ditches along the canal that caused the total water use to
be much higher than required by the demand.
4 CONCLUSIONS AND FUTURE DEVELOPMENT OF GROUNDWATER
RESOURCES
Our study shows that conjunctive use pattern significantly varies with the surface water sit-
uation, season, aquifer characteristics and irrigation-rainfed area. The agricultural sector is
Table 3. Conjunctive use ratio of five basins during times of
normal surface water situation.
Basin Water demand:SW:GW
Irrigation area Rain-fed area
Mae Nam Ping 1:0.54:0.08 1:0.83:0.17
Mae Nam Yom 1:0.90:0.01 1:0.87:0.13
Mae Nam Nan 1:0.53:0.30 1:0.97:0.03
Mae Nam Chaophraya 1:1.00:0.02 1:0.99:0.01
Mae Nam Sakaekrang 1:0.83:0.01 1:0.99:0.01
All basins 1:0.62:0.17 1:0.93:0.07
Table 4. Groundwater use for different aquifer types.
Aquifer types Pump yield Average ratio of GW
m
3
/hr use to water demand
Flood deposits 1020 17%
Low terrace deposits 512 7%
High terrace deposits 110 2%
Table 5. Conjunctive use ratio in pilot study
area during dry season.
Water situation Water demand:SW:GW
Wet 1:0.49:0.27
Normal 1:0.32:0.36
Dry 1:0.47:0.52
Drought 1:0.63:0.68
the major user of groundwater resources in the study area (91% of the total groundwater
use). Groundwater supplements 25% of the water-demand in the wet-season and 2982%
in the dry season when there is increased surface water shortage. The major area of ground-
water use is that of the irrigation project (60% of the groundwater use). Moreover, ground-
water demand in the groundwater basin was also examined by using the conjunctive use
ratio to predict the groundwater use. Figure 10 shows three lines obtained by using differ-
ent calculation methods for the pumping rate, namely, (1) the average pumping rate, (2) an
increasing pumping rate that reflects growth and, (3) pumping rates based on the conjunc-
tive use ratio of this study, with a climate, wet and dry season, and water situation, wet,
drought etc., as observed in the past. The groundwater use was set up as a constraint and,
using the simulation model, the groundwater levels are predicted. The model results are
depicted in Fig. 10 and show the different water levels encountered with these three
approaches. One can clearly make out the one with the lowest water levels, i.e. the conjunc-
tive use ratio approach. As the groundwater drawdown is an important factor in the consid-
eration of the impact of groundwater extraction and, namely, sustainability, the groundwater
levels in the pilot area are examined with the conjunctive use ratio approach, using the same
future cyclic water demand and water situations as in the past. The simulation results for the
groundwater level in the Mae Nam Yom basin, where groundwater is abundantly extracted,
are illustrated in Fig. 11. One observes that the water levels would possibly decline by about
10m from a wet season to a drought year, as the one mimicked for year 2018.
Finally, the future fate of the groundwater table in the study ahead was predicted using
the conjunctive use ratio. Figure 12 shows that in the year 2026, the water table will have
declined by an average of 23m in a dry season under a normal water situation, relative to
that of the dry season in 2003. Therefore, the conjunctive use pattern is the key factor for
the estimation of future groundwater consumption and may assist in the proper conjunc-
tive planning, especially in the future, in order to mitigate water shortages and sustain the
groundwater resources for years to come.
33
34
35
36
37
38
39
40
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 Year
G
r
o
u
n
d
w
a
t
e
r

L
e
v
e
l

(
m
.
M
S
L
)
gw level with average pumping rate approach (1)
gw level with 0.7% growth-rate approach (2)
gw level with conjunctive pattern approach (3)
(3) conjunctive pattern
(1) average pumping rate
(2) 0.7% increasing
pumping rate
Figure 10. Groundwater levels resulting from the use of different pumping calculations.
ACKNOWLEDGEMENTS
The authors wish to thank the staff at the Water Resources System Research Unit,
Chulalongkorn University and the project staff of the Conjunctive Use between Groundwater
Groundwater Level in Mae Nam Yom basin
17
19
21
23
25
27
29
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
G
r
o
u
n
d
w
a
t
e
r

L
e
v
e
l

(
m
.
M
S
L
)
Groundwater Level
Moving average
12-month moving average
Figure 11. Groundwater level at the pilot study area examined with the conjunctive use ratio approach.
Figure 12. Simulated development of the groundwater table using the conjunctive use ratio.
and Surface Water in the upper Central Plain of Thailand for assisting with the data compi-
lation. We also acknowledge the assistance of the Royal Irrigation Department for providing
useful information on the study area. The paper could not be finished without the financial
support of the Department of Groundwater Resources, Ministry of Natural Resources, for
which we are very grateful.
REFERENCES
Anderson, M.P., Woessner, W.W. (1992) Applied Groundwater Modeling, Academic Press, San
Diego.
Azaiez, M.N., Hariga, M. (2001) A single-period model for conjunctive use of ground and surface
water under severe overdrafts and water deficit. Eur. J. Oper. Res. 133 (3), pp. 653666.
Belaineh, G., Peralta, R.C., Hughes, T.C. (1999) Simulation/optimization modeling for water
resources management. J. Water Res. Plann. Manage. 125 (3), pp. 154161.
Chulalongokorn University (1998) Engineering Feasibility Study for Groundwater Rehabitating in
Sukhothai Province.
Chulalongokorn University (1998) Engineering Feasibility Study for Groundwater Rehabitating in
Kampangpet Province.
Chun, R.Y.D., Mitchell, L.R., Mido, K.W. (1964) Ground water management for the nations future
optimum conjunctive operation of ground water basin. J. Hydraul. Div., ASCE 90 (4), pp. 7995.
Harbaugh, A.W., Banta, E.R., Hill, M.C., McDonald, M.G. (2000) MODFLOW-2000, The
U.S.Geological Survey Modular Groundwater model user guide to modularisation concepts and
the groundwater flow process. US Geological Survey Report 00-92.
Jindasagnon, S. (1997) Groundwater Modeling in the Kampangpet Province, Master Thesis, Faculty
of Engineering, Chulalongkorn University, Bangkok, Thaialnd.
Koontanakulvong, S. et al. (2002) Groundwater Potential and Demand Study for Groundwater
Management in the Northern Part of Lower Central Plain, Chulalongkorn University.
Koontanakulvong, S., Siriputtichaikul, P. (2002) Determination of Recharge Rate from Soil
Classification Map in GW Modeling. 17th World Congress of Soil Sciences, Bangkok, Symposium
No.4, Paper No.1845, pp. 1845.16.
Koontanakulvong, S., Siriputtichaikul P. (2003) Groundwater Modeling In the North Part of the
Lower Central Plain, Thailand, International Conference On Water and Environment, Bhopal,
India, Vol.Ground Water Pollution No.19, pp. 180187.
RID Royal Irrigation Department (2005) Feasibility study for improvement of the irrigation sys-
tem in Phitsanulok Area, Kingdom of Thailand: Interim Report, Eurostudio, Spanish
Development Assistance Fund.
Siriputtichaikul, P. (2003) Modeling of Groundwater Flow in the upper part of Lower Central Plain.
Master Thesis, Faculty of Engineering, Chulalongkorn University, Bangkok, Thailand.
CHAPTER 14
Modelling of the Saq aquifer system (Saudi Arabia)
Yves Barthlemy
1
, Olivier Bon
1
, Yves-Michel le Nindre
2
, Syed Munaf
3
,
Dominique Poitrinal
2
, Alexis Gutierrez
2
, Michel Vandenbeusch
1
,
Abdullah Al Shoaibi
3
& Marcus Wijnen
2
1
Antea, Direction Technique; Orlans, France
2
Brgm, Service Eau, Orlans, France
3
Abunayyan Trading Corporation, Water Resources Department, Riyadh, Saudi Arabia
ABSTRACT: From 1981 to 1985 BRGM has carried out a large-scale (375,000km
2
) hydrogeo-
logical study of the north-western part of the Kingdom of Saudi Arabia. Based on data collected dur-
ing field surveys and exploration drilling, a numerical groundwater model of the Saq aquifer and
overlying aquifers was developed, providing guidelines for irrigation-development plans. Since then,
major agricultural development has taken place and adverse drawdown affects several aquifers. In
20052006, two new models were developed to improve the geological and hydrogeological under-
standing of the area. Developing simulation models covering such a large surface area with a com-
plex hydrogeological structure requires many technical difficulties to be overcome as well as large
scale field- and remote-sensing data campaigns. Despite such difficulties, the 2006 mathematical
groundwater model appears to be an effective tool for understanding groundwater flow patterns in
the Saq and overlying aquifers.
1 BACKGROUND
The geology of Saudi Arabia comprises Precambrian basement, which is exposed in the
west, bordering the Red Sea, around which to the northeast, east and southeast are draped
Phanerozoic sedimentary rocks that decrease in age eastward. Geographically, the
Precambrian is mountainous and high, whereas the Phanerozoic forms undulating plains,
locally interrupted by ridges and scarps, that gently dip toward the Gulf and occupy the
largest part of the country (Lloyd, 1969, Lloyd and Pim, 1990).
Semi-arid climatic conditions occur in the southwest, but conditions elsewhere are
mostly arid. The average annual rainfall rarely exceeds 100mm and is normally around
50mm or less. Surface-water resources are extremely limited: perennial rivers do not exist,
flood runoff events of any significance are few, and today only scarce springs occur.
The Phanerozoic to the east of the mountain ranges consists of an extremely thick suc-
cession of sedimentary rocks one of the worlds largest sedimentary basins that contain
major sandstone and limestone aquifers. These aquifers have received recharge in the past
under wetter climatic conditions than prevail today, and thus contain large reserves of
groundwater. The sedimentary formations decrease in age towards the Gulf, but increase
substantially in thickness; they have an overall dip to the northeast, with some block fault-
ing and localized folds that are of importance to the regional oil reserves.
From 1981 to 1985, BRGM has carried out a large-scale hydrogeological study of the
north-western part of the Kingdom of Saudi Arabia (Figure 1) with the objective of assess-
ing the groundwater and the agricultural potential of this huge area of 375,000km
2
. The
study (BRGM, 1985) included extensive soil, geological, geophysical and hydrogeologi-
cal investigations, a water-point inventory, a major exploratory drilling programme
(16,000m of wells) and the creation of monitoring networks. All these activities provided
data that were synthesized in a numerical groundwater model of the Saq aquifer and
its overlying aquifer units. This model was used for simulating various abstraction scen-
arios and the results of these simulations provided guidelines for irrigation-development
plans.
Huge changes have occurred in this area during the past 20 years. The population has
grown significantly, large new areas are now being irrigated, but at the same time irriga-
tion has slowed down or stopped in other areas whilst intense groundwater drawdown is
reported from many places.
In order to provide the Saudi Arabian Ministry of Water and Electricity (MoWE) with
an accurate and efficient decision-support tool, it was necessary to update the original Saq
groundwater model. To this end, new large-scale (452,000km
2
) hydrogeological investiga-
tions (Figure 1) were conducted between 2004 and 2006 and two new models were developed.
The first one is a 3D geological model that is used for defining the horizontal and vertical
Figure 1. Location of the Saq project area.
boundaries of the geological units involved in groundwater flow. The second one is an
improved 3D groundwater model, based on the revised geological model, and on new data
on hydrogeology, the present-day status of groundwater resources and an updated assess-
ment of water demand. Its fine grid and wide extent have been defined so as to include all
natural features influencing groundwater flow as realistically as possible. The unsteady
state calibration covers the period from 1960 till 2005, and highlights the past 20 years
during which a monitoring network of 80 observation wells was operational.
2 GEOLOGY AND 3D GEOLOGICAL MODELLING
The numerical groundwater model required an accurate definition of the geological units
composing the aquifer system in terms of the geometry of the different horizons (eleva-
tion, thickness, faulting), and the lithology and related physical properties controlling the
hydraulic behaviour of these horizons.
As a first step, the thick stack of sedimentary layers composing the cover of the base-
ment rocks was subdivided into 26 more-or-less homogeneous units based on their
hydraulic role as either aquifer or aquitard. The stratigraphic nomenclature was taken from
Sharland et al. (2004) for subsurface units and from Saudi Geological Survey (formerly
Deputy Ministry for Mineral Resources, DMMR) publications for outcropping units.
Modelling of the Saq aquifer system (Saudi Arabia) 177
N
200
kilometre
100 50 0
44
46
24
42
30
28
26
40 38
34
36
Dawmat-al-J andal
Sakakah
Unayzah
Buraydah
Baq'a
Qibah
AL Ula
Tayma
Tabuk
J ubbah
Ha'il
Al Qurayyat
Sajir
Ad Dawadimi
Tawil
Sandstone
Harrats
Saq Sandstone
Wadi Sirhan
graben
Aruma
Formation
32
Figure 2. Geological map of the Saq Project area (BRGM, 2005).
To construct the complex 3D geometry of these 26 geological units, data originating
from numerous sources have been collected, before being processed by data processing
and for 3D modelling software packages. Respectively, these were:
GDM (Geologic Data Management, BRGM) database software for processing bore-
hole data, interpolations, geostatistics and 2D display (maps, cross-sections, etc.).
EarthVision (Dynamic Graphics Inc.), a 3D geological modeller used for computing

and viewing top and bottom surfaces of each horizon.
The volume to be modelled for the Saq Project (1200km1000km15km) required
information that had mostly already been interpreted and published by ARAMCO
(Arabian-American Oil Co.), USGS (United States Geological Survey), and BRGM (e.g.
Figure 2), since the 1980s. Surface data were published as geological maps with explana-
tory notes (e.g. Delfour et al., 1982; Janjou et al., 1997; Wallace et al., 1998), scientific
papers (e.g. Mahmoud et al, 1992, Vaslet et al., 2005), or books (Le Nindre et al., 1990,
Vaslet, 1990).
Subsurface data have been published as papers (e.g. Al-Jallal, 1994, Aoudeh and
Al-Hajri, 1994, Konert et al., 2001, Al-Husseini, 2004) and as DMMR reports (e.g.
van Eck et al., 1984). In a first stage, all the data were controlled, compiled, digitized,
geo-referenced, compared and combined, using GIS software and geo-databases. In a sec-
ond stage, they were processed with the 3D geological modeller to create, by an iterative
process, a set of unit top surfaces for the 26 horizons of main interest from a geological
and/or hydrogeological viewpoint (Figure 3).
Figure 3. Top Basement structure. Thickness of the colour-scale unit 1000m.
3 CONCEPTUAL GROUNDWATER MODEL
The conceptual groundwater model was defined according to the following guidelines:
Where possible, the modelled area was extended to the actual boundaries of the hydro-
geological system in order to avoid artificial boundary conditions. Where impractical
because the aquifer system extends far outside the Project area, remote boundaries
were defined at such large distances from the main abstraction areas that the latter
could be assumed not to have a perceptible effect.
A simplified, but representative, hydrogeological succession was defined. The object-

ive was to reduce the 26 units derived from geological modelling to a stack of 7 to 15
layers with contrasting hydraulic properties. This was achieved through:
Mapping of the 26 units defined by the geological modelling. For each unit three
maps were drawn: top elevation above mean sea level, depth to top below ground
level, and thickness of the unit (Figure 4). In addition, 26 east-west cross sections
(every 50km) and 10 north-south cross sections (every 100km) show the lateral
evolution of the geological structure.
Identification of the formations that play main roles in groundwater flow because
they are either major aquifers or major aquitards.
Grouping of the successive units that have similar hydraulic properties (pervious or
impervious), in order to create an alternating succession of aquifers and aquitards.
As a result of thorough geological and hydrogeological analyses:
A northwest extension was included in the groundwater model, extending into Jordan
where data could be found in previous studies and publications (Lloyd, 1969; Lloyd
and Pim, 1990; Haiste Kirkpatrick International, 1995; Mull, 2005).
N
200 100
kilometre
50 0
48
46
26
24
32
44
30
28
48
46
42
44
42
24
38
34
36
40
34
36 38
28
40
32
30
26
-
1
0
0
0
m
-
1
5
0
0
m
5
0
0

m
-500 m
5
0
0
m
0
m
0
m
-2
5
0
0
m -2
0
0
0
m
-
3
5
0
0
m
-1000 m
-1500 m
-500 m
-
4
5
0
0

m
500 m
-
1
5
0
0

m
-
4
0
0
0

m
-
3
5
0
0

m
-
3
0
0
0

m
-
2
5
0
0

m
-
2
0
0
0

m
0
m
5
0
0
m
-
5
0
0

m
-
1
5
0
0

m
-
1
0
0
0

m
-
5
0
0

m
0
m
5
0
0

m
-
1
0
0
0

m
-
5
0
0
m
0
m
Contour Lines
(m msl)
Main cities
Top elevation of Quwarah-Sarah
(mmsl)
1000 to 1500
500 to 1000
0 - 500
-500 - 0
-1000 to -500
-1500 to -1000
-2000 to -1500
-2500 to -2000
-3000 to -2500
-3500 to -3000
-4000 to -3500
-4500 to -4000
-5000 to -4500
Figure 4. Top elevation of the Quwarah-Sarah aquifer unit.
Thirteen units (seven aquifers separated by six aquitards) were selected for their
hydraulic contribution to groundwater flow.
The boundary conditions were carefully defined, and the main inflow and outflow
areas were precisely located and differentiated. The first covers recharge through rain-
fall and wadi channels, and the second includes abstraction from wells and natural
overflow (springs, sabkhas and wadi drainage).
4 AGRICULTURAL ABSTRACTION
The main groundwater abstraction in the Saq Project area is for irrigation. Abstraction is
difficult to monitor and, as a result, its variation over time during the past 20 years is not
precisely known. For this reason, it was necessary to reconstruct the withdrawal time series
and its spatial distribution.
Between 1960 and 1984, comprehensive field inventories were made of the wells and
boreholes used for domestic, industrial and agricultural purposes. These provided data on the
approximate drilling dates of the wells and the aquifers penetrated. A compilation of these
data led to an understanding of the distribution of abstraction rates between the aquifers, and
of the withdrawal-yield time series for the main abstraction areas in each aquifer from 1960
to 1982.
As irrigation sharply increased after 1980, these data could not be extrapolated to more
recent years. A fresh assessment of agricultural groundwater abstraction during the past 20
years was thus necessary through combining satellite-image processing and field surveys,
using a five-step approach:
For three periods over the past 20 years (198485, 199091, 200003), 68 satellite
images were processed to precisely map the irrigated areas and identify the types of
crop. The area covered by satellite imagery is over 375,000km
2
(Figure 5). Six crops
were distinguished: wheat and barley, maize, fodder (alfalfa), vegetables, palm trees,
and other orchards. To identify crop types, images of three seasons were analysed for
each period in order to compare the cultivated areas against the seasons.
Between these three periods, data were interpolated based on statistics of the Ministry of
Agriculture that are available for the administrative regions. The study area covers much
of seven such regions. This interpolation process was applied to each crop, in each region.
Different sources of information were used for assessing the average pumping yields
for each type of crop, in each region: field enquiries on farms, pumping data reported
in the well inventory, and national and international reference data on crop-water
requirements.
To transfer the data related to irrigation withdrawals into the model grid, a preliminary
step determined which cells of the model represented irrigated areas. The horizontal
position of the respective cells could easily be determined in a GIS from the geograph-
ical location of the irrigated area. But for associating a model layer to an irrigated area,
it was necessary to determine the aquifer tapped through several types of data: aquifer
name declared by farmer, well- and screen-depth data collected in the field, a local
geological log, water temperature, and chemical analyses.
The abstraction time series in each grid cell then were computed as follows: crop areas
were distributed into the model grid year by year; after that in each cell and for each
year, the crop areas were multiplied by the corresponding pumping rate. As irrigated
sectors cover an extremely large area, more than 14500 cells of the groundwater model
were given individual abstraction time series for the 19602005 period.
5 GROUNDWATER MODEL DESIGN AND NUMERICAL CODES
Spatial discretization in the horizontal plane is based on a regular grid of 2 2km cells.
The grid contains 458 rows (956km), 507 columns (1014km) and 13 layers, in accor-
dance with the conceptual groundwater model. This spatial discretization leads to 232,206
cells per layer. The total number of cells is 3018678, of which 1468922 are active. Figure
6 shows a 3D view of the modelling grid where the south-western corner has been
removed to show the internal structure of the layers. For the transient calibration, two dif-
ferent time-steps were used: a 1-year time-step for the period 19601983 and a 6-month
time-step for 19842005, to account for seasonal variations in agricultural abstraction.
Two computer codes, both running with finite-difference algorithms, were used:
MODFLOW 2000 (USGS) with the GMS 6.0 pre- and post-processor, a groundwa-
ter modelling computer code already used at the MoWE for modelling other aquifer
systems.
Figure 5. Crop identification from satellite images.
MARTHE 6.7, a BRGM computer code for groundwater modelling, used for its
powerful calculation capacity and its very effective pre- and post-processing functions
(Thiery, 1990; Thiery, 1993).
6 MODEL CALIBRATION STRATEGY
The numerical model was calibrated under steady-state and transient flow conditions over
two successive 20 to 25-year periods:
For the most part, steady-state conditions existed until 19601970, before significant
agricultural and urban development took place. The groundwater situation described by
Parsons Basil in 1968, after a first extensive field inventory, can be supposed to have
been close to the initial natural situation. It was mainly characterized by the existence of
numerous artesian wells and large artesian areas.
The first transient period used for calibration, from 1960 to 1985, corresponds to the ini-
tial development of modern farms with numerous irrigation pivots. During this period,
the groundwater table evolution was not monitored because there was no efficient piezo-
metric network. However, an extensive well inventory was carried out in 198284, to
assess the then current groundwater abstraction and to draw water table maps for the var-
ious aquifers. This inventory showed that even though many artesian wells still existed
(Figure 7) major drawdown was already reported from the main irrigated perimeters.
Besides, it appeared that many wells drilled in the upper aquifers were tapping shallow
groundwater, namely in the valley of the Wadi ar Rimah near Unayzah, in the depressions
of Tabuk and Al Jawf, and along the southern fault of the Wadi Sirhan graben (Figure 8).
Under arid climatic conditions, such shallow groundwater areas unquestionably induce
significant evaporation outflow. Therefore, these zones most probably were the natural
outlets of groundwater flow until the recent water table decline.
The second transient period, from 1985 to 2005, was marked by the intensive develop-
ment of modern irrigated perimeters. As the water used for irrigation has mainly
Figure 6. 3D grid of the groundwater model.
N
200
kilometres
100 50 0
48
46
26
24
32
44
30
28
48
46
42
44
42
24
38
34
36
40
34
36

38
28
40
32
30
26
Artesian heads
proving the Jauf
aquitard effect
Artesian heads proving
the Ash Shiqqah
aquitard effect
Ad Dawadimi
Sajir
AL Ula
Unayzah
Buraydah
Sakakah
Dawmat-al-J andal
Tayma
Al Qurayyat
Tabuk
J ubbah
Ha'il
Baq'a
Qibah
Head Level in Artesian Wells in 1985
(magl)
0 to 20
20 to 40
40 to 60
60 to 80
80 to 100
100 to 120
Figure 7. Aquitard effect proved by artesian wells inventoried in 198284.
N
200
kilometre
100 50 0
48
46
26
24
30
32
28
44
48
46
42
44
42
24
34
36 38
38
34
36
40
28
40
32
30
26
Potential evaporation :
300 km x 10 km x 0.01 to 0.1 m/a
= 1 to 10 m3/s
Buraydah
Unayzah
Baq'a
Tabuk Tabuk Tabuk
Tabuk
Tabuk
Tabuk Tabuk Tabuk Tabuk
Sakakah
Ad Dawadimi
Qibah
Sajir
Ha'il
Tayma
J ubbah
Al Qurayyat
Dawmat al J andal
AL Ula
Shallow Groundwater Wells inventoried in 1985
Water Table Depth (m.b.g.l.)
2.5 - 5
0 - 2.5
Figure 8. Shallow groundwater tables measured during the 198284 field inventory.
been tapped from local aquifers, very large drawdown has resulted from such pumping.
Hydrographs from the MoWE monitoring network show that cones of depression now
commonly exceed 100m, and reach even 150m in the main irrigated areas (Figure 9).
7 MAIN HYPOTHESES USED FOR THE MODEL CALIBRATION
In order to limit the effects of possibly misleading hydraulic hypotheses, the model cali-
bration was based on the following guidelines:
No-flow and drainage conditions were assigned to most of the aquifer boundaries,
i.e. very few fixed-head cells were used because there are no perennial rivers in the
model area. Moreover such boundary conditions make calibration easier but they can
erroneously influence the results where a cone of depression reaches the boundary.
Estimated recharge fluxes and drainage conditions were assigned to the outcrops, i.e.
no fixed-head cells because these would not allow dewatering on the outcrops.
As far as possible, homogeneous hydraulic-parameter distributions were assigned to

each layer in agreement with the results of field investigations. In other words,
hydraulic conductivity, specific yield and specific storage were given (close to) uni-
form values in each layer. As a consequence, the simulated groundwater flow mainly
results from the spatial (3D) structure of aquifer and aquitard distribution, derived
from the geological modelling, rather than from a questionable heterogeneous distribu-
tion of hydrogeological parameters resulting from calibration.
N
Baq'a
200 100
kilometres
50 0
48
46
26
24
28
44
32
30
48
46
42
44
42
24
38
34
36
40
34
36 38
28
40
32
30
26
Baq'a
Tabuk
AL Ula
Ha'il
Ad Dawadimi
Qibah
Tayma
J ubbah
Al Qurayyat
Dawmat-al-J andal
Sakakah
Water Table Decline (m)
overthe 1985-2005 period
1 to 5
5 to 20
20 to 60
60 to100
100 to163
Main irrigated
areas
Figure 9. Location of the main irrigated areas and water table decline from 1983 to 2005.
With these basic rules, the calibration was guided by comparison between simulated, and
measured or estimated data, specifically through piezometric maps representing the 1985
and 2005 groundwater situations in the various aquifers, and through water-level time
series recorded in the MoWE network (80 observation wells of which 59 are used for tran-
sient calibration).
8 MAIN RESULTS OF THE MODEL CALIBRATION
The various adjustments and sensitivity analyses performed on the horizontal and vertical
hydraulic conductivity, recharge, specific yield and specific storage, progressively led to
the following results:
The horizontal and vertical hydraulic-conductivity values that can be assigned to the
layers are significantly constrained by the geometry of the aquifer system, i.e. by the
hydrogeological structure (aquifer/aquitard sequence defined during the geological
modelling) and the ground-elevation distribution. Particularly, the large variations in
thickness of the aquitards modulate the spatial distribution of exchanges between the
stacked aquifers.
The south-western faults limiting the Wadi Sirhan graben probably act as a vertical
conductive plane that allows deep groundwater to seep to the surface (Figure 10).
It is most unlikely that any significant deep groundwater flow reaches the Dead Sea
despite its low elevation of 420m below sea level. The great rise of the basement act-
ing as a threshold along the eastern ridge of the Jordan Valley most probably prevents
any significant westward groundwater flow.
As usually reported for such arid climatic conditions, the actual recharge is assumed to
be very low, less than 2mm/a over 69% of the modelled area, 3mm/a over 17% and
5mm/a over 6%. The remaining 7% of the outcrops would receive between 6mm/a and
15mm/a. A few very restricted areas, covering less than 1% of the total surface, could
receive between 15mm/a and 40mm/a; these very high values might be explained by
the infiltration of wadi floods where such high-recharge areas are located. Even if more
Figure 10. East-west cross section at the latitude of the Dead Sea.
humid climatic conditions would occur, the effective deep infiltration should not be
very different. The recharge flux is limited by the vertical conductivity of the aquitards
that have to be crossed by the streamlines originating from the outcrops and flowing
towards the discharge zones.
Figure 11. Groundwater-head distribution and streamlines simulated in the Saq aquifer for 1983.
Northern Central Qassim
495
515
535
555
575
595
615
635
655
675
695
1960 1963 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005
Year
G
r
o
u
n
d
w
a
t
e
r

H
e
a
d

(
m
)
1Q135S-Meas
(BU9135) - Saq
1Q135S-Simul
1Q164S-Meas
(BU9164) - Saq
1Q164S-Simul
1Q210S-Meas
(BU9210) - Saq
1Q210S-Simul
1Q205T-Meas
(BU9205) - Kahfah
1Q205T-Simul
1Q136S-Meas
(BU9136) - Saq
1Q136S-Simul
Figure 12. Simulated against observed hydrographs in north-central Qassim.
Under natural conditions, there was considerable evaporation in the Wadi Sirhan graben
as well as in the other topographic depressions (Dawmat-al-Jandal, Sakakah, Wadi ar
Rimah, etc.). This probably explains the existence of human settlements for thousands of
years in these low areas.
As there is very little aquifer recharge, any significant abstraction is clearly ground-
water mining, mainly by dewatering of water-saturated rock.
The groundwater-head contour maps simulated by the model in 1960, 1985 (Figure 11)
and 2005 are in good agreement with the maps made from measurements done at these
times. In particular, the model fairly reproduces the large cones of depression meas-
ured in the irrigated perimeters of Tabuk, Qassim and Buraydah.
The 59 groundwater-head time series simulated by the model fit with the observed
time series in most of the observation wells (Figure 12).
The calibrated aquifer and aquitard properties are in good agreement with the values
derived from field measurements and pumping tests, as well as with the standard val-
ues usually reported for similar formations. In the Saq aquifer for example, a homoge-
neous permeability of 1.10
5
m/s was attributed to most of the formation (83%), and
a specific storage of 2.10
6
m
1
to 94% of it, which corresponds to the average val-
ues obtained from test pumping in this aquifer.
A model with 13 layers gives much better results than a model with 9 or 11 layers,
because the latter cannot satisfactorily simulate the artesian head levels in the Al Jawf
and eastern Qassim areas. In Dawmat-al-Jandal and Sakakah for example, the impact
of the Jawf Formation on the groundwater flow pattern is dual: it is an aquifer in the
horizontal direction, but an aquitard in the vertical sense. The latter role is proved by
the artesian wells that tap the underlying Tawil Formation (Figures 13 and 7). This dual
function, due to the existence of shale beds within a succession of limestone and sand-
stone, is modelled with a vertical/horizontal anisotropy factor.
After calibration, several uncertainties still affect the 2006 model. They concern mainly
rainfall recharge, storage coefficients, the hydrodynamic behaviour of the Tawil aquifer,
and the supposed existence of a linear low-permeability structure north of Tabuk.
Figure 13. East-west cross section at the latitude of Sakakah.
Compared to the earlier numerical models developed for the Saq Project area, both in
Saudi Arabia and in Jordan, the 2006 model has the two-fold advantage of covering a
large area and of having a detailed 3D geology, to which is added a precise distribution
of pumping time-series in the various aquifers. These advantages enable us to assign
accurate boundary conditions to the model or, when this criterion could not be respected,
conditions that do not skew the simulation results under transient-flow conditions.
CONCLUSIONS
Despite real difficulties due to the very large scale of the Saq Project and the complexity
of the geological and hydrogeological context, the 2006 groundwater mathematical model
appears to be an effective tool for understanding the groundwater flow pattern in the Saq
and overlying aquifers. The keys reasons of its accuracy and representativeness can be
summed up as follows:
The high quality of the geological background data imported from a precise 3D geo-
logical model, built with a powerful geological modeller.
The spatial extension of the modelled area to the actual -or to remote hydrogeological
boundaries in order to avoid arbitrary boundary conditions that could distort results.
The definition of a realistic conceptual groundwater model involving a stack of 13

aquifers and aquitards. The choice of the vertical discretization, i.e. the number of layers
to differentiate in the model, is clearly a strategic decision, because an inadequate lay-
ering can definitely alter the representative value of the model.
An accurate reconstruction of the agricultural-abstraction time series in each cell of the

groundwater model through a complex procedure involving remote-sensing analyses, sta-
tistics of the Ministry of Agriculture, field enquiries, and database and GIS processing.
A long but effective calibration process, verified by two sets of measured contour maps for
the years 1983 and 2005, and 59 piezometric time series covering the 19832005 period.
REFERENCES
Al-Husseini M.I. (2004) Pre-Unayzah unconformity, Saudi Arabia. GeoArabia, Special
Publication n
o
3 Carboniferous, Permian and Early Triassic Arabian Stratigraphy. pp. 1559, 2004
Gulf PetroLink, Bahrain.
Al-Jallal I.A. (1994) The Khuff Formation: Its regional reservoir potential in Saudi Arabia and
other Gulf Countries; depositional and stratigraphic approach. Geo 94. Selected Middle East
Papers from the Middle East Geoscience Conference. April 2527, 1994, Bahrain, pp 103119;
ed.: M.I. Al-Husseini, Gulf PetroLink, Manama, Bahrain.
Aoudeh S.M. and Al-Hajri S.A. (1994) Regional distribution and chronostratigraphy of the
Qusaiba Member of the Qalibah Formation in the Nafud Basin, northwestern Saudi Arabia. Geo
94, The Middle East Petroleum Geosciences. Volume 1. pp 143154. April 2527, 1994, Bahrain;
ed. M.I. Al-Husseini, Gulf PetrolLink, Bahrain.
BRGM (1985) Water, agriculture and soil studies of Saq and overlying aquifers. Ministry of
Agriculture and Water, Kingdom of Saudi Arabia.
Delfour J., Dhellemmes R., Elsass P., Vaslet D., Brosse J.M., Le Nindre Y.M. and Dottin O. 1982
Geologic Map of the Ad Dawadimi Quadrangle. Sheet 24G, Kingdom of Saudi Arabia. Scale
1:250.000. Geoscience Map GM-60A, Jiddah, Saudi Arabia A.H. 1403. Ministry of Petroleum
and Mineral Resources.
Haiste Kirkpatrick International; Scott Wilson Kirkpatrick (1995) Final Report on Long Term
Management of Aquifer Resources. Vol II: Study of the Rum Aquifer of South Jordan. Ministry
of Water and Irrigation. Water Authority of Jordan.
Janjou D., Halawani M.A., Brosse J.M., Al-Muallem M.S., Becq-Giraudon J.F., Dagain J., Genna A.,
Razin P., Roobol M.J., Shorbaji H. and Wyns R. (1997b) Explanatory notes to the geologic map
of the Tabuk quadrangle, Kingdom of Saudi Arabia. Geoscience Map GM-137, scale 1:250,000,
sheet 28B. Deputy Minstry for Mineral Resources, Ministry of Petroleum and Mineral Resources,
Kingdom of Saudi Arabia. 49 p.
Konert G., Afifi A.M., Al-Hajri S.A. and Droste H.J. 2001. Paleozoic stratigraphy and hydrocarbon
habitat of the Arabian Plate. GeoArabia, 6, 407442.
Le Nindre Y.M., Vaslet D. and Manivit J. (1990) Le Permo-Trias dArabie Centrale. Histoire
gologique de la Plate-forme Arabe. Volume 3. Document du BRGM n
o
193. Editions BRGM,
Orlans, 1990. 290 p., 2 ann. 9 plates, 51 fig., 4 tables, 11 photo plates.
Lloyd J.W. (1969) Investigation of the Sandstone Aquifers of East Jordan. The Hydrogeology of the
Southern Desert of Jordan. UNDP. FAO.
Lloyd J.W. and Pim R.H. (1990) The hydrogeology and groundwater resources development of the
Cambro-Ordovician sandstone aquifer in Saudi Arabia and Jordan.
Mahmoud M.D., Vaslet D. and Al-Husseini M.I. (1992) The Lower Silurian Qalibah Formation of
Saudi Arabia: an important hydrocarbon source rock. AAPG Bulletin, 76, 14911506.
Mull R. (2005) Central Water supply project for Tabuk and other coastal cities and towns.
Resources of Saq aquifer in northern Saudi Arabia Wawi (University of Hannover).
Parsons Basil (1968) Agricultural and water resources. The great Nafud sediment basin. Kingdom
of Saudi Arabia.
Sharland P.R., Casey D.M., Davies R.B., Simmons M.D. and Sutcliffe O.E. 2004 Arabian plate
sequence stratigraphy, GeoArabia, 9, 199214.
Thiery D. (1990) Software MARTHE. Modelling of aquifers with a rectangular grid in transient
state for hydrodynamic calculations of heads and flows. Release 4.3. BRGM report 4S/EAU n
o
R
32548.
Thiery D. (1993) Modlisation des aquifres complexes Prise en compte de la zone non sature
et de la salinit. Calcul des intervalles de confiance. Revue Hydrogologie, 325336.
Van Eck M., Riddler G.P., McHugh J.J. and Farasani A.A. 1984. Sirhan-Turayf phosphate project.
The discovery of a significant phosphorite resource in the Al Jalamid area. Open File Report RF-
OF-04-19. Ministry of Petroleum and Mineral Resources, DMMR, Jiddah, 1404AH. 30 p., 12 fig.
Vaslet D. (1990) Le Palozoque (ant-Permien suprieur) dArabie saoudite. Histoire gologique
de la bordure occidentale de la plate-forme arabe, volume 1. Document du BRGM n
o
191. 209 p.
Vaslet D., Le Nindre Y.M., Vachard D., Broutin J., Crasquin-Soleau S., Berthelin M., Gaillot J.,
Halawani M. and Al-Husseini M.I. (2005) . The Permian-Triassic Khuff Formation of central
Saudi Arabia. GeoArabia, 10, 77134.
Wallace C.A., Dini S.M. and Al-Farasani A.A. (1998) Geologic map of Ash Shuwayhitiyah quad-
rangle. Ministry of Petroleum and Mineral resources, Deputy Ministry for Mineral resources.
CHAPTER 15
Holocene and present recharge of the Saharan aquifers:
A numerical modelling study
Mohamedou Ould Baba Sy
1
& Mustapha Besbes
2
1
Observatoire du Sahara et du Sahel, BP 31, 1080 Tunis, Tunisie
2
Ecole Nationale dIngnieurs de Tunis, LMHE- BP 37, 1002 Tunis, Tunisie
ABSTRACT: The extremely arid conditions in the Sahara suggest that there is currently no recharge
to the Saharan aquifers. This assumption is based on work carried out on the North Western Sahara
Aquifer System flow model (NWSAS). The model reproduces the natural non-influenced reference
state of the aquifer for the year 1900. An attempt is made to reproduce this state after a 10,000 year
simulation, starting from the beginning of the Holocene period (10,000BP), during which zero
recharge is assumed to occur. The initial conditions are based on a pre-Holocene humid climate,
which has filled the reservoirs up to their overflow level. From 10,000 BP to the present, the system
behaves purely as a depletion regime. The results of the model indicates that, the present zero recharge
hypothesis is plausible in the southern part, which, according to the model, contributes a relatively
small quantity to the recharge of the NWSAS, but it is not the case for the northern part, notably on
the Saharan Atlas outcrops, which provide the main share of the present recharge. This suggests that
there has been a high recharge rate for the whole Holocene period as well as for the present, other-
wise, it is impossible to replicate the observed behaviour of the aquifer system as it is seen today.
1 INTRODUCTION
The North Western Sahara Aquifer System (NWSAS) extends across Algeria, Tunisia and
Libya over an area of more than 1 million km
2
. It contains enormous groundwater reserves,
which certain authors consider to be very old (Burdon, 1977; Pizzi and Sartori, 1984), i.e.
the observed hydraulic gradients are fossil gradients (Lloyd and Farag, 1978), or residual
ones inherited from wet periods in the Holocene. The present day recharge is zero and the
aquifers are today suffering a pure depletion regime. The water at present circulating in
this aquifer system is very old. Similar groundwater ages are found within all very large-size
aquifers, even those located in temperate regions (Raoult et al., 1998), but does this mean,
that they are no longer recharged today?
The NWSAS model (Besbes et al., 2003; OSS, 2003b; OSS, 2003c; Mamou et al., 2006;
Besbes and de Marsily, 2006), whose parameters have been calibrated on substantial historical
time-series is used to study the zero recharge hypothesis. Starting from the NWSAS model
reference state (1950) it is possible to establish: a) a natural, non influenced regime (i.e.
without withdrawals) and assumed to be that of the year 1900, b) and an initial steady state
representing the beginning of the Holocene (10000 BP). Next a transient-state simulation
can be made in order to study the system behaviour since the beginning of the Holocene,
given zero recharge. This makes it possible to track the depletion of the aquifer over the last
10,000 years. Our objective was to reproduce the natural state (1900) starting from an ini-
tial situation characterized by ancient climatic conditions (the beginning of the Holocene).
The long-term simulation technique used to assess the impact of paleoclimatic situations,
as a key to understand present conditions, has already been used by Dieng et al. (1990),
Coudrain et al. (2001) and Cedare (2001).
2 THE CLIMATE MODEL
For the last 2.5 million years, the global climate has been subject to a repetition of glacial/
interglacial cycles, with a 40,000 year periodicity between 2.5 and 0.7 million years BP and
100,000 years for the last 700,000 years BP (Bonnefille, 1993). The last glacial period
reached its peak in the interval 20,00018,000 BP. In spite of a relatively stable climate dur-
ing the last 10,000 years (the Holocene), there have been continental-scale hydrologic balance
variations of considerable amplitude, varying from a millennium time scale to a seasonal one.
The wettest forest types are known to have existed between 10,000 and 6000 BP (Kendall,
1969) and the Sahara was green from 11000 to 5000 BP according to Gasse (2003).
Forests were replaced by drier vegetation around 6000 BP (Jolly, 1992), when the current
climate set in (Rognon, 1976).
The peak of the last recharge period in the Saharan aquifers occurred around 10,000 BP.
The adopted climate model assumes that since then, climate changes have caused a pro-
gressive reduction in the rainfall which has led to the establishment of the current arid
period (Burdon, 1977; Pizzi and Sartori, 1984).
3 HYDROGEOLOGY
The NWSAS includes two principal deep aquifer layers: the Continental Intercalary (CI)
formation, which is the deepest (Horn, 1964; Salem and Baruni, 1990), and the Terminal
Complex (TC) (Bel and Demargne, 1966; Pallas, 1980; Mamou, 1990). This system has
been investigated by numerous authors since the 1960s (Geopetrole, 1964; UNESCO, 1972;
de Marsily et al., 1978; Rab, 1987; Besbes and Zammouri, 1988; Zammouri, 1990; Geomath,
1994; Besbes et al., 2003), who have helped to describe and define the precise extent of
the main aquifer formations (Fig. 1).
3.1 Main formations and basin limits
By examining the geological outcrops (Fig.1) and the cross-section (Fig. 2), it is possible
to define the Continental Intercalary (CI) as all the continental formations ranging between
the Hercynian orogeny, which drove the sea out of the Saharan platform, and the marine inva-
sion of the Upper Cretaceous. This unit includes mainly the continental Lower Cretaceous
clayey sandstone formations, associated with marine or lagunar post-Palaeozoic and pre-
Cenomanian sediments interbedded within the CI (Busson, 1970; Fabre, 1976).
This definition of the CI, which constitutes the most extensive aquifer formation of the
region, determines the limits set for the study zone of the NWSAS. These limits, based
also on the study of geological outcrops and stratigraphic logs are: a) in the north-west, the
southern slope of the Saharan Atlas, b) in the west and the south, the limits of paleozoic
outcrops, c) in the north, the southern Atlasic structure to the north of the chotts (salt water
filled depressions), and towards the Gulf of Gabs (El Hamma Mdenine Fault), d) in
the north-east, outcrops of the CI on Dahar and Jebel Nefusa, and e) in the east, and within the
Syrte Basin, the CI water becomes brackish: this transition was adopted as a limit of the
CI aquifer (OSS, 2003b).
The Terminal Complex (TC) is the name used for a group of several aquifers located in
different geological formations (Senonian and Eocene limestone and dolomite, and the
detrital Mio-Pliocene or Continental Terminal), because these aquifers are part of the
same hydraulic unit (Bel and Cuche, 1969; Eress, 1972a; OSS, 2003a, b and c; Babasy,
2005). Connections between the Senonian, Eocene and Mio-Pliocene occur across the whole
basin, except near the chotts where the impermeable Middle and Upper Eocene is present.
The Turonian aquifer is different because there is an impermeable cover of lacustrine
deposits of Senonian age, but its piezometric levels agree with those of the Senonian and
Mio-Pliocene on the border of the basin.
Holocene and present recharge of the Saharan aquifers: A numerical modelling study 193
Figure 1. Geological map of the NWSAS (OSS, 2003a and 2003c).
Figure 2. West-east stratigraphic correlation crossing the whole basin (OSS, 2003c).
3.2 Hydrodynamics of the CI aquifer
It is generally considered (Oss, 2003a, b and c) that the situation of the NWSAS aquifers
in 1950 represents a steady state that is not significantly influenced by pumping. The avail-
able published maps (Cornet, 1964; Eress, 1972b; Pallas and Bufila, 1978; Geomath, 1994)
and the 136 values registered in the NWSAS data base (OSS, 2003b) have been used to draw
up a piezometric map of the entire CI aquifer (Fig. 3) which describes a quasi-natural
hydrodynamic situation. The map highlights the recharge zones: a) the southern Atlasic
outcrops in the north-west, b) Tinrhert in the south, c) Dahar in the east, d) Jebel Nafusa in
the north-east, and e) Jebel Hassawna in the south.
The discharge zones, suggested by the end-points of the flow lines (Fig. 3) are: a) the
foggaras (springs) of Touat-Gourara and Tidikelt, b) the Tunisian outlet represented by the
El-Hamma Fault, and c) the Libyan outlet in the Ain-Tawargha area. The piezometric
anomaly on the Amguid mount can be explained by vertical leakage through faults towards
the Terminal Complex. This was confirmed by isotopic signatures for groundwater in the
area (Moulla and Guendouz, 2003).
3.3 TC aquifer hydrodynamics
The piezometric map for the Terminal Complex is derived from successive studies carried
out over the last forty years, since the publication of the first piezometric map covering the
whole Saharan Continental Terminal by Cornet (1964). Among the most significant
contributions, are those by Bel and Cuche (1969), Eress (1972a), Levassor (1975), Gefli
(1978), Pallas (1978), Idroteneco (1982), Srivastava (1981), Armines-Enit (1984), Mamou
(1990), Geomath (1994), Oss (2003a, b and c). These studies have made it possible to draw
up a piezometric map of an initial or little influenced state covering the entire north-western
Sahara basin (Fig. 4). The relatively old water levels used for this purpose (118 values)
come from the cited references and from the three national and NWSAS project data bases
(Oss, 2003a and b). The TC piezometric map (Fig. 4) highlights the recharge zones, namely:
(a) the Saharan Atlas in the north-west, (b) Dahar in the east, (c) Jebel Nafusa in the north-east,
Figure 3. Piezometric reference map (1950) of the Continental Intercalary (Oss, 2003a and b).
and (d) Tinrhert in the south. Discharge zones are concentrated on Algerian and Tunisian
chotts and the Gulf of Syrte.
3.4 Recharge and paleo-recharge
The rainfall data available at Bni-Abbs, in the Saoura watershed, contains a small con-
tent of heavy isotopes without having to account for contributions from wetter climate
periods (Conrad and Fontes, 1970; Fontes, 1976). In addition, significant tritium contents
were recorded in the Western Great Erg: 2W to 2E, 31 to 34N (Conrad and Olive, 1972;
Conrad et al., 1975), where both the CI and TC aquifers crop out and indicative of a frac-
tion of recent water. These authors admit that recharge to the Continental Intercalary and
Terminal Complex aquifers in the Western Erg cannot be strictly paleo-climatic. There are
certainly underground reserves inherited from former wet periods, but they would be
recharged intermittently by exceptional rainfall events and by ephemeral stream floods.
4 THE NWSAS MODEL
The simplified structure adopted to represent the NWSAS (Oss, 2003b) includes three
aquifer layers (Complexe Terminal, Turonian, Continental Intercalary) separated by two
aquitards (lagoonal Senonian, Cenomanian) (Fig. 5).
The model describes a square grid of 12.5 12.5km
2
, or 16523 cells representing a
developed aquifer surface area of almost 2,580,000km
2
. The aquitard layers are represented
by vertical leakage flow across them, controlled by adjacent aquifer head losses. The model
is Processing Modflow (Chiang and Kinzelbach, 2001); it is quasi three-dimensional, based
on the multi-layer assumption, according to which flow is vertical in aquitards whereas flow
in major aquifers is horizontal. This assumption is amply verified when the permeability
difference between adjacent formations (aquifer/aquitard) is significant. That is the case in
the Sahara where studies carried out (in Algeria, Tunisia or Libya) have estimated the verti-
cal permeability of the Cenomanian and lagunar Senonian to 10
10
to 10
13
m/s.
Figure 4. Piezometric reference map (1950) of the Terminal Complex (Oss, 2003a and b).
The NWSAS model was calibrated and validated by representing as accurately as pos-
sible in steady state, the variables of the system, respectively: a) the overall piezometric maps
drawn up for CI and TC, representing a steady state corresponding approximately to the
year 1950, b) the specific piezometric values observed or estimated around this date, c) the
flow of natural resurgences measured at that time and of the Algerian foggaras for CI. In
the transient state the model produced a good replication of the reference historical draw-
down series and outlet discharge series: Foggaras in Algeria, Sources of Nefzaoua and
Djerid in Tunisia, Ain Taouargha in Libya.
5 THE 1900 NATURAL STEADY STATE
The natural steady state of the two main aquifers (CI and TC) was defined as the year 1900,
corresponding to the steady state defined by the NWSAS model (Oss, 2003b) where pump-
ing is set at zero as actual pumping was almost non-existent. A simulation is carried out start-
ing from the modified NWSAS model (all pumping set at zero) to establish this reference
state. The results (Table 1) allow a comparison between the water balance of the states in
1950 and 1900 and give rise to the following observations: a) flows calculated in discharge
areas show no significant differences for the CI aquifer, because withdrawals were low in
1950 (0.3m
3
/s); and b) for the TC aquifer, where pumping was more intense (7.5m
3
/s), its
elimination caused a clear increase in spring flow and in AlgerianTunisian chotts.
An analysis of piezometric differences between 1950 and the 1900 natural state shows
that eliminating pumping causes no significant increase in water levels of the CI aquifer
Figure 5. Structural sketch of the NWSAS model (after Oss, 2003c).
Table 1. The main elements of steady state in the 1900 and 1950 water balances, calculated by the
model.
Continental intercalary 1950 1900 Terminal complex 1950 1900
Inflow (m
3
/s) Inflow (m
3
/s)
Recharge (boundaries) 9.4 9.4 Recharge (boundaries) 18.2 18.2
Turonian leakage 0.5 0.5 Turonian leakage 5.4 4.10
CambrianOrdovician supply 2 2
Total Inflow 11.9 11.9 Total Inflow 23.6 22.3
Outflow (m
3
/s) Outflow (m
3
/s)
Withdrawals 0.3 0 Withdrawals 7.5 0
Turonian Leakage 3.85 3.49 Turonian Leakage 2.4 1.07
Saoura 0.486 0.64 Ain Tawargha Springs 2.2 2.2
Foggaras 3.718 3.84 Chotts and Sebkhas; 8.2 13.84
AlgeriaTunisia
Tunisian outlet 3.89 3.92 Kaam and Sidra 1.30 1.29
Gulf of Sidra 0.105 0.105 Nefzaoua Djerid Springs 2 3.96
Total Outflow 11.9 11.9 Total Outflow 23.6 22.3
(about 10m at El Gola Touggourt in Algeria, above 12m locally in the north). As for the
TC aquifer, where pumping was more intense in 1950, eliminating it generated a consid-
erable increase, over 50m in the area of Mghaier, in Algeria.
6 THE INITIAL STATE AT THE BEGINNING OF THE HOLOCENE
We simulated the state of the aquifer at the beginning of the Holocene by assuming that
rainfall was sufficiently abundant to saturate all of the aquifer up to the level of outcrop.
To do that, we started with the NWSAS model, in its 1900 state, on which the hydraulic heads
were initialised with the piezometric heights at the beginning of the Holocene, which cor-
responded to ground elevations, where aquifer formations are outcropping. Pumping was
maintained at zero. The steady state simulation stabilized at the beginning of the Holocene
(10,000 years BP) and provided the water balance presented on Table 2.
The comparison between the 10,000 BP and the 1900 balance reveals: a) that aquifer
recharge was indeed more significant 10,000 years ago: two and a half to three times that
of today, for both CI and TC, but no more; and b) discharge at the outlets was up to four times
higher, particularly from the foggaras, reflecting the vast area of outcrop in the area. However,
the flow of the Tunisian outlet into the Gulf of Gabs did not change; this discharge area
is very far from the outcrop and recharge zones and the general gradient has not varied
significantly.
Figures 6 and 7 present the piezometric maps calculated for the beginning of the Holocene.
The analysis of piezometric differences between the start of the Holocene and those of the
natural state in 1900 allows us to conclude that, for the CI (Fig. 6), the piezometry at the
beginning of the Holocene was 100 to 150m higher than that of the natural state in 1900,
in the centre of the basin. Upstream, this increase can reach 200m on the Saharan Atlas,
Dahar, Jebel Nefusa and Adrar Ben Drich. For the TC aquifer (Fig. 7), the piezometry at
the beginning of the Holocene was 50 to 100m higher than that of the natural state in 1900,
in the area of the Grand Erg Oriental. This increase can reach 200m in the North of the
chotts and at Jebel Nefusa.
Table 2. Main Water balance elements of the NWSAS respectively in 10000 BP and 1900.
Continental intercalary 10,000 BP 1900 Terminal complex 10,000 BP 1900
Inflow (m
3
/s) Inflow (m
3
/s)
Recharge (boundaries) 28.42 9.4 Recharge (boundaries) 43.38 18.2
CambrianOrdovician supply 0.59 2
Outflow (m
3
/s) Outflow (m
3
/s)
Withdrawals 0 0 Withdrawals 0 0
Saoura and S. Timimoun 2.25 0.64 Ain Tawargha Springs 3.65 2.2
Foggaras 17.37 3.84 Chotts and Sebkhas 29.09 13.84
AlgeriaTunisia
Gulf of Sidra 1.09 0.105 Nefzaoua Djerid Springs 13.96 3.96
Total Outflow 29.9 11.9 Total Outflow 53.2 22.3
7 ANALYSIS OF AQUIFER DEPLETION DURING THE HOLOCENE AND
THE ASSUMPTION OF ZERO RECHARGE
Burdon (1977) was one of the first to propose a pure depletion model with an assumption
of zero recharge for the Saharan aquifers. This assumption was considered by Idrotecneco
(1982) who notes that the natural state of the aquifer system results from a very long
period (10,000 years) of natural depletion (see also Pizzi and Sartori, 1984 and Thorweihe &
Heinl, 1996; 2002) and Cedare (2001) working in the Nubian sandstone aquifer system.
The natural hydraulic gradients of these aquifers were fossil gradients, or residual ones,
inherited from the wet periods at the beginning of the Holocene, when aquifers were sat-
urated until overflowing at the outcrops (Fig. 8a). The current recharge was zero (Fig. 8b)
and the aquifers were in a simple depletion regime (Fig.8c). This hypothesis has been
checked using simulations with the NWSAS model.
Figure 6. Piezometric map of the CI at the beginning of the Holocene.
Figure 7. Piezometric map of the TC at the beginning of the Holocene.
7.1 Simulation of the Holocene period
Based on the initial state at the start of the Holocene, the zero recharge assumption means
that the aridity was established, the recharge was zero, and the aquifers evolved through a
period of natural depletion for the whole of the Holocene period. The model was also run
by disconnecting the prescribed heads at the outcrops and running the simulation for
10,000 years. The results of are reported in Table 3, Figs. 9 and 10. The end of the deple-
tion phase, which lasted from the beginning of the Holocene, should replicate the 1900 ref-
erence state, described above.
The comparison of the two water balances: end of the 10,000 year depletion (Table 3)
and natural steady state (1900) indicates, in addition to zero recharge, that flow in the dis-
charge areas has decreased regularly by a half for CI (Tunisian outlet and foggaras), and
Figure 8. a, b and c: Diagram of the CI aquifer in pure depletion regime during the Holocene.
Table 3. Main water balance elements of the NWSAS, respectively at the end of depletion over
10,000 years and at the natural steady state of 1900.
Continental Transient; Steady Terminal complex Transient; Steady
intercalary End of State End of State
10000 years 1900 10000 years 1900
Inflow (m
3
/s) Inflow (m
3
/s)
Recharge (boundaries) 0 9.4 Recharge (boundaries) 0 18.2
CambrianOrdovician 2.68 2
supply
Reservoir depletion 4.80 0 Reservoir depletion 2.23 0
Outflow (m
3
/s) Outflow (m
3
/s)
Withdrawals 0 0 Withdrawals 0 0
Saoura 0.03 0.64 Ain Tawargha Springs 2.16 2.2
Foggaras 1.81 3.84 Chotts&Sebkhas 2.49 13.84
AlgeriaTunisia
Gulf of Sidra 0.66 0.10 Nefzaoua Djerid 0.34 3.96
Springs
Total outflow 7.59 11.9 Total Outflow 6.76 22.3
more radically for the TC where the Tunisian springs of Nefzaoua-Djerid have disap-
peared, and the chotts become ephemeral but remain as the main discharge system of the
aquifer. Otherwise, piezometric maps calculated at the end of depletion (Figs. 9 and 10)
indicate completely different flow configurations from those actually observed in the cur-
rent situation.
In fact:
For the CI aquifer: Figure 9 indicates the main flow paths from the area of Bechar
towards the Tunisian outlet. The hydraulic gradients originating in the Saharan Atlas have
completely disappeared. This design is in perfect contradiction with the configuration
observed in the natural state (see Fig. 3). Thus, an assumption of zero recharge cannot
be considered either for the Saharan Atlas, or Dahar and Jebel Nefusa. Elsewhere the
piezometric contour lines and the calculated flow lines show a configuration similar to
the current one, in particular on the southern border of the basin at Adrar Ben Drich,
Figure 9. End of depletion piezometric map of the CI (corresponding to the year 1900).
Figure 10. End of depletion piezometric map of the TC (corresponding to the year 1900).
Tinrhert and Tademat. In these areas, the current long-term average rainfall is extremely
small (less than 10mm in Adrar Ben Drich). The zero recharge assumption thus appears
acceptable in these areas.
For the TC aquifer: Figure 10 shows flow, or rather a flux, converging towards Chott
Melrhir. The Chott Djerid is no longer a principal outlet and the artesian flow has com-
pletely disappeared from the whole area of the Chotts. This configuration is not at all
in conformity with the current situation, see the water balance in Table 3.
7.2 Depletion of the outlets and the hydraulic heads
Table 2 shows the decrease of flow to discharge areas of the CI (foggaras and Tunisian out-
lets) and the TC (chotts and springs). It shows clearly that the maximum flow rates that the
discharge system can produce are: a) 17m
3
/s in the foggaras, 5m
3
/s at the Tunisian outlet
(CI), and b) 29m
3
/s for Algerian-Tunisian chotts and 14m
3
/s for the Nefzaoua-Djerid
springs (TC). The same table indicates, at the end of the 10,000 year simulation, flow rates
estimated at nearly 2m
3
/s in the foggaras, 2m
3
/s at the Tunisian outlet, 2m
3
/s in Algerian-
Tunisian chotts and almost nothing at the Nefzaoua-Djerid springs.
7.3 Comparison between the 1900 steady state and simulated arrival of depletion
The differences in the piezometric distributions calculated by the model, between the 1900
steady state and the end of the depletion period starting at the beginning of the Holocene
were examined.
CI Aquifer: On the principal zones of aquifer recharge: Saharan Atlas, Tunisian Dahar
and Jebel Nefusa, the elimination of recharge causes very heavy hydraulic head decreases
after the depletion phase and piezometric heights are considerably lower than those of
the 1900 reference state (Fig. 11). These differences are 300m on the Saharan Atlas
and Tunisian Dahar, and 150m on Jebel Nefussa. The smaller hydraulic head differences
Figure 11. Computed piezometric differences for CI (1900 state end of depletion).
in the discharge zones (the foggaras, Saoura valley, Tunisian outlet and Tawargha) are
justified by the nature of the draining boundary condition imposed there. Anomalies
(negative differences reaching 100m) noted in the Tinrhert and Adrar Ben Drich, are
explained by the low transmissivities in this area, and should be revised in any future
refinement of the NWSAS model. Therefore, using zero recharge conditions, the model
cannot reproduce the 1900 reference state for the CI aquifer, particularly on the Atlas
recharge areas, Dahar and Jebel Nefussa.
TC aquifer: The map of hydraulic head differences between steady state in 1900 and
end of depletion reveals, for the TC (Fig. 12), that piezometric heights at the end of
depletion are always lower than those of the 1900 reference state. Very significant dif-
ferences are noted in the north-west, where they reach 400m, and about 200m on all
other upstream boundaries, e.g. in the west (Mzab), in the east on Dahar, the south of
Libya and north on Jebel Nefussa. Down gradient, at the chotts area or at Ain Tawargha
the piezometric surface is lower (50m) but never nil. With zero recharge conditions,
the model cannot reproduce the 1900 reference state, particularly in the traditional
recharge zones of the Atlas, Mzab and Dahar. The small differences noted in the dis-
charge zones are plausible and justified by draining conditions, similar to fixed heads,
imposed in these zones.
8 DISCUSSION
The model raises a number of questions. One is the level and impact of the model reliabil-
ity and the validity of parameter calibration, i.e., mainly the spatial distribution of permea-
bility, the accuracy of the piezometry, and the spatial distribution and intensity of recharge.
Another is that of the temporal variability of recharge during the Holocene, or its variation
between zero and the presumed present value.
The initial transmissivities incorporated in the NWSAS model derive from pumping tests,
past studies and models applied to various parts of the system; the NWSAS model these,
and the calibration process attempted to reproduce the reference piezometric map while
Figure 12. Computed piezometric differences for TC (1900 state end of depletion).
varying model parameters (transmissivities and leakage coefficients). There are no true
values of horizontal and vertical permeability, and the derived solution cannot be unique
but rather a first attempt at modelling all the Saharan aquifers within the same entity. The
data incorporated into the piezometric maps are not completely synchronous and were
gathered to allow an overall vision of the regional hydrodynamics the annual variability
in head is neglected in the 1950 initial state. The possibility of direct rainfall recharge on
all the outcrop areas of the CI and TC, and the eastern and western Ergs should not be
rejected (Babasy, 2005). However, in the NWSAS model, we chose to retain only recharge
from external boundaries in accordance with the classically adopted scheme in the area,
which is compatible with the outcrop distribution.
However, the results are based on the simulation of zero recharge from 10,000 BP to
1900. With a non-zero recharge (equivalent to that imposed on the NWSAS model) and
starting from the state in 10,000 BP, the state of 1900 would be created more rapidly.
Comparison with simulations using intermediate rates of recharge would involve too many
speculations both on the rates and their variation throughout the Holocene period. The
objective here was to check the feasibility of the specific hypothesis, that of zero recharge
in order to demonstrate that it is flawed. A simulation could be run for a transient state
between 1900 and 1950. However, as the NWSAS model was calibrated on the time series
19502000, the simulation was not possible without inducing a recharge component to
attain the natural current state (1900).
9 CONCLUSION
This study has explored the pure depletion regime of the aquifers in the north-western
Sahara during the Holocene given the assumption that no recharge took place. Starting from
the piezometric situation for the beginning of the Holocene, the depletion of the aquifer in the
subsequent 10,000 years was simulated without recharge. At the end of the simulation,
the piezometric configuration obtained on the southern side of the basin (Adrar Ben Drich,
Tinrhert, Tademat) is similar to the current one, which may indicate that the assumption
of zero recharge could be feasible in these southern sectors.
However, on the whole northern and central side of the NWSAS, Atlas, Dahar, Jebel
Nefussa and Mzab areas, the derived piezometric configuration is completely different
from that currently observed. Computed piezometric maps and main flow lines at the end
of depletion are wholly incompatible with the current observations, where the Saharan
Atlas is the principal origin of recharge to the Continental Intercalaire and Complexe
Terminal. This simulation shows that the assumption of current zero recharge is unaccept-
able for the Saharan Atlas, Dahar, Jebel Nefusa and Mzab areas. Recharge is necessary to
simulate the observed flow pattern and piezometry.
ACKNOWLEDGMENTS
The authors would like to thank the reviewers: A. Coudrain, P. Lachassagne and V. Durand
for their care in reviewing this paper and their constructive comments for its revision.
Thanks are also addressed to G. de Marsily for his interest and his encouragement in our
publishing the results of this work, and to the Observatoire du Sahara et du Sahel for the
financial support and the facilities to carry out the research.
REFERENCES
Armines, Enit (1984) Modle mathmatique du Complexe Terminal Nefzaoua Djerid ; Min. Agr.
Tunisie. 19 fig., 26 table, 87p.
Babasy M (2005) Recharge et palorecharge du systme aquifre du Sahara septentrional. Thse de
Doctorat en Gologie. Facult des Sciences de Tunis. Tunisie. 277p.
Bel F, Cuche D (1969) Mise au point des connaissances sur la nappe du Complexe Terminal; ERESS;
Ouargla. Algrie. 3 fig., 17 planches, 20p.
Bel F, Demargne F (1966) Etude gologique du Continental Terminal; DEC, ANRH, Alger, Algrie,
24 planches, 22p.
Besbes M, Zammouri M (1988): Extension en Lybie du modle du Continental Intercalaire Algero-
Tunisien. In Computer Methods and Water Resources. 1st Int. Conf. Morocco, 1988, Vol 1. Ed.
Ouazar & Brebbia, Comp. Mech. Pub.
Besbes et al. (2003) Systme aquifre du Sahara septentrional. Gestion commune dun bassin trans-
frontire. La Houille Blanche/N 52003. pp.128133.
Besbes M, De Marsily G (2006) Les aquifres du SASS, in Les Eaux Continentales , Acadmie
des Sciences, Institut de France, sous la direction de G.de Marsily, EDP.d.
Bonnefille R (1993) Afrique, paloclimats et deforestation. Scheresse n4, vol.4, pp.22&231.
Burdon DJ (1977) Flow of fossil groundwater. Q. J. Eng. Gol., 10, 97124.
Busson G (1970) Le Msozoque saharien. 2me partie : Essai de synthse des donnes des
sondages algro-tunisiens. Edit., Paris, Centre Rech. Zones Arides , Gol., 11, 811p. Ed.
C.N.R.S.
Cedare (2001) Regional Strategy for the Utilisation of the Nubian Sandstone Aquifer System.
Ground Water Model ; Vol. III. 91p.Cedare ed. Cairo.
Chiang WH, Kinzelbach W (2001): 3D groundwater modeling with PMWIN. A simultion system
from modelling groundwater flow and pollution. Springer, Berlin.
Conrad G, Fontes JC (1970) Hydrologie isotopique du Sahara Nord-Occidental. Isotope Hydrology
1970, A.l.E.A., Proceed. Symp. Vienne, 405419.
Conrad G, Olive P (1972) Premires donnes sur les teneurs en tritium des nappes aquifres du
Sahara nord-occidental et central (Algrie); consquences hydrogologiques. C.R. Acad. Sc.
Paris, 274, 20192022.
Conrad G, Marce A, Olive P (1975) Mise en vidence par le tritium, de la recharge actuelle des
nappes libres de la zone aride saharienne (Algrie) Journ. Hydrol., 27, 207224.
Cornet A (1964) Introduction lhydrogologie saharienne ; Rev. de gogr. phys. et de gol. dyn.
(2),vol. VI, fasc.1; 572.
Coudrain A, Talbi A, Ledoux, E, Loubet, M, Vacher, J, Ramirez E (2001) Subsurface transfer of
chloride after a lake retreat in the central Andes. Ground Water 39, 5, 751759.
De Marsily G et al. (1978) Modeling of large multilayered aquifer system, theory and applications,
J. of Hydrology, 36, 133.
Dieng B, Ledoux E, De Marsily G (1990) Palaeohydrology of the Senegal sedimentary basin: a ten-
tative explanation of the piezometric depressions. J. of Hydrol., 118, 357371.
Eress (1972a) Nappe du Complexe terminal. Plaquette 3, App., 7 cartes, 2 planches. OSS, Tunis ; 59p.
Eress (1972b) : Nappe du Continental intercalaire. Plaquette 2, App., 5 cartes, 4 pl.. OSS, Tunis; 46p.
Fabre J (1976) : Introduction la gologie du Sahara algrien. SNED, Alger ; Algrie. 422p.
Fontes JC (1976) Les isotopes du milieu dans les eaux naturelles.Houille blanche.Nspcial3/4,
205221.
Gasse F (2003) Palohydrologie continentale et paloclimat lHolocne. Acadmie des sciences
Colloque sur leau Water Symposium.
Gefli (1978) Survey for the development of the Central Wadi Zone & Golf of Sirte ; Groudwater
Resources ; Final Synthesis Report ; Text & App 2&3- tome1 : 151p ; tome2 : pp152418. App2:
drill.rep ; App3: water analyses, hydrogeol. cross sections, maps.- General Water Authority,
ref.:AL-WR-205. Tripoli, Libye.
Geomath (1994): Western Jamahirya System ; Hydrogeological Modelling of aquifers & well fields;
Final Report; Text & pl. General Water Authority, ref.: AW-MI-579. Tripoli, Libya.
Geopetrole (1964) Etude analogique de la nappe du Continental Intercalaire saharien, pour lorg.
Mise en valeur sous-sol saharien. Direction Gnrale des Ressources en Eau, Tunis.
Idrotecneco (1982) Hydrgeological study of Wadi Ash Shati , Al Jufrah and Jabal Fezzan area.
Annex 3 : Construction of the Model, Final Report. General Water Authority. Tripoli, Libye. 93p.
Jolly D, Bonnefille R, Roux M (1992) Numrical interpretation on a hight resolution Holocene
pollen record from Burundi. Papaeogeogr, Palaeoclim, Palaeoecol, special volume.
Kendall RL (1969) An ecological history of a lake victoria basin. Ecol Monogr; 39; pp.121176.
Levassor A (1975) Etablissement dun modle de gestion et de mise en valeur des nappes aquifres
du Complexe Terminal. Rapp. int. ENSMP-INRH. ANRH, Alger, Algrie.
Lloyd JW, Farag MH (1978): Fossil groundwater gradients in arid regional sedimentary basins,
Ground Water, 16, 388393.
Mamou A (1990) Caractristiques et valuation des ressources en eau du Sud tunisien. Thse
Doctorat es-Sciences. Univ. Paris Sud.
Mamou A, Besbes M, Abdous B, Latrech D, Fezzani C (2006) North Western Sahara Aquifer System
(NWSAS); in Non renewable groundwater resources, S. Foster & D.P. Loucks ed.; UNESCO
series on groundwater N10.
Moulla AS, Guendouz A (2003) Etude des ressources en eau souterraines en zones arides (Sahara
algrien) par les mthodes isotopiques. Hydrology of the Mediterranean and Semi-arid Regions
(Proceedings of an international symposium held at Montpellier, April 2003). IAHS. n. 278.
pp.3542.
OSS (2003a) Systme Aquifre du Sahara Septentrional. Volume 2: Hydrogologie. Projet SASS.
Rapport interne. Coupes. Planches. Annexes. Tunis, Tunisie. 275p.
OSS (2003b) Systme Aquifre du Sahara Septentrional. Volume 4: Modle Mathmatique. Projet
SASS ; Rapport interne. Annexes. 229p.
OSS (2003c) Systme Aquifre du Sahara Septentrional; Rapport de synthse; 129p.
Pallas P (1978) Water resources of the socialist peoples Libyan Arab Jamahiriya. Secretariat of
dams and water resources. S.P.L.A.J., General Water Authority. Tripoli, Libya, 55p.
Pallas P (1980) Water Resources of the Socialist Peoples Arab Libyan Republic, In Salem &
Busrewille (ed): The Geology of Libya; Ac. Press, Vol. II
Pallas P, Bufila T (1978) Tawurgha spring. The socialist peoples Libyan Arab Jamahiriya
(S.P.L.A.J.). Secretariat of dams and water resources. Tripoli, Libya.
Pizzi G, Sartori L (1984) Interconnected groundwater systems simulation (IGROSS) Description
of the system and a case history application. J. Hydrol., 75, 255285.
Rab, UNDP (1987) Ressources en eau dans les pays de lAfrique du Nord, Proj. RAB 80/100.
RaoultY, Lauverjat J, Boulegue J, Olive P, Bariac T (1998) Etude hydrogologique dune ligne d-
coulement de laquifre de lAlbien dans le bassin de Paris entre Gien-Auxerre et Paris ; Bulletin
de la Socit Gologique de France; May 1998; v. 169, 3, 453457.
Rognon P (1976) Essai dinterprtation des variations climatiques au Sahara depuis 40,000 ans.
Revue de gographie physique et de gologie dynamique. (2) Vol. XVIII , Fasc. 2.3, pp.251282
Paris.
Salem O, El Baruni S (1990) Hydrogeology of the Kikla aquifer in NW Libya. International confer-
ence on groundwater in large sedimentary basins. Perth, Australia. pp.438447.
Srivastava ML (1981) Report on regional hydrogeological study in Ghadames Derj 4 Sinawen
area; Tripoli. 155 p. 18 pl. HT ; ref.: AW-GA-682.
Thorweihe U, Heinl M (1996) Groundwater resources of the Nubian aquifer system. Aquifers of
Major Basins non-renewable Water Resource, Sahara and Sahel Observatory; Tunis, 82p.
Thorweihe U, Heinl M (2002) Groundwater resources of the Nubian aquifer system. Aquifers of
Major Basins non-renewable Water Resource, Modified synthesis; OSS, Tunis , 24p.
Unesco (1972) Etude des ressources en eau du Sahara Septentrional, Rapport sur les rsultats du
Projet REG-100, UNESCO, Paris (1972).
Zammouri M (1990) Contribution une rvision des modles hydrogologiques du sud tunisien,
Thse doct. Fac. Sc. Tunis.
CHAPTER 16
Vadose zone characterisation of a hydrogeologic system in
a mountain region: Serra da Estrela case study (Central
Portugal)
Jorge Espinha Marques
1
, Joo M. Duarte
2
, Alfredo T. Constantino
2
,
Armindo Afonso Martins
3
, Carlos Aguiar
4
, Fernando T. Rocha
5
, Manuela
Incio
5
, Jos M. Marques
6
, Helder I. Chamin
7,5
, Jos Teixeira
5
, Javier
Samper
8
, Frederico S. Borges
1
& Jos M. Carvalho
7, 5
1
Dep. Geologia (CGUP), FCUP, Universidade do Porto, Porto, Portugal
2
Agroconsultores, Lda., Lisbon, Portugal
3
Dep. Edafologia, UTAD, Vila Real, Portugal
4
Centro de Investigao da Montanha (CIMO), IPB, ESA, Bragana, Portugal
5
Centro de Minerais Industriais e Argilas (MIA), Univ. Aveiro, Aveiro, Portugal
6
Dep. Eng
a
Minas e Georrecursos, IST, Lisbon, Portugal
7
Dep. Eng
a
Geotcnica (LABCARGA), ISEP, Porto, Portugal
8
ETSI Caminos, Canales y Puertos, Univ. Corua, A Corua, Spain
ABSTRACT: Understanding the role of the vadose zone is essential to accurately assess hydrogeo-
logical systems and their respective groundwater resources. The study area (Manteigas Nave de Santo
Antnio Torre sector, Serra da Estrela Mountain, Central Portugal), presents specific geological,
morphotectonic and climatic characteristics which influence the hydrogeological regime. The vadose
zone has particular features that contribute to control both the quantity and the quality of the ground-
water resources. The regional characterisation of this zone was carried out in terms of the structure, the
soil, including the broad physical, chemical and mineralogical features and the soil hydraulics. The study
included field work focused on geological and pedological features, soil permeability field tests, labora-
tory tests (including soil water retention at different pressure heads and clay mineralogy) and mathemat-
ical modelling. Water retention data were used to derive the parameters of van Genuchtens water
retention curve. Unsaturated hydraulic conductivity was estimated using the Gardner mathematical
model.
1 INTRODUCTION
Mountain areas usually support water resources of both of good quality and of significant
social and economic importance. The seasonality and spatial variability of groundwater
signatures as well as the complex role of soils, geomorphology, geology, climate, land cover
and human activities in the hydrology of mountain areas is difficult to model, even when the
relevant data are available. Understanding the role of the vadose zone is essential to assess
hydrogeological systems and the respective groundwater resources accurately (e.g. Dingman,
1994; NAP, 2001). Consideration should be given to soil formation factors such as parent
material, topography, climate, organisms and human action, as these may also control the vol-
ume of infiltration and groundwater recharge.
The study area is the Manteigas Nave de Santo Antnio Torre sector, located in the
Serra da Estrela mountain (Central Portugal), that is, the drainage basin of the River Zzere
upstream of Manteigas village (Figure 1). The regional aquifer system presents specific
geological, morphotectonical and climatic characteristics, which contribute to control the
local hydrogeological regime. Therefore, the corresponding vadose zone has particular fea-
tures that control both the quantity and the quality of the groundwater resources.
The study was performed between 2003 and 2005 within the scope of the multidiscipli-
nary hydrogeologic R&D Project HIMOCATCH (Espinha Marques et al., 2005). This
regional-scale vadose zone characterisation was supported by geological, pedological and
hydrogeomorphological analyses and focused on the physical, chemical and mineralogical
features of the soil, vadose zone structure and the soil hydraulics. Soil hydraulic and min-
eralogical features in this sector result from the particular way in which the pedogenetic fac-
tors act in a mountain environment.
Figure 1. Geological framework of Serra da Estrela region (adapted from Geological Map of
Portugal, 1/50,0000, 5th Edition; Oliveira et al. 1992).
Vadose zone characterisation of a hydrogeologic system in a mountain region 209
The tasks carried out included fieldwork focused on geological and pedological features,
field and laboratory tests and mathematical modelling. Soil samples were collected at
selected sites. Soil permeability field tests were performed. Laboratory tests were conducted
in order to determine, among other aspects, water retention at different pressure heads and
clay mineralogy. Unsaturated hydraulic conductivity was estimated with the Gardner math-
ematical model. Water retention data were used to derive the parameters of the van Genuchten
water retention curve. With the resulting characteristic curves it was possible to compare
the Serra da Estrela soil with soils from other locations. Additionally, the effect of texture and
organic matter on soil water retention was evaluated.
2 HYDROGEOLOGIC FRAMEWORK
The Serra da Estrela region is part of the Central-Iberian Zone of the Iberian Massif (Ribeiro
et al., 1990). The main lithotypes found in the region are (Figure 1): i) Variscan granitic rocks;
ii) Precambrian-Cambrian metasedimentary rocks; iii) alluvium and Quaternary glacial
deposits. The largest regional tectonic structure is the NNE-SSW Bragana-Vilaria-
Manteigas fault zone (BVMFZ). According to Ribeiro et al. (1990), the origin of the Serra da
Estrela Mountain is connected to an uplift process related to the reactivation of the BVMFZ
megastructure during Cenozoic times, by the Alpine compressive tectonics, together with the
reactivation of major ENE-WSW trending reverse faults (such as the Seia-Lous fault).
An important issue related to the infiltration and aquifer recharge in the Serra da Estrela
region is the identification of areas of prevailing fractured or porous media (Espinha
Marques et al., 2005). In particular, porous media occur in alluvium and Quaternary glacial
deposits as well as in the weathered and brittle/sheared granites and metasedimentary rocks.
Porous media usually occur at shallow depths (typically less than 50m). On the other hand,
fractured media occur in weakly weathered rocks. Such media may be present very close to
the surface (especially in areas where granite outcrops dominate, with thin or absent sedi-
mentary cover) or below the shallow porous materials. The regional hydrogeological units
correspond closely to the major geological features (Table 1): i) sedimentary cover, includ-
ing alluvium and Quaternary glacial deposits; ii) metasedimentary rocks, including schists
and graywackes; and iii) granitic rocks.
Serra da Estrela is the highest mountain in the Portuguese mainland (with an altitude
reaching 1993m a.s.l.) and is part of the Cordilheira Central, an ENE-WSW mountain range
Table 1. Main hydrogeological features of the Serra da Estrela mountain region.
that crosses the Iberian Peninsula. This region shows distinctive climatic and geomorphologic
characteristics that play an important role in the local water cycle. The river Zzere drainage
basin upstream of Manteigas has an area of around 28km
2
with an altitude ranging from
875m a.s.l., at the streamflow gauge measurement weir of Manteigas, to 1993m a.s.l., at the
Torre summit (Figure 2). The relief of the study area consists mainly of two major plateaus,
separated by the NNE-SSW valley of the Zzere River (Vieira 2004): the western
TorrePenhas Douradas plateau (14501993m a.s.l.) and the eastern Alto da PedriceCurral
do Vento plateau (14501760m a.s.l.). Late Pleistocene glacial landforms and deposits are
distinctive features of the upper Zzere catchment, since the majority of the plateau area was
glaciated during the Last Glacial Maximum (e.g. Daveau et al., 1997; Vieira, 2004).
Espinha Marques et al. (2005) proposed 9 hydrogeomorphologic units, based on the spatial
distribution of lithological, geomorphological and climatic features (Figure 2): (i) the eastern
plateau; ii) Zzere valley eastern slopes; iii) Lower Zzere valley floor; iv) Nave de Santo
Antnio col; v) upper Zzere valley floor; vi) Zzere valley western slopes; vii) Cntaros
slopes; viii) lower western plateau; ix) upper western plateau.
The Serra da Estrela climate has a Mediterranean climate, i.e., dry and warm summers, and
the wet season extends from October to May, and the mean annual precipitation is about
2500mm on the Torre summit and more than 2000mm on the plateaux (Daveau et al., 1997;
Vieira and Mora, 1998). Precipitation seems to be controlled mainly by the slope orientation
and the altitude. According to Vieira and Mora (1998), the warmest month is July and the
coldest is January. Mean annual air temperatures are below 7C in most of the plateaux area
and, in the Torre vicinity, they may be as low as 4C. The spatial and temporal irregularity of
snow-related phenomena has been analysed in earlier studies (e.g., Mora and Vieira, 2004).
3 VADOSE ZONE CHARACTERISATION
Since soils contribute to control both the volume of recharge and the water chemistry in an
aquifer system, they were the main target of the regional-scale vadose zone study at Serra
da Estrela. Soil features greatly influence the volume of water that infiltrates during a pre-
cipitation event, so controlling aquifer recharge as well as the short-term stream response.
The most important factors usually identified that affect the balance between infiltration
and overland flow are the amount and temporal distribution of precipitation (or irrigation), the
physical and chemical soil features (e.g. texture, organic-matter content, saturated hydraulic
conductivity at the surface, clay mineralogy, presence of water-repellent substances), previous
soil water saturation, surface slope and roughness, land cover, land use and amount of evapo-
transpiration (e.g., Dingman, 1994). The vadose zone study included several field studies car-
ried out in 2004 and 2005. During these studies, soil samples were collected for physical,
chemical and mineralogical characterisation (Figure 2). Detailed vadose zone studies were
carried out at specific sites. Table 2 describes each study site in terms of parent material, relief,
land cover and dominant soil profile.
The parent material of all studied soils is granite or granite derived glacial deposits. These
soils are coarse-textured, as shown in Figure 3, obtained from analyses of A and C horizon sam-
ples. In the A horizon, particles coarser than 2mm (mainly fine and medium sized pebbles)
correspond to an average value of 34% by weight (with a standard deviation of 12%). Other A
horizon features (bulk density and total porosity, determined on soil cores, and pH and organic
matter content, determined in the fine earth fraction) are presented in Table 3. The soil cores
Figure 2. Hypsometric features of the river Zzere drainage basin upstream of Manteigas; vadose
zone study sites; hydrogeomorphologic units: Eastern plateau (1); Zzere valley eastern slopes (2);
Lower Zzere valley floor (3); Nave de Santo Antnio col (4); Upper Zzere valley floor (5); Zzere
valley western slopes(6); Cntaros slopes (7); Lower western plateau (8); Upper western plateau (9).
were obtained with metal cylinders following the recommendations of Topp et al. (1993) in
order to get representative samples and to minimize soil disturbance during collection and
transportation. Additionally, Table 3 includes the volumetric water content () measured in the
relevant soil cores through the pressure-plate method (e.g., Topp et al., 1993) in A horizon sam-
ples at different pressure heads (h): 50cm, 100cm, 500cm and 15850cm. The soil
cores were saturated inside their metal cylinders prior to insertion into the pressure plate device,
with precautions taken to avoid air entrapment. Thus, the resulting values of volumetric water
content at saturation were very close to the total porosity.
Table 2. Main features of the vadose zone study sites.
Site Parent material Topography/Geoform Land cover Prevailing
soil profile
1 Glacial deposit Base of slope Maritime pine woodland A-C
2 Glacial deposit Base of slope Genista florida and
Cytisus sp.pl. scrubland A-C
3 Glacial deposit Valley floor Meso-hygrophilous grassland A-C
4 Glacial deposit Base of slope Meso-xerophilous grassland A-C
5 Glacial deposit Col Nardus stricta grassland A-C
6 Granite Base of slope Heathland A-C or A-C-R
7 (I, II) Glacial deposit Base of slope Heathland A-C
8 Granite Plateau Nardus stricta grassland A-R
9 Granite Plateau Common juniper shrubland A-R
10 Granite Slope Quercus pyrenaica forest A-B-C-R
Figure 3. Soil texture classification; texture classes: sand (I), loamy sand (II) and sandy loam (III).
The spatial distribution of soil organic matter in the A horizon depends principally on
altitude (Figure 4), mainly due to a climatic effect well known in Northern and Central
Portugal (e.g., Agroconsultores and Geometral, 2004; Pereira et al., 2006). Higher soil
organic matter contents are usually found in upper regions characterised by minimum mean
yearly air temperature and maximum precipitation. In fact, soils from the Torre area (Figure
2) present the highest measured organic-matter contents (240.5g kg
1
), whereas soils
from lower areas have measured values ranging from 48.8 to 84.1g kg
1
. A significant
effect of organic matter versus bulk density and total porosity (with important implications
for soil hydraulic properties) was observed: more organic soils tend to have lower bulk den-
sity and higher total porosity. However, this is a regional trend which does not exclude the
influence of other factors (such as land cover, land use or geomorphology among others),
especially relevant on a local scale.
Mineralogical analyses of the fine soil fractions (silt and clay) reveal a clearly detrital min-
eralogy, rich in phyllosilicates (mainly micas), quartz, plagioclases and K-feldspars, with, as
accessory minerals, siderite, opal C/CT, anhydrite, hematite, ilmenite, anatase, zeolites
Table 3. Main features of the A horizon.
Site Bulk Total pH Organic matter (% of volume) at different h values
density porosity (g kg
1
)
(g cm
3
) (%) 50cm 100cm 500cm 15850cm
1 1.25 44.7 4.6 71.7 23.3 19.9 14.3 12.3
2 1.26 49.6 4.6 57.4 29.3 23.8 15.7 12.2
3 1.16 52.7 4.8 71.6 34.4 30.8 22.4 18.6
4 1.21 42.7 4.4 48.8 23.5 19.5 14.3 11.5
5 0.94 49.7 4.1 133.1 42.0 33.7 25.5 22.2
6 1.07 45.2 4.7 103.8 29.3 23.8 17.8 16.4
7I 0.88 51.9 4.1 187.5 44.7 37.2 26.8 25.2
7II 1.09 53.4 4.3 71.1 35.7 28.5 21.5 19.1
8 0.68 62.2 4.3 240.5 57.3 50.9 40.8 38.1
9 0.88 55.6 4.3 188.4 46.1 37.7 28.7 26.9
10 0.96 48.4 4.8 84.1 31.2 26.0 18.8 15.8
Average 1.03 50.6 4.5 114.3 36.1 30.2 22.4 19.8
Figure 4. Relation between soil organic matter and altitude.
(mainly heulandite-clinoptilolite), gibbsite, rozenite and jarosites. Samples related to gran-
ites show higher amounts of phyllosilicates.
As a general rule, illite is the predominant clay mineral, but samples related to granites
show a trend toward a relative decrease of illite contents whereas kaolinite and, in some sam-
ples, vermiculite-Al and/or smectite, increase. Significant amounts of vermiculite-Al occur at
sites 1, 3, 4 and 10 (the sampling sites situated at lower altitudes), particularly in their A hori-
zons. On the other hand, the expansive clay minerals (smectite and irregular mixed-layers
illite-smectite) show a relative increase at sites 2, 6, 8, 9 and 10 (essentially, soils developed
on granite).
According to the soil map of the Serra da Estrela region (Agroconsultores and Geometral,
2004) the following pedologic units occur in this sector: (i) Humic, Leptic and Skeletic
Umbrisols; (ii) Lithic and Umbric Leptosols; (iii) Umbric Fluvissols; (iv) Rock outcrops.
Field observations produced further information concerning the pedologic units, their spa-
tial distribution according to the hydrogeomorphologic framework and their hydrologic clas-
sification by means of the Hydrologic Soil Groups system (from low runoff potential soils,
group A, to high runoff potential soils, group D) (e.g. USSCS 1964, Langan and Lammers,
1991; Boulding, 1993).
Some of the most distinctive features of the vadose zone in this sector of Serra da Estrela
are the wide distribution of: (i) granitic rock outcrops and (ii) an umbric A horizon and
consequently, Umbrisols and Umbric Leptosols thus reflecting the high organic-matter
content in the upper part of the soil profile. On the plateau and slopes of hydrogeomorpho-
logic units 1 and 2 (see Figure 2) similar sets of pedologic units occur: Leptic Umbrisols
dominate (especially in unit 2); rock outcrops and Umbric Leptosols are secondary.
Skeletic Umbrisols are dominant in units 3 (valley floor) and 4 (col); Humic Umbrisols
and Umbric Fluvisols are subdominant. In unit 5, rock outcrops prevail, whereas Umbric
Leptosols and Fluvisols are subdominant; minor occurrences of Leptic Umbrisols were
observed. Rock outcrops are also dominant in the remaining units (6 to 9) but in the upper
part of unit 6 slopes, Lithic and Umbric Leptosols are subdominant, while in the lower part
of the slopes these soils are replaced by Skeletic and Leptic Umbrisols. In the rocky slopes
of unit 7, there are minor occurrences of Umbric and Lithic Leptosols. In the plateau area of
units 8 and 9 Lithic and Umbric Leptosols are subdominant whilst smaller areas of Leptic
Umbrisols were found in unit 8.
Four types of vadose zone structures were identified in the region:
Type (i) Composed of a single granite layer with very thin or absent soil cover; present
in granitic outcrop areas of plateaus and slopes. Water circulation in fractured medium.
Included in soil hydrology group D.
Type (ii) Composed of a soil layer typically less than 0.5m thick overlying a continuous
and hard granitic layer; described at sites 6, 8 and 9 (see Table 2); present on plateaus,
especially above 1600m a.s.l., and slopes. Coexistence of porous and fractured media.
Corresponding to Lithic and Umbric Leptosols (both included in soil hydrology group D).
Type (iii) Composed of a soil layer frequently between 0.5 and 1.0m thick overlying a
weathered granite layer and/or a slope deposit; described at sites 6 and 10; it is present
on lower altitude slopes and plateaux (where chemical weathering processes are more
active) or along tectonised zones. Both porous and fractured media. Corresponding
essentially to Leptic Umbrisols with C horizon composed of weathered granite and/or a
slope deposit. These soils are included in hydrology group C.
Type (iv) composed of a soil layer frequently over 1.0m thick overlying a glacial
deposit; described at sites 1 to 5 and 7; present at the base of a slope, col and valley floor
areas; porous medium. Prevailing Skeletic and Humic Umbrisols (A, B or C hydrology
groups) and subdominant Umbric Fluvisols (C or D hydrology groups).
The importance of in situ evaluations of saturated hydraulic conductivity (K
s
) for accurate
determination of water movement in the field related to infiltration and runoff is a well estab-
lished fact (Reynolds, 1993). K
s
measured in the vadose zone is usually referred to as field-
saturated hydraulic conductivity (K
fs
). As complete saturation is not achieved in such field
tests, K
fs
can be a factor of 2 (or more) below K
s
(Bouwer, 1978; Reynolds and Elrick, 1987).
The soil permeability study consisted of a set of 40 field tests carried out at sites 1, 2, 3,
5, 6, 8, 9 and 10 (Figure 5) by means of the constant-head well permeameter also known
Figure 5. The constant-head well permeameter: general view (a), well head scale (b), water reser-
voir (c), tripod (d) and soil auger set (e).
as the Guelph permeameter method (e.g. Reynolds and Elrick, 2002). In this study, a bore-
hole 15m deep and 6cm in diameter (manually excavated with an auger) was used together
with two successively ponded heads (5 and 10cm). All tests refer to A horizon. The follow-
ing expressions were used to calculate K
fs
:
K
fs
(G
2
Q
2
) (G
1
Q
1
)
where
and
G
1
G
2
(H
2
C
1
)/ (H
1
C
2
); Q
1
AR
1
; Q
2
AR
2
H
1
and H
2
(L) are the two successively ponded heads (H
1
H
2
); r is the well radius (L); C
1
and C
2
are shape factors corresponding to f(H
1
/r) and f(H
2
/r), respectively (see Reynolds
and Elrick, 2002); R
1
and R
2
are the steady rates of fall of the water in the permeameter
reservoir (LT
1
) corresponding to H
1
and H
2
, respectively; A is the reservoir cross-
sectional area (L
2
).
The measured K
fs
values are presented in Table 4, each value representing the average
of five field tests. In all cases the permeability is high (assuming K
s
2K
fs
) according to
the SSDS (1993) classification.
An estimate of the unsaturated hydraulic conductivity (K) was also obtained using the
empirical function of K(h) by Gardner (1958):
K(h) K
fs
exp[a(h-h
e
)] if 0 a and h h
e
0
K(h) K
fs
if h h
e
h is the pressure head (L); a (L
1
) is a slope parameter that depends mainly on soil texture
and structure; h
e
is an entry pressure head which represents the air entry pressure head
for drainage from field saturation and the water entry pressure head for wetting up to field
saturation. In most natural soils h
e
0 and a a* (L
1
) (Reynolds et al., 2002). a* repre-
sents the ratio of gravity to capillary forces during infiltration or drainage. Figure 6 illustrates
the unsaturated hydraulic conductivity as a function of pressure head assuming the average
K
fs
value for Serra da Estrela soils (7.8544cm/h) and the value of a* indicated by Elrick et al.
(1989) for coarse soils (0.36cm
1
).
The hydraulic description of the vadose zone also included the determination of soil char-
acteristic curves. For this purpose, the analytical model proposed by van Genuchten (1980)
was fitted to the measured water retention values (see table 3) by means of the RETC v6.0
code (van Genuchten et al., 1991; Yates et al., 1992) available at http://www.ars.usda.gov/
internet site.
G
H C
2H H H H r H C H C
2
1 2
1 2 2 1
2
1 2 2 1

( ) ( )
l
l
ll
Van Genuchten (1980) defined the relationship between pressure head (h) and the soil
volumetric water content () by the expression:
where
n
1
1 m
and
1
h
2 1
b
1/m
1 m
( )

h
1 h
r
s r
n
m
( )
( )
l
l
l
l
Table 4. Measured values of field saturated
hydraulic conductivity.
Site Measured K
fs
(cm/h)
Average Standard deviation
1 7.7864 5.5269
2 4.7399 1.5484
3 4.1947 1.8637
5 3.3135 2.3662
6 12.7172 4.0201
8 4.6740 1.8532
9 12.8208 7.9673
10 12.5890 3.6269
All field tests 7.8544 5.4920
Figure 6. Unsaturated hydraulic conductivity (K) as a function of pressure head (h) for Serra da
Estrela soils.
and h are as defined earlier;
s
is the water content at saturation;
r
is the residual water
content; h
b
is the bubbling pressure head; mis an empirical constant affecting the shape of
the retention curve.
Table 5 presents the estimates of van Genuchten parameter values resulting from the
application of RETC code to observed volumetric water contents. The best results were
obtained considering
s
values equal to the total porosity, assuming the procedure pro-
posed by Fetter (1999); R
2
values stand for the regression of observed versus fitted val-
ues. These results are compared to average values of coarse-textured soils obtained
through experimental means by Carsel and Parrish (1988).
Figure 7 illustrates the contrast between the hydraulic behaviour of Serra da Estrela soils
and the reference soils of Carsel and Parrish (1988). It is noticeable that for analogous h values
Table 5. Estimated values of the van Genuchten soil parameters.
Site
r
(1)
(% vol.)
s
(2)
(%vol.) n
(1)

(1)
R
2
1 11.8 44.7 1.6253 0.1017 0.996
2 11.6 49.6 1.6371 0.0604 0.999
3 17.5 52.7 1.5129 0.0739 0.992
4 11.1 42.7 1.5950 0.0901 0.999
5 22.5 49.7 2.0139 0.0198 0.996
6 16.3 45.2 1.9233 0.0423 0.999
7I 25.0 51.9 2.2104 0.0170 0.999
7II 19.0 53.4 1.9001 0.0394 0.999
8 38.0 62.2 2.0875 0.0148 0.999
9 26.9 55.6 2.1947 0.0205 0.999
10 15.4 48.4 1.6710 0.0523 0.999
Average studied soils 19.6 50.6 1.8519 0.0484
Average coarse-textured soils
(3)
5.6 41.7 2.2833 0.1147
(1) estimated; (2) assumed equal to total porosity according to Fetter (1999);
(3) after Carsel and Parrish (1988).
Figure 7. Water-retention curves: Serra da Estrela soils (1) and average coarse-textured soils (2).
Serra da Estrela soils present higher values. This effect becomes more apparent in the curve
region corresponding to greater suctions (more negative pressure heads), that is, where most
water is retained in micropores. Since these soils have relatively high organic-matter and
low clay contents (see Table 3 and Figure 3) the observed trend should be explained by the
former feature. The correlation coefficients (R) of organic matter versus water retention
at 50cm and 15850cm (0.92 and 0.94, respectively) support this idea.
4 CONCLUDING REMARKS
Given that the role of the vadose zone is essential to understanding the genesis and evolution
of groundwater resources, a multidisciplinary approach supported by geological, pedological
and hydrogeomorphological analyses was used , providing important information to achieve
a regional scale characterization of the vadose zone in the hydrogeological system of the
Manteigas- Nave de Santo Antnio-Torre sector. This description focused on the broad phys-
ical, chemical and mineralogical soil features, vadose zone structure, and soil hydraulics.
Although coarse textures and low pH tend to be common soil features in Serra da Estrela,
organic-matter content and, therefore, bulk density and total porosity, are related to altitude
due to the climatic effect. The mineralogical soil study focused on clay mineralogy. In fact,
fine- fraction soil mineralogy is closely related to a detrital origin (with absolute predom-
inance of quartz, mica/illite and feldspars). Nevertheless, some distinctive features are evident,
particularly: samples related to granites which show relatively higher amounts of phyllosili-
cates (but a decrease in illite whereas kaolinite increases).
A distinctive feature of the vadose zone in this sector of Serra da Estrela is the wide distri-
bution of granitic rock outcrops and an umbric A horizon and, consequently, Umbrisols
and Umbric Leptosols, thus reflecting the high organic-matter content in the upper part of
the soil profile.
Four types of vadose zone structures were identified, with a spatial distribution closely
controlled by geology and hydrogeomorphology. High runoff potential soil hydrology
groups (C and D) are dominant in spite of the high permeability of the A horizon, meas-
ured by means of the constant-head well permeameter. Unsaturated hydraulic conductiv-
ity was estimated with the Gardner equation.
The parameters of van Genuchten water-retention curve were derived from water-retention
data in A horizon samples. According to the resulting characteristic curves, it is noticeable
that for analogous pressure head values Serra da Estrela soils present higher water-retention
values than the average coarse-textured soils, especially in the curve region corresponding
to greater suction (in which case water is mainly retained in micropores). This contrast is
explained by the high organic- matter content of the studied soils.
ACKNOWLEDGMENTS
This study was performed within the scope of the HIMOCATCH R&D Project granted
by the Portuguese Foundation for Science and Technology (FCT) and FEDER EU funds,
contract POCTI/CTA/44235/02. Constructive comments by the two anonymous reviewers
greatly improved the manuscript.
REFERENCES
Agroconsultores, Geometral (2004) Elaborao da Carta de Solos e de Aptido das Terras da Zona
Interior Centro (Making of the Soil Map and Land Suitability of Zona Interior Centro). Instituto
de Desenvolvimento Rural e Hidrulica, Lisbon, 313pp
Boulding JR (1993) Subsurface characterization and monitoring techniques. In: The vadose zone,
field screening and analytical methods. EPA/625/R-93/003
Bouwer H (1978) Groundwater Hydrology. McGraw-Hill, Inc, New York, 480 pp
Carsel FF, Parrish RS (1988) Developing joint probability distributions of soil water retention char-
acteristics. Water Resources Research, 24,5, 755769
Daveau S, Ferreira AB, Ferreira N, Vieira G (1997) Novas observaes sobre a glaciao da Serra da
Estrela (New observations on the Serra da Estrela glaciation). Estudos do Quaternrio 1:4151
Dingman SL (1994) Physical hydrology. Prentice Hall, New Jersey, 575 pp
Elrick DE, Reynolds WD and Tan KA (1989) Hydraulic conductivity measurements in the unsat-
urated zone using improved well analyses. Ground Water Monit. Rev., 9, 3, 184193
Espinha Marques J, Marques JM, Chamin HI, Afonso MJ, Carreira PM, Fonseca PE, Cabral J,
Monteiro Santos FA, Vieira GT, Mora C, Gomes A, Teixeira J, Samper J, Pisani BJ, Aguiar C,
Gonalves JA, Almeida PG, Cavaleiro V, Carvalho JM, Sodr Borges F, Aires-Barros L, Rocha FT
(2005) Hydrogeological study of a high mountain area (Serra da Estrela, Central Portugal): a mul-
tidisciplinary approach. Cadernos Lab Xeol Laxe, 30, 145166
Fetter CW (1999) Contaminant hydrogeology. Prentice-Hall, Inc, New Jersey 500 pp
Gardner WR (1958) Some steady-state solutions of the unsaturated moisture flow equation with
application to evaporation from a water table. Soil Sci., 85, 228232.
Langan LN, Lammers DA (1991) Definitive criteria for hydrologic soil groups. Soil Survey
Horizons, 32, 3, 6977
Mora C, Vieira GT (2004) Balance radiactivo de los altiplanos de la Sierra de Estrella (Portugal) en
una maana de invierno. Metodologa y primeros resultados (Radiation balance of the plateaus of
Sierra de Estrella (Portugal) in a winter morning. Methodology and first results). Bol. R. Soc. Esp.
Hist. Nat. (Sec Geol), 99, 14, 3745
NAP National Academy Press (2001) Conceptual models of flow and transport in the fractured
vadose zone. Washington DC, 374 pp
Oliveira JT, Pereira E, Ramalho M, Antunes MT, Monteiro JH ,Coords (1992) Carta Geolgica de
Portugal a escala 1/500 000 (Geological Map of Portugal, scale 1/500 000). 5th Edition.
Portuguese Geological Survey, Lisbon.
Pereira SP, Correia AV, Correia AC, Ferreira MT, Onofre N, Freitas H, Godinho F (2006) Florestas e
Biodiversidade (Forests and Biodiversity): In: Santos FD, Miranda P (eds) Alteraes climticas
em Portugal; cenrios, impactos e medidas de remediao, Projecto SIAM II (Climate change in
Portugal; scenarios, impacts and adaptation measures, SIAM II Project), Gradiva, Lisbon, pp
305343
Reynolds WD (1993) Saturated hydraulic conductivity: field measurement: In: Carter MR (ed) Soil
sampling and methods of analysis, Lewis Publ, Boca Raton, pp 599613
Reynolds WD, Elrick DE (1987). A laboratory and numerical assessment of the Guelph permeame-
ter method. Soil Sci., 144, 282299
Reynolds WD, Elrick DE (2002) Constant head well permeameter (vadose zone). In: Dane JH,
Clarke Topp G (ed) Methods of soil analysis, Soil Science Society of America, Inc., part 4,
Physical methods, 844858.
Reynolds WD, Elrick DE, Youngs EG, Amoozegar A, Booltink HWG and Bouma J (2002) Saturated
and field-saturated water flow parameters: In: Dane JH, Topp GC (eds) Methods of soil analysis,
part 4, Physical methods. Soil Science Society of America, Inc, Madison, pp 797878
Ribeiro A, Kullberg MC, Kullberg JC, Manuppela G, Phipps S (1990) A review of Alpine tectonics
in Portugal: Foreland detachment in basement and cover rocks, Tectonophisics, 184, 357366
SSDS, Soil Survey Division Staff (1993) Examination and description of soils: In: Soil Survey
Manual. Agricultural Handbook 18, chapter 3
Topp GC, Galganov YT, Ball BC, Carter MR (1993) Soil water desorption curves: In: Carter MR
(ed) Soil sampling and methods of analysis, Lewis Publ, Boca raton, pp 569579
USSCS, U. S. Soil Conservation Service (1964) Hydrology: In: SCS National Engineering
Handbook, Section 4
Van Genuchten MT (1980) A closed-form equation for predicting the hydraulic conductivity of
unsaturated soils. Soil Sci. Soc. Am. J. 44, 892898
Van Genuchten MT, Leij FJ, Yates SR (1991) The RETC code for quantifying the hydraulic functions
of unsaturated soils. U S Environmental Protection Agency 85 pp
Vieira GT (2004) Geomorfologia dos planaltos e altos vales da Serra da Estrela: ambientes frios do
Plistocnico Superior e dinmica actual (Geomorphology of the plateaus and high valleys of Serra
da Estrela: cold environments from Late Pleistocene and present dynamics). Unpublished PhD
Thesis, Department of Geography, University of Lisbon, 724pp
Vieira GT, Mora C (1998) General characteristics of the climate of the Serra da Estrela: In Vieira GT
(ed), Glacial and Periglacial Geomorphology of the Serra da Estrela, Guidebook for the field-trip,
IGU Commission on Climate Change and Periglacial Environments, CEG and Department of
Geography, University of Lisbon pp 2636
Yates SR, van Genuchten MT, Warrick AW, Leij FJ (1992) Analysis of measured, predicted and esti-
mated hydraulic conductivity using the RETC computer program. Soil Sci. Soc. Am. J., 56, 347354
B. Large aquifers, water salinity and its
evolution with exploitation
CHAPTER 17
Reconstructing natural groundwater flow and geochemical
processes in a perturbed multi-layer aquifer system
Dallas Baird
1,2
, Corinne Le Gal La Salle
2,3
, Andrew Love
2,4
&
Craig Simmons
1,2
1
Flinders University, Adelaide, SA, Australia
2
Centre for Groundwater Studies, Adelaide, SA, Australia
3
Centre Universitaire de Formation et de Recherche de Nmes, LaboGIS, Nmes, France
4
Dept. of Land, Water and Biodiversity Conservation, Adelaide, SA, Australia
ABSTRACT: A combined hydraulic and hydrogeochemical approach is helping to improve our
understanding of a perturbed multi-layer aquifer system in the Northern Adelaide Plains, South
Australia. Excessive groundwater pumping from a confined Tertiary carbonate aquifer has reversed the
historical upward hydraulic gradient. Although significant contamination has occurred by leakage
through leaky wells, large scale leakage through a confining layer does not appear to be significant.
Carbon-14 activities verify the original groundwater flow direction. Stable isotopes of water reveal that
although subsurface flow from the adjacent mountain ranges is the dominant recharge mechanism,
areas missing an important confining layer also receive a significant amount of evaporated water. The
stable isotopes of water, together with Cl concentrations, reveal climatic changes over the past tens of
thousands of years. Groundwater chemistry is predominantly affected by evapotranspiration. Carbon-
13 and strontium isotopes indicate that carbonate mineral dissolution is also an important geochemical
process.
1 INTRODUCTION
Developing accurate conceptual models of natural groundwater systems is crucial when
determining the groundwater residence time and recharge rate, modelling the system, and
assessing the sustainable yield of the resource. Disturbances to the natural groundwater sys-
tem, by excessive groundwater extraction, for example, can hinder our understanding of
the natural groundwater flow. In situations where the present day potentiometric surface no
longer represents the natural system, historical data and geochemical studies can provide the
necessary information to reconstruct natural, undisturbed conditions and develop accurate
conceptual models (e.g. Schlehuber et al, 1989; Edmunds et al, 2002). However, contamination
of aquifers by leakage through confining layers or leaky wells can limit the usefulness of geo-
chemical tracers. It is essential when using tracers such as carbon-14 to determine groundwa-
ter recharge rates that these natural conditions are understood by identifying the geochemical
processes that affect the concentration of geochemical tracers. In addition, the effects of any
contamination by leakage through confining layers or leaky wells must also be accounted for.
One area greatly affected by excessive groundwater extraction is the multi-layered aquifer
system of the Northern Adelaide Plains (NAP), South Australia. The region relies heavily on
groundwater for an important horticulture industry. The multi-layer aquifer system consists
of a sequence of Quaternary age clays, silts, sands and gravels containing relatively saline
groundwater, overlying confined Tertiary age limestone and calcareous sandstone deposits
which provide good quality groundwater for irrigation. Prior to significant groundwater
extraction, the confined carbonate aquifers were artesian (Smith, 1979; Gerges, 2001).
Several decades of excessive groundwater extraction have seen a significant cone of depres-
sion develop in the centre of the region. This has resulted in radial groundwater flow
towards the cone of depression and reversing the vertical hydraulic gradient so that saline
groundwaters in overlying aquifers may flow downwards into the fresh carbonate aquifers.
This problem is exacerbated by numerous leaky boreholes in the region.
This paper first builds on the present understanding of the groundwater hydrology using an
environmental tracer approach. The potential for leakage is then examined, and the ground-
water chemistry is assessed for contamination. A hydrogeochemical model is then discussed
for a 30km long fresh groundwater corridor in the most heavily exploited aquifer. By deter-
mining the location and extent of anthropogenic effects, the undisturbed groundwater chem-
istry can be better understood, allowing us to reconstruct the pre-irrigation conditions.
2 HYDROGEOLOGICAL SETTING
Centred approximately 40km north of Adelaide, South Australia, the NAP is a gently sloping
to flat alluvial plain, bordered by the southern Mount Lofty Ranges (MLR) to the east, and
Gulf St Vincent to the west (Fig. 1). The area has a low relief, sloping gently from the foothills
in the east towards the coast. An escarpment in the east is formed by the Para Fault, above
which the MLR rise to more than 600m. A number of watercourses occur on the escarpment,
however, only two water courses, the Gawler River and the Little Para River, reach the
coastline. Even these are seasonal, only flowing to the sea during wetter months or after heavy
rainfall events.
The region enjoys a Mediterranean climate, with hot, dry summers and cool, wet winters.
Rainfall across the plains is less than 500mm/a, but increases to more than 800mm/a in
the more elevated areas of the adjacent hills. Potential evaporation greatly exceeds rainfall
for most of the year.
The Adelaide Plains sub-basin is a graben formed by one in a series of block faults in
Precambrian rocks of the Adelaide Geosyncline. These rocks consist primarily of siltstones,
quartzites, dolomites and finer grained sedimentary rocks deposited in largely marine envir-
onments. On the plains, Quaternary clays, sands and gravels deposited in alluvial environ-
ments overlie Tertiary limestones and calcareous sandstones (Fig.2). In the Gawler area in the
northeast of the study area, the Alma Fault Block (AFB) is formed by the Alma and Para
Faults. Here, the Tertiary aquifer is comprised of undifferentiated siliceous sediments. The
sediments in the basin form a wedge shape, which thins to the north; the thickness of
the sediments increases from approximately 100m in the north to approximately 400m in
the south.
The aquifers are numbered in order of increasing depth. The Quaternary aquifers (Q1 to Q4)
have salinities ranging from 1,000 to 15,000mg/L, with freshest groundwaters occurring near
the major watercourses. The upper Tertiary aquifers (T1 and T2) are also generally freshest
beneath the rivers, with salinities ranging from about 700 to 3,000mg/L. Two lower Tertiary
aquifers (T3 and T4) are highly saline, however, they are separated from the fresher aquifers by
thick clays and silts. The most important aquifer in the region is the T2 aquifer, a well-
cemented limestone generally 80 to 120m thick. The aquifers in the Quaternary sediments are
separated from each other and the Tertiary sediments by varying thicknesses of clay. In some
areas, these confining layers can be thin or absent. The main confining unit in the region is the
Munno Para Clay, a 68m thick layer separating the T1 and T2 aquifer. This confining layer is
missing to the north of the Gawler River and in the Gawler area in the north-east of the NAP.
Recharge to the sedimentary aquifers occurs predominantly as subsurface flow (referred to
as mountain block recharge; Wilson and Guan, 2004; Manning and Solomon, 2005).
Groundwater in fractured rock aquifers in the MLR flows into confined aquifers that abut
onto the fractured rock under the ground surface (Gerges, 1986; 2001). Hydraulic connection
across these faults was demonstrated by Gerges (1999) in an area to the south of the NAP.
The T2 aquifer provides the majority of groundwater for irrigation, particularly around the
Virginia area. Approximately 18,000ML/a is extracted from the Tertiary aquifers in the NAP
region, about 14,000ML/a from the T2 aquifer alone (Gerges, 2001). This greatly exceeds
natural recharge. A cone of depression up to 70m below natural groundwater levels has devel-
oped, resulting in radial groundwater flow. Leakage through leaky boreholes has resulted in
significant rapid local increases of salinity. Within the 750km
2
NAP area, there are approxi-
mately 4400 licensed, disused or abandoned wells, as many as 56% of which may be leaking
(Schuster, 2000). Leaky wells are therefore a significant threat to the long-term sustainability
of the fresh groundwater resource. They are also a potential source of contamination for
detailed groundwater chemistry investigations, which must be identified and corrected for.
Reconstructing natural groundwater flow and geochemical processes 227
Figure 1. Location of the study area, showing the northern limit of the Munno Para Clay, the boundary
of the T2 fresh groundwater corridor, the Para and Alma Faults, and the location of Gawler and Virginia.
3 METHODOLOGY
Data used in this investigation include historical data and samples collected specifically for
this study. The historical data provided major ions and general parameters, and enabled a more
complete coverage of the NAP and the fresh groundwater corridor. Well data and chemical
analyses are available from the Department of Water, Land and Biodiversity (DWLBC)
Drillhole Enquiry System (online database; https://info.pir.sa.gov.au/des/desHome.html).
Data from as early as the 1930s is available, when the Tertiary aquifers were still artesian.
Samples displaying anomalous chemical data such as poor ion balance or unrealistic values,
or with unsuitable well construction, were not included.
In addition to the historical data, 47 samples across the NAP were also collected from moni-
toring wells and private wells to examine other tracers in addition to major ion chemistry.
Samples were analysed for major ions, stable isotopes
2
H,
18
O and
13
C, radioactive tracers
14
C and tritium, and
87/86
Sr. All stable isotopes were analysed using stable isotope mass spec-
trometer. Approximately half of the carbon-14 samples were analysed by Liquid
Scintillation Counting (LSC) at CSIRO in Adelaide, while the remainder were analysed by
Accelerated Mass Spectrometry (AMS) at ANSTO in Lucas Heights. Tritium was analysed
after electrolytic enrichment by LSC at ANSTO Environment. Strontium isotopes were
analysed at Adelaide Universitys School of Earth and Environment Sciences.
Samples collected for this study had pH determined by a calibrated pH meter connected
to a flow cell at the well head, and display pH values consistently 0.20.4pH units lower than
the historical data in similar areas. This has significant implications for the calculation of sat-
uration indices for carbonate minerals, and is discussed further below. While the data were
collected for the entire NAP region, most of the geochemical evolution discussed here refers
to the fresh T2 corridor extending from Gawler through Angle Vale and Virginia to the coast.
4 CONCEPTUAL MODEL
The present potentiometric surface is greatly disturbed, so while the current direction for
potential groundwater flow is known, this does not reflect the natural (undisturbed) condi-
tions of the NAP. Two conceptual models are presented for the NAP: the hydrogeological
Figure 2. Simplified conceptual models for pre-1940s (a) and present (b) systems along the T2
fresh groundwater channel. Block arrows indicate lateral groundwater flow, small arrows indicate
potential vertical flow between the Quaternary and Tertiary aquifers. Q Quaternary aquifers.
T Tertiary aquifers. C confining layers. MLR Mount Lofty Ranges (fractured rock).
RF Redbank Fault. AF Alma Fault. PF Para Fault.
reconstruction of pre-1940s natural system (Fig. 2a), and the present day system in response
to several decades of groundwater extraction (Fig. 2b). Although the precise hydraulic heads
of the various aquifers are unknown before significant extraction began, the Tertiary aquifers
were flowing prior to the mid-1940s (Smith, 1979). The key features of the pre-1940s con-
ceptual model are that groundwater flowed from the Mount Lofty Ranges towards Gulf St
Vincent and groundwater had potential for upward flow. Although regular monitoring of water
levels did not begin until after significant disturbances to the system, Shepherd (1971) pre-
sented a potentiometric surface for the Tertiary aquifers for September, 1959. This indicated
groundwater flowing across the NAP in a west-south-westerly direction, perpendicular to the
coastline. The natural groundwater flow also had an upward hydraulic gradient (Gerges, 1999;
2001), with the potentiometric surface of the Tertiary aquifers being higher than the
Quaternary aquifers. In some areas, the Q2 aquifer had a higher potentiometric surface than
the Q1 aquifer. These historical hydraulic gradients, combined with salinity profiles, led
Gerges (1986) to suggest that groundwater recharge to the NAP aquifers occurred as subsur-
face flow from the adjacent mountain ranges. The present day conceptual model shows that
radial flow towards the main pumping centre at Virginia, and potential for downward flow,
now dominates.
4.1 Recharge and groundwater flow
Using environmental tracers, it has been possible to refine the conceptual model for ground-
water flow and recharge. Carbon-14 can provide information on the geochemical processes
groundwater has undergone during recharge and flow, as well as provide an indication as to
the age of the water due to radioactive decay. Carbon-14 activities across the NAP range from
1.4 to 90.3pmC (Fig 4a). Highest activities occur in the Gawler region and decrease along the
flowpath towards the coast. The uncorrected activities suggest residence times up to tens of
thousands of years. The distribution of
14
C activities confirms the direction of groundwa-
ter flow indicated by the 1959 potentiometric surface, with modern activities decreasing
with increasing distance from the MLR. A relatively high value (4.36pmC) near Virginia is
sampled from a suspected leaky well which also contains measurable tritium. The Quaternary
aquifers have generally higher
14
C activities, suggesting that they must receive some com-
ponent of modern recharge. If upward leakage from the Tertiary aquifers were the only mech-
anism of recharging the deeper Quaternary aquifers, the
14
C activities would be similar.
The stable isotopes of water,
18
O and
2
H, were examined to provide information regard-
ing the source of water, the processes it has undergone prior to recharge, and of mixing
between waters with differing signatures. In addition to the samples collected from the NAP,
existing data from samples in the MLR are also utilised for this interpretation. Data for the
MLR is from two sources, both approximately 35km south-east of Virginia. Ivkovic et al
(1998) investigated groundwaters in the Piccadilly Valley, while Radke et al (2000) examined
the much larger Upper Onkaparinga region. Although these groundwaters are located on the
eastern side of the topographic divide,
2
H values for MLR groundwaters just south of the
NAP (Dighton et al, 1994) have a similar range (
2
H between 30.0 and 24.8), justify-
ing the use of the Ivkovic and Radke data (
2
H between 31.5 and 23.2). The Ivkovic
data were sampled at a higher average elevation than the Radke data, and are slightly more
depleted due to altitude and rainfall amount effects.
The
2
H and
18
O values for all groundwaters range from 31.5 to 18.6 and 6.22
to 3.50, respectively. Most groundwaters plot along or near the Local Meteoric Water
Line (LMWL, A. Herczeg, pers.comm; Fig. 3), indicating that groundwaters are of a mete-
oric origin, and no connate or trapped seawater remains in the aquifers. Plotting along the
LMWL also suggests that minimal fractionation due to evaporation occurred prior to
recharge due to rapid infiltration of water to the water table.
Four broad, overlapping yet distinct groups of waters are observed in Figure 3. Group A
(
18
O 6.22 to 3.75;
2
H 31.5 to 21.9) represents the MLR groundwaters, which
is the group of most depleted groundwaters. These groundwaters plot along a relatively low
slope when compared with the Adelaide LMWL. However, a LMWL determined for Chain
of Ponds in the MLR (Ivkovic et al, 1998) also has a lower slope than the Adelaide LMWL.
The lower slope of the Chain of Ponds LMWL and the MLR groundwaters may be due to the
site being more inland or elevated than the Adelaide LMWL site (Ivkovic et al, 1998; Radke
et al, 2000). Evaporation of falling rain, particularly in arid environments, can lead to lower
slopes (Clark and Fritz, 1997).
Group B (
18
O 4.97 to 4.10;
2
H 28.3 to 20.1) contains the bulk of the NAP
groundwaters (except for groundwaters in the Gawler region). They are more enriched than
the MLR groundwaters although the two groups overlap, and they also plot along the LMWL.
NAP groundwaters are more depleted further along the groundwater flowpath (Fig. 4b). It is
unlikely that the depletion in
2
H along the flowpath can be due to mixing with saline ground-
waters from overlying or underlying aquifers as a consistent increase in Cl
, which is not
observed, would also be expected. It is also unlikely that groundwater from varying elevations
in the Mount Lofty Ranges contribute in varying amounts to the groundwater along the same
flow line. A similar trend was observed by Love et al (1994) for a coastal aquifer system in the
Otway Basin, also located in SE Australia. Possible causes for the variation included lower
temperatures resulting from climatic changes and continental effects caused by sea level fluc-
tuations, which resulted in a coastline 200300km further west during the last glacial maxi-
mum. Similar processes would also have influenced rainfall in the NAP region.
Group C (
18
O 3.88 to 3.50;
2
H 23.6 to 18.6) comprises of groundwaters
in the Gawler area of the NAP where the Munno Para Clay is absent. They are enriched, and
appear to have been subjected to slight evaporation, or have mixed with evaporated water.
Group D (
18
O 3.42 to 3.05;
2
H 21.8 to 17.5) includes all but two of the
-32.0
-28.0
-24.0
-20.0
-16.0
-6.50 -6.00 -5.50 -5.00 -4.50 -4.00 -3.00 -2.50
2
H
MLR Groundwaters
MLR Surface Waters
Qty
Q4
T1
T2
T3
Adelaide LMWL
Chain of Ponds LMWL
A
B
C
D
18
O
-3.50
Figure 3.
18
O and
2
H values ( VSMOW) values for groundwaters and surface waters in the MLR
and NAP.
MLR surface waters and show considerable evaporation effects relative to the groundwaters.
The two other MLR surface waters are depleted and plot with the MLR groundwaters, pos-
sibly reflecting baseflow contributions or sample collection shortly after rainfall.
With the exception of groundwaters in the Gawler area, the bulk of NAP samples (Group
B) overlap the MLR groundwaters, but are notably more enriched. This enrichment does
not appear to be due to evaporation, as they plot along the LMWL. This suggests that
recharge to the NAP aquifers is predominantly by subsurface flow from the MLR, however,
this recharge must come from lower elevations in the MLR where the groundwater isotope
signature is less depleted. Groundwaters in the more elevated parts of the MLR (Group A)
have relatively depleted values, and must discharge elsewhere.
Groundwaters in the Gawler area (Group C) plot between NAP groundwaters and surface
waters, suggesting either mixing between surface waters and MLR groundwaters, or contri-
bution from direct recharge across the plains. Heavy groundwater extractions in this area
combined with thin confining layers means that downward leakage is possible. It is difficult
to confirm this, however, as the stable isotope values for the Quaternary aquifer ground-
waters do not all display such enriched signatures. It may also be possible that infiltration
of stream and river water runoff from the Para Fault scarp face, either at the fault or in the
foothills above the fault, provides this enriched signature. This suggests that an evaporated
source of water is important to the water balance of this area.
4.2 Potential for leakage
The present day conceptual model (Fig. 2b) shows that there is potential for downward leak-
age to the T2 aquifer across the entire NAP. It is essential to determine whether leakage is
occurring in order to use
14
C ages to establish groundwater recharge rates. If recent
14
C has
been introduced, recharge rates estimated will be greater than the actual rate. Leakage to the
confined aquifers is known to occur through leaky wells, where sudden increases in TDS or
Cl concentration are the first indicators of such leakage. The presence of tritium or nitrate
can also indicate contamination, and may occur where subtle changes in TDS or Cl concen-
tration are not detected. Conversely, the absence of nitrate or tritium does not necessarily
mean contamination has not occurred, as they may be removed over time by denitrification
and tritium radioactive decay.
Measured tritium values in the NAP range from 0.4 to 1.4 TU. Although tritium was
recorded in wells across the NAP, indicating contamination, the groundwaters in the
Gawler area also showing enriched stable isotope signatures all contained measurable tri-
tium. This may indicate contamination by leakage or it could be due to natural recharge.
Tritium concentrations in the MLR indicate relatively rapid movement to considerable
depth. Ivkovic et al. (1998) report tritium values greater than 1.4 TU at 200m depth.
Therefore, high tritium concentrations in the Tertiary aquifers near to the subsurface
recharge boundary may result from subsurface flow. Measurable nitrate levels in the NAP
ranged from 0.01 to 114mg/L, although most contained less than 6mg/L. While 39% of
all Tertiary groundwater samples contained NO
3
(50% of T1, 33% of T2), nearly 71% of
all Quaternary aquifer groundwater samples contained measurable NO
3
. Approximately
20% of T2 wells specifically sampled for this study contained measurable nitrate. These
values reveal the extent of leakage that is occurring in the NAP. A combination of Cl con-
centration increases, the presence of NO
3
or anomalously high
3
H or
14
C could be used to
indicate potential contamination.
Gerges (2001) used salt balances to determine that lateral through-flow has replenished
groundwater removed from storage, and not downward leakage from the relatively saline
Quaternary aquifers. Leakage from the Quaternary aquifers to the Tertiary aquifers
through leaky wells was estimated at between 600 and 1600ML/yr. It is also possible to
calculate how long downward leakage may take from the overlying saline aquifers to the
T2 aquifer through the main confining layer using Darcys Law:
V
z
K
v
i
v
/n
e
where V
z
is the vertical average linear velocity, K
v
is the vertical hydraulic conductivity of
the confining layer, i
v
is the vertical hydraulic gradient across the layer and n
e
is the effective
porosity of the confining layer. Along the T2 fresh groundwater corridor in the most heavily
pumped area, the T2 aquifer is separated from the more saline T1 aquifer by 6m of Munno
Para Clay. The average K
v
for this unit is 2.110
6
m/day, and the minimum recorded poros-
ity is 0.37 (Gerges, 1999). Using the peak head difference of approximately 40m between
the T1 and T2 aquifers in the centre of the cone of depression, it is estimated that it would
take more than 400 years for relatively saline T1 groundwater to enter the T2 aquifer. This
estimate is valid for the centre of the cone of depression, and does not reflect average values
across the NAP. Smaller head differences and higher porosities observed elsewhere would
mean that more time is required for leakage to occur. Given that the potential for downward
flow has only been possible for a maximum of 50 years, and that the Munno Para Clay is
an effective confining layer where present (Gerges, 2001), contamination of the confined T2
aquifer is a result of leakage through wells. Where the Munno Para Clay is absent (e.g. in the
Gawler area), less effective confining layers may allow some downward leakage to occur.
The relatively low transmissivity of the T2 aquifer (Shepherd, 1971; Gerges, 2001) means
that contamination would initially be relatively localised.
Chloride concentrations along the T2 flowpath are shown in Fig.4c and 4d. The Cl concen-
tration initially decreases slightly across the Alma Fault from around 225 to 180mg/L. This
could be due to fresher groundwater recharge from fractured rock aquifers on the AFB or
palaeoclimate changes. The Cl concentration then increases to around 300mg/L near
Virginia before decreasing to around 250mg/L further along the flowpath. The sample
located at 24km has a higher Cl value as it is located just outside the fresh groundwater chan-
nel, and is used to investigate isotope trends to this point along the flowpath. There are no low
Cl sources available for mixing which could be responsible for the decrease beyond Virginia.
The highest Cl concentrations along the fresh groundwater channel occur beneath the area of
greatest groundwater extraction and bore density, and might appear to be the result of large
scale downward leakage in the area. However, historical records from prior to 1950 (1935 to
1949) indicate that the same concentrations and trends existed at this time. As the pre-1950s
records were taken prior to significant groundwater extraction, when the number of leaky
wells was negligible and the T2 aquifer still had artesian pressures, it appears that leakage has
not significantly altered the groundwater chemistry over a large scale in the region.
Contamination is likely to be localised to wells at this point in time, and vertical leakage
through the aquitards is unlikely over such a large area.
Love et al (1994) also observed decreasing Cl concentrations (from around 1600 to
210mg/L) along a groundwater flowpath in a confined aquifer in the Otway Basin.
Minimum Cl concentrations occurred between 18ka and 10ka, during a cooler and more
arid period. Higher evapotranspiration rates since this period have resulted in increasing Cl
concentrations. Similar trends in stable isotope and Cl concentrations have been observed in
both the Otway Basin and the NAP. The changes in Cl concentration along the T2 fresh
groundwater flowpath are therefore not due to leakage, and most likely reflect changes in
palaeoclimate over the past tens of thousands of years.
5 GEOCHEMICAL EVOLUTION
5.1 Origin of solutes
Most groundwaters in the NAP are dominated by Na and Cl. Relatively fresh groundwaters
are generally Na-Ca-Cl-HCO
3
type groundwaters, and tend towards a Na-Cl type as salin-
ity increases. Evapotranspiration is the dominant process controlling groundwater chem-
istry in the MLR and the NAP. Accession of sea salts, dry deposition and evapotranspiration
of rainfall during summer months allow salts with a largely marine signature to accumu-
late in the soil zone. During winter when groundwater recharge occurs, infiltrating waters
dissolve some of these salts and transport them to the water table.
The ratio of conservative ions Br/Cl follows a seawater dilution line (SWDL) when plot-
ted against Cl (Fig. 4e), suggesting that the source of solutes is marine. Dissolution of halite
within the formation is not a source of Cl, which would be indicated by considerably lower
Br/Cl ratios. Combined with the stable isotope data, which show that all groundwaters are of
a meteoric origin, it is apparent that the solutes in the NAP groundwaters reflect the recharge
history, not the depositional history, of the aquifer.
5.2 Geochemical processes
Ca and HCO
3
concentrations indicate that carbonate mineral dissolution is an important
water-rock interaction occurring in the aquifer. HCO
3
concentrations increase from
150mg/L on the AFB to approximately 270mg/L in the carbonate aquifer. HCO
3
concentra-
tions then vary between 225 and 300mg/L along the rest of the flowpath. Ca concentrations
increase from 25mg/L on the AFB to around 75mg/L and remain relatively constant.
The saturation indices for carbonate minerals were determined using PHREEQC
(Parkhurst and Appelo, 1999). Most groundwaters tend to be considerably supersaturated
with respect to calcite and dolomite. CO
2
degassing during sampling would result in carbon-
ate mineral supersaturation and precipitation, along with a pH increase (e.g. Langmuir, 1971;
Mahlknecht et al, 2006). NAP groundwaters are nearer to saturation when pH is reduced by
0.5pH units (e.g. Edmunds et al, 2003). The pH of historical data used in this study was pre-
sumably not measured in field. This is suggested by the pH levels of historical data being con-
sistently 0.20.4pH units higher than samples measured in the field. Samples collected in the
field have probably also been subjected to some degassing, but not to the same extent as the
historical data. The result is that calculated saturation indices for NAP groundwaters differ
from the true in-situ values, and provide only an approximate guide as to whether a mineral
will dissolve or precipitate. It is likely that NAP groundwaters are at saturation.
Calcite saturation indices for groundwaters along the T2 fresh groundwater corridor range
from 1.42 to oversaturated. Lowest values are in the undifferentiated sediments on the AFB
indicating that carbonate dissolution can occur near the recharge zone. Saturation indices
rapidly approach zero along the flowpath, and groundwaters exceed saturation approximately
30
28
26
24
22
20
0 5 10 15 20 25
Distance along Flowpath (km)
2
H

(
V
-
S
M
O
W
)
WEST EAST
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Virginia
b.
150
200
250
300
350
0 5 10 15 20 25
C
l
-

(
m
g
/
L
)
WEST
WEST
EAST
EAST
d.
P
a
r
a

F
a
u
l
t
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Virginia
Virginia
12
10
8
6
0 5 10 15 20 25
1
3
C

(
V
-
P
D
B
)
f.
0.712
0.716
0.720
0.724
0 5 10 15 20 25
Distance along Flowpath (km) Distance along Flowpath (km)
8
7
/
8
6
S
r
WEST EAST h.
Virginia
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
0
5
10
15
0 5 10 15 20 25
K
+

(
m
g
/
L
)
WEST EAST j.
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Virginia
0
20
40
60
80
100
0 5 10 15 20 25
1
4
C

(
p
m
C
)
WEST EAST
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Virginia
a.
0
250
500
750
1000
1250
0 5 10 15 20 25
C
l
-

(
m
g
/
L
)
WEST EAST
c.
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Virginia
0.0000
0.0005
0.0010
0.0015
0.0020
0 50 100 150
mCl
m
B
r
/
m
C
l
Q13
Q4
T1
T2
T3/4
e.
Seawater dilutionline
Halite dissolution
0.0
0.2
0.4
0.6
0.8
1.0
12 10 8 6
13
C (V-PDB)
S
r
2
+

(
m
g
/
L
)
g.
0
0.25
0.5
0.75
1
1.25
0 5 10 15 20 25
m
N
a
+
/
m
C
l
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
Q13
Q4
T1
T2
WEST EAST
i.
Seawater dilutionline
Virginia
P
a
r
a

F
a
u
l
t
A
l
m
a

F
a
u
l
t
Figure 4. Geochemical trends for the NAP groundwaters.
5km along the flowpath. It is more likely that equilibrium is obtained at some point further
along the flowpath than indicated, and is maintained rather than exceeded greatly saturation.
Carbonate mineral saturation indices for the Quaternary aquifers are generally lower than the
Tertiary aquifers due to a relatively lower concentration of carbonate minerals in the
Quaternary sediments.
Carbon-13 can provide information regarding the processes affecting dissolved carbonate
species in groundwater. The
13
C values across the NAP region range from 15.1 to 6.0,
and generally become more enriched along the flowpath (
13
C from 10.7 to 6.2; Fig.
4f). The
13
C of soil zone CO
2
in SE Australia is approximately 23. Fractionation
processes between soil zone CO
2
and carbonate matrix dissolution can explain the range of
13
C values encountered in the NAP.
Carbonate mineral dissolution is also evidenced by Sr
2
concentrations and
87/86
Sr ratios.
Unpublished data for siltstone, regolith and soil at Scott Creek in the MLR indicate high
87/86
Sr ratios between 0.728464 and 0.815905 (R. Cranswick, pers.comm.). As a result of
these high host rock ratios, groundwaters from the same area also have high values compared
with the seawater ratio, ranging from 0.726173 to 0.753041. Although measurements of
87/86
Sr were not performed on the NAP aquifer matrix, these units were deposited during the
same period as the Murray Group sediments in the Murray Basin (Drexel and Preiss, 1995).
Dogramaci and Herczeg (2002) found that the overall
87/86
Sr ratio (carbonate and silicate) for
these deposits were between 0.70869 and 0.70934, with values in carbonate minerals ran-
ging between 0.7084 and 0.7087. This is equivalent to seawater ratios for Tertiary age marine
sediments of between 0.7077 and 0.7089 (Burke et al, 1982; Palmer and Elderfield, 1985).
Carbonate minerals in the Tertiary aquifers of the NAP can be expected to have similar val-
ues. Dogramaci and Herczeg (2002) found that
87/86
Sr ratios for groundwaters from the
Murray Group Aquifer to range between 0.7084 and 0.7112. These values resulted from car-
bonate mineral dissolution, however, higher values down-gradient resulted from upward
leakage from a silicate aquifer.
The
87/86
Sr ratio for groundwaters in the NAP range from 0.709951 to 0.724204, although
most samples are between 0.7125 and 0.7160. These values are intermediate between the
values for the MLR and expected values for the Tertiary limestone. Sr concentrations along
the fresh T2 channel increase from 0.23 to 0.86mg/L, and increase with increasing
13
C
(Fig. 4g). This suggests that carbonate mineral dissolution is the likely source of Sr in NAP
groundwaters. It may be possible to quantify carbonate mineral reactions using Sr isotopes,
which may be beneficial in correcting
14
C ages (e.g. Bishop et al, 1994). The relatively
radiogenic
87/86
Sr value at the start of the flowpath indicates recharge from the MLR.
Dissolution of the carbonate matrix along the flowpath leads to progressively lower values
(Fig. 4h). The values level out at approximately 0.7120.713, which might suggest some
contribution from silicate mineral weathering. Minor variations probably result from aquifer
heterogeneity. Mixing with groundwaters from the Quaternary aquifers is unlikely, as an
increase in Cl concentration would be expected with an increase in
87/86
Sr values.
While carbon and strontium isotopes indicate that carbonate mineral dissolution occurs
along the groundwater flowpath, there needs to be a mechanism which drives the reaction at
saturation, otherwise no dissolution will occur. Possible mechanisms include mixing between
two different waters, ion exchange, gypsum dissolution encouraging dedolomitisation or
dynamic equilibrium. Calcite dissolution at saturation may be driven by Ca
2
removal by
ion exchange (e.g. Plummer et al, 1990; Mahlknecht et al, 2006). The molar Na/Cl ratio
for the T2 aquifer groundwaters along the fresh groundwater corridor (Fig. 4i) shows an
excess of Na over the seawater ratio. This may be partly explained by Na-silicate dissolu-
tion, however, exchange for Ca in solution may also be responsible. In the Quaternary
aquifers, the Na/Cl ratio is less than the seawater ratio, suggesting reverse ion exchange.
The trend of K along the flowpath (Fig. 4j) suggests that its concentration is affected by
ion exchange. In the T2 aquifer, K concentrations decrease slightly between Gawler and
Virginia from 6 to 5.6mg/L, while all other major cations increase, suggesting ion
exchange where K is removed from solution in exchange for Ca. Near Virginia, K concen-
trations increase dramatically to approximately 10mg/L. This may be an ion-exchange
front associated with the increasing salinity of palaeowaters.
Increasing sulphate concentrations along the groundwater flowpath indicate water-rock
interactions are contributing sulphate. This may be due to gypsum dissolution or pyrite oxi-
dation. XRD and lithological logs in the NAP do not reveal the presence of gypsum or anhyd-
rite (Rattray, 1999; Vanderzalm, 2004), while pyrite has been observed in some layers.
Saturation indices for gypsum suggest that gypsum is not present in the system. Preliminary
modelling with NETPATH (Plummer et al, 1994) is inconclusive as to the source of SO
4
.
Another possible mechanism may be that the older groundwaters further along the flow-
path were recharged with higher SO
4
/Cl ratios. Kayaalp (1999) has shown that SO
4
/Cl ratios
in rainfall increase further inland in the region. Changes to the coastline location may have
resulted in changes to the SO
4
/Cl ratio in rainfall falling over the MLR over time.
6 CONCLUSIONS
The historical groundwater flow direction across the NAP, from the MLR to the coast, has
been verified using
14
C activities. Stable isotopes of water have been used to establish end-
members for MLR groundwater, NAP groundwaters, Gawler area groundwaters and surface
waters. The majority of NAP groundwaters show little indication of evaporation, suggesting
that subsurface recharge is the dominant source of recharge to the confined Tertiary aquifers.
However, groundwaters in the Gawler area show some evaporation effects, indicating that an
evaporated water source is an important part of the groundwater balance in the area. In
terms of contamination, groundwaters displaying significant increases in Cl concentration,
the presence of nitrate or tritium, or anomalously high
14
C activities can be considered as con-
taminated. Due to effective confining layers, leakage through wells is the dominant source of
contamination across most of the NAP, which is localised to near the well. Downward leak-
age may be occurring where the Munno Para Clay is absent in the Gawler area, as indicated
by enriched stable isotopes of water and the presence of tritium. The Cl concentration peak
under the most heavily exploited area appears to be a relict of the natural system, and not
regional contamination. Together with the stable isotopes of water becoming more depleted
along the flowpath, they reveal climatic variations from the previous tens of thousands of years.
Groundwaters are dominated by Na and Cl, resulting from evapotranspiration of rainfall
and dry deposition of marine salts in the MLR. Carbonate mineral dissolution is the dom-
inant geochemical process, resulting in increasing
13
C and Sr concentrations, and decreas-
ing
87/86
Sr ratios. Increasing SO
4
concentrations may result from water-rock interactions or
reflect changes in SO
4
/Cl ratios in rainfall as the coastline has moved.
Future work will involve modelling of the reactions occurring in the system and corrections
to carbon-14 ages. Possible contamination means that the
14
C activities for some samples may
not be reliable. It may be possible to correct the
14
C activities based on the tritium content
(e.g., Mazor et al, 1986). However, the non-conservative behaviour of these tracers may limit
their usefulness. It may also be possible to better constrain carbonate reactions using Sr
isotopes, in order to obtain more accurate corrected ages. However, it should be considered
that gypsum dissolution may also provide an additional source of Sr to the groundwaters.
There is the potential with the stable isotopes of water to establish a relationship between ele-
vation and
2
H and
18
O. From this, it may be possible to determine the elevation in the MLR
which predominantly contributes to NAP recharge, and possibly determine the depth of circu-
lation of the MLR groundwaters. This would be an important tool in understanding ground-
water flow in similar fractured rock environments, and will be investigated more thoroughly.
Further work is also required to complete the geochemical model provided here.
ACKNOWLEDGEMENTS
The authors thank J. Safta, H. Zulfic and other staff at the South Australian Department of
Water, Land and Biodiversity (DWLBC) for assistance with sampling and analyses and
advice regarding the NAP. Staff at CSIRO Land and Water and ANSTO provided stable iso-
tope and radiocarbon analyses. J. Foden and D. Bruce at Adelaide University School of Earth
and Environmental Sciences provided Sr isotope analyses. J. Pritchard, J.L. Michelot and
L.Ma are thanked for review of this manuscript. This study was supported with the aid of
AINSE grants 03/072 and 04/097 for tritium and radiocarbon analyses. DB was supported
by an APA Scholarship and CGS stipend.
REFERENCES
Bishop, P.K., Smalley, P.C., Emery, D. and Dickson, J.A.D., 1994. Strontium isotopes as indicators
of the dissolving phase in a carbonate aquifer: implications for
14
C dating of groundwater. Journal
of Hydrology, 154, 301321.
Burke, W.H., Denison, R.E., Hetherington, E.A., Koepnick, R.B., Nelson, H.F. and Otto, J.B. 1982.
Variation of seawater
87
Sr/
86
Sr throughout Phanerozoic time. Geology, 10, 516519.
Clark, I.D. and Fritz, P., 1997. Environmental Isotopes in Hydrogeology. Lewis Publishers, Boca
Raton, Florida, 328 pp.
Dighton, J.C., Herczeg, A.L., Leaney, F.W., Lennard, R.P., Love, A.J. and Gerges, N.Z. 1994. Stable
isotope and radiocarbon data for groundwaters from the Adelaide Metropolitan Area, Centre for
Groundwater Studies Report No.56.
Dogramaci, S.S. and Herczeg, A.L., 2002. Strontium and carbon isotope constraints on carbonate-
solution interactions and inter-aquifer mixing in groundwaters of the semi-arid Murray Basin,
Australia. Journal of Hydrology, 262, 5067.
Drexel, J.F. and Preiss, W.V., 1995. The Geology of South Australia. Vol. 2, The Phanerozoic. South
Australia. Geological Survey. Bulletin 54.
Edmunds, W.M., Carrillo-Rivera, J.J. and Cardona, A., 2002. Geochemical evolution of ground-
water beneath Mexico City. Journal of Hydrology, 258, 14, 124.
Edmunds, W.M., Guendouz, A.H., Mamou, A., Moulla, A., Shand, P. and Zouari, K. 2003.
Groundwater evolution in the Continental Intercalaire aquifer of southern Algeria and Tunisia:
trace element and isotopic indicators. Applied Geochemistry, 18, 6, 805822.
Gerges, N.Z., 1986. Underground Water Resources of the Adelaide Metropolitan Area with the
Latest Understanding of Recharge Mechanism, Proceedings of the International Conference of
Groundwater Systems Under Stress, Brisbane. Australian Government Publishing Service,
Canberra., pp. 165175.
Gerges, N.Z., 1999. The Geology and Hydrogeology of the Adelaide Metropolitan Area, PhD Thesis,
Flinders University, Adelaide, 243 pp.
Gerges, N.Z., 2001. Northern Adelaide Plains groundwater review. DWR Report 2001/013,
Department for Water Resources, Adelaide.
Ivkovic, K.M., Watkins, K.L., Cresswell, R.G. and Bauld, J., 1998. A Groundwater Quality
Assessment of the Fractured Rock Aquifers of the Piccadilly Valley, South Australia. AGSO
Record 1998/16, Australian Geological Survey Organisation, Canberra.
Kayaalp, A.S., 1999. Application of rainfall chemistry and isotope data to hydro-meteorological
modelling. PhD Thesis, The Flinders University of South Australia, Adelaide, 251 pp.
Langmuir, D., 1971. The geochemistry of some carbonate ground waters in central Pennsylvania.
Geochimica et Cosmochimica Acta., 35, 10231045.
Love, A.J., Herczeg, A.L., Leaney, F.W., Stadter, M.F., Dighton, J.C. and Armstrong, D. 1994.
Groundwater residence time and palaeohydrology in the Otway Basin, South Australia:
2
H,
18
O
and
14
C data. Journal of Hydrology, 153,14, 157187.
Mahlknecht, J., Grfias-Solis, J., Aravena, R. and Tesch, R., 2006. Geochemical and isotopic inves-
tigations on groundwater residence time and flow in the Independence Basin, Mexico. Journal of
Hydrology, 324, 14, 283300.
Manning, A.H. and Solomon, D.K., 2005. An integrated environmental tracer approach to character-
izing groundwater circulation in a mountain block. Water Resources Research, 41,12, 18pp.
DOI:10.1029/2005WR004178
Mazor, E., Jaff, F.C., Fluck, J. and Dubois, J.D., 1986. Tritium corrected
14
C and atmospheric noble
gas corrected
4
He applied to deduce ages of mixed groundwaters: Examples from the Baden
region, Switzerland. Geochimica et Cosmochimica Acta, 50, 8, 16111618.
Palmer, M.R. and Elderfield, H., 1985. Sr isotope composition of sea water over the past 75 Myr.
Nature, 314, 526528.
Parkhurst, D.L. and Appelo, C.A.J., 1999. Users Guide to PHREEQC (Version 2) A Computer
Program for Speciation, Batch-Reaction, One-Dimensional Transport and Inverse Geochemical
Calculations, Water-Resources Investigations Report 994259. US Geological Survey, Denver,
Colorado, 312 pp.
Plummer, L.N., Busby, J.F., Lee, R.W. and Hanshaw, B.B., 1990. Geochemical Modeling of the
Madison Aquifer in Parts of Montana, Wyoming, and South Dakota. Water Resources Research,
26, 9, 19812014.
Plummer, L.N., Prestemon, E.C. and Parkhurst, D.L., 1994. An Interactive Code (NETPATH) for
Modeling NET Geochemical Reactions Along a Flow PATH. Version 2.0. Water-Resources
Investigations Report 944169, U.S. Geological Survey, Reston, Virginia.
Radke, B.M., Ivkovic, K.M., Watkins, K.L., Cresswell, R.G. and Bauld, J., 2000. A Groundwater
Quality Assessment of the Upper Onkaparinga Region, Southern Mt Lofty Ranges, South
Australia. Australian Groundwater Quality Assessment Project Report No. 9, Bureau of Rural
Sciences, Kingston, ACT.
Rattray, K.J., 1999. Geochemical Reactions Induced in Carbonate Bearing Aquifers Through
Artificial Recharge. MSc Thesis, Flinders University of South Australia, Adelaide, 304 pp.
Schlehuber, M.J., Lee, T.-C. and Hall, B.S., 1989. Groundwater level and hydrochemistry in the San
Jacinto basin, Riverside County, California. Journal of Hydrology, 106, 12, 7998.
Schuster, C.D., 2000. A preliminary report on the geophysical testing of wells that have abnormally
high salinity, NAP Corroded Casing Survey 1999/2000. PIRSA RB 2000/00014, Primary
Industries and Resources South Australia, Adelaide.
Shepherd, R.G., 1971. The Hydrogeology of the Northern Adelaide Plains Basin. Mineral Resources
Review, South Australia, 134, 5261.
Smith, D.L., 1979. Land use and groundwater history of the Northern Adelaide Plains. Engineering
and Water Supply Department, South Australia, Adelaide, 70 pp.
Vanderzalm, J.L., 2004. Hydrogeochemical Processes during Aquifer Storage and Recovery (ASR)
with Reclaimed Water in a Carbonate Aquifer. PhD Thesis, Flinders University of South Australia,
Adelaide, 317 pp.
Wilson, J.L. and Guan, H., 2004. Mountain-block hydrology and mountain-front recharge. In: J.F.
Hogan, F.W. Phillips and B.R. Scanlon (Eds.), Groundwater Recharge in a Desert Environment:
The Southwestern United States. American Geophysical Union, Water Science and Applications
Series, v. 9, Washington, D.C., pp. 113137.
CHAPTER 18
The Guaran Aquifer System: state-of-the-art in Argentina
Ofelia Tujchneider
1,2
, Marcela Perez
1
, Marta Paris
1
&
Mnica DElia
1
1
Universidad Nacional del Litoral, Santa Fe, Argentina
2
CONICET, Santa Fe, Argentina
ABSTRACT: The Guaran Aquifer System (G.A.S.) is one of the worlds most important fresh
groundwater reservoirs. It lies under four South American countries: Argentina, Brazil, Paraguay
and Uruguay, which use this resource for several purposes with very different exploitation levels.
This great transboundary aquifer is situated in aeolian and fluvial sandstone of Triassic-Jurassic age,
usually covered by basalt formations from the Cretaceous, which provide a high degree of confine-
ment. Due to the vast extent of the aquifer, this paper focuses on the border area between Argentina
and Uruguay. An analysis of the geological and hydrochemical data available for this area made it
possible to delineate zones with different characteristics. In addition, it was possible to evaluate the
feasibility of deep thermal groundwater being present towards the southwest portion of the G.A.S.
in Argentina, where the boundary has not yet been defined.
1 INTRODUCTION
The Guaran Aquifer System (G.A.S.) is one of the worlds most important fresh ground-
water reservoirs due to its estimated areal extent (1200000km
2
) and volume (40000km
3
).
Underlying a region of approximately 24000000 inhabitants, the G.A.S. is shared by four
South American countries: Argentina, Brazil, Paraguay and Uruguay, which use the
resource for very different purposes at varying exploitation levels (Figure 1).
In Brazil, the G.A.S. lies under the eight most developed states of the country. In So Paulo
State, the most industrialized, the aquifer is increasingly exploited. Nowadays, it is used mainly
for domestic and industrial water supply, irrigation, therapeutic baths and mineral water.
In Argentina and Uruguay, the main uses are balneotherapy and recreation. In 1994, the
first Argentinian infrabasaltic well was drilled in Federacin city (Entre Ros province).
Since that time, eight other hydrothermal facilities have been built. They are located prin-
cipally in the north central part of the province. In Uruguay, the G.A.S. is also used for
water supply and irrigation. During the 1990s, seven deep pumping wells were drilled in
this country. In Paraguay, the main use of the aquifer is to supply water to rural villages
located in the eastern part of the country.
This paper presents an analysis of the geological and hydrochemical information (exist-
ing and generated data) from currently functioning wells across an area of approximately
78000km
2
, (Figure 1). It includes part of the Argentina-Uruguay border as well as the
north central part of the Entre Ros province (Argentina). The objective is to study the
Argentine zones where the deep wells have been pumping water with high saline contents.
The water samples obtained from these wells were first related to the sediments in the
water-bearing formation. Then, they were compared with the low-salinity samples charac-
teristic of the water from elsewhere in this aquifer.
From the analysis of the data, it was possible to consider that deep hot groundwater
occurred on the right bank of the Paran River along its central reach (Santa Fe province).
The results represent an advance in the knowledge of the western boundary of the Guaran
Aquifer System in Argentina, where its extent has not yet been defined.
2 GEOLOGICAL AND HYDROGEOLOGICAL SETTING
The stratigraphic sequence comprising the Guaran Aquifer System is described below
(Table 1). The geological formations are known under different names in each country.
Figure 1. Location of Guaran Aquifer System and of the study area.
This immense aquifer is situated in aeolian and fluvial sandstones of continental origin
deposited in Triassic and Jurassic times. The sandstones are generally covered by Cretaceous
basaltic formations that provide different degrees of confinement. Covering approximately
6065% of its extent, the basalt thickness ranges from 200 to more than 1000m. Therefore,
the aquifer system is both confined and unconfined in different places (Montao et al., 2004).
The basaltic formation outcrops in the north-eastern part of Argentina, where it is heavily
fissured. It dips westwards and becomes thinner both westwards and southwards (Figures 2
and 3).
The pattern of the sandy sediments depends on: the Paran Sedimentary Basin bound-
aries, the structural configuration of the geologic basin and the basaltic deposits that cover
the sandstones.
The thickness of the aquifer system ranges from a few meters up to 800m in Brazil and
600m in Uruguay. Its depth varies throughout and can reach up to 1800m. It is not surpris-
ing that in view of the vast extent of this groundwater-bearing formation, the hydraulic
parameters exhibit great spatial variability. The hydraulic conductivity ranges from 0.3 to
5.0m/day. Generally, in the unconfined sandstone the effective porosity is approximately
15% and in the confined zones the storage coefficient is 10
4
.
Other distinctive characteristics of the G.A.S. are: artesian pressures and high yields at
many points (500m
3
/h where the aquifer is thickest), good quality groundwater and tem-
peratures ranging from 38C to 60C by geothermal gradient.
Generally, the water in this aquifer system is of good quality, with an average salinity of
300mg/l. The unconfined zones are characterized by water of a calcium-bicarbonate com-
position whilst the confined ones produce water with a sodium-bicarbonate composition.
According to Fili and Tujchneider in Montao et al. (1998), it is possible to identify two
subsystems within the overall system in Argentina: one with low salinity water and higher
The Guaran Aquifer System: state-of-the-art in Argentina 241
Table 1. Stratigraphic sequence of the G.A.S.
Period Geological Description
Formation
Cretaceous Serra Geral
(1)
It is one of the worlds biggest basaltic effusions. It is
Arapey
(2)
composed of tholeitic basalts, with dominant fine grain
Curuz Cuati
(3)
size, aphanitic, seldom porphyrical and dark grey to black
colour. The successive basalt flows are of very variable
thickness. The greatest thickness known (1,930m) is at the
Cuiab Paulista Region (So Paulo State, Brazil).
Jurassic Botucat
(1)
Set of sandstones of fluvial and aeolian origin. They are
Tacuaremb
(2,3)
generally well sorted, reddish, from friable to well-lithified,
Misiones
(4)
fine- to medium-grain size, with frequent diagonal
lamination. They were probably accumulated under desert
conditions.
Triassic Piramboia
(1)
White and reddish sandstones, fine grain size with clayey
Buena Vista
(2)
lenses. Lacustrine, fluvial and aeolian origin.
Yaguar
(2)
Permian San Gregorio
(2)
Permian sediments of low permeability deposited in
Tres Islas
(2)
shallow marine environments. Due to their variable saline
Chacabuco
(3)
contents they would certainly affect the groundwater
quality.
Note:
(1)
Brazil,
(2)
Uruguay,
(3)
Argentina,
(4)
Paraguay.
Figure 2. West East cross section. (Source: Fili and Tujchneider, 1997. In Montao et al. 1998).
Figure 3. South North cross section. (Source: Fili and Tujchneider, 1997. In Montao et al. 1998).
temperatures, located in the north-central part of the Entre Ros province, the other with
salty water and lower temperatures, and located on the southern boundary of the aquifer
system.
Similarly, Montao et al. (2004) defined two subsystems for the aquifer in Uruguay. The
areas where these authors focused their investigations correspond to the boundary between
Argentina-Uruguay and Brazil-Uruguay. They called typical G.A.S. the aquifer system situ-
ated in the sediments of the Tacuaremb, Yaguar and Buena Vista geological formations.
The water stored in this subsystem can be used as drinking water. The other subsystem is
called Permian G.A.S.. It lies under the typical G.A.S. and consists of Permian sediments
deposited in marine environments, and it is these that condition the water quality.
Due to the extent of the G.A.S., the piezometric surface is hard to define. Gilboa et al.,
(in Araujo et al. 1999) prepared the first piezometric map. Since that time, several water-
level contour maps have been drafted by different authors, many of them showing only a
sector of the aquifer system and its behaviour. For instance, Sracek and Hirata (2002) drew
Figure 4. General groundwater flow directions (Source: Tujchneider et al. 2003).
preferential flow lines for So Paulo State (Brazil). They expressed important reserves
about the validity of the piezometric and hydraulic information because many of the
reported values did not consider the water density and viscosity-temperature correction.
Nevertheless, it seems that the direction of groundwater flow in the northern sector of
the Guaran Aquifer System (Brazil) runs from the outcropping areas of the Botucat for-
mation to the axis of the basin. Therefore, the general direction of the groundwater flow is
from the north-east to the south-west. The groundwater elevation ranges from 300 to
700m. It also appears that the area between the Paran and Uruguay rivers is probably the
discharge area of this aquifer.
Silva Busso (1999) performed an analysis of the confinement pressures related to the
depth of the aquifer formation in the Entre Ros province (Argentina) and north-west of
Uruguay. According to this author, the recharge areas for Uruguay are located between 110
to 120m above sea level. The hydraulic potentials decrease from the north-east towards the
south-west.
However, there is little agreement so far on an accepted flow pattern within this aquifer.
This is a subject of current research. Figure 4 shows a simplified scheme of the main flow
direction according to the authors mentioned above (Tujchneider et al., 2003).
3 MATERIALS AND METHODS
The regional and local background information on the geology, hydrogeology and hydro-
chemistry (Montao et al., 1998; Fili 2001; Tujchneider et al., 2003; Tujchneider and
Tineo, 2005) was revised and reinterpreted.
In order to evaluate the possibility of the presence of the Guaran Aquifer System on the
right margin of the Paran River alluvial valley, geophysical exploration using electrical
resistivity was undertaken. Four vertical electrical soundings (VESs) were carried out in
the Garay Department (Santa Fe province, Argentina).
The hydrochemical data set was composed of 48 analyses of calcium, magnesium,
sodium, potassium, bicarbonate, chloride, sulphate (all of them expressed in meq/l), Total
Dissolved Solids (expressed in mg/l) and pH. These records correspond to water samples
collected at different times (between 1997 and 2004) in the boreholes that penetrate the
aquifer system in the study area.
After evaluating the analytical error using the ionic balance, the data were categorized
according to the Piper-Hill classification.
To obtain the statistics and frequency distributions of these ionic series, a univariate stat-
istical analysis was performed. Then, R-mode and Q-mode Cluster Analyses were applied
to identify homogeneities in the data sets. R-Pearson and Euclidian distance coefficients
were used respectively. The linkage criterion was un-weighted average pairs. Natural loga-
rithm and standard transformations were applied to the series.
4 RESULTS
The geological formations comprising the G.A.S. are known in Argentina, Brazil, Paraguay
and Uruguay under different names, but well-known in the region as Botucat, Tacuaremb
or Misiones sandstones. They are confined by the Serra Geral or Arapey basalt formation.
In Argentina, the sandstones occupy part of the Chaco Paranense Sedimentary Basin,
underlying the provinces of Misiones, Corrientes, Entre Ros, north-central part of Santa
Fe, eastern Crdoba, southern Chaco and the east-central sector of Santiago del Estero. In
Corrientes province, the sandstones outcrop in some small areas located in the central and
south sectors of the province, but prevail in the subsurface (Figure 1).
For the medium reach of the Paran River alluvial valley, Table 2 shows the regional
stratigraphic sequence from the rocks that compose the crystalline basement to the more
recent sediments according to Fili (2001), and Tujchneider and Tineo (2005). The VESs
conducted in Santa Fe province indicate a depth to the top of the basalt of approximately
690m and a thickness of 685m. Underlying the basalt formation, at a depth of approxi-
mately 1300m, sediments that could be saturated were identified. Figure 5 presents a
regional stratigraphic cross-section summarizing all this information.
The analyses were filtered using an admissible error margin given by the ionic balance.
According to the Piper-Hill classification, all water samples in the cation triangle are of
the sodium type. However, in the anionic triangle, 44% are of the bicarbonate type and the
remaining 56% belong to the chloride type. Figure 6 shows the spatial locations of the
samples.
Table 3 summarizes the main statistics calculated for the variables considered in the
analyses. With regard to the coefficient of variation, skewness and kurtosis as well as the
mean and median values of the chemical data series, bicarbonate and pH are normally dis-
tributed.
Water samples collected from Mara Grande, La Paz and Villaguay (Entre Ros
province, Argentina) deep confined pumping wells have high TDS values that fall outside
the 75 percentile. Sulphate, chloride, calcium and sodium contents show a similar behav-
iour to the TDS values in the same wells.
Table 2. Regional stratigraphical sequence.
Age Formation Lithology
Holocene Undistinguishable recent
and current alluvial deposits Sands and clayey silts
Late Pleistocene Pampa Group Clays, silts and loess
Middle Pleistocene
Early Pleistocene Ituzaing (Puelches sands) Ochre, quartzose, fluvial sands
Late Pliocene
Late Miocene Paran Green clays and Grey marine sands
Early Miocene Olivos/Fray Bentos Reddish, calcareous and gypsiferous clays
Eocene
Paleocene Mariano Boedo/Yeru Calcareous sandstones Claystones
Late Cretaceous
Guarani aquifer system
Early Cretaceous Serra Geral Basalt
Late Jurassic Tacuaremb Quartzose sandstones
Late Triassic Buena Vista Quartzose sandstones
Early Permian Chacabuco Claystone and tuffite
Late Carboniferous Charata Sachayoj Siltstones and sandstones
Lutaceous claystone and sandstones
Precambrian Basement rocks Granite-gabbro and metamorphic rocks
Figure 5. Regional stratigraphic cross-section (Source: Tujchneider et al. 2005).
According to Seoane and Silva Busso (2005), different authors have mentioned the
presence of fluoride in groundwater samples collected in the G.A.S. Brazilian aquifer
areas. They consider three possible causes of the fluoride. The first one relates to rocks
originating from basaltic and alkaline magma intrusion, sometimes rich in fluoride, sec-
ondly, hydrothermal deposits rich in fluoride derived from igneous rocks and finally,
remobilization of fluoride distributed in rocks along the flow path. In the border area
between Argentina and Uruguay, the fluoride contents range from 0.01 to 3.8mg/dm
3
. The
possible geochemical origin could be the basalt deposits of the Serra Geral formation.
Figure 6. Spatial location of data set according to Piper-Hill classification.
Table 3. Statistics.
TSD pH CO
3
H SO
4
Cl Ca Mg Na K
(mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l)
Mean 12094.5 7.87 196.6 1503.6 3318.1 164.8 16.8 2744.3 21.5
Median 743.0 7.80 185.0 149.0 183.0 29.5 5.7 294.4 8.2
Max 126151.0 8.80 549.0 11930.0 28810.0 1061.0 92.3 23500.0 145.8
Min 256.0 7.15 29.0 10.7 17.1 1.2 0.3 72.0 0.9
Lower 523.4 7.62 87.7 51.5 42.9 9.6 2.1 163.4 4.0
quarter
Upper 11324.8 8.01 279.3 2429.8 4934.5 237.3 27.3 3909.0 19.5
quarter
Standard 26805.8 0.43 129.6 2463.0 6000.6 252.0 22.8 4743.4 35.2
deviation
Variation 2.2 0.05 0.7 1.6 1.8 1.5 1.4 1.7 1.6
coefficient
Skewness 3.3 0.90 0.7 2.8 3.0 2.3 1.8 3.1 2.7
Kurtosis 11.3 0.54 0.2 9.7 10.5 5.6 3.0 11.7 6.5
Figure 7 shows the dendrogram for the R-mode cluster analysis. Note the considerable
influence of sulphate, sodium, chloride and calcium in the TDS composition and variabil-
ity as well as the marked relationship between bicarbonate concentrations and pH as an
indicator of water sample acidity-alkalinity.
Using the same variables, Figure 8 shows the dendrogram for the Q-mode cluster analy-
sis. This plot allows the individualization of two main water sample groups, which are
coincident with those, identified by the Piper-Hill classification. Group 1 is defined by
bicarbonate waters. Group 2 is defined by chloride-sulphate waters. With a higher similar-
ity level it is possible to identify four subgroups whose distribution is shown in Figure 9.
This figure shows that the data collected from the Mara Grande, Villaguay and La Paz
wells (Entre Ros province, Argentina) constitute a set with internal homogeneity but with
a clear heterogeneity regarding all other observations.
5 DISCUSSION
From the analysis of both the available information and the new data generated by the field
studies, the depths and thickness of the geological formation identified in the study area were
corroborated. For the right bank of the Paran River in the Garay Department (east-central
Santa Fe province, Argentina), the top of the basalt elevation is clearly identified at a depth
of 683 to 693m, with an average thickness of 692m. In this area, underlying the basalt for-
mation, sediments that might be saturated were identified at a depth of approximately
1360m. These sediments belong to the Tacuaremb Formation (sandstone of Tacuaremb).
From the analyses of the chemical data, the differences between the wells pumping fresh
groundwater (Group 1) and those that pump salty groundwater (Group 2) are clearly visible:
Group 1 Subgroup 1A involves water samples from: Federacin, Coln, Concordia

(Argentina), Arapey, Daymn and Guaviy (Uruguay). In this group the samples
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
CO
3
H
PH
K
MG
CA
NA
CL
SO
4
TDS

1
-
r

P
e
a
r
s
o
n
Figure 7. R-mode dendrogram.
belonging to Federacin, Concordia, Daymn and Guaviy are of good quality. It is
considered that these wells pump the Guaran Aquifer groundwater.
Group 1 Subgroup 1B is represented by Coln and Arapey groundwater samples with

higher salinity.
Group 2 Subgroup 2A is defined by water samples of Villa Elisa, Gualeguaych,

Concepcin del Uruguay (Argentina), Guichn, Almirn, and Paso Ulliestie
(Uruguay), whose salinity (TSD) is around 2600 and 15000mg/l and sulphate concen-
tration between 600 and 2700mg/l.
Group 2 Subgroup 2B involves water samples from Mara Grande, La Paz and
Villaguay (Argentina), whose TDS lies between 60000 and 100000mg/l and sulphate
content is 4000mg/l.
More saline groundwater would belong to the layered permeable levels intercalated in
the basalt formation overlying the water-bearing sandstone of the aquifer (which stores
fresh groundwater).
It is also important to notice how data collected in the same place show changes in
chemical composition with time. Figure 10 shows an example of these changes in Dayman
(Uruguay).
E
u
c
l
i
d
e
a
n

d
i
s
t
a
n
c
e
0
2
4
6
8
10
C
_
2
8
C
_
3
0
C
_
2
7
C
_
2
9
C
_
3
2
C
_
3
1
C
_
2
3
C
_
2
2
C
_
1
8
C
_
2
0
C
_
1
9
C
_
2
1
C
_
1
5
C
_
1
4
C
_
1
3
C
_
6
C
_
7
C
_
5
C
_
2
C
_
2
4
C
_
1
7
C
_
1
6
C
_
1
1
C
_
2
6
C
_
2
5
C
_
1
2
C
_
1
0
C
_
8
C
_
9
C
_
4
C
_
3
C
_
1
C_1 C_12 Dayman C_23 Villa Elisa 12/04
C_2 Federacin 04/95 C_13 G. Almirn 10/88 C_24 Coln 03/97
C_3 Federacin 12/97 C_14 G. Almirn 03/97 C_25 Concordia 11/97
C_4 Federacin 03/99 C_15 G. Almirn C_26 Concordia 03/99
C_5 Arapey 08/95 C_16 Guaviy 06/96 C_27 La Paz 12/04
C_6 Arapey 11/97 C_17 Guaviy 03/97 C_28 Mara Gde 12/04
C_7 Arapey C_18 Villa Elisa 03/97 C_29 Paso Ulliestie
C_8 C_19 Villa Elisa 03/97 C_30 Villaguay 03/04
C_9 Dayman 01/92 C_20 Villa Elisa 04/97 C_31 Gualeguaych 03/99
C_10 Dayman 07/92 C_21 Villa Elisa 07/97 C_32 C. del Urug. 04/99
C_11 Dayman 08/97 C_22 Villa Elisa 08/98
Subgroup 1A
Group 1
Group 2
Subgroup 1B
Subgroup 2A
Subgroup 2B
Federacin12/94
Dayman 07/85
Figure 8. Q-mode dendrogram.
6 FINAL REMARKS
Undoubtedly, the Guaran Aquifer System is one of the most complex in the world. Due to
its great spatial variability, more and better hydrogeological information is required, both
Figure 9. Locations of groups and subgroups identified in the Q-mode cluster analysis.
0
50
100
150
200
250
300
350
400
450
500
j
u
l
-
8
5
j
u
l
-
8
6
j
u
l
-
8
7
j
u
l
-
8
8
j
u
l
-
8
9
j
u
l
-
9
0
j
u
l
-
9
1
j
u
l
-
9
2
j
u
l
-
9
3
j
u
l
-
9
4
j
u
l
-
9
5
j
u
l
-
9
6
j
u
l
-
9
7
j
u
l
-
9
8
sodium chloride sulphate bicarbonate
Figure 10. Concentration (mg/l) vs. time Dayman deep well (Uruguay).
at a regional and a local scale. The homogeneity that many authors mention is due to a
lack of data or detailed studies.
In the Santa Fe province, this work is the first step in a study whose purpose is to estab-
lish the outline of the western border of the G.A.S. and reduce the uncertainties associated
with developments that could be undertaken to produce deep hot groundwater.
Using the analyses of both the available information at local and regional scale and the
new data generated by the field studies, the depth and thickness of the geological forma-
tion identified in the study area were corroborated.
For the right bank of the Paran River (east-central Santa Fe province, Argentina), the
top of the basalt formation and its thickness were clearly identified. In this area sediments
were found that might be saturated underlying the basalt formation. These sediments
belong to the Tacuaremb Formation (sandstone of Tacuaremb).
Based on comparisons between lithological well log descriptions of several deep
pumping wells drilled in the Entre Ros province, permeable sediments interbedded in the
basalt formation might also be found in the Santa Fe province. These layered materials
would, hydraulically, behave as aquifers and could store warm salty water.
The differences between the wells pumping fresh groundwater and those that pump salty
groundwater are clearly visible. Brackish groundwater is considered to belong to the layered
permeable levels interbedded in the basalt formation overlying the G.A.S. water-bearing
sandstone, which stores fresh groundwater.
Some environmental problems have been detected in the Entre Ros Province as a con-
sequence both of the inadequate well construction procedures and the bad management of
the saline effluents on the surface. Consequently, emphasis must be put on the need for
more efficient controls of well-design features and construction practices as well as of well
performance and operation. Furthermore, it is recommended that lithological logs, sedi-
ment samples for mechanical analyses, water samples for chemical analyses, information
on groundwater levels, types of aquifers penetrated should all receive more attention in the
design of such wells.
A better knowledge of the characteristics of the Guaran Aquifer System is a key issue
for the protection and management of this transboundary aquifer.
ACKNOWLEDGMENTS
The authors wish to thank the reviewers of this paper for their detailed and constructive
comments which helped them to improve the text. The authors also wish to express their
immense gratitude to the Editor in Chief, Professor Gh. De Marsily for his helpful and
wise suggestions.
REFERENCES
Araujo, LM, Frana, AB, Potter, PE (1999). Hydrogeology of the Mercosul aquifer system in the
Paran and Chaco-Paran Basisns, South America and comparison with the Navajo-Nugget
aquifer system, USA. Hydrogeology Journal 7(3): 317339.
Fil, MF (2001). Sntesis Geolgica e Hidrogeolgica del noroeste de la Provincia de Entre Ros,
Repblica Argentina. Boletn Geolgico y Minero 112: 2536.
Montao, J, Tujchneider, O, Auge, M, Fil, M, Paris, M, DEla, M, Perez, M, Nagy, MI, Collazo, P,
Decoud, P (1998). Acuferos Regionales en Amrica Latina. Sistema Acufero Guaran. Captulo
Argentino-Uruguayo. Centro Internacional de Investigaciones para el Desarrollo de Canad
(C.I.I.D.) y Centro de Publicaciones de la Universidad Nacional del Litoral. Santa Fe, Argentina.
Santa Fe, Argentina.
Montao, J, Da Rosa Filho, E, Hernandez, M (2004). Caractersticas hidrogeolgicas del Sistema
Acufero Guaran. http://www.alhsud.com/castellano/articulos_listado.asp. Cited 01 Abr 2004.
Seoane Borracer, N, Silva Busso, A (2005). Aspectos sobre la distribucin de fluoruros en el Sistema
Acufero Guaran en Argentina y Uruguay. In: Blarasn, Cabrera & Matteoda (Compiladores)
Proc. IV Congreso Argentino de Hidrogeloga 1: 171180.
Silva Busso, A (1999). Contribucin al conocimiento de la geologa e hidrogeologa del Sistema
Acufero Termal de la Cuenca Chacoparanense Oriental Argentina. PhD Universidad de Buenos
Aires, Argentina. (in Spanish).
Sracek, O, Hirata, R (2002). Geochemical and stable isotopic evolution of the Guaran Aquifer
System in the state of Sao Paulo, Brazil. Hydrogeology Journal 10(6): 643655.
Tujchneider, O, Perez, M, Paris, M, DEla, M (2003). Sistema Acufero Guaran: un recurso com-
partido por cuatro pases. In: Basile, Mancinelli, Pouey, Riccardi, Zimmermann (Eds) Proc. III
Congreso Argentino de Hidrogeologa 2: 7191.
Tujchneider, O, Tineo, A (2005). Geohydrology in plain areas. A conceptual model of a complex sys-
tem Los Saladillos, Santa Fe Province, Argentina. In: Bocanegra, Hernndez, Usunoff (Eds)
Groundwater and Human Development. Balkema. 12: 145158.
CHAPTER 19
Defining groundwater flow and mass vertical
exchanges in a regional aquifer system through
geochemical modelling Aquitaine Basin, France
Laurent Andr
1
, Michel Franceschi
2
& Olivier Atteia
2
1
BRGM Water Division, avenue C. Guillemin, Orlans Cedex, France
2
Institut EGID University Michel de Montaigne Bordeaux 3, Pessac, France
ABSTRACT: Hydrogeological and geochemical data together with geochemical modelling were
used to identify main flow directions and the origin of water mineralization in a regional deep aquifer
system located in the Aquitaine Basin. Four geochemical processes were identified (dissolution, redox,
acidbase reaction, and ion exchange), which explain most of the observed spatial variability in
groundwater composition. Combining geochemical modelling and hydrologic data allowed the identi-
fication of the most likely water pathways at the aquifer scale. In addition, concentrations of several
ions were used to highlight aquitardaquifer interactions. A complementary study was performed to
estimate mass fluxes between the two units. Processes such as leakage, diffusion and mineral dissolu-
tion followed by diffusion from an overlying aquitard were investigated. Results suggest that the com-
position of the major aquitards may impact significantly on the chemistry of the aquifer water.
1 INTRODUCTION
The Eocene sands aquifer, a deep regional aquifer located in the southern portion of the
Aquitaine Basin, constitutes an important water resource used for various purposes (drinking
water, agriculture, thermal spa and gas storage). Water usage conflicts have prompted
detailed hydrogeological studies of the aquifer (Labat, 1998). Hydrodynamic characterisation
allowed identification of some main flow lines and a better understanding of this aquifer
system near particular geological structures.
However, to obtain a better representation of the overall behaviour of the aquifer, a geo-
chemical study has to be carried out. Recent studies have presented individual aspects of
the geochemical properties of the Eocene sands aquifer. The spatial variation of the chemical
composition of the waters was investigated by Blavoux et al. (1993) as well as their resi-
dence time on the basis of
14
C activities. More recently, new data on sulphur isotopes from
dissolved sulphates were used to develop a geochemical model that analyses the processes
leading to the chemical composition of each type of water (Andr, 2002; Andr et al.,
2002). At the aquifer scale, four main geochemical processes dissolution, redox, acidbase
reaction, and ion exchange were identified which could explain most of the observed
spatial variability in groundwater composition.
After a summary and a review of all the previous studies, this paper presents a new
approach which goes one step further through the combined use of geochemical indicators
and modelling, both linked to structural or deep sedimentological features in order to identify
main flow pathways and potential mass transfers. Geochemical modelling results were
combined with hydrogeologic data to identify potential flow pathways at the aquifer scale.
Similar regional approaches were previously used by others to interpret the geochemical
evolution in regional aquifers (e.g., Hendry and Schwartz, 1990; Weaver and Bahr, 1991;
Gerla, 1992; Sracek and Hirata, 2002). This coupled modelling allowed the identification of
flow barriers at the regional scale and a more precise definition of water circulation paths in
complex geological structures. This approach, which involves analyses of geochemical
reactions, also highlights aquitardaquifer interactions and mass exchange between these
two units.
Studies using different tracers such as noble gases (Castro and Goblet, 2003),
36
Cl
(Love et al., 2000), and CO
2
(Chiodini et al., 1999) have demonstrated the presence of
vertical fluxes within aquitards. To better understand mass transfer between aquifers and
aquitards we suggest the use of a method based on the chemical composition of ground-
water. Processes such as diffusion, dissolution and leakage in the aquitard are analyzed in
detail. Such analyses together with consideration of the geological and sedimentological
context allow us to subsequently identify the dominant process taking place in the
aquitards.
2 HYDROGEOLOGICAL AND GEOLOGICAL SETTINGS
The Aquitaine Basin located in south-western France (Fig. 1) is limited in the east by the
foothills of the Montagne Noire, in the south by the North Pyrenean Piedmont, in the west by
the Atlantic Ocean and in the north by the Poitou Plateau. The Eocene sands aquifer, part of a
multi-layered aquifer system, is located in the southern portion of the basin, south of the
Garonne River, and extends over 150km from east to west and 200km from south to north.
Its outcrop is limited in extent, and cross-formational flow both from under- and overlying
formations probably represents a major water source for this aquifer. The average thickness of
the quartz sand deposit is about 50m, its porosity is 2035%, and its average hydraulic
conductivity, estimated from pumping tests and numerical modelling, is 3 10
5
m/s
(Labat, 1998). Using a gradient of 0.0015 (a hydraulic gradient of 160m between wells 23
and 8 for a distance of 100km) and an effective porosity of 20% leads to a pore velocity
of 5 m/yr. Groundwater flow is from SE to NW but outflow from the aquifer is not com-
pletely identified (Fig. 1).
The sand deposits consist mainly of quartz, some calcite, and, occasionally, dolomite and
K-feldspars (Andr et al., 2002). Detrital sediments were eroded from the Massif Central,
the Montagne Noire and the emerging Pyrenees and deposited in vast marshy plains. Local
geological structures (e.g., domes of Garlin or Saint-Mdard) as well as larger ones (e.g.,
Audignon anticline, on the western border of the aquifer; and the Celt-Aquitaine flexure
which appears to divide the aquifer into two distinct zones) have affected this surface.
Most of the structures appear to be related to the deformation of deep Triassic sands (Rey,
1995; Serrano, 2001).
Clay deposits of variable thickness (10100m) separate the underlying calcareous and
dolomitic DanoPaleocene aquifer from the Eocene sands. The latter is overlain by several
hundreds of metres of Tertiary molasses. The aquitard mineralogy, although less well
known, is more varied, with quartz, feldspars, micas, several clay types (detrital limestone
layers with sandy-clay deposits) and crystallised gypsum commonly observed in cuttings
at the base of the molasse formation.
3 GROUNDWATER CHEMISTRY
Water from the Eocene sands aquifer has a dominant calcium bicarbonate facies. Sodium
bicarbonate water with low total dissolved solids (e.g., well 13) or moderately
mineralised water (e.g., well 17) is rare. Sodium sulphate water is found only in a
few wells. Most of the water has a total dissolved solid concentration lower than 1000mg/l
(Table 1). The water presents a redox potential ranging between 130mV to weakly
positive values except near the outcrops (Eh of well 23 is close to 360mV) (Andr,
2002).
The distribution of sulphate and chloride concentrations shows a regional zonation. In
the south, weakly mineralised water has sulphate concentrations that vary from 0.20meq/l
near the outcrop to 1.50meq/l near discharge areas. In the north, concentrations lie
between 2 and 23meq/l.
Sulphur is present in various forms, the most common being dissolved sulphate.
Sulphide can also be present but concentrations of this species are always relatively low
(1mg/l), and represent at most 10% of the total sulphur. Sulphide results mostly from
local bioreduction phenomena.
A more accurate study of sulphur isotopes from dissolved sulphates in the waters suggests
a division of the aquifer into four distinct zones (Table 2, Fig. 2; see Andr et al., 2002):
Defining groundwater flow and mass vertical exchanges in a regional aquifer system 255
Figure 1. Map of the studied region and piezometric data of the Eocene sands aquifer.
T
a
b
l
e

1
.
C
h
e
m
i
c
a
l

c
o
m
p
o
s
i
t
i
o
n

o
f

w
a
t
e
r

f
r
o
m

t
h
e

E
o
c
e
n
e

s
a
n
d
s

a
q
u
i
f
e
r
.
N
o
W
e
l
l

n
a
m
e
s
T

(
C
)
p
H
C
a
2
M
g
2
N
a
(
m
g
/
l
)
K
H
C
O
3
C
l
S
O
4
2
(
m
g
/
l
)
(
m
g
/
l
)
(
m
g
/
l
)
(
m
g
/
l

H
C
O
3
)
(
m
g
/
l
)
(
m
g
/
l

S
O
4
)
1
B
a
r
b
o
t
a
n

1
0
2
3
7
.
2
7
.
0
7
5
0
.
6
0
6
.
8
8
1
2
.
1
6
5
.
8
3
1
7
0
.
3
0
9
.
4
8
4
0
.
0
1
2
B
a
r
b
o
t
a
n

1
0
3
3
6
.
9
7
.
2
5
4
8
.
2
9
6
.
7
0
1
1
.
7
4
5
.
9
9
1
7
6
.
0
0
8
.
7
1
3
1
.
6
2
3
B
a
r
b
o
t
a
n

L
o
t
u
s

2
3
1
7
.
1
0
5
4
.
4
8
8
.
1
0
1
1
.
3
7
5
.
8
0
1
8
3
.
7
0
8
.
3
8
5
1
.
3
9
4
L
e
c
t
o
u
r
e
4
2
.
5
7
.
2
0
4
4
.
6
4
1
3
.
4
9
5
4
8
.
2
0
2
0
.
3
0
3
6
7
.
7
0
3
2
1
.
2
0
6
5
1
.
8
0
5
L
u
s
s
a
g
n
e
t

5
7
4
5
.
4
7
.
2
9
4
3
.
7
1
7
.
1
2
9
.
3
7
5
.
5
2
1
7
8
.
5
4
6
.
5
9
2
0
.
3
2
6
I
z
a
u
t
e

5
3
5
.
5
7
.
3
5
4
1
.
8
5
5
.
2
3
9
.
3
3
4
.
8
7
1
6
3
.
6
9
7
.
9
5
1
5
.
2
2
7
N
o
g
a
r
o

2
5
1
.
3
7
.
3
5
3
8
.
4
6
4
.
9
4
1
4
.
3
0
5
.
0
2
1
5
5
.
0
6
8
.
1
6
2
8
.
1
9
8
G
o
n
d
r
i
n
4
2
.
4
7
.
4
5
3
1
.
0
6
7
.
4
4
2
3
.
1
9
5
.
7
6
1
6
3
.
0
0
7
.
1
0
2
7
.
6
7
9
D
e
m
u
5
2
.
3
7
.
3
2
3
2
.
8
7
4
.
6
4
1
7
.
1
1
6
.
2
4
1
5
9
.
7
0
7
.
4
1
7
.
9
3
1
0
B
e
a
u
c
a
i
r
e
3
4
.
3
7
.
4
4
1
3
4
.
3
5
0
.
6
3
1
6
1
8
2
0
5
.
4
4
4
9
.
9
3
1
0
6
9
1
1
C
a
s
t
r
a
-
V
e
r
d
u
z
a
n
3
0
.
4
7
.
4
8
4
4
.
3
2
1
4
.
5
5
2
1
.
3
8
6
.
8
5
1
7
8
.
6
0
7
.
0
9
7
9
.
2
0
1
2
P
l
h
a
u
t
2
7
.
9
7
.
4
5
4
2
.
4
8
1
0
.
6
2
1
8
.
5
8
8
.
9
8
1
6
7
.
1
1
6
.
9
8
4
9
.
1
1
1
3
E
u
g
n
i
e

l
e
s

B
a
i
n
s

2
1
9
.
0
7
.
9
4
2
2
.
7
8
6
.
2
6
4
2
.
8
8
7
.
1
6
2
0
1
.
5
2
7
.
4
4
1
0
.
9
8
(
E
L
B

2
)
1
4
G
e
a
u
n
e
3
0
.
9
7
.
4
3
5
3
.
0
8
5
.
0
9
9
.
2
0
4
.
3
2
1
8
1
.
6
3
5
.
8
3
1
3
.
5
3
1
5
G
e
a
u
n
e

P
c
o
r
a
d
e

1
0
1
3
2
.
4
7
.
4
1
4
9
.
8
4
4
.
1
3
6
.
0
8
3
.
5
2
1
8
5
.
3
0
5
.
6
5
1
2
.
7
7
1
6
G
e
a
u
n
e

B
a
t
s
3
2
.
5
7
.
3
1
5
0
.
6
0
4
.
2
7
5
.
2
7
3
.
6
9
1
8
4
.
3
5
5
.
2
7
1
1
.
3
2
1
7
B
l
a
g
n
a
c

(
p
i
s
c
i
n
e
)
4
9
.
6
7
.
2
0
1
2
.
2
2
3
.
2
3
2
8
4
.
6
0
1
0
.
2
8
4
2
5
.
4
6
1
2
6
.
3
0
1
3
1
.
0
0
1
8
L
a
l
b
a
r
e
d
e
2
5
.
7
7
.
7
7
1
2
.
2
2
6
.
8
2
1
6
5
.
2
0
5
.
4
0
2
8
1
.
4
0
8
0
.
1
5
5
8
.
3
8
1
9
G
a
r
l
i
n
2
7
.
2
7
.
5
0
3
4
.
3
9
8
.
5
8
2
5
.
5
0
3
.
6
3
1
8
8
.
1
6
1
5
.
0
3
4
.
6
3
2
0
L
e
s
p
i
e
l
l
e

1
2
6
.
5
7
.
3
9
4
2
.
1
9
5
.
1
0
9
.
9
3
2
.
8
2
1
7
3
.
5
5
7
.
4
6
1
0
.
8
8
2
1
L
a
m
a
z
r
e
5
0
.
1
7
.
3
1
3
2
.
1
8
9
.
9
7
5
7
.
1
5
1
0
.
3
3
2
9
2
.
3
8
7
.
2
6
2
5
.
9
0
2
2
S
a
i
n
t

M
d
a
r
d
2
0
.
6
8
.
4
2
2
2
.
3
0
1
1
.
9
1
1
5
.
6
0
5
.
2
6
1
4
3
.
0
6
1
1
.
7
1
1
9
.
1
6
2
3
B
o
r
d
e
s

3
1
3
.
4
7
.
4
6
8
7
.
1
6
4
.
0
8
5
.
0
2
1
.
4
0
2
2
3
.
7
4
1
1
.
9
2
1
4
.
4
3
2
4
G
r
i
g
n
o
l
s
2
1
.
4
7
.
0
8
1
2
3
.
9
4
5
.
6
1
4
1
.
9
0
1
4
.
7
2
0
2
.
5
2
1
6
7
.
4
0
4
2
3
.
5
T
a
b
l
e

2
.
I
s
o
t
o
p
i
c

c
o
m
p
o
s
i
t
i
o
n

o
f

w
a
t
e
r

f
r
o
m

t
h
e

E
o
c
e
n
e

s
a
n
d
s

a
q
u
i
f
e
r
.
W
e
l
l

n
u
m
b
e
r
1
8
O
(
H
2
O
)
1
3
C
A
1
4
C

(
p
.
c
.
m
.
)
1
8
O
(
S
O
4
)
3
4
S
(
S
O
4
)
3
4
S
(
H
2
S
)
(

S
M
O
W
*
)
(

P
D
B
*
*
)
(

S
M
O
W
*
)
(

C
D
T
*
*
*
)
(

C
D
T
*
*
*
)
1
8
.
9
1
2
5
.
5
0
.
3
4
.
4
1
2
0
.
1
2
4
.
2
8
2
8
.
9
1
2
5
.
1
0
.
5
5
.
3
3
1
6
.
1
8
N
M
3
8
.
6
1
1
.
9
2
4
.
2
0
.
4
3
.
9
5
1
5
.
8
3
2
.
2
5
4
7
.
2
9
.
9
8
2
.
5
0
.
2
1
6
.
1
3
1
2
.
2
3
/
5
9
.
4
1
1
.
2
2
.
3
0
.
3
9
.
1
9
3
.
2
1
1
5
.
4
8
6
8
.
7
8
1
0
.
9
6
8
.
9
3
0
.
1
1
9
.
1
1
1
.
6
3
2
4
.
0
5
7
8
.
7
1
3
.
5
3
.
3
0
.
2
1
2
.
6
1
8
.
7
6
2
0
.
1
7
8
1
2
.
9
2
.
1
0
.
3
1
5
.
4
1
1
.
5
6
/
9
7
.
6
9
1
4
.
4
4
.
5
0
.
4
1
2
.
8
4
1
8
.
9
3
1
6
.
3
3
1
0
7
.
2
2
6
.
8
1
4
.
9
4
0
.
0
8
1
3
.
6
6
1
5
.
0
5
N
M
1
1
8
.
1
8
1
0
.
0
5
4
.
7
5
0
.
1
3
1
5
.
1
5
1
0
.
3
2
N
M
1
2
8
.
2
1
3
.
8
4
.
5
0
.
5
1
5
.
2
8
9
.
7
6
/
1
3
9
.
5
6
1
2
.
5
8
5
.
2
4
0
.
1
2
1
4
.
1
4
3
3
.
1
8
1
0
.
5
5
1
4
8
.
6
1
1
.
4
6
.
9
0
.
5
8
.
3
3
.
2
/
1
5
7
.
2
9
1
1
.
2
9
9
.
7
2
0
.
1
1
5
.
9
8
4
.
0
1
N
M
1
6
7
.
5
6
1
1
.
5
1
2
.
8
1
0
.
1
3
6
.
6
3
2
.
9
8
N
M
1
7
7
.
3
5
8
.
4
1
1
.
4
4
0
.
0
7
1
5
.
2
3
1
3
.
2
3
7
.
8
6
1
8
7
.
3
2
1
1
.
9
8
2
.
8
9
0
.
0
6
1
7
.
2
3
1
3
.
0
6
1
0
.
7
8
1
9
8
.
1
1
4
.
5
0
.
7
1
2
.
8
5
1
3
.
0
1
2
8
.
0
5
2
0
1
4
.
3
3
.
0
0
.
8
1
1
.
9
6
5
.
7
9
N
M
2
1
8
.
5
6
3
.
6
0
.
2
1
4
.
9
3
1
7
.
5
/
2
2
7
.
3
1
8
.
4
8
1
3
.
0
9
0
.
2
1
4
.
2
7
2
4
.
3
7
N
M
2
3
7
.
3
1
1
.
9
7
7
.
3
1
.
0
5
.
1
3
4
.
8
7
/
2
4
7
.
5
3
4
.
1
5
0
.
9
6
0
.
0
7
1
2
.
8
3
1
5
.
7
0
/
N
M
:

N
o
t

M
e
a
s
u
r
e
d
;

*
S
M
O
W
S
t
a
n
d
a
r
d

M
e
a
n

O
c
e
a
n

W
a
t
e
r
;

*
*
P
D
B
P
e
e

D
e
e

B
e
l
e
m
n
i
t
e
;

*
*
*
C
D
T
C
a
n
y
o
n

D
i
a
b
l
o

T
r
o
i
l
i
t
e
.
Zone A, along the edge of the Pyrenees, in the outcrop area. Here,
34
S is close to
values found for sulphates in rainwater (3.2
34
S 8.2 CDT) (Pearson and
Rightmire, 1980).
Zone B, the largest zone, extends from the east toward the centre of the basin.
34
S and
18
O(SO
4
) are close to values measured in gypsum crystals sampled in the overlying
molasse, at the top of the aquifer (
34
S 12.72 1.20 CDT and
18
O 14.9
SMOW). These values agree with those for sulphate evaporites (Claypool et al., 1980).
Zone C, an elongated area oriented SWNE. Low to very low negative values is
observed (from 2.98 to 20.1 CDT) with a decrease from the south northward.
18
O(SO
4
) values follow a similar decreasing trend. These values seem to reflect an ori-
gin of dissolved sulphates distinct from that of zone B. Points 1,2,3 are very depleted
in
34
S and
18
O which may be due to oxidation of sulphide minerals (Dazy et al., 1980).
Points 5,6, 1416 present
34
S values close to 0 and
18
O values that are typical for air
pollution. Nevertheless, due to the aquifer depth (more than 600m), this assumption
can not be considered and the most probable explanation of these values is oxidation of
mineral sulphides.
Zone D, in the west, is represented by a single point, close to the aquifer boundary. This
value presents a significant enrichment in sulphur 34 (
34
S 33 CDT), probably
linked to bioreduction processes. Although bioreduction is particularly well identified
for this well, this process is also present elsewhere in the basin (e.g. wells 9 and 19).
The bioreduction phenomenon is confirmed by the calculated fractioning coefficient,
which ranges from 1.032 to 1.044 for these three wells. This value is commonly
Figure 2. Map of sulphur-34 values in dissolved sulphates.
accepted for sulphate reduction by Desulfovibrio desulfuricans bacteria (Krouse,
1980).
4 GEOCHEMICAL MODELLING STRATEGY
Geochemical modelling is a tool which added to hydrogeological data makes it possible to
determine accurately the main groundwater pathways. In this study, the geochemical model-
ling, performed with the geochemical code Phreeqc (Parkhurst and Appelo, 1999), was
made in two steps: firstly, the relative magnitude of each chemical reaction at each sampled
well was quantified; secondly, flow between different wells was assumed and the most
probable ones were used to define the geochemical pathways within the aquifer.
During this first geochemical modelling stage, only the system Na, Ca, Mg, Cl, S, CO
3
, and
H was considered. Si and K concentrations were not included due to lack of information on
the composition of the clay fraction of the aquifer, which was not forthcoming from the
borehole cuttings. The Eocene sands mineralogy is poorly understood with only few minerals
identified as quartz, calcite, feldspars, kaolinite, mixtures of iron oxides/hydroxides and
pyrite. Gypsum and anhydrite are present at only one place in the extreme east of the basin. In
places the sandy series include argillaceous or argillaceous-sandstone interbeds. Calculations
of saturation indices with the geochemical code Phreeqc (Parkhurst and Appelo, 1999) indi-
cate that the water is generally at equilibrium with respect to calcium carbonate minerals
and slightly under-saturated compared to dolomite. All the waters are under-saturated with
respect to sulphur-containing minerals (gypsum, anhydrite and pyrite) and near saturation
with respect to siderite. All other mineral phases which could control NaCl in a sedimentary
system are largely under-saturated. According to these results, calcite, dolomite, gypsum,
halite, siderite and pyrite are the mineral phases taken into account to represent our system.
To obtain the original infiltrating water composition used for simulations, the sim-
plest assumption is generally to take into account the lowest observed concentration of
each ion within all groundwater samples. This relies on the fact that most of the water-rock
interaction processes tend to increase ion concentrations. Thus, the average concentrations
in the infiltration water, is deduced from both the lowest observed concentrations and elem-
ental concentrations in precipitation throughout historical time. The initial equilibrium
pCO
2
of this infiltrating water was assumed to be 10
2.5
atm.
Water interactions between original water and mineral phases can provide a basis for the
reactions that may occur in the aquifer. Based on facies and sulphur isotopic data, three
distinct hydrochemical processes responsible for the water composition were identified:
Calcium carbonate equilibrium. A dominant Ca-HCO

3
facies extending from recharge
to discharge areas in which dissolution of carbonate minerals such as calcite appears to
control the equilibrium. Equilibrium with calcite and potential dolomite dissolution
was used in the geochemical modelling to define the Ca-Mg-CO
3
system.
Pyrite oxidation and gypsum dissolution (Andr et al., 2002). These two processes can
explain variations of the sulphur isotopic data and observed continuous increase of sul-
phur concentrations along water pathways. Although gypsum or anhydrite dissolution
constitutes a well-known geochemical process, pyrite oxidation must be detailed accur-
ately in order to model it. As the Eocene sand water contains small amounts of oxygen
(00.15mg/L) and redox potentials range between 50mV and 0mV, indicating a
slightly reduced medium, the following reactions involving iron oxides and pyrite
(Holmes and Crundwell, 2000) were considered as one of the possible combinations
able to reproduce the Fe, H
2
S and SO
4
concentrations in solution:
Fe(OH)
3
3 H
:Fe
3
3 H
2
O
FeS
2
14 Fe
3
8 H
2
O :15 Fe
2
2 SO
4
2
16 H
Fe
2
HCO
3
:H
FeCO
3
The overall reaction is given by:
FeS
2
14 Fe(OH)
3
11 H
15 HCO
3
:2 SO
4
2
34 H
2
O 15 FeCO
3
Cation exchange between the water and reservoir rocks. This process explains the
observed sodium bicarbonate facies encountered in some waters as well as an increase
in Na
+
concentration without an associated increase of Cl
(see, e.g., Back, 1966;

Freeze and Cherry, 1979; Thorstenton et al., 1979; Chapelle and Knobel, 1983; Appelo
and Postma, 1993). In an aquifer where carbonate minerals are present, cation exchange
may be accompanied by calcite dissolution (and dolomite) as Ca
2
(Mg
2
) is removed
from the solution and replaced by Na
. In the Eocene sands, clay beds of variable

thickness are occasionally found which can play the role of ion exchangers. Under
these conditions, ion exchangers are considered in the modelling approach and used to
reproduce the Na-HCO
3
facies present in some waters.
Under the chemical conditions considered here, excluding redox reactions in a first
approach, seven conditions are imposed: equilibrium with calcite, equilibrium or addition
of dolomite, addition of gypsum and halite, ion exchange, starting pCO
2
of infiltration
water and electro neutrality. Seven ion concentrations are considered, i.e. Ca, Mg, Na,
CO
3
, SO
4
, Cl and H. This is, therefore, a fully determined system which should have a
unique solution. Although the modelling system by itself may be consistent, the assump-
tions might be wrong; for instance, we may be able to solve a system by adding a small
amount of dolomite, while Mg may come from another source. To overcome this difficulty,
validation variables are needed to confirm or reject our hypotheses.
Among many parameters, the pH was chosen to be one of these variables. The attention
paid to the pH measurements in the field allows us to consider this variable as a good val-
idation variable. Moreover,
13
C can also be used as a validation variable because its rela-
tive concentration in groundwater will depend only on the geochemical processes cited
above (Clark and Fritz, 1997). Thus, if the calculated amounts of dissolved calcite and
dolomite are the right ones, the
13
C values should match.
If the first step of the geochemical modelling consists in determining the reactions able
to reproduce the chemical composition of the water at each well, the second step tries to
define the links between all of the wells. In fact, a similar composition of two waters is not
a proof of hydrodynamic connection. However, if a hydraulic connection is assumed, one
must be able to explain the chemical processes that lead from one point to the other. This
is the basis of our use of the geochemical model to delineate groundwater flow paths:
some water paths are not possible and some preferential flow directions can be drawn up
if one compound increases along the flow line.
5 GEOCHEMICAL MODELLING RESULTS AT THE WELL SCALE
From these three processes and using a limited number of mineral phases identified in the
solid, a step-by-step modelling approach was developed which allowed us to successfully
reproduce the major ion concentrations at each sampling point. The geochemical processes
controlling the chemical composition of the Eocene sands aquifer are summarized in Fig. 3.
As deduced from waters facies, calcite equilibrium is one of the major factors influen-
cing pH evolution together with pyrite oxidation. This result is very convincing because,
due to the number and complexity of the involved geochemical processes, the pH appears
as a rather sensitive variable which reflects the chemical reactivity of the system. Measured
and computed
13
C values are also in agreement, with absolute deviations of 4 PDB.
Moreover, the overall dissolved-solid concentrations can be obtained mainly from gypsum
dissolution and minor halite (NaCl) dissolution at some locations. Although this process is
in agreement with the water composition, the waters are largely undersaturated with
respect to both minerals, indicating that gypsum and halite are not present within the
aquifer, a conclusion consistent with the depositional age of the aquifer and its fairly rapid
renewal. We have, therefore, to postulate gypsum dissolution (shown by the
34
S-
18
O val-
ues) outside the aquifer and a process that transfers dissolved sulphate into the Eocene
aquifer. We investigated the potential source of this gypsum within the overlying aquitard.
Throughout the aquifer small amounts of dissolved dolomite were required to reproduce
the observed Mg concentrations. Because most waters are under-saturated with respect to
dolomite, Mg -rich water may incorporate the Eocene sand waters through vertical leak-
age rather than originating in the reservoir rocks. As expected, Na variations are influenced
by ion exchange processes. Despite our poor knowledge of the clayey phase, the typical
ion exchange constant used seems to reproduce the measured concentrations. Although
ion exchange occurs only in certain areas, it significantly modifies the local groundwater
composition.
Figure 3. Geochemical processes governing the chemical composition of waters from the Eocene
sands aquifer.
6 GEOCHEMICAL AND HYDROGEOLOGICAL MODELLING AT
THE AQUIFER SCALE
Based on the local reconstruction of the chemical composition of the water, geochemical
modelling was carried out to define geochemical pathways. This led to the identification
of two distinct zones, one in the south and the other in the northern portion of the aquifer
(Fig. 4).
The latter presents a clear trend of increasing salt content from east to west. Main flow
directions obtained through geochemical modelling agree with those indicated by hydraulic
gradients. In the southern area, the potentiometric surface indicates a major flow direction
from south-east to north-west. In the eastern area, some trace elements such as fluoride, stron-
tium and lithium are more concentrated (Andr, 2002). On the other hand, isotopic studies
(Andr et al., 2002) have shown that in the western part a significant portion of the sulphates
arise from pyrite dissolution. The water composition shows large bands of similar composi-
tion oriented south-north. In the western part, the evolution of
34
S isotopes also suggests a
south-north trend of pyrite dissolution. All geochemical arguments thus converge to indicate
a south-to-north flow direction. The composition of some waters also indicates that their
chemistry is mainly influenced by deep geological structures. The concept of open versus
closed structures might be applicable to this region. This concept would enlarge the classi-
cal discussion of faults acting as barriers versus conduits for water flow (Andr et al., 2005).
Figure 4. Major flow direction derived from modelling results. The main pathways are built by tak-
ing into account the hydrogeological data (piezometric lines) combined with geochemical data (in
particular
34
S to delineate the water in the western part of the aquifer).
7 CAN VERTICAL EXCHANGES EXPLAIN MINERALISED WATERS IN
AN AQUIFER?
In the previous sections we were able to explain the chemical composition of groundwater
by geochemical modelling. Although pyrite and calcite can be found in the Eocene sands,
NaCl or CaSO
4
bearing minerals are absent. We have shown elsewhere (Atteia et al., 2005)
that these types of highly soluble minerals are rapidly dissolved at the geological
timescale. On the other hand, investigations on deep layers of the molasse deposits showed
the presence of evaporitic layers. As these layers could deliver significant amounts of ions,
we detail in this section the potential vertical mass transfer between aquitard and aquifer.
This approach is not quantitative because it would need perfect knowledge of the whole
basin and the available data obtained from sampled wells are too scattered to reach this
goal. That is why this study focuses on the qualitative vertical exchanges between
aquiclude and aquifer.
Several processes can explain this vertical sulphate transfer, the most significant ones
being leakage and diffusion. Leakage is used with increasing frequency in regional mod-
elling of multi-layered aquifer systems (Tth, 1999). The quantification of such vertical
fluxes is critical for the management of aquifer exploitation at the regional scale. It con-
sists of a vertical flow through an aquitard, in this case the molasse, downward, to the
Eocene sands. Water present in the molasse can be saturated with respect to one mineral
(i.e. gypsum) leading to large concentration variations in this water over short distances.
On the other hand, diffusion from concentrated brines in the aquitard (molasse) will lead
to a cumulative flow of ions into the diluted medium (the Eocene sands) and a concentra-
tion increase along the flow direction. Dissolution of mineral phases existing in the
aquitard generates the same type of concentration gradient except that it may remain over
a longer time scale due to the larger amounts of matter available in the solid phase.
8 QUANTIFICATION OF VERTICAL EXCHANGES
Downward leakage is considered between the two previously described units. For a fully
saturated aquitard, which crops out, hydraulic head values in this area roughly correspond
to the topographic level. In this situation and at high elevations, downward leakage to the
underlying aquifer may occur. If longitudinal dispersion is negligible, fluxes in the aquitard
and aquifer can be estimated (Fig. 5). Ion concentrations in the aquifer are obtained through
mixing, with an initial concentration C
0
in the aquitard being diluted in the aquifer. Solutes
are added along pathways through leakage. An example of such flow and its impact on the
head and ion concentrations in the aquifer is shown for variable aquitard properties and a
negligible ion concentration at the upstream boundary of the aquifer (Fig. 5; see also
Atteia et al., 2005). Results indicate that a small vertical flux leads to large concentration
variability in the aquifer. Under such assumptions, the cumulative leakage flux can reach
10 to 35% of the total groundwater flux at an outcrop distance of 5km. Therefore, concen-
tration increase in certain areas along water flow paths may indicate leakage from an
aquitard containing mineralized water.
If neighbouring aquifers and aquitards have distinct chemical compositions, diffusion
may occur. If the aquitard contains a large amount of dissolved salts, the latter may diffuse
toward the less mineralised water. To illustrate this, we consider an aquifer overlain by a
very thick aquitard with an original ion concentration C
0
. The aquitard contains pore fluid
such as seawater but the aquifer is fresh water dominated. The diffusive flux between the
aquitard and aquifer is calculated using Ficks law and diffusion coefficients in pure water
at infinite dilution (e.g., Lasaga, 1998). A small residual ion diffusive flux may lead to a
slight but visible increase in concentration of the resulting ions along the flow direction.
The residual flux from diffusion may remain in the system over geological time scales.
After 50Myr (5000 pore volume flushes), Na concentrations are still significant (around
2meq/l of Na, see Atteia et al., 2005).
Figure 5. (a) Boundary conditions for an analysis of leakage between x
1
and x
2
. P
a
and P
b
are heads
in aquifer and in aquitard, at the ground surface, respectively. K
a
and K
b
are permeabilities whereas
Z
a
and Z
b
are the thickness of aquifer and aquitard, respectively; (b) Distribution of head values with
distance; (c) Distribution of the flux along the aquifer; (d) Evolution of the C/C
0
ratio, where C is
concentration in the aquifer and C
0
is concentration in the aquitard. Thick line, Z
b
250m,
K
b
10
9
ms
1
; dashed line, Z
b
750m, K
b
10
9
ms
1
; thick shaded line, Z
b
250m,
K
b
10
10
ms
1
.
Where soluble minerals are embedded in the aquitard, diffusion is responsible for trans-
fer of dissolved ions from the aquitard to the aquifer. This flux tends to deplete the first
few meters of the aquitard pore water in contact with the aquifer and favour local dissol-
ution of minerals. A dissolution front will result and move from the aquifer-aquitard inter-
face into the aquitard. This case is distinct from diffusion alone since the reservoir of
diffusing ions is much larger, as it is in a solid form. The boundary condition in the
aquitard is given by the vertical position of the dissolution front. Over a long distance,
within a homogeneous aquifer, results show a saturation of the downstream part of the
aquifer along the flow path leading to a decreased diffusive flux from the aquitard. This
phenomenon, as for diffusion alone, leads to an increase in solute concentrations down-
flow, an increase that persists over geological time scales. Despite the fast water renewal
within the aquifer (10kyr for 50km), depletion of minerals in the aquitard is slow. For
example, the dissolution front in the aquitard is only 12m away from the aquifer-aquitard
interface after 5Myr of steady-state flow. This explains why a slow but significant dissol-
ution remains and influences aquifer chemistry even after 50Myr (see Fig. 6).
Under the boundary conditions of interest, the three investigated processes (diffusion,
dissolution and diffusion, leakage) lead to a linear increase in groundwater concentrations
with distance.
Fig. 5 shows that a reduced zone of leakage can lead to a large increase of ion concen-
tration in the aquifer. On the contrary, diffusion from the aquitard generates low concen-
trations of sodium. These concentrations, despite being low compared to the C
0
value in
the aquitard, are however easily measurable.
As mentioned above, there is no source of sulphate in the aquifer material. Therefore,
the two processes described above might be the origin of the transfer of sulphate from the
aquitard to the aquifer. Diffusion occurs everywhere but leads to small concentration
increases that may be around 10mg/l of sulphate for 20kyr old waters (see Fig. 6). It is
thus clear that, in the northern area, leakage must be present and may deliver significant
amounts of salt to reach an amount of total dissolved solids of 2g/l as measured at well 4
(Table 1). The presence of leakage in the northern area might also be at the origin of the

0 2 4 6 8 10
0
0.04
0.08
0.12
0.16
0.20
0.24
0.28
x/v (ky)
C/C
eq
1
5
25
10
50
Figure 6. Longitudinal distribution of average ion concentration in an aquifer where diffusion
occurs from an aquitard containing a soluble mineral; C
eq
is the equilibrium concentration for this
mineral (if the considered mineral is gypsum, C/Ceq 0.04 is approx. 5mg/L SO
4
). Circles give
dissolution time in millions of year.
low hydraulic gradient between points 17 and 4 (see Fig. 1). In fact, as shown in Fig. 5, the
presence of leakage tends to decrease the hydraulic gradient upward from the leakage area.
In the southern area, leakage cannot be excluded but it may be very low, with a contri-
bution similar to the one of diffusion. This difference between the south and the north
corresponds to the sedimentology features of the region. In the north, the molasse deposits
are 200 to 300m thick while the thickness lies between 700 and 1000m in the south. The
deposits of the southern area are known to be clayey, while they contain more sand
(erosion from the Massif Central) in the north.
9 CONCLUSIONS
The results of this study show that detailed hydrochemical data coupled with geochemical
modelling can help to elucidate the hydrologic and geologic factors controlling water
chemistry within a regional basin.
In the context of the Aquitaine Basin, geochemical modelling has proven its ability to effi-
ciently constrain the groundwater flow in an aquifer where few data are available. However,
the approach shows that the use of validation variables and information on mineral phases
present in the solid are required for a detailed analysis of each reaction. On the other hand,
the concentration increase for one dissolved species alone may be attributed to the wrong
reaction. We also show that a geochemical model was able to provide four types of hydrody-
namic information: identification of regional flow directions, existing flow barriers, mixing
between aquifer waters and interactions between aquifer and aquitard. However, this infor-
mation remains qualitative: the only solution to reach more quantitative flow patterns is to
use 3D multilayered models, including hydrodynamics, transport and chemistry.
Throughout this paper, we also attempt to estimate the effects of aquifer/aquitard inter-
actions on the resulting concentration gradients in the aquifer. This approach shows the
very different patterns resulting from leakage, dissolution and diffusion. The results may
help to better understand the evolution of chemistry in deep groundwater. Our results also
show that important changes in concentration in the aquifer are observed when leakage is
significant. This may explain the high Na and Cl concentrations in the northern area of the
basin where the molasse is not very thick and might be more permeable than in the south.
By contrast, and despite the presence of gypsum, the low mineralization in the southern
zone tends to indicate an absence of leakage. Ion exchange from the aquitard, able to pro-
vide Na over a geological time scale, seems to be the major geochemical process at spe-
cific places. These areas are often located close to regions where, owing to changes in
sedimentation regimes, numerous clay layers are present in the aquifer. The concomitant
work on geochemical modelling of longitudinal flow in the aquifer and mass exchange
between aquitard and aquifer highlights the major influence of geological structures on
groundwater chemistry.
REFERENCES
Andr L (2002) Contribution de la gochimie la connaissance des coulements souterrains pro-
fonds. Application laquifre des Sables Infra-Molassqiues du Bassin Aquitain. Thse,
Universit de Bordeaux 3, France [in French] 228 p.
Andr L, Franceschi M, Pouchan P, Atteia O (2002) Origines et volution du soufre au sein de
laquifre des Sables Infra-Molassiques du Bassin Aquitain. C.R. Gosciences, 334, 749756.
Andr L, Franceschi M, Pouchan P, Atteia O (2005) Using geochemical data and modelling to
enhance the understanding of groundwater flow in a regional deep aquifer, Aquitaine Basin,
SouthWest France. Journal of Hydrology 305, 14, 4062.
Appelo CA, Postma D (1993) Geochemistry, Groundwater & Pollution, A. A. Balkema,
Rotterdam/Brookfield/1993, 536 p.
Atteia O, Andr L, Franceschi M, Dupuy A (2005) Contributions of diffusion, dissolution, ion
exchange, and leakage from low-permeability layers to confined aquifers. Water Resources
Research 41(9):W09412, doi:10.1029/2003WR002593.
Back W (1966) Hydrochemical facies and groundwater flow patterns in northern part of Atlantic
Coastal Plain. U.S. Geological Survey Professional Paper 498-A, 42 p.
Blavoux B, Dray M, Fehri A, Olive P, Grning M et al. (1993) Paleoclimatic and hydrodynamic
approach to the Aquitaine Basin deep aquifer (France) by means of environmental isotopes and
noble gases. International Symposium on Application of Isotope Technics, (Proc. Symp. Vienna,
1993), IAEA-SM-329/69, 293305.
Castro MC, Goblet P (2003) Calibration of regional groundwater flow models working toward
a better understanding of site-specific systems. Water Resources Research 39(6):1172,
doi:10.1029/2002WR001653.
Chapelle FH, Knobel LL (1983) Aqueous geochemistry and the exchangeable cation composition of
glauconite in the Aquia aquifer, Maryland. Ground Water, 21, 3, 343352.
Chiodini G, Frondini F, Kerrick DM, Rogie J, Parello F, Peruzzi L, Zanzari AR (1999) Quantification
of deep CO
2
fluxes from central Italy. Examples of carbon balance for regional aquifers and of soil
diffuse degassing. Chemical Geology 159, 14, 205222.
Claypool GE, Holser WT, Kaplan R, Zak I (1980) The age curves of sulphur and oxygen isotopes in
marine sulphate and their mutual interpretation. Chemical Geology, 28,199260.
Dazy J, Rochat J, Olive Ph (1980) Nouvelles donnes gochimiques et isotopiques sur les eaux ther-
males dUriage-les-Bains (Isre). 105
me
Congrs National des Socits Savantes, Caen, 1980,
Sciences II:111122.
Freeze RA, Cherry JA (1979) Groundwater: Englewood Cliffs, N.J., Prentice-Hall, 604 p.
Gerla PJ (1992) Pathline and geochemical evolution of groundwater in a regional discharge area, Red
River Valley, North Dakota. Ground Water, 30, 743754.
Hendry MJ, Schwartz FW (1990) The chemical evolution of groundwater in the Milk River Aquifer,
Canada. Ground Water, 28, 253261.
Holmes PR, Crundwell FK (2000) The kinetics of the oxidation of pyrite by ferric ions and dissolved
oxygen: an electrochemical study. Geochimica & Cosmochimica Acta, 64, 2, 263274.
Krouse H (1980) Sulphur isotopes in our environment, in: P. Fritz, J.-C. Fontes (Eds.), Handbook of
Environmental Isotopes. Geochemistry, pp. 435471.
Labat N (1998) Rle des particularits sdimentaires et structurales sur le comportement des sables
sous-molassiques soumis aux fluctuations induites par les stockages souterrains de gaz.
Application ltude de leur influence sur lhydrodynamisme des mergences locales. Thse,
Universit de Bordeaux 3, France [in French] 228 p.
Lasaga A (1998) Kinetic Theory in the Earth Sciences, University Press (Princeton).
Love AJ, Herczeg AL, Sampson L, Cresswell RG (2000) Sources of chloride and implications for
36
Cl dating of old groundwater southwestern Great Artesian Basin, Australia. Water Resources
Research, 36, 6, 15611574.
Parkhurst DL, Appelo CAJ (1999) A computer program for speciation, batch-reaction, one dimen-
sional transport and inverse geochemical calculations. US Geol. Surv. Water Resour. Invest. Rep.
994259, 312 p.
Pearson FJ, Rightmire CT (1980) Sulphur and oxygen isotopes in aqueous sulphur compounds In
Handbook of Environmental Isotopes Geochemistry. P. Fritz and J. Ch. Fontes, Elsevier, Chapter
6, 227258.
Rey J (1995) Le Bassin dAquitaine: composition, volution et structure In Du Lias Nord aquitain
aux molasses miocnes Gnralits. Livret Guide dexcursion, STRATA, Actes du laboratoire
de Gologie Sdimentaire et Palontologie de lUniversit Paul Sabatier Toulouse, 1995, Srie
2, Mmoire, 7140.
Serrano O (2001) Le Crtac Suprieur/Palogne du bassin davant-pays Nord-Pyrnen (Bassin
dAquitaine): sdimentologie, stratigraphie et volution godynamique. Thse, Universit de
Rennes, France [in French].
Sracek O, Hirata R (2002) Geochemical and stable isotopic evolution of the Guarani Aquifer System
in the state of Sao Paulo, Brazil. Hydrogeology Journal, 10, 6, 643655.
Thorstenton DC, Fisher DW, Croft MG (1979) The geochemistry of the Fox Hills-Basal Hell Creek
aquifer in southwestern North Dakota and northwestern South Dakota. Water Resources
Research, 15, 6, 14791498.
Tth J (1999) Groundwater as a geologic agent: an overview of the causes, processes and manifest-
ations. Hydrogeology Journal, 7, 114.
Weaver TR, Bahr JM (1991) Geochemical evolution in the Cambrian-Ordovician Sandstone aquifer,
Eastern Wisconsin. 2. Correlation between flow paths and groundwater chemistry. Ground Water,
29, 510515.
CHAPTER 20
Hydrogeology of Bni Abbs: potential, hydrodynamics
and influence on the palm field (Valley of Saoura, Algerian
South-West)
Touhami Merzougui
1
, Abderrahmane Mekkaoui
2
, Hamidi Mansour
3
&
Khadidja Graine-Tazrout
4
1
GTZ Algeria. Project, Water Resources Integrated Management Oasis of Bni Abbs-. N 39,
Cit El Ouahda Bni Abbs, Bchar, Algeria
2
Institute of Architecture, Civil Engineering and Hydraulics, B.P: 417, University of Bchar,
Bchar, Algeria
3
FSTGAT, Department of Earth Sciences, University of Oran, Oran, Algeria
4
FSTGAT, Department of Earth Sciences, U.S.T.H.B, BP:32, El Alia, Algiers, Algeria
ABSTRACT: The oasis of Bni Abbs has a large groundwater resource potential. There are four
aquifers, the Great Western Erg being the largest, which are in hydraulic continuity to form a single com-
plex aquifer system. Each aquifer has its own discrete properties. The Great Western Erg offers good-
quality water and represents the principal resource in the aquifer system; it supplies both drinking water
and half as much again for irrigation of the palm field area (40ha of agricultural land). The supply is
drawn from the spring of Sidi Othmane. The alluvial terraces are formed of sand and gravel (alluvial
deposits) in the Saoura Valley and form a superficial aquifer. The aquifer at Hamada of Guir has a mod-
erate potential. Finally the Palaeozoic aquifer is probably multi-layered. The piezometry shows a general
flow direction from east to west. The Saoura Valley, formerly very active, currently receives very little
recharge due to the construction of the Djorf Torba dam on Wadi Guir. The wells that withdraw water
from the alluvial deposits inhibit the natural communication between the Great Western Erg aquifer and
the alluvial terraces, and accelerate the arrival of saline water. Salinity increasingly invades the alluvial
deposits causing serious problems for irrigation. A good knowledge of the aquifer system is necessary
to safeguard the water resources and to recommend appropriate management practices.
1 INTRODUCTION
The Saoura valley harbours several oases (Bni Abbs, Taghit, Igli), which appear luxuri-
ant with abundant water. However, the water balance is in deficit (Bennadji et al., 1998).
The social and economic development and the requirements of modern life impose many
constraints on the water resources of these oases, demands for drinking water and irriga-
tion are increasing steadily while the resources remain limited. The aim of the present
study is to define the groundwater potential around the village of Bni Abbs and to inves-
tigate the dynamics of the local aquifer system, as well as the problems of salinity which
affect it.
2 GEOGRAPHIC LOCATION
The Bni Abbs oasis is situated in south-western Algeria, 250km south of the main city
in the administrative region of Bechar, 355km north of Adrar and 880km south-west of
Algiers (Fig. 1a). This village, on the left bank of Wadi Saoura, is the largest in the Saoura
Valley. It is constrained to the north and east by the dunes of the Great Western Erg, to the
south by the region of Merhouma and to the west by the Gulf of Ouarourout and the
plateau of the Hamada of Guir. The climate of this region is typical of a desert, charac-
terised by little rainfall, estimated at 33mm/year (Mekideche et al., 1995), (Fig. 1b).
3 GEOLOGY AND WATER RESOURCES
3.1 Geology
The Bni Abbs oasis is situated on a plateau at an altitude of 500m, which belongs to the
old Saharan platform (Fig. 2). The area is adjacent to the Ougarta Mountains and consists
of a range of formations, from Precambrian to Recent in age (Menchekoff, 1933; Alimen,
1957). Locally, there are examples of upper Devonian, discordantly overlain by Neocene
strata, also known as Hamadas. Regarding the lithology, the section at Gara Diba is the
most representative (Fig. 3) and includes, from the bottom upwards:
Weathered, chocolate-coloured Devonian clay.
A hard conglomeratic sandstone lens.
Sand with cross-bedding.
About ten meters of Torba (an earth-like formation) with a flint layer in the middle.
Three silicified limestone benches separated by two calcareous sandstone ones. Together,
these benches compose the lower Hamada layer.
Apart from the Aeolian accumulations in the Great Western Erg, the Quaternary period
(Fig. 4) includes the following:
(a) Old alluvial terraces: (Pliocene Villafranchian) consisting of a detrital series overlain
by conglomerates and sands.
32 32 32 32 32 32 32 32 32
00 00 00 00 00 00 00 00 00
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
Grand erg occi.
30 30 30 30 30 30 30 30 30
00 00 00 00 00 00 00 00 00
29 29 29 29 29 29 29 29 29
00 00 00 00 00 00 00 00 00
1 00 1 00 1 00 1 00 1 00 1 00 1 00 1 00 1 00 E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
E
r
g

C
h
e
c
h
2 00 2 00 2 00 2 00 2 00 2 00 2 00 2 00 2 00
31 31 31 31 31 31 31 31 31
00 00 00 00 00 00 00 00 00
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
R
a
o
u
i
Kerzaz Kerzaz Kerzaz Kerzaz Kerzaz Kerzaz Kerzaz Kerzaz Kerzaz
Ougarta Ougarta Ougarta Ougarta Ougarta Ougarta Ougarta Ougarta Ougarta
Zghamra Zghamra Zghamra Zghamra Zghamra Zghamra Zghamra Zghamra Zghamra
Ksi ksou Ksi ksou Ksi ksou Ksi ksou Ksi ksou Ksi ksou Ksi ksou Ksi ksou Ksi ksou
3 00 3 00 3 00 3 00 3 00 3 00 3 00 3 00 3 00
Boukas Boukas Boukas Boukas Boukas Boukas Boukas Boukas Boukas
Bassin de Bassin de Bassin de Bassin de Bassin de Bassin de Bassin de Bassin de Bassin de
Bchar Bchar Bchar Bchar Bchar Bchar Bchar Bchar Bchar
Abadla Abadla Abadla Abadla Abadla Abadla Abadla Abadla Abadla
E
r
E
r
E
r
E
r E
r
E
r
E
r
E
r
E
r
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
H
a
m
a
d
a
d
u
G
u
ir
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g

Ig
u
id
i
E
r
g
E
r
g
E
r
g
E
r
g
E
r
g
E
r
g
E
r
g
E
r
g
E
r
g
Tabelbala Tabelbala Tabelbala Tabelbala Tabelbala Tabelbala Tabelbala Tabelbala Tabelbala
20 km 20 km 20 km 20 km 20 km 20 km 20 km 20 km 20 km
Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra Hamada du Dra
00000000 0
4 00 4 00 4 00 4 00 4 00 4 00 4 00 4 00 4 00
Anti Atl as Anti Atl as Anti Atl as Anti Atl as Anti Atl as Anti Atl as Anti Atl as Anti Atl as Anti Atl as
ori ental ori ental ori ental ori ental ori ental ori ental ori ental ori ental ori ental
G
u
ir
G
u
ir
G
u
ir
G
u
ir
G
u
ir
G
u
ir
G
u
ir
G
u
ir
G
u
ir
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
Z
o
u
s
f
a
n
a
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
S
a
o
u
r
a
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
M
e
s
s
a
o
u
d
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
o
.

D
a
o
u
r
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
c
h
a
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
O
.

A
b
i
o
d
M

o

n
t

s

d

'
O

u

g

a

r
t
a
M

o

n

t

s

d

'
O

u

g

a

r
t
a
M

o

n

t

s

d

'
O

u

g
a

r
t

a
M

o

n
t

s

d

'
O

u

g

a

r
t
a
M

o

n

t

s

d

'
O

u

g
a

r
t

a
M

o

n
t

s

d

'
O

u

g

a

r
t
a
M

o

n

t

s

d

'
O

u

g

a

r
t
a
M

o

n

t

s

d

'
O

u

g
a

r
t

a
M

o

n

t

s

d

'
O

u

g

a

r
t
a
20 20 20 20 20 20 20 20 20
30 30 30 30 30 30 30 30 30
40 40 40 40 40 40 40 40 40
50 50 50 50 50 50 50 50 50
100 100 100 100 100 100 100 100 100
Igl i Igl i Igl i Igl i Igl i Igl i Igl i Igl i Igl i
(a) (b)
study area
Figure 1. Situation of Bni Abbs. (a): geographical location, (b): rainfall contour map of south-
western Algeria (rainfall in mm/year).
(b) Alluvial terraces: (Pleistocene), locally called Saourian, which consist of a series of
hard layers overlain by fluvial-aeolian sands interbedded with marls.
(c) Upper alluvial terraces: (Holocene or Guirian). This is a formation of alternating sand
and gravel layers with subordinate fine-grained sand.
All these strata offer excellent reservoir conditions.
3.2 Groundwater resources
Bni Abbs is well provided with groundwater although not all the strata have characteris-
tics that are suitable for exploitation.
Hydrogeology of Bni Abbs: potential, hydrodynamics and influence on the palm field 271
Figure 2. Outline geology of Beni Abbes.
0 m
10
20
30
40 m
Unconsolidated
Sandy limestone, silicified at the top
Sandy limestone bench with small flints
Interbedded sand and limestone
Sandy limestone
Cavernous flagstone
Torba with angular detrital flints
limestone nodules at the top
Torba
Bedded aeolian sand
Devonian over argillaceous shist
Figure 3. Cross-section of Gara Diba (Roche,
1973).
V
i
l
l
a
f
r
a
n
c
h
i
e
n

-

Q
u
a
t
e
r
n
a
i
r
e
G
u
i
r
i
e
n
S
a
o
u
r
i
e
n
0 m
1
23 m
Sand - grey yellow or brown
Sandy marl
Grey sands (grey, yellow or brown)
Sandy marls - sand grey yellow or
brown
Sandy marls - sand grey yellow or
brown
Sandy marl
Sandy marls - sand grey, yellow or
brown
Sandy marl
Sandy marls - and grey, yellow or
brown
Marls
Marls and sands
Figure 4. Schematic cross-section of the
Quaternary deposits (Roche, 1973).
3.2.1 The Great Western Erg aquifer
This is the largest aquifer in the Saoura Valley; it is vast, limited to the North by the southern
flexure of the Atlas. In the north-east, it extends towards the Daias (northern part of Ghardia)
and in the south-west, it abuts the Tadmait continental aquifer (Fig. 5). The piezometric map
shows flow towards Saoura and the south (Schoeller, 1945; Cornet, 1964) (Fig. 6). The
Figure 5. Groundwater resources of the Algerian South.
Figure 6. Piezometric map of the Great Erg (Roche, 1973).
aquifer is recharged mainly by the northern wadis with a flow rate of 8.48m
3
/s (Ould Baba Sy,
2005, Ould Baba Sy and Besbes, 2007) supplemented by infiltration of meteoric water
into the sand and by the floods at djebel el Kehla (Schoeller, 1962; Roche, 1973). The total
flow rate is between 5 and 6.5m
3
/s (Conrad and Fontes, 1970). The great spring of Sidi
Othmane is an important source of drinking-water for the city and for the irrigation of the
palm field. It constitutes a main outlet of the aquifer in this area, with a flow rate of 33l/s.
3.2.2 Alluvial terrace aquifers and underflow
The alluvial terrace is formed of tiers sand and gravel (alluvial deposits) along the Valley
of Saoura, and is known as Saourian (upper Pleistocene) and Guirian (Holocene). The
piezometric map shows a general east-west flow direction, which means that flow to the
alluvial deposits is taking place from the Great Erg aquifer (transmissivity: 6 10
4
m
2
/s
to 4. 10
2
m
2
/s and a storage coefficient of 1.3 to 6%). The flow beneath the Valley of
Saoura reflects the groundwater base level. It is fed by leakeage from the ephemeral
Saoura Wadi. The depth to the water table varies from 0 (Guelta) to 3m below the ground
surface. The hydraulic gradient is roughly the same as the slope of the longitudinal profile
of the Valley of Saoura, i.e. 0.65 from Marhouma to Sebkha el Melah (Roche, 1973).
3.2.3 The Palaeozoic aquifers
These aquifer are poorly understood, but the various catchments in the villages of Zghamra
and Ougarta (springs, wells, drillings) suggest it is a multi-layer aquifer, probably a set of
superposed fossil aquifers, recharged during wet Quaternary periods.
3.2.4 Aquifer in the Hamada of Guir
The Hamada is located on the right bank of the Saoura Valley. It is a slightly inclined plateau,
extending from Boudnibe (1150m) to Ougarta (650m). It is 110km wide and 200km
long. The plateau has a north-west to south-east orientation and it contains a limestone
aquifer. Recharge occurred during the wet periods in the Quaternary, and it also receives
water from a system of wadis (Oued Aicha, Oued El Abiod, Oued Alarfedj). The general
groundwater flow direction is along the axis of the Hamada (Combes, 1966), (Fig. 7).
sss
s
s
s s s s
5
0
0
5
0
0
5
0
0
5
0
0
5
0
0
5
0
0
5
0
0
5
0
0
5
0
0
5
5
0
5
5
0
5
5
0
5
5
0
5
5
0
5
5
0
5
5
0
5
5
0
5
5
0
6
0
0
6
0
0
6
0
0
6
0
0
6
0
0
6
0
0
6
0
0
6
0
0
6
0
0
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
G
u
i
r
u
ed
O
u
ed
O
u
ed
u
ed
O
u
ed
u
ed
O
u
ed
O
u
ed
O
u
ed
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
Z
e
g
a
g
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
A
b
io
d
*
*
*
*
O
u
ed
O
u
ed
O
u
ed
O
u
ed
O
u
ed
O
u
ed
O
u
ed
O
u
ed
O
u
ed
6
5
0
6
5
0
6
5
0
6
5
0
6
5
0
6
5
0
6
5
0
6
5
0
6
5
0
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
re
fd
ji
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
A
c
h
a
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
M
a
r
o
c
7
0
0
7
0
0
7
0
0
7
0
0
7
0
0
7
0
0
7
0
0
7
0
0
7
0
0
9
0
0
9
0
0
9
0
0
9
0
0
9
0
0
9
0
0
9
0
0
9
0
0
9
0
0
E
l
E
l
E
l
E
l
E
l
E
l
E
l
E
l
E
l
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d
O
u
e
d

Sand dunes
Quaternary and Tertiary
Tertiary
Cretaceous
Carboniferous
Siluro- Devonian
Cambro- Ordovician
Volcanic rocks
Piezometric line
Flow direction
Legend
Figure 7. Map of the Hamada of Guir aquifer.
Groundwater abstraction from traditional wells reflects weak flow rates, and a low resource
potential.
4 AQUIFER SYSTEM AND HYDROCHEMICAL EVOLUTION
4.1 Functioning of the aquifer system
The aquifers collectively form a complex aquifer system: The Saoura Valley separates this
system into two compartments: a passive western compartment that includes the Hamada
of Guir aquifer (which, in some zones, is in contact with the Upper Palaeozoic), and an east-
ern compartment which drains groundwater from the Great Western Erg aquifer to the allu-
vial terraces. The latter are in contact with the Palaeozoic aquifers at some points (Fig. 8).
The geological cross-section (Fig. 9) through the Saoura Valley shows the aquifers in the
eastern compartment: upper Devonian layers, characterised by broad folds, overlain in
angular unconformity, by the Neocene formations. The lithology of the Neocene formations
allows groundwater to flow down from the higher areas allowing natural drainage from the
Great Western Erg aquifer to the alluvial deposits. A system of foggaras (drainage galleries)
has been installed. The Hamada of Guir aquifer contributes very little to the recharge of the
groundwater flow beneath the Saoura and the Palaeozoic aquifers even less.
4.2 Water resources and rate of exploitation (Fig.10)
The reservoir of the Great Western Erg aquifer is very large and is exploitable around the
oases in the Saoura Valley at the base of the Great Western Erg. This aquifer has a flow rate
estimated at 500 million m
3
/y in the Bni Abbs region. Its productivity is estimated at 32
million m
3
/y, including 25 million m
3
from the great spring and the remainder from two
boreholes in the Mio-Pliocene (7 million m
3
/y). The hydrodynamic characteristics of the
aquifer in the alluvial deposits make it possible to calculate the flow rate for use in the
Figure 8. Schematic diagram of the water-bearing system of Bni Abbs.
palm field, i.e. approximately 140 million m
3
/y. The same characteristics were used to esti-
mate the water resources at 0.6m
3
/s by studying the flow on a 5-km front (Bouhouche and
Ghazi, 1993).
The alluvial terrace aquifers and the flow beneath the valley (underflow) represent a
capacity of approximately 80,000m
3
, recharged mainly by the natural drainage and infil-
tration of water from the Wadi Saoura. It is estimated that the groundwater in the alluvial
Figure 9. Hydrogeological cross-section in the area of Beni Abbes through the Saoura valley.
Wadi loss and Infiltration
Groundwater of
the Great Western
Surface water
(Saoura Oued)
6 millions de m
3
/ y
FLOW =
500 millions de
m
3
/y
140
millions m
3
/y
Evaporation
Flow exploited
Use
32 millions m
3
/yr
Consumption
(Drinking water)
28 millions m
3
/y
Irrigation
(Palm plantation)
10 millions m
3
/y
4,2
millions
m
3
/y

20 000 m
3

Brines

60 000 m
3

Brackish water

20 000 m
3

Fresh water
Tank of the groundwater of the alluvial soils
Underflow
Contaminant flow
Naturel drain
Figure 10. Water balance of the aquifer system in the Bni Abbs oasis.
terraces that is contaminated by salinisation represents 70% of the total volume and is
increasing.
4.3 Hydrochemical evolution and threat of salinity increase:
The water in the Great Western Erg aquifer does not show any chemical evolution. It is
characterized by a neutral pH and a very low mineral content. Similarly, that of the
Hamada of Guir aquifer does not show any changes (Table 1).
However, groundwater in the alluvial terrace aquifer and the longitudinal flow beneath
the valley (underflow) has an alarming trend (Fig. 11). In 1966 the valley flow had a TDS
of 5.1g/l and the alluvial deposits did not exceed 3g/l. In 2005, the water in the alluvial-
terrace aquifers and the valley flow had a TDS of between 4 and 20g/l. A compilation of
available hydrochemical data enables contour maps of the TDS concentration to be pre-
pared for any given time. In the 1960s, the palm field was situated in fresh water (Fig.
12a). Today, a west-east evolution is visible, accompanied by hydrochemical zoning with
a from over 15g/l at the centre to less than 6g/l (Fig. 12b).
Table 1. Chemical characteristics of the groundwater of the Great Western Erg and the Hamada of
Guir (Roche, 1973; Yousfi and Have-Ahmed, 1992; Merzougui, 1998).
pH TDS Ca
Mg
Na
Cl
SO
4
HCO
3
(mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l) (mg/l)
Water of Great Western 7.6 530 75 30 50 15 90 600 285
Erg aquifer
Water of Hamada of Guir aquifer 7.1 1200 75 48 287 191 441
0
5000
10000
15000
T
D
S

m
g
/
l
20000
25000
Groundwater
of the Great
Western Erg
TDS mg/l 1966
TDS mg/l 2005
Groundwater
of Alluvia 1
Groundwater
of Alluvia 2
Groundwater
of Alluvia 3
Groundwater
beneath the
valley
Figure 11. Evolution of the Total Dissolved Solids (TDS) in the Bni Abbs area.
Comment:
The spatial distribution of the TDS concentration shows a progressive increase in the
direction of groundwater flow from the Erg towards the wadi. It reaches the highest values
at the edge of the wadi (Fig. 12b). This salinity is caused by:
Losses from Wadi Saoura, which formerly received the floods of the Guir (200Mm
3
/y)
and the Zouzfana (6Mm
3
/y), (Mekideche et al., 1995). The construction of the Djorf
Torba dam on the Guir is responsible for this phenomenon.
Dissolution of the salts in the Quaternary sand formations.
Intense evaporation (2153mm/y).
Overexploitation promotes a rapid rise of saline water from the valley flow.
Reduction in input from the Erg caused by pumping.
Wells around the palm field used to compensate for the reduction of the proportion of
water from the Great Spring allotted to irrigation.
Diffusion.
5 DISCUSSION AND CONCLUSION
The aquifer system in the Bni Abbs region includes four inter-connected aquifers. The
Great Western Erg aquifer is the largest, and is connected to the alluvial deposits and the val-
ley underflow. The decrease in flow along Wadi Saoura due to the construction of the Djrof
Torba dam coupled with the high evaporation rate in the Bni Abbs region are both major
factors in the salinisation of the groundwater. This, in turn, leads to an imbalance in the
aquifer system, and to serious irrigation problems for the palm field. The overexploitation of
Figure 12. Map of the TDS in groundwater in the alluvial deposits at Beni Abbes (around the palm
field).
wells around the palm field (alluvial terraces) brings up saline water from below. In addition,
the salinity is transported through diffusion from high concentration towards lower concen-
tration areas. Withdrawal of groundwater from the alluvial terrace aquifer inhibits flow and
accelerates the diffusion of the salinity. Measures can be taken in order to limit the increase
in salinity:
A protection zone between the Great Western Erg and the palm field would allow water
from the Great Western Erg to flow down the length of the valley.
Allowing a release of water from the Djorf Torba dam at the same time as there are
floods in the Zousfana.
Recycling of waste water and using it to irrigate the palm field.
Construction of a dam perpendicular to the flow direction of the Erg towards the wadi.
REFERENCES
Alimen, H. (1957) Tertiary and Villafranchien in the North-Western Sahara. C.R..Geological Survey,
France , pp 23840.
Beaucaire, C., Gassama, N., Tresonne, N., Louvat, D. (1999) Saline groundwaters in the Hercynian
granites (Chardon mine, France) : geochemical evidence for the salinity origin.. Report,
(IPSN/DPRE/SERE/LRC, CEA) 67/84
Bennadji, A., Bennadji, H., Cheverry, C., Bounaga, N. (1998) Bni-Abbs or the deterioration of a
palm field. Scheresse, V. 9, n
o
2, pp 131137
Bouhouche, Z., Ghazi, F. (1993) Contribution to the hydrogeologic study of the area of Bni-Abbs.
Thesis Ing. State, Univ. Sci. Tech. H.B. Algiers, 146 p.
Chalbaoui, M. (2000) Vulnerability of groundwater and underflow in southwestern Tunisia. Science
et changements plantaires. Scheresse. V. 11, n
o
2, 8591.
Cornet, A. (1962) Test on the hydrogeology of the Great Western Erg and its bordering areas. Willaya
of the Foggaras . Work of the Saharan Research Institute,. VII: 71122.
Cornet, A. (1964) Introduction to Saharan hydrogeology; Rev. de Gogr. Phys. Et de Gol. Dyn. (2),
V. VI, n
o
1; 572.
Combes, M. (1966) Hydrogeologic study of the Hamada of Guir. Commune of Bechar . Willaya of
the Saoura. Study S.E.S.: 68/DH.1M.1I/C. of the Circonscription des T.P.H. des Oasis et Saoura,
8 surveys maps, 4 pl., 14 p.
Conrad, G., Fontes, J.C. (1970): Isotopic hydrology of the Sahara North-Westerner. Isotope
Hydrology 1970, A.I.E.A., Proc. Symp. Vienna, pp 405419.
Haddou, K., Bensalem, S. (2004) Hydrogeologic contribution to the study of the quality of the
groundwater of Bni Abbs Report Ing. State University of Oran.
Halwani, J., Ouddane, B., Crampon, N. (2001) Contamination saline dans les eaux souterraines de la
plaine dAkkar au Liban (Saline contamination of groundwater of the Akkar plain in Lebanon).
J.Europen dHydrologie, ISSN 1023-6368, V. 32, n
o
1, pp. 93108 (1 p.1/4).
Houpert, A. (1993) Salinit des eaux souterraines de lAltiplano : rapport de stage de deuxime
anne, du 25 juin au 8 septembre 1993. La Paz: ORSTOM, 31 p.
Mansour, H. (1993) Isotopes in the water of the Algerian North-Western Sahara (Great Western Erg
- Hammada of Guir). Second Engineering Days of Geology, Sfax, pp. 88111.
Mansour, H., Safa, A., Bekhedda, B. (1998) The Southern Atlas flexure and its hydrodynamic role in
the recharge of Saharan aquifers with Bni Ounif. Study days on arid regions, retroperspectives,
stakes and realities. Adrar
Margat, J. (1985) Hydrology and water resources of the arid regions. Bull. Soc. Geol. France, (8),
T. 1, n
o
7, pp.10091020.
Margat, J. (1992) Fossil water. Contemporary Africa, N
o
161 (Special) 1st Quarter.
Mekideche, D., Sai, N., Touat, S., Younsi, N. (1995) Hydrogeologic chart of the area of Bchar.
Explanatory leaflet, 73p.
Merzougui, T. (1998) Valorization of the water resources of the high Valley of Saoura (between
Taghit and Kerzaz) Thesis Ing state, Univ. Sci. Tech. of Oran, Algeria, 175 p.
Mohamed, M., Hind, M. (1998) Study of the annual rains and day in the Algerian Sahara.
Scheresses, V.9, n
o
3, page 193199.
Ould Baba Sy, M. (2005) Refill and paleorecharge of the aquifer system of the septentrional Sahara,
Doctorate thesis, university of Tunis El Manar, 261 p.
Ould Baba Sy, M., Besbes, M. (2005) Holocene and present recharge of the Saharan aquifers : a
numerical modelling study. This volume.
Roche, M.A. (1973) Hydrogeology of High Saoura (the Sahara north-Westerner). Paris: Publ CNRS,
Gol series, 91p.
Scholler, H. (1945) The hydrogeology of part of the valley of Saoura and the large one of Western
Erg, Somm, camp, geol.fr.5. T.XV, page 563585.
SOGREAH (1970) Hydrogeologic study of the water tables of the Western Erg and Piedmont of the
Atlas. Grenoble. ANRH, Algiers, Algeria. UNESCO, Paris, 1972
Stephensen, G.R., Zuzel, J.F. (1981) Groundwater recharge characteristics in a semi-arid environ-
ment of southwest Idaho, J.. Hydrol., (53), pp 213227.
Subiyani, A.M. (2005) Hydrochemical identification and salinity problem of groundwater in Wadi
Yalamlam basin, Western Saoudi Arabia, J.of Environment 60, p5366.
UNESCO (1972) Study of the water resources of the Septentrional Sahara, Report/ratio on the
results of Project REG-100, UNESCO, Paris (1972)
Wright, E.P., Benfield, A.C., Edmunds, W.M., Kitchen, R. (1982) Hydrogeology of the Kufra and
Sirte basins, eastern Libya. Q. Journal. Eng. Geol., 15, pp83103.
Yousfi, N., Hait-Ahmed, C. (1992) Contribution to the hydrogeological study of the Great Source
and the palm field of Bni-Abbs (the Western Sahara). Thesis Ing. State, Univ. Sci. Tech. of
Oran-Algeria, p 440.
Zouari, K., Mamou, A. (1992) Aquifer systems of the Tunisian South: Hydrochemical and isotopic
characteristics of refill the conditions and optimization of management. Proc. of the International
Congress on Hydrogeology. Marrakech, Morocco, 4144.
CHAPTER 21
Groundwater system compartmentalized by a tectonic zone
in the Kanto plain, central Japan
Masaya Yasuhara
1
, Akihiko Inamura
1
, Masaaki Takahashi
1
,
Takeshi Hayashi
1
, Hiroshi Takahashi
1
, Masahiko Makino
1
,
Hiroko Handa
1
& Toshio Nakamura
2
1
Geological Survey of Japan, Tsukuba, Japan
2
Center for Chronological Research, Nagoya University, Chikusa, Japan
ABSTRACT: In the Kanto plain, the largest Quaternary groundwater basin in Japan, water samples
were obtained from 130 water supply boreholes to characterize the groundwater system. The spatial vari-
ation in chloride (Cl
) concentration indicates that groundwater with high Cl
concentrations of more
than 100mg/l (up to 216mg/l) occurs at the centre of the plain and at depths between 200m and 450m
below ground surface. The area with the Cl
-rich groundwater, spreading from the northwest to south-

east, corresponds with the so-called Motoarakawa tectonic zone (8km wide by 35km long) bounded by
the faults on its longer sides. We have found the Cl
-rich groundwater is also characterized by low D,

low
18
O, small d-parameter, and low
14
C values. Both the chemical and isotopic evidence strongly sug-
gests that the Motoarakawa tectonic zone divides the groundwater system in the Kanto plain into three
distinct hydrologic sub-areas. Two faults, which delineate the Motoarakawa tectonic zone, act as bar-
riers to the southward and eastward regional flow of groundwater in the Kanto plain. As a result, the
Motoarakawa tectonic zone has been under an isolated hydrologic environment for a long period of
time, resulting in the occurrence of groundwater with anomalous hydrochemistry. With regard to the ori-
gin of isotopically-depleted groundwater in the tectonic zone, a potential source is assumed to be precip-
itation in a cooler climate than the present. Admixture of interstitial water (residual sea water) squeezed
out of the adjacent aquitards and/or upward movement of the deep-seated Na-Cl groundwater along the
faults is likely to account for the elevated Cl
concentrations of groundwater in the tectonic zone.

1 INTRODUCTION
Much attention has been paid to the impact of faults on the regional groundwater flow systems
in sedimentary basins (e.g. Bense and Person, 2006). Faults often form preferential flow paths
for vertical fluid flow (e.g. Mozley and Goodwin, 1995), while they can also act as barriers to
horizontal fluid flow compartmentalizing regional groundwater flow systems (e.g. Knipe,
1997; Bense and Van Balen, 2003). The present study is focused on the impact of faults on the
groundwater flow regime in the Kanto plain, central Japan (Fig. 1). The Kanto plain (about
8,000km
2
in area), comprises the city of Tokyo in its southeastern part, and is the largest
Quaternary sedimentary basin in Japan. From the few previous studies based on the hydraulic
head pattern, it has been suggested that some faults trending approximately northwest to
southeast have an important effect on regional groundwater flow (e.g. Miyake and Saito,
2003). In addition, groundwater with chloride concentrations as high as 100150mg/l occurs
in the central part of the Kanto plain (Ikeda, 1984). A better understanding of the relationship
between the faults and anomalous groundwater hydrochemistry is essential for future man-
agement of the groundwater resource from both quality and quantity points of view. In this
study, the combination of major elements, and H, O and C isotopes reveals the spatial distribu-
tion and origins of the Cl
-rich groundwater in more detail, and helps characterize the geolog-

ically-controlled dynamic scheme of the groundwater flow system in the Kanto plain.
2 STUDY AREA
The study area (Fig. 1) is centred on the city of Saitama (population 1,050,000) and includes
most of the cities of the southeastern parts of Saitama Prefecture. It lies on the central part of
the Kanto plain and is undergoing rapid urban growth. The study area has a temperate climate
Figure 1. The study area and location of sampled boreholes. The city of Saitama consists of four
districts named Omiya, Yono, Urawa, and Iwatsuki (in parentheses).
with an average precipitation of 1,338mm/year at the city of Saitama (the period from 1979
to 2000; AMeDAS data of the Japan Meteorological Agency). The geomorphology, with an
elevation range 1530m a.m.s.l., is characterized by thick alluvial deposits. The uplands are
divided into several terraces of different ages and altitudes, generally the height difference
between upland and lowland being 510m. Regional land surface dips are generally very
shallow, being on the order of 0.1 to the south(east) or east. The bedrock is covered by up to
3,500m of thick Neogene and Quaternary sediments. Upper some 5001000m of these sed-
iments are referred to as the Kazusa and Shimousa groups of late Pliocene to Quaternary age.
Sand and gravel layers in the Kazusa and Shimousa groups, which are interbedded with a lot
of thick, extensive clay layers, form excellent confined aquifers. The aquifers generally slope
to the south(east) or south with almost the same gentle dips as the land surface. With the
exception of the city of Kurihashi in the northern parts of the study area and the adjacent areas,
land subsidence problems due to excessive pumping of deep groundwater from these con-
fined aquifers (100400m deep) has already been put under control.
From northwest to southeast, there is the 35-km long, by 8-km wide Motoarakawa tectonic
zone, which is bounded by the Ayasegawa and Kuki faults on its long sides (Shimizu and
Horiguchi, 1981). Recent geologic studies on the Ayasegawa fault suggests a 911m vertical
displacement to the depth of some 30m below the present ground surface (Katsumata et al.,
2005). Further down, a vertical displacement is assumed to be 150200m with an average slip
rate of 0.1m per 1,000 yrs. (Kiyohide Mizuno, personal communication), but details are not
known about the Kuki fault 8km northeast across the Motoarakawa tectonic zone. Hydraulic
head is obviously low in the tectonic zone compared with that in the periphery. Extreme
hydraulic head difference of 1025m is known to exist across the Ayasegawa fault, while
hydraulic head drop across the Kuki fault is smaller with a difference of less than 5m (Miyake
and Saito, 2003).
3 WATER SAMPLING AND ANALYSES
The diversity of the hydrogeological setting of the area results in boreholes of widely different
depths ranging from 20 to 450m and screen settings. Among these, 130 water supply bore-
holes typically with 200400m depths and fully screened were sampled (Fig. 1). These bore-
holes are screened extensively in the upper aquifers of the Kazusa group. Groundwater
sampling took place in 2003 and 2004 and water samples were analysed for water chemistry
and isotopes. At the Geological Survey of Japan, water chemistry was determined by ion chro-
matography and stable isotopes were analysed by mass spectrometry following standard
preparation methods (CO
2
equilibration for
18
O, reduction with zinc shot for D, and acidifi-
cation with phosphoric acid for
13
C-DIC). The precision of
18
O and D analyses is 0.1
and 1, respectively, and that of
13
C analysis is 0.1. For the
14
C activity analysis, dissolved
inorganic carbon (DIC) was precipitated to SrCO
3
by adding carbon-free SrCl
2
solutions. CO
2
reacted from SrCO
3
precipitation was reduced to graphite using H
2
. The
14
C activity was then
measured by AMS system at the Nagoya University with a precision of 0.51.5pmc.
4 CHLORIDE CONCENTRATIONS AND ISOTOPE COMPOSITION
The spatial variation in chloride (Cl
) concentration (Fig. 2) clearly indicates that ground-

water with elevated Cl
concentrations of more than 100mg/l occurs in Shobu, Ina, Shiraoka,

Groundwater system compartmentalized by a tectonic zone in the Kanto plain 283
Hasuda, and Kasukabe and the adjacent areas. Groundwater with the very high Cl
contents
is found in Shiraoka (216mg/l and 184mg/l in two different 310m deep boreholes) and
Kasukabe (163mg/l, 160mg/l, and 153mg/l in 350m, 270m, and 350m deep boreholes,
respectively). Although the distribution of Cl
-rich groundwater is in good agreement with

that of Ikeda (1984), the Cl
concentration observed in this study is generally increased by

1020% during the last 20 years.
Interestingly, this Cl
-rich groundwater zone, spreading from northwest to southeast,

roughly corresponds with the Motoarakawa tectonic zone outlined above (Fig. 2). The Cl
contents of groundwater outside the Motoarakawa tectonic zone (such as in Washimiya,

Satte, Sugito, Kurihashi, Otone, and Kitakawabe to the north, and Ageo, Omiya, Yono,
Urawa, Kawaguchi, Warabi, Toda, Hatogaya, and Soka to the south) are in the range
between 5 and 50mg/l, being one or two orders of magnitude lower than those inside the
tectonic zone. The abrupt change in Cl
concentration across the Ayasegawa and Kuki faults

may suggest that the confined groundwater system in the Motoarakawa tectonic zone is
not hydraulically interconnected with the aquifers outside the tectonic zone. Figure 3
shows the distribution of bicarbonate (HCO
3
) concentration. Unlike the Cl
concentra-
tion, no obvious difference in HCO
3
concentration is observed between the groundwater

in the tectonic zone and that in the periphery. Overall, groundwater in the tectonic zone
proved to be of Na-Ca-Cl-HCO
3
type, whereas, with some exceptions, that in the periph-
ery of Ca-Na-HCO
3
-Cl type.
Figure 4 shows the
18
O-D relationship for the groundwater samples in the study area.
The groundwaters plot in the range 73 to 53 for D and 10.5 to 8.2 for
18
O.
All points fall between the local meteoric water line with a d-parameter (D8*
18
O)
of 15 and the world meteoric water line defined by Craig (1961) (d-parameter 10).
Groundwaters collected inside the tectonic zone (filled blue circles) show more depleted
isotopic composition than the others collected outside the tectonic zone (filled red circles).
It is also obvious from Fig. 4 that there is a tendency for groundwater to show a smaller
d-parameter toward the depleted end of the range.
Figure 2. Spatial variation of Cl
concentra-
tion (mg/l) in the study area.
Figure 3. Spatial variation of HCO
3
concen-
tration (mg/l) in the study area.
The D values indicate significant variations over the study area (Fig. 5). In the
Motoarakawa tectonic zone, groundwater is generally characterized by the low D values
of 70 to 65. The most isotopically-depleted groundwater with the D values of less
than 70 were found in Shiraoka, Hasuda, and Ina in the middle and southeastern parts
of the tectonic zone. Outside the tectonic zone, by way of contrast, the groundwater to the
south has the more enriched D values by 510 as in Ageo and Omiya, and the distri-
bution shows an increase in D from 65 in the immediate area of the Ayasegawa fault
to 55 in Yono, Urawa, Kawaguchi, Warabi, and Toda some 5km away from the fault.
This is also the case for the Kuki fault. Groundwater in Washimiya, Kurihashi, Otone and
Kitakawabe collected outside the tectonic zone shows the enriched D values of more
than 60. The isotopically-depleted groundwater in the Motoarakawa tectonic zone (Fig.
4) has a small d-parameter in comparison with that outside the tectonic zone. These dis-
tinct differences in D and d-parameter on either side of the Ayasegawa fault and the Kuki
fault strongly suggest that groundwater in the Motoarakawa tectonic zone does not com-
municate with that in aquifers outside the tectonic zone.
The carbon isotope data (
14
C in Fig. 6 and
13
C in Fig. 7) supports the assertions made
above and allow further constraints to be made on the groundwater system in the study area.
The
14
C values in Fig. 6 range from 2.4 to 5.3pmc in the tectonic zone and from 11.8 to
21.4pmc in the periphery, with the intermediate values of 6.4 and 8.7pmc on the faults. The
high
14
C value of 18.9pmc is found for the groundwater northwest of the tectonic zone,
whereas southeast of the tectonic zone the groundwater shows the low
14
C values of 3.3 and
4.0pmc. At this point, it must be checked whether a
13
C correction should be applied to the
14
C values representative of groundwater residence times. In the
13
C-
14
C plot (Fig. 8), which
is based on Figs. 6 and 7, there is no clear relationship between
13
C and
14
C. This indicates
that
14
C dilution by interaction with calcareous materials (CaCO
3
dissolution) does not seem
to be a quantitatively significant process along the down-gradient flow in the aquifers,
although it probably contributes to some extent. Therefore, in the study area, it is assumed that
there is no need to apply the
13
C correction. As a result, the observed low
14
C values in the
Figure 4.
18
O- D diagram. Figure 5. Distribution of D values () in
the study area.
Motoarakawa tectonic zone (Fig. 6) indicate longer groundwater residence in agreement
with the assumption of groundwater flow discontinuity between the tectonic zone and the
periphery.
5 MOTOARAKAWA TECTONIC ZONE AND GROUNDWATER SYSTEM
On a regional scale, groundwater in the northern and western Kanto plain recharged by pre-
cipitation on the mountains flows to the south and to the east respectively (e.g. Nirei and
Furuno, 1988), and flow lines converge at the central parts of the plain, near where the
Figure 6. Spatial variations of
14
C values
(pmc) in the study area.
Figure 7. Distribution of
13
C values () in
the study area.
Figure 8.
13
C as a function of
14
C in the study area.
Motoarakawa tectonic zone exists. As discussed above, groundwater in the Motoarakawa
tectonic zone has anomalous hydrochemistry characterized by the high Cl
concentration,
low D, low
18
O, and small d-parameter. In addition, a marked difference in hydraulic head
on either side of the Ayasegawa fault and the Kuki fault is reported by Miyake and Saito
(2003). These observations suggest that the two faults delineating the Motoarakawa tectonic
zone impede groundwater flow across the faults. The Ayasegawa and Kuki faults act as bar-
riers to southward and eastward regional movements of groundwater in the Kanto plain,
dividing the groundwater system in the plain into three distinct sub-areas. Consequently, the
groundwater in the Motoarakawa tectonic zone is separated from the regional groundwater
flow system, and is virtually stagnant. The longer residence times for groundwater in
the Motoarakawa tectonic zone (Fig. 6) agree with the hypothesis that the groundwater in the
tectonic zone was separated from the actively-circulating regional groundwater system of the
Kanto plain and is put under an isolated hydrologic environment for a long period of time.
Southeast of the tectonic zone, groundwater also shows low
14
C values (3.3 and 4.0pmc
in Fig. 6). This may indicate a possible extension of the Motoarakawa tectonic zone further
southeast toward the city of Tokyo. Although the Ayasegawa and Kuki faults generally
form strong barriers to the regional groundwater flow, they can reduce the sealing effect as
shown in Fig. 5 in the upper reach of the Ayasegawa fault and the lower reach of the Kuki
fault. A better understanding of the hydrogeological implications of this kind of structure
is essential to the management of the groundwater resource in the Kanto plain.
The origin of the isotopically-depleted groundwater in the tectonic zone is likely to be pre-
cipitation under a cooler climate than the present. According to Dansgaard (1964), a tempera-
ture dependence of the D of precipitation, or temperature effect, is 5.6/C. Therefore,
the observed depletion of some 15D for the groundwater in the tectonic zone (Figs. 4 and
5) can be accounted for by recharge due to precipitation in a period when the mean annual
temperature was 3C or so colder than at the present. The isolated, stagnant groundwater sys-
tem in the tectonic zone contains this fossil water. Admixture of interstitial water from the
adjacent aquitards and/or upward movement of the Cl
-rich deep-seated groundwater is

likely to account for the elevated Cl
concentration of the groundwater in the tectonic zone.

As the Shimousa and Kazusa groups were mainly marine (Kimura et al., 1991), it is thought
that interstitial water of clay and silt layers in these groups was once abundant in Cl
which
had been incorporated at the time of deposition. A part of Cl
still remains unflushed and, in

response to a recent pressure drop accompanying the overpumping of groundwater, the Cl
-
rich interstitial water (residual sea water) had the opportunity to squeeze out of the aquitards
to contribute to groundwater in the aquifers (e.g. Nagai, 1985). There is also a possibility that
the origin of Cl
is below these aquifers and has migrated upward along the Ayasegawa and
Kuki faults to invade the shallow aquifer system. Stoessell and Prochaska (2005) demon-
strated vertical movement of deep formation fluids into shallow groundwater aquifers along
the fault in South Louisiana in response to freshwater removal exceeding surface recharge.
In the Kanto plain, occurrence of the Na-Cl groundwater at depths (Cl
5,00015,000mg/l
mainly at the depth between 1,000 and 1,500m) is discussed by Marui and Seki (2003). The
existence of Cl
-rich fluids at depths, as well as the active faults and ongoing groundwater
removal, indicates the study area is under the hydrologic condition similar to that in South
Louisiana. It may, therefore, be reasonable to think the same scenario takes place in the
Motoarakawa tectonic zone. If this is the case, the Ayasegawa and Kuki faults may have a
dual behaviour toward the groundwater flow: as a barrier against horizontal groundwater
flow and as a conduit for vertical migration of groundwater.
6 CONCLUSIONS
The study has focused on the impact of faults on the groundwater system in the Kanto
plain, central Japan. The groundwater in the Motoarakawa tectonic zone proved to be char-
acterized by the high Cl
concentration, low D, low

18
O, small d-parameter, and low
14
C
values. Water chemistry and isotope data provided evidence that the two faults delineating
the Motoarakawa tectonic zone impede groundwater interconnection across the faults,
groundwater in the tectonic zone being isolated from the regional groundwater flow sys-
tem in the Kanto plain and contained in an isolated hydrologic environment for a long
period of time.
ACKNOWLEDGEMENTS
The authors thank the three anonymous reviewers for their helpful comments on the man-
uscript. The assistance of Dr. Kohei Kazahaya and Dr. Noritoshi Morikawa in improving
the clarity and accuracy of this paper is also highly acknowledged.
REFERENCES
Bense V, Van Balen R (2003) Hydrogeological aspects of fault zones on various scales in the Roer
Valley Rift System. J Geochem Explor., 7879, 317320.
Bense VF, Person MA (2006) Faults as conduit-barrier systems to fluid flow in siliciclastic sediment-
ary aquifers. Water Resour Res., 42, W05421, doi:10.1029/2005WR004480.
Craig H (1961) Isotopic variations in meteoric waters. Science, 133, 17021703.
Dansgaard W (1964) Stable isotopes in precipitation. Tellus, 16, 436469.
Ikeda K (1984) Chemical characteristics of groundwater in the central part of the Kanto Plain.
Hydrology, 14, 3649. (in Japanese with English abstract)
Katsumata N, Sugai T, Hachinohe S, Mizuno K (2005) Geomorphological development of the
Omiya upland. Proc. Gen Meeting Assoc Jap Geog., 67, 259. (in Japanese)
Kimura T, Hayashi I, Yoshida S (1991) Geology of Japan. Univ Tokyo Press, Tokyo, 287 pp.
Knipe RJ (1997) Juxtaposition and seal diagrams to help analyze fault seals in hydrocarbon reser-
voirs. Am Assoc Petrol Geol Bull., 81, 185197.
Marui A, Seki H (2003) Deep groundwater in the Kanto plain. J Jap Assoc Hydrol Sci., 33, 149160.
Miyake N, Saito M (2003) Assessing confined groundwater flow in mainly alluvial area of Tokyo
and Saitama. J Jap Assoc Hydrol Sci., 33, 185196. (in Japanese with English abstract)
Mozley PS, Goodwin LB (1995) Patterns of cementation along a Cenozoic normal fault: A record of
paleoflow orientation. Geology, 23, 539542.
Nagai S (1986) Hydrochemical investigation of ground water in subsidence area -Toride and
Ryugasaki cities, Ibaraki Prefecture-. Indust. Water, 321, 2334. (in Japanese)
Nirei H, Furuno K (1988) Monitoring of groundwater basin. Urban Kubota, 27, 2026. (in Japanese)
Shimizu Y, Horiguchi M (1981) Motoarakawa tectonic belt in northeastern part of Omiya upland,
Kanto plain, central Japan. Mem Geol Soc Japan, 20, 95102. (in Japanese with English abstract)
Stoessel RK, Prochaska L (2005) Chemical evidence for migration of deep formation fluids into
shallow aquifers in South Louisiana. Trans Gulf Coast Assoc Geol Soc., 55, 794808.
C. Karstic and carbonate aquifer systems
CHAPTER 22
Hydrogeological consequences of the Messinian Salinity
Crisis in the Rhone River basin, France
Rmi de La Vaissire, Frdric Lalbat & Bernard Blavoux
Laboratoire dHydrogologie Universit dAvignon, Avignon, France
ABSTRACT: At the end of the Miocene, the Rhone River valley was subjected to an exceptional
event, the Messinian Salinity Crisis. The oscillation of the Mediterranean Sea base level created
deep canyons which were filled during the Pliocene. The most spectacular hydrogeological conse-
quences of this was the extensive karstification of the Cretaceous limestone in the Rhone River val-
ley. These affected all the aquifers which are in contact with Messinian karst or the Pliocene fill
material in the Rhone River canyons. This paper gives three examples from the foreland basins along
the Rhone River valley (the basins of Valence, Valras and Carpentras). It shows that deep ground-
water resources exist at the base of the Pliocene filling and emphasizes the hydraulic connections
between karstic, basinal and Pliocene aquifers. These results are characteristic of all peri-Mediterranean
river basins.
1 INTRODUCTION
The Messinian Salinity Crisis in the Mediterranean area was an exceptional event in geo-
logical history. This crisis had a geomorphological impact on the entire Mediterranean
basin because of the Messinian entrenchment and the filling of the Pliocene valleys. In the
Rhone River valley, the Cretaceous limestone karstification forms a most spectacular
hydrogeologic setting. The famous Fontaine-de-Vaucluse spring near Avignon is assumed
to derive from this period (Julian and Nicod, 1984; Puig, 1987). The consequences of this
crisis must also be taken into account when considering any of the aquifers in contact with
the Messinian karsts or Pliocene rias. This paper gives three examples from the foreland
basins along the Rhone River valley (Valence, Valras and Carpentras basins).
2 GEOLOGICAL HISTORY
The studied area is located in south-eastern France in the Rhone River valley between the
Alpine chain to the east and the European craton (Massif Central) to the west (Fig. 1). This
area contains a succession of foreland basins created by the orogenesis of the Alps during
the Miocene (Carpentras, Valras, Valence). The basins are filled with multilayered sands,
sandstones, clays and marls with a maximum thickness of 600m. At the end of the Miocene,
Figure 1. The Pliocene Rhone ria. Location of the three foreland basins, the Codolet borehole and
the seismic profile.
before the Messinian salinity crisis, the Rhone River and its tributaries flowed into the
Mediterranean Sea. During the Messinian crisis the rivers cut down into the rock to reach
the new Rhone River base level following a sharp fall in the level of the Mediterranean Sea
(Clauzon, 1982; Gargani, 2004).
The end of the crisis was marked by a rapid rise in the Mediterranean Sea level (about
5.32Ma) when the Gibraltar strait was created (Blanc, 2002). The paleo-canyons were
subsequently flooded by the Mediterranean Sea. The sea progressed up the Rhone valley
to a point south of Lyon, approximately 250km inland from the current shoreline (Fig. 1).
3 MORPHOLOGICALGEOLOGICAL IMPACT OF THE CRISIS
At the same time as this dramatic event, the canyons were filled with coarse detritical
materials termed infra-Pliocene by Ballesio (1972). In the Avignon area, 70km inland
from the Mediterranean coast, high-resolution seismic reflection investigations have now
demonstrated the existence of these sediments (Fig. 2).
When the canyons changed into rias, they were gradually filled with sediments accord-
ing to a Gilbert-delta type sedimentation (Clauzon et al., 1995). The Codolet borehole,
located in the Ceze ria near Avignon, offers an example of this sedimentation with a basal
sandy and then a clayey layer above the infra-Pliocene coarse horizon (Fig. 3).
The rapid drop in the base level affected the karst development of the Cretaceous car-
bonate sediments in the Rhone River valley (Fig. 4). The deep karst drainage was blocked
by the filling of the ria, forcing groundwater to rise through Vaucluse-type conduits such
as the Fontaine-de-Vaucluse spring.
4.1 Valence basin
The Valence basin is located in the middle of the Rhone River valley, 180km inland from
the Mediterranean coast. Its covers an area of approximately 720km
2
and the aquifer con-
sists of sand, sandstone and marl and has a maximum thickness of 500m. It borders on the
Hydrogeological consequences of the Messinian Salinity Crisis 293
Figure 2. Seismic cross-section profile of the Pliocene Rhone ria (modified from Schlupp et al.,
2001).
Alpine front in the east (Vercors plateau). Three paleo-tributaries joined the Messinian
Rhone River (Fig. 5), and the Rhone canyon was filled by marine Pliocene clay and sea
sand over 300m thick.
Figure 5 indicates the positions of the marine sedimentary unit and shows that three tribu-
taries of the Rhone River flowed from the plain. Surprisingly, no marine sandy clay is in
contact with the Vercors plateau. However, the Urgonian limestone on the Vercors plateau
contains a large karst development. In the eastern part of the plain a borehole penetrates an
active karst feature beneath 407m of Tertiary cover at an elevation of 61m asl. In add-
ition, at the northern end of the Vercors plateau, a Vaucluse-type karst spring (Thais cave)
was explored along a flooded, 715m long conduit, reaching a depth of 91m asl as com-
pared with the spring level, i.e. 79m asl (Billaud, 2004). Karstification related to the
entrenchment of a Messinian canyon by tributaries of the Rhone River is a plausible hypoth-
esis for explaining the existence of karstic phenomena on the western side of the Vercors.
If the Messinian hydrographic network really entrenched the western side of the Vercors,
the base level of the hypothetical canyons can be estimated at between 81 to 27m asl
(de La Vaissire et al., 2006). These levels are compatible with the observed karst levels in
the spring and borehole. While the Thais cave presently functions as an outlet for the karst
system, it may have functioned like Fontaine-de-Vaucluse, which was formerly an input
or swallow point of the karst aquifer, but presently acts as an outlet of the system (Audra
et al., 2004).
The piezometric map for the basin aquifer shows a regional flow from the east to the
west of the plain in sympathy with the surface relief (Fig. 5). Two major discharges of the
Figure 3. Gamma-ray log of the Codolet borehole (modified from Ferry et al., 1997).
Miocene aquifer are located below the city of Valence and the Drme River valley verti-
cally above the location of the Pliocene rias. The Briffaut (near Valence city) and Allex (in
the valley of the Drme river) boreholes cross-cut the Pliocene horizon. They have high
specific capacity values (Q/s), respectively 2.6 and 5m
3
/h.m. In addition, work on the
south-eastern high-speed rail link (see Fig. 5) has shown that a horizon of infra-Pliocene
breccias lies between the Pliocene clays and the Miocene deposits. This layer is 100 times
more permeable than the basinal molasse deposits (breccias 10
5
m/s versus molasse
10
7
m/s), corresponds to the Messinian event.
Radiocarbon dating was carried out in the basinal aquifer and showed an increase of the
residence time from east to west related to an increase in the magnesium concentration.
The groundwater age increases from modern to 4000 years along a flow path of 10km.
Magnesium can be used as a surrogate indicator of the residence time of the groundwater.
Downstream in the basinal aquifer, the magnesium concentration is in the range of
2030mg/l under the Isre valley and to the east of Valence city (respectively rectangles 1
and 2 Fig. 6). However, the magnesium concentration in the shallow groundwater (Fig. 6)
shows a difference from rectangle 1, with an average of 13mg/l, to rectangle 2, with an
average concentration of 6mg/l.
This difference means that, in the Isre River valley, relatively old groundwater dis-
charges from the underlying basinal aquifer. However, groundwater discharge is from the
Figure 4. Diagrammatic model of karst drainage evolution corresponding to the Messino-Pliocene
eustatic impulse (Audra et al., 2004).
underlying basinal aquifer into the shallow aquifer in the Valence plain is of small volume.
Part of the discharge system of the basinal Valence aquifer is thus formed by the Pliocene
and the infra-Pliocene permeable layers.
4.2 Valras basin
The second basin is located in the South of the Rhone River valley, 100km inland from the
coast. The area covers approximately 500km
2
and the aquifer comprises multilayered
sands, sandstones, clays and marls, with a maximum thickness of 600m. A Pliocene ria is
located in the southern part of the basin (the paleo-Aigues). It reaches a thickness of 250m
and it is infilled with marine Pliocene clay and marine sand.
Groundwater flows from the northeast to the southeast of the Valras basin (Huneau,
2000a). The piezometric map shows that the main discharge area is located vertically
above the ria in the paleo-Aigues axis (Fig. 7). The basinal aquifer becomes confined
under the Pliocene clays. In each case boreholes reach the Miocene beneath the Pliocene
Figure 5. The Valence plain (geomorphological and piezometric maps).
cover and are artesian, but the boreholes in the Pliocene sediments are only poorly product-
ive. The Pliocene clayey-sandy deposits in the Aigues ria act as a hydraulic barrier.
An isotopic study of the basinal aquifer was carried out in 1997 (Huneau, 2000a).
Radiocarbon dating along a 20km flow path revealed a strong increase in the residence
time (from modern to more than 30000 years bp) in the confined part of the aquifer. Stable
isotopes show a strong depletion and it may, therefore, be assumed that they have infil-
trated during the Wrm glacial stage (Fig. 8). Abrupt changes in the isotopic composition
of groundwater may reflect discontinuities and indicate a lack of flow from the unconfined
to the confined aquifer. The confined part of the system may have only a restricted or a
non-existent drainage through the Pliocene ria. Therefore, the main discharge pathway is
to the unconfined shallow aquifers in the alluvium.
4.3 Carpentras basin
The Carpentras basin, the third example, is located 70km inland from the coast. Its area
and its Miocene filling are almost the same as in the Valras basin. Both (Valras and
Carpentras basins) are situated close to each other and their geology is similar, but not
their hydrogeology.
Figure 6. Distribution of the magnesium content of the surface-water table.
The Pliocene ria of paleo-Ouvze crosses the northern part of the Carpentras basin (Fig.
9). The Pliocene deposits are clayey and the sandy horizons are thin and rare. Thus, there
are very few boreholes in the ria area and those that there are have a low transmissivity.
Therefore, the piezometry in the ria is not very well known but the piezometric map (Fig.
9) seems to show a groundwater divide along the ria. The hydraulic gradients are low
around Camaret and Jonquires because there is almost no flow through the ria.
In the North of the ria, groundwater flows westwards and leaves the basin in the Late-
Cretaceous sandy sediments near Orange and Serignan. In the south, the flow lines focus
on the middle of the basin. The direction of this central axis (NNE-SSW) is the same as
that of the main fault which marks the Cretaceous substratum (Nmes fault).
In the Carpentras basin an Early-Cretaceous karstic aquifer is in contact with the basi-
nal aquifer. It is the western extension of the Fontaine-de-Vaucluse system. Very few bore-
holes reach the karst under the Miocene cover and a piezometric map cannot be drawn.
Nevertheless, water levels in the karst are known locally and they are always above the level
of the Miocene aquifer, i.e. groundwater could flow upwards from the karst to the basinal
aquifer. The flow rate is less than 1m
3
s
1
because the hydraulic budget of the Fontaine-
de-Vaucluse system is balanced (Puig, 1987; Couturaud, 1993).
The karst under the Carpentras basin is probably a pre-Messinian network reused during the
Messinian crisis (Audra et al. 2004). Its outlet was at the level of the Messinian hydrographical
Figure 7. Situation of the Valreas basin geomorphological and piezometric maps (from Huneau,
2000a).
network and it was sealed by the Pliocene clayey and sandy filling. Presently groundwater in
the concealed karst has the chemical characteristics of long-residence time water compared
with the active karst water (Couturaud, 1993). Its conductivity is around 700S cm
1
(around
400S cm
1
in the active karst of Fontaine-de-Vaucluse), the Mg/Ca ratio is around 0.8
(around 0.1in the active karst).
5 DISCUSSION
The Messinian Salinity Crisis had an impact on the Rhone River aquifer systems and the
hydraulic connection between karstic, Miocene and Pliocene reservoirs.
Marine Pliocene forms the impermeable base of the non-marine Pliocene and Quaternary
aquifers. From the basins that have been studied, it is known that the marine Pliocene filling
plays an important part in how the basinal aquifers behave. Up to now, the first assumption
was to regard the marine Pliocene filling as a hydraulic barrier to groundwater flow because
of its argillaceous nature. The facies in the rias of paleo-Aigues and paleo-Ouvzes of the
Valreas and Carpentras basins confirm this interpretation. However, the data from drilling
in the rias of the Valence basin and the Cze River reveal considerable aquifer resource
potential for freshwater supply and small irrigation from the sandy Pliocene horizons.
Figure 8. Isotopic variation along the axis 1 flow line direction (from Huneau, 2000b).
Moreover, the conglomeratic layer located at the bottom of the rias may also act as a dis-
charge route for the aquifers. These suggest a new role of the Pliocene rias in groundwater
discharge systems.
This possibility requires a reinterpretation of the hydrogeological data for the Valras
basin. The low hydraulic gradients indicate the existence of flow in the infilling sediments
of the Pliocene ria. Furthermore, if it is a hydraulic barrier, groundwater in the Valras
basin would be as old as the Pliocene filling. Even if the measured ages are very signifi-
cant for such a small basin, they imply some renewal of groundwater. This renewal is
attributed to groundwater flow occurring through the conglomeratic layer at the base of the
paleo-Aigues ria.
The role of the Messinian crisis in the karstic development of south-eastern France is a
spectacular phenomenon. Many examples exist where active circulation occurs derived
from the low sea level during the Quaternary Period: the Fontaine-de-Vaucluse (Puig,
1987; Gilli and Audra, 2004), the underwater spring of Port-Miou (Blavoux et al., 2004)
or the spring of Font Estramar (Aunay et al., 2003). But these can all be assigned to the sig-
nificant lowering of the Mediterranean base level during the Messinian event.
At the time of the crisis, the karstic network of Fontaine-de-Vaucluse had a system of
discharge into the Rhone Messinian canyon. This network was identified under the
Carpentras basin (Couturaud, 1993). Besides, the hydraulic gradients between the karst
and the Miocene aquifer, which favour such exchanges, would use the fracturing (Nmes
fault) or re-use the ante-Miocene karstic network. In this way, an enhanced discharge sys-
tem would exist between the karstic system and the Pliocene ria.
Figure 9. Piezometric map of Carpentras basin. The contour lines indicate hydraulic heads in
meters above sea level (Faure, 1982). The bold grey line is the basin limit. The limit of the Pliocene
ria comes from Clauzon (CEREGE, unpublished data, 2005). Projection: French Lambert II coord-
inates system.
The groundwater flow patterns in the karstic systems and porous aquifers are very var-
ied. Exchanges between these systems are limited only by the permeability of the Miocene
and Pliocene filling. The volumes involved might be small in comparison to the overall
groundwater budget in karstic systems.
6 CONCLUSION
The findings presented in this paper have an important bearing on the aquifer potential of
the Pliocene filling. The clayey filling above the sandy and coarse horizons is a very valu-
able barrier against surface pollution. The lower aquifer of the Pliocene rias offers a con-
siderable freshwater resource and should be subjected to rigorous investigation.
The approach could be implemented in five stages. The first step is to locate the Pliocene
ria by using the existing data (boreholes, electric loggings). The second step would describe
the shape of the Messinian canyons by means of high resolution seismic profiles. The result
would establish the location of the base of the Messinian canyons. An exploration borehole
is required to fully penetrate the Pliocene filling. Concurrently, sedimentology and sequence
stratigraphy studies can be undertaken by using well logging. The fourth step would be
assessment of the sustainable yield of the filling based on pumping test data. If the yield is
adequate then attention will be turned to groundwater quality.
Groundwater resources (quantitative and qualitative) are stressed in the majority of the
peri-Mediterranean countries. In the future, this stress will increase as demand increases.
Demand is related to the population growth and maybe also to global climate change (trend
to aridity in the Mediterranean area). New resources must be sought and exploited. In this
context, the aquifers of the Pliocene fillings of the Messinian canyons must be explored.
The Rhone and the Nile Rivers valleys are the sectors where the Messinian canyons are
known, but their groundwater resources remain to be evaluated. Moreover, all the many
other Messinian paleo-fill areas around the Mediterranean Sea should also be investigated
for new water supplies.
REFERENCES
Audra P, Mocochain L, Camus H, Gilli E, Clauzon G, Bigot JY (2004) The effect of the Messinian
Deep Stage on karst development around the Mediterranean Sea. Examples from Southern
France. Geodinamica Acta, 17, 6, 2738
Aunay B, Drfliger N, Le Strat P, Ladouche B, Bakalowicz M (2003) Evolution gologique, mise en
place de la karstification et thermalisme des aquifres karstiques pri-mditerranens. Exemple
du karst des Corbires dOpoul (Geological evolution, karstification and thermalism of per-
imediterranean karstic aquifers. Example of the Corbires dOpoul karst). In: 10me journe
technique du Comit Franais de lAIH. Circulations hydrothermales en terrains calcaires,
Carcassone, 1522
Ballesio R (1972) tude stratigraphique du Pliocne rhodanien. (Stratigraphic study of the
Rhodanian Pliocene). PhD, University of Lyon I, France [in French]
Billaud Y (2004) Rapport dexploration de la grotte de Thas (Exploration report of the Thas cave).
Spelunca, 96, 57
Blanc PL (2002) The opening of the Plio-Quaternary Gibraltar Strait: assessing the size of a cata-
clysm. Geodinamica Acta, 15, 303317
Blavoux B, Gilli E, Rousset C (2004) Alimentation et origine de la salinit de la source sous-marine
de Port-Miou (Marseille-Cassis). Principale mergence dun rseau karstique hrit du Messinien
(Watershed and origin of the salinity of the karstic submarine spring of Port-miou (Marseille-
Cassis, France). Main outflow of a Messinian karstic system). C.R. Geoscience, 336, 523533
Clauzon G (1982) Le canyon messinien du Rhne: une preuve dcisive du desicated deep-basin
model [Hs, Cita et Ryan, 1973] (The Messinian Rhone canyon: a conclusive proof of the des-
icated deep-basin model [Hs, Cita et Ryan, 1973]). Bull. Soc. Gol. France, 24, 7, 597610
Clauzon G, Rubino JL, Savoye B (1995) Marine Pliocene Gilbert-type fan deltas along the French
Mediterranean coast. ASF Publications, 23, 145222
Couturaud A (1993). Hydrogologie de la partie occidentale du systme karstique de Vaucluse.
(Hydrogeology of the Western part of the Fontaine-de-Vaucluse karstic system). PhD, University
of Avignon, France [in French]
de La Vaissire R, Faure G, Clauzon G, Blavoux B (2006) Impact of the Messinian Salinity crisis on
the hydrogeology of the Valence plain (South-East France). In: Duran, Andreo, Carrasco (Eds)
Karst, climate change and groundwater. Hidrogeologia y Aguas Subterraneas, 18, 385393,
Spain, Madrid
Faure G (1982) tude hydrogologique du bassin Miocne de Carpentras Relation entre laquifre
molassique et le substratum exploitabilit. (Hydrogeology of the Carpentras molassic basin.
Relation between molassic and undercover aquifers exploitability). PhD, University of Lyon I,
France [in French]
Ferry S, Ballesio R, Monier P (1997) Modalits du remplissage sdimentaire de la ria pliocne du
Rhne. Problmes en suspens aprs les premiers travaux de lANDRA. (Sedimentary process of
filling of the Pliocene Rhone canyon. Abeyance issues after the ANDRA works) In: Journes sci-
entifiques ANDRA, Bagnoles sur Czes
Gargani J (2004) Modelling of the erosion in the Rhone valley during the Messinian crisis (France).
Quaternary International, 121, 1322
Gilli E, Audra P (2004) Les lithophages pliocnes de la fontaine de Vaucluse (Vaucluse, France). Un
argument pour une phase messinienne dans la gense du plus grand karst noy de France (The
Pliocene Lithophagia of the Fontaine-de-Vaucluse karst spring (Vaucluse, France). An argument
for a Messinian stage in the genesis of the largest French phreatic karst). C. R. Geoscience, 336,
14811489
Huneau F (2000a) Fonctionnement hydrogologique et archives paloclimatiques dun aquifre pro-
fond mditerranen. Etude gochimique et isotopique du bassin miocne de Valras (Sud-Est de
la France). (Hydrogeological functioning and paleoclimatic records of a deep mediterranean
aquifer. Geochemical and isotopic study of the molassic basin of Valras (South-East, France)).
PhD, University of Avignon, France [in French]
Huneau F (2000b) Isotopic hydrogeology within the Miocene basin of Carpentras-Valras (south-
eastern France). In: Dassargues A (ed) Proc. of TraM2000 Tracers and Modelling in
Hydrogeology, Lige, Belgium, 433438
Julian M, Nicod J (1984) Palokarst et palomorphologie nogne des Alpes Occidentales et rgions
adjacentes (Paleokarst and Neogene paleomorphology of the Western Alps and the nearest area).
Karstologia, 4, 1118
Puig JM (1987) Le systme karstique de la Fontaine de Vaucluse. (Karstic system of the Fontaine-
de-Vaucluse). PhD, University of Avignon, France [in French]
Schlupp A, Clauzon G, Avouac JP (2001) Mouvement post-messinien sur la faille de Nmes: impli-
cations pour la sismotectonique de la Provence (Nimes fault post-messinian movement: sismotec-
tonic implications in the Provence area). Bull. Soc. Gol. France, 172, 6, 6977119
CHAPTER 23
Hydrodynamic behaviour during pumping test and
modelling of the Cent Fonts karst system
Bernard Ladouche
1
, Nathalie Drfliger
1
, Jean-Christophe Marchal
1
,
Patrick Lachassagne
1
, Michel Bakalowicz
2
, Irinia Valarie
3
&
Philippe Lenoir
3
1
BRGM, EAU-RMD, Montpellier
2
CREEN-ESIB, Facult dIngnierie (HydroSciences CNRS Montpellier), Riad El Solh BP
BEYROUTH (Liban)
3
Conseil Gnral de lHrault, Direction Eau et Milieux Aquatiques, Ple Environnement,
Eau rue dAlco, Montpellier Cedex
ABSTRACT: Aware of the need for a sustainable water resources management policy, the Conseil
Gnral of Hrault, together with BRGM and CNRS, is conducting a research project on the hydro-
geology of the binary Cent Fonts karst system (60km
2
). This aquifer system is partially fed by
Buges river water losses. A methodological approach using hydrographical analysis was previously
used to characterize the structure and functioning of the saturated zone and evaluate the respective
role of the epikarst and Buges water losses. The aquifer storage was estimated at 8.6 0.4 million
m
3
. In order to determine the resources available within the system, pumping tests were carried out
during the summer of 2005. Since classical analytical solutions cannot effectively reproduce the
hydrodynamic behaviour of this system, a specific mathematical model combining a reservoir
model and an analytical solution for groundwater flow from the surrounding fissured blocks was
developed.
1 INTRODUCTION
Aware of the need for a sustainable water resources management policy, the Conseil
Gnral of Hrault (France), in close scientific collaboration with BRGM and CNRS, is
conducting a research project on the characterisation of the structure and functioning of
the Cent Fonts karst aquifer. Located in the Hrault River basin, an area predicted to have
a deficit in drinking water resources in 2015, this aquifer system is part of one of the major
karst aquifers of the Rhne-Mditerrane-Corse basin and has been classified by the
Water Agency as a priority aquifer for study.
A methodology developed specifically for characterizing karst systems (Bakalowicz,
1999) was used to estimate water resources and propose water management policies. It
includes four major steps:
1. Identification. The aim of this first step is to identify the type of system (fissured, func-
tioning or non-functioning karst), identify the presence the reserves within the saturated
zone, and determine the boundaries of the system. The catchment is delineated by means
of various complementary tools such as the water budget, and artificial and natural
tracers. Tools used to study tectonics and lithostratigraphy as well as methods specific to
karst hydrogeology such as the analysis of recession curves and the inventory of karst
features are also used.
2. Characterization: The system is characterized by analysing both its hydrodynamics and
its conduit flow system. Infiltration functions, storage capacity and karst drainage effi-
ciency are assessed. The degree of karstification and the hydrogeological role played
by paleokarst structures are important during this phase.
3. Validation or demonstration: The system is modelled by means of black- or grey-box
models (reservoir models, inverse modelling) and, depending on the available data, by
a distributed model. These models are based on the conceptual model synthesising the
interpretation at the end of step 2. The aim of the modelling is to test the conceptual
model and related assumptions, and to simulate system behaviour under various con-
ditions such as seasonal over-pumping in a borehole or a spring, aquifer storage recov-
ery (ASR) or climate change effects. Pumping tests may be carried out during this
phase to verify the water resources assessment and demonstrate the ability of the sys-
tem to provide water during a set period.
4. Assessment or management: This step involves pumping, protecting and managing the
resource. The characterization of the system must, therefore, be supplemented by
approaches taking into account all constraints on the system directly impacting its real
potential, water quality, actual use of the resource, aquifer vulnerability, soil cover and
principal pollution risks.
The recent studies of this karst system began in 1997 with the two initial phases involving
identification and characterisation of the structure and functioning based on geological,
hydrogeological and hydrochemical analyses. Classical geological approaches and data
acquisition of parameters such as spring discharge and chemical content were carried out.
The swallow hole in the Buges streambed was also monitored, as it is one of the two main
components of recharge, along with rainfall infiltration, in the karst catchment. In order
to estimate the resource in this binary karst system, pumping tests were carried out during
the summer of 2005. The main objectives of this third demonstration or validation phase
were to:
(i) improve our knowledge of the karst system and, in particular, estimate water
reserves,
(ii) characterize the ability of the borehole intercepting the conduit of the underground
drainage network to mobilize groundwater reserves,
(iii) propose groundwater management scenarios,
(iv) assess the environmental impact of the groundwater exploitation from the aquifer on
groundwater, surface water and the microfauna living in the networks of caves.
However, the impact on microfauna is not discussed here.
The main hydrodynamic behaviour of the Cent Fonts karst system during pumping tests
and the modelling approach used for the interpretation of the long-term pumping test are
described below. The mathematical model developed specifically for this system com-
bines a reservoir model and an analytical solution for the groundwater flow from the sur-
rounding fissured blocks.
2 SITE DESCRIPTION AND PREVIOUS RESULTS
2.1 Site description
The Cent Fonts karst system is located north of Montpellier (Hrault, Southern France) in
a series of thick calcareous and dolomite massifs (Middle and Upper Jurassic) extending
from the Montpellier fold (North Pyrenean Thrust) to the southern border of the Massif
Central shield. The Buges valley and the Cvennes fault mark the boundaries of the
catchment in the north and northwest, and the Hrault valley marks the boundary in the
southeast (Fig. 1). Morphologically, the area is defined as a plateau representing relicts of
an Oligocene erosion surface that was uplifted during the late Quaternary to an altitude of
200 to 500m. This plateau is strongly incised by the Hrault River, which represents the
base level of the karst system.
Hydrogeologically, the saturated zone of the Cent Fonts karst system is located in
Bathonian dolomite layers (Middle Jurassic, Upper Dogger, between 150 to 300m thick) and
possibly in the underlying Aalenian-Bajocian (Middle Jurassic, Lower Dogger) formation.
Hydrodynamic behaviour during pumping test and modelling of the Cent Fonts karst system 305
Figure 1. Simplified geological framework and location map of the Cent Fonts karst system catch-
ment (Hrault, Southern France).
The water infiltrates through either the upper Jurassic layers (Upper Oxfordian and
Kimmeridgian) or directly by continuous water seepage from the Buges riverbed (swallow
hole) into the Bathonian calcareous dolomite.
2.2 Previous results
Discharge from the Cent Fonts spring measuring since May 1997. Annual mean discharge
is close to 1100l s
1
, while during the summer (July and August) the mean discharge
ranges from 290l s
1
to 340l s
1
.
The relationship between the Buges stream losses (occurring in the segment of the
stream between the upstream (B1) and downstream (B2) gauging stations, Fig. 1) and the
Cent-Fonts spring was determined by a tracer experiment using artificial and natural tracers.
Discharge measured at the two gauging stations enabled to estimate the flow that recharges
the saturated zone of the Cent Fonts karst system. The swallow hole in the Buges stream
contributes about 50% of the Cent Fonts spring discharge, mainly during storm flow
events. During the summer, Buges losses are total with a mean discharge ranges from
147l s
1
and 87l s
1
(Ladouche et al., 2002, 2005). In August, the inter-quartile
(0.250.75) interval is 296l s
1
348l s
1
and 45 l s
1
130l s
1
for the Cent Fonts
spring and Buges losses, respectively. The recharge area of the binary karst system has a
surface area of about 60km
2
and the effective rainfall contribution (550mm/year) corres-
ponds to a recharge area of close to 30km
2
.
During phase 1 of the study, a geochemical study was carried out during flood and low-
flow stages (Aquilina et al., 2005, 2006). There is a decrease in Mg and an increase in Ca
in the Cent Fonts spring during flood events. During low-flow periods, the Mg concentra-
tion increases whilst Ca remains more or less constant. This temporal evolution has been
used to characterize the functioning of the karst system in terms of transit and transfer with
an inverse model (Ladouche et al., 2002). The inverse model uses rainfall and flux data
(discharge, calcium and magnesium fluxes) to calculate unit hydrographs and impulse
responses of fluxes and thereby to separate hydrographs. Fluxes have been defined as the
product of discharge and calcium and magnesium of spring water. In this approach, ini-
tially developed by Pinault et al., (2001), the magnesium content enables us to characterise
the pre-event component (i.e. the water coming from the saturated zone in the Bathonian
formation). Calcium is used to characterise the event component (i.e. the component
related to the effective rainfall in epikarst recharge). The relative contribution of the two
discharge components (event and pre-event water) is estimated to be 20% (recharge by
effective rainfall) and 80% (discharge from the saturated zone reservoir). These percent-
ages represent the mean contribution of the two components. The event-component
impulse response is rapidfollowing an effective rainfall event, a discharge peak appears
after a lag of about 1.5 days. At this time (1.5 days after the rainfall event), pre-event con-
tribution is minimum (around 45%) and spring discharge is composed mainly of event
water. Three days after the discharge peak, the event contribution has dropped to around
25% and the spring discharge is composed mainly of pre-event water. The mean residence
time of the impulse response of the event water is around 15 days. For a lag of up to 45
days, spring discharge is composed mainly of pre-event water. Characteristics of event-
component impulse response indicate that the epikarst zone is well connected to the satur-
ated zone as a result of extensive karstification. The analysis of 8 recession curves
(19982005) suggests that the dynamic volume, which is similar to the amount of water
stored in the saturated zone, is significant: 8.6 0.4 million m
3
.
The Cent-Fonts spring, located on the right bank of the Hrault River, is the main dis-
charge point for the massif known as the La Selle plateau. There are 11 outlets (8 perman-
ent and 3 temporary) along the river bank (Fig. 2) with a total discharge of between
200 l s
1
during severe low-water stages and more than 12000l s
1
during peak flow
periods in the winter or spring. The Cent Fonts springs are the outlet of a saturated karst
network that has been partially explored and mapped by divers (Fig. 2).
Two boreholes intercept the karst network near the springs (Fig. 2). One is about 60m
deep and was drilled in the 1990s by CGE. The other, drilled by FORACCO for the
Conseil Gnral de lHrault in 2005, is located further upstream and reaches the largest
explored cavity 120m deep (Fig. 2). This recent borehole (total depth: 128m, diameter:
1066mm to cavity 125m below ground level) is equipped with a 675kW submerged pump
producing a flow rate of 518l s
1
under 130m water head. A temporary pipeline carries
Figure 2. Location of the pumping and monitoring boreholes near the Cent Fonts springs.
the pumped water to the Hrault River downstream of the last outlet of the Cent Fonts
springs (Fig. 2).
A monitoring network of 14 sampling points was installed 4 collecting surface water
(from the Hrault and Buges rivers) and 10 collecting groundwater (6 in the Cent Fonts
catchment (CF(F3), CGE, P3, P5, P6, P7) and 4 in an adjacent karst system Buges spring,
Fontanilles spring, Embougette and St-Andr de Buges wells (Figs. 1 and 2)). The
Hrault River is monitored near the outlets to measure the discharge of the Hrault River,
measure the output of the pumping test, estimate spring discharge, and assess Hrault
infiltration towards the karst system during the pumping tests.
4 RESULTS
4.1 Cent Fonts spring
Time series of discharge from the Cent Fonts springs (Fig. 3) were calculated using a gauging
curve taking into consideration the water level in the CGE borehole and gauging of the
springs for various hydrological conditions. The error of the gauging curve is estimated to
be 20% for discharge lower than 10 000l s
1
and 40% for discharge higher than
10 000l s
1
. The configuration of the site does not allow gauging high discharge.
The spring discharge ranges between 282l s
1
(56l s
1
) and 254l s
1
(51l s
1
)
between 01/07/2005 and 01/08/2005 (Fig. 3), date of the beginning of pumping tests. The
discharge from the outlets along the Hrault was measured during step pumping tests. The
0
2000
4000
6000
8000
10000
12000
01/07/05 19/07/05 06/08/05 24/08/05 11/09/05 29/09/05
D
i
s
c
h
a
r
g
e

(
l
/
s
)
Upstream gauging (H1) Downstream gauging (H2)
Cent Fonts discharge
Hrault river discharge
(upstream station: H1)
Spring dried-up periods
Out flow of pumping
Figure 3. Mean hourly discharge of the Cent Fonts spring, Hrault River (upstream station: H1)
and outflow of pumping (into the Hrault River, upstream of the H2 gauging station) between
01/01/2005 and 06/10/2005.
pumping rate was higher than the discharge from the springs (254l s
1
) and flow from
the outlets stopped after about 45 to 85 minutes.
4.2 Hrault and Buges rivers
Five differential gauging were done between 20/07/2005 and 12/08/2005 (Fig. 3) at two
gauging stations on the Hrault River near the Cent Fonts springs (Fig. 2). Discharge in the
Hrault at the upstream station (H1) ranged between 1511l s
1
(76l s
1
) and
1861l s
1
(93l s
1
), and discharge at the downstream station (H2) ranged between
1927l s
1
(96l s
1
) and 2235l s
1
(76l s
1
). The differential discharge (D
H2
-
D
H1
) measured before the pumping test (20/07/05: 262l s
1
26l s
1
) was similar to
the natural spring discharge (259l s
1
52l s
1
) estimated from the Cent Fonts gauging
curve. During the pumping test, when the Cent Fonts springs were dry (Fig. 3), the differ-
ential discharge generally corresponds to the pumping discharge, taking into consideration
the margin of error in the pumping rate and discharge measurements. The differential dis-
charge approach is not accurate enough to estimate the water loss in the Hrault that
occurred during the pumping tests. Water level monitoring at the upstream station (H1)
gives discharges in the Hrault upstream from the Cent Fonts outlets ranging between
1960l s
1
and 1570l s
1
during the pumping tests. The mean electrical conductivity in
the Hrault River was 350s/cm during the pumping tests. The mean daily temperature of
the river water influenced by daily variations in air temperature is between 22.9 and 27.3C.
The minimum and maximum values are 21.6 and 28.6C, respectively.
The Buges stream level is monitored at two gauging stations (B1 and B2, Fig. 1) and
the discharge was calculated using gauging curves. During the JulyAugust period, no dis-
charge was measured at the downstream gauging station (B2), all of the Buges stream-
flow having been lost. Discharge at the upstream station (B1) ranged between 80 and
15 l s
1
in July 2005 and was steady at around 15l s
1 in
August 2005. The Buges
losses measured during the summer of 2005 were very low compared to those measured
during the previous study. All of the water lost from the Buges stream recharges the Cent
Fonts karst system. This was estimated to account for 6% of the natural discharge of the
Cent Fonts spring observed before the long-term pumping test (254l s
1
) began. The
mean value of the electrical conductivity of Buges water during the observation period
ranged between 374 and 354s/cm. The temperature of the Buges stream ranged between
17.9 and 24.1C, the mean value being 21.1C.
4.3 Pumping tests
The first stage of the pumping tests began with a step pumping test. This was done at the
end of July 2005 (27/07/200530/07/2005) during a severe low water stage (Q CF
254l s
1
). Pumping was done in four steps with flow rates of 200, 300, 500 and 400l s
1,
each lasting 6 hours with pumping halted between each step until the water level had risen
to the initial level in the borehole.
The long-term pumping test began with a 400l s
1
flow rate on August 1, 2005 at 9.06 am
(French Summer Time GMT2) for more than one month (01/08/200506/09/2005).
Pumping stopped four times during the test, twice on purpose (09/08/2005 and 02/09/2005),
once due to electrical problems (22/08/2005 for less than one hour) and once due to heavy
rainfall and a storm event (06/09/2005 for 14 days). Heavy rainfall at the beginning of
September caused peak flow at the spring and consequently the temporary halting of the
pumping test. The pumping test began again on 20/09/2005 and continued until the end of
September at the following flow rates 300l s
1
(duration [d] 27 hours), 500l s
1
(d 26 hours) and 400l s
1
(d 168 hours, Fig. 4).
The water level in the karst drain dropped 19.11m between 01/08/2005 and 09/08/2005
(first programmed halt of the pumping test, Fig. 4). The daily drawdown was 4.59m on
day 1, 2.65m to 2.18m on days 25, and 1.95m to 1.5m on days 69. The water level rose
3.43m in 6 hours after the first halt. It dropped 52.12m between 09/08/2005 and
02/09/2005 (second halt). The daily drawdown ranged between 4.27m (day 1) to 1.09m
(day 24). Although the monitored water level followed the same evolution in F3, F2 and
CGE boreholes between 01/08/2005 and 18/08/2005, the water level at the CGE borehole
stopped decreasing on 18/08/2005 and stabilized at around 41.6m NGF (35m of
Figure 4. Pumping rates and water level time series at the observation and pumping boreholes.
drawdown). The total volume of water pumped during the period (27/07/2005 to 30/09/2005)
was 1 493 496m
3
.
The influence of the pumping test on the piezometers (Fig. 6) in the catchment was as
follows: weak for P3 and P7 (13cm/month vs. 4cm/month before the pumping test) and
stronger for P5 (4.8m/month before the pumping test and 7.5m/month during the pump-
ing test) and P6 (1.08m/month before the pumping test and 2.8m/month during the pump-
ing test). Piezometer P3 seems to be representative of a karst conduit or an annex drainage
system (ADS) poorly connected to the karst conduit (CGE and CF boreholes). P7, located
in the water loss area of the Buges stream, seems to be located within fissured blocks
connected to a nearby karst conduit. The influence of the pumping test on piezometers P5
and P6 seems to indicate the hydrogeological functioning of fissured blocks. They are
not located in the immediate vicinity of a conduit of the Cent Fonts karst conduit network.
The range in water levels measured in P6 during the pumping test was small (around 5m),
compared to the range measured in the Cent Fonts borehole (around 54m). In the simula-
tion, the hydraulic head measured in P6 is assumed to represent the hydraulic head of
reservoir 2 (Fig. 6).
80
90
100
110
120
130
140
30/06/05 18/07/05 05/08/05 23/08/05 10/09/05 28/09/05
P
i
e
z
o
m
e
t
r
y

(
m
,

N
G
F
)
P7
P6
P5
P3
Start of long
duration
pumping
tests
Probe out
of water
20
30
40
50
60
70
80
90
30/06/05 18/07/05 05/08/05 23/08/05 10/09/05 28/09/05
P
i
e
z
o
m
e
t
r
y

(
m
,

N
G
F
)
Cent Fonts
borehole (F3)
Start of long
duration
pumping
tests
Figure 5. Temporal evolution of piezometry in piezometers P3, P5, P6, P7 and the Cent Fonts bore-
hole (F3) during the summer of 2005.
The electrical conductivity, temperature and turbidity of the water within the karst drain
during the pumping tests were monitored in the CGE and CF boreholes. Electrical conduct-
ivity before the long-term pumping test was 454 (1) S/cm. The mean temperature of
the water within the karst drain was around 13.5 (0.1) C. During the long-term pump-
ing test, electrical conductivity in the CGE borehole decreased after 5 hours and 35 min-
utes, then continued to decrease to a value of 360 S/cm after more than 52 hours of
pumping. The temperature started to rise after 17 hours and 25 minutes of pumping to a
maximum of 20C after 161 hours of pumping. These measurements in the CGE borehole
evidence the infiltration of Hrault river water into the Cent Fonts karst system during the
long-term pumping test. The electrical conductivity of the pumped water ranged between
454 and 446S/cm between 01/08/2005 and 03/08/2005, and between 446 and 458S/cm
during the 03/0802/09/05 period, with a maximum close to 460 s/cm observed on
25/08/05.
The turbidity of the water in the karst drain before the pumping test was very low: 0.5
(0.1) NTU. It started to increase after 9 hours of pumping and reached a peak value of
33 NTU on 03/08/2005 (1.20 pm). Several turbidity peaks were observed between
03/08/2005 and 13/08/2005. After the peak flow (09/09/2005) with a maximum value of
over 40 NTU, the turbidity decreased to reach a value of 2.5 NTU (20/09/2005). About
three hours after the pumping test stopped, the turbidity decreased to 1.1 NTU.
4.3 Mathematical model
The karst system functioning was conceptualized by two interacting free-surface water
reservoirs (Fig. 6): reservoir 1 corresponds to the drain network and reservoir 2 corres-
ponds to the matrix and the annex drainage system (ADS). The development of the math-
ematical model is described in Marchal et al. (2006 b).
Reservoir 1 corresponds to the drain network in the karst aquifer. It is supplied by flow
from losses (Q
L
) and the river (Q
R
), while pumping flow rate (Q
P
) withdraws water from
the reservoir. Exchanges with reservoir 2 depend on the respective hydraulic heads in the
two reservoirs. The hydraulic head (H
D
) is assumed to be the same throughout the entire
network (quadratic head losses have been assumed to be weak in the conduit network) and
limited by a threshold value (Th
D
) that is the elevation of the spring, above which the reser-
voir overflows. This causes a discharge at the spring (Q
S
) when the head in the drain is
Figure 6. Sketch of the 2 reservoirs model and volume conservation equations (Marchal et al.
2006 b).
higher than the elevation of the spring. Under specified initial and boundary conditions
and parameters, reservoir 1 calculates the hydraulic head in the drains and the discharge
rate at the spring.
Reservoir 2 corresponds to the matrix and ADS drained by the conduit network. It is sup-
plied by recharge (R). Exchange diffuse flow (Q
E
) between reservoirs 1 and 2 is computed
according to the difference in hydraulic head between the reservoirs. The hydraulic head
in reservoir 2 (H
B
) is assumed to be the same throughout the basin.
The diffuse flow in the matrix/ADS towards the drains varies according to the fluctuations
in the difference in hydraulic head between the drains and matrix. As suggested by using
Barkers analytical solution (1988) (not shown), the geometry of the solicitation of diffuse
flow to conduits during the pumping test is probably linear with a flow-dimension close to
1 (Marchal et al., 2006 b). Therefore, the analytical solution for reservoir 2 is one-dimen-
sion flow towards a trench (Marchal et al., 2006 b). It was assumed that drawdown in the
reservoir 2 matrix during pumping tests is negligible compared to drawdown in the con-
duit network.
In karst systems, since there is an initial difference in hydraulic head between matrix
and drains that produce the discharge flow at the spring in the absence of pumping, accord-
ing to the principle of superposition (de Marsily, 1986), the total exchange flow from the
matrix/ADS is equal to the sum of the initial flow Q
before pumping due to initial

hydraulic head difference between matrix/ADS and drains (H
b0
H
d0
) and the additional
flow due to the pumping expressed as a function of the additional drawdown.
5 DISCUSSION
The evolution of electrical conductivity, temperature and turbidity time series within the
karst drain through the CGE borehole during the first phase of the long-term pumping test
is probably directly related to Hrault River infiltration into the karst system. In karst
aquifers, flood stages with peak flow at springs and pumping tests are likely to remove
intra-karst sediments and unplug clay in fractures or cavities. Although the drawdown due
to the pumping tests is great, sediment mobilisation and clay unplugging do not appear to
be a major phenomenon. The monitoring of electrical conductivity at the pumping bore-
hole (F3) and at the Hrault River station enabled us to estimate the contribution of the
Hrault River during the pumping tests at between 0 and 37l s
1
18l s
1
. The amount
of infiltration water of Hrault River is equal to zero as long as the hydraulic head in the
karst system is higher than the water level of the Hrault River. This is in agreement with
the geochemical data (
87
Sr/
86
Sr vs. Cl/Sr ratios) used to calculate the water mixture ratio
of karst water and Hrault water (Ladouche et al., 2005). Under the hydrological condi-
tions prevailing during the summer of 2005, pumping tests with a pumping rate higher
than 200l s
1
caused infiltration from Hrault River toward the karst system. The mean
value of the infiltration is estimated at 30l s
1
. This infiltration occurs near the outlets,
possibly within the riverbed.
The comparison of observed and simulated hydraulic heads in the drain during the long-
term pumping test shows how well the model matches the monitored data (Fig. 7). This is
particularly visible during the recovery observed after pumping is halted at the beginning
of September and the second step at 300l s
1
. The models major parameters, initial and
boundary conditions (Table 1) are the following:
(i) Water supply of the karst network by water losses from the Buges River and by
water from blocks; infiltration of Hrault River up to 30l s
1
(estimated with geo-
chemical data); no recharge by rainfall (R).
(ii) Out storage within karst drain network: 1900m
3
/m of drawdown, non-function of the
water head within the drain network. The sensitivity analysis of this parameter
(Marchal et al, 2006b) showed that it is well constrained: 1900 100m
3
/m.
(iii) The length of the drain network: unknown. It is assumed to be about 10km on the
basis of the shape of the water catchment.
(iv) Drawdown in reservoir 2 matrix (P6 measurements) during the pumping test is assumed
to be negligible compared to drawdown in the conduit network (Cent Fonts borehole).
The obtained transmissivity (T 1.6 (0.1)*10
5
m
2
/s) of the reservoir 2 matrix is quite
low and responsible for the large drawdown measured in the Cent Fonts borehole during
the pumping test. This value, determined for a reservoir matrix intersecting a 10-km-long
drain network, implies that the hydraulic connection between the reservoir matrix and
drain network is not very efficient.
0
10
20
30
40
50
60
70
80
90
0
100
200
300
400
500
W
a
t
e
r

l
e
v
e
l

N
G
F

(
m
)
P
u
m
p
i
n
g

r
a
t
e

(
l
/
s
)
27/07 01/08 06/08 11/08 16/08 21/08 26/08 31/08 05/09
Date (2005)
Observed simulated Pumpingrate
Figure 7. Temporal evolution of the piezometry simulated by the model during the pumping test.
Table 1. Parameters, initial and boundary conditions of the model giving the best match of
observed drawdown (grey boxes: calibrated parameters).
Data Reservoir 1 (drains) Reservoir 2 (matrix and ADS)
Parameters S
DRAIN
(m
2
) 1900 S
BASSIN
(km
2
) 30
Th
D
(m, NGF) 76.9 S () 0.007
T (m
2
/s) 1.6 10
5
l (m) 10 000
Initial conditions H
D0
(m, NGF) 76.9 H
b0
(m, NGF) 115
Boundary conditions Q
L
(m
3
/s) 0.015 R 0
Q
R
(m
3
/s) if H
D
75m 0.030
Q
P
(m
3
/s) Series
The numerical model enables us to determine the temporal evolution of the various
sources contributing to flow during the pumping test: the total discharge is made up of the
natural diffuse contribution of matrix/ADS, additional diffuse flow from matrix/ADS
caused by pumping, Hrault River contribution, the Buges stream losses and the dewater-
ing of the karst drain network. During the long-term pumping test (400l.s
1
), 64% of the
water comes from natural out storage from blocks, 19% from reserves (Matrix / ADS
solicitation), 4% from Buges River losses, 7% from Hrault River infiltration and 6%
from drain dewatering, with a total discharge of 1 106m
3
.
CONCLUSION AND PERSPECTIVES
Karst aquifers are complex aquifers in which there are dual or triple flow systems with
localized and often turbulent flow in conduits and Darcian flow in the fissured matrix
(blocks). Pumping tests in boreholes located in the blocks appear to be well suited for
estimating hydraulic parameters. Pumping tests in boreholes intercepting karst conduits,
on the other hand, are difficult to interpret because the geometry of drains and connections
to the fissured matrix require different hypotheses.
Past attempts to interpret pumping test data in karst systems have involved only a study
of hydrochemical aspects in order to evaluate the renewal of the exploitable water
(Reynaud et al., 1999) or a qualitative interpretation of drawdown (Bakalowicz et al.,
1994). The only methodology available for quantitative interpretation (Marsaud 1997) is
particularly well-suited but also limited to the case of pumping in a well located in the
matrix/ADS linked to the main spring of the system and based on the interpretation of
spring discharge rate fluctuations during pumping.
The interpretation proposed here, based on the conceptual model of the Cent Fonts karst
system using a two-reservoir model one for the karst conduit network and another for the
groundwater flow within the matrix toward the conduits is unique. It enables us to repro-
duce the drawdown and recovery phases and take into consideration input such as river
losses and infiltration. This is a novel modelling approach that can be used for long-term
pumping tests (72 hours) and applied to karst systems under active management. This
model is being used to generate water management scenarios that are necessary and assessed
along with others, before decision are made on the design of an active exploitation of this
karst system.
ACKNOWLEDGMENTS
This study is part of the Cent Fonts project funded by the Conseil Gnral of Hrault and
by BRGM within the framework of the COMPLEXAQUI project (EAUR15).
REFERENCES
Aquilina, L., Ladouche, B., Drfliger, N. (2005) Recharge processes in karstic systems investigated
through the correlation of chemical and isotopic composition of rain and spring-waters. Applied
Geochemistry, 20, 21892206.
Aquilina, L., Ladouche, B., Drfliger, N. (2006) Water storage and transfer in the epikarst of karstic
systems during high flow periods, J.of Hydrology, 327, 472485.
Bakalowicz, M. et al. (1994) High discharge pumping in a vertical cave: fundamental and applied
results. In: M. Bakalowicz and N. Crampon (Editors), Basic and applied hydrogeological research
in French karst areas, Montpellier-Millau, 93110.
Bakalowicz, M., (1999) Connaissance et gestion des resources en eaux souterraines dans les rgions
karstiques. Guide technique N
o
3, Agence de leau Rhne Mditerrane & Corse.
Barker, J.A. (1988) A generalized radial flow model for hydraulic tests in fractured rock. Water
Resour. Res., 24, 10, 17961804.
Ladouche, B., Drfliger, N., Pouget, R., Petit, V., Thiery, D., Golaz, C. (2002) Caractrisation du
fonctionnement des systmes karstiques nord-montpellirains. Rapport du programme
19992001 Buges. BRGM RP-51584-FR, 200p. 111 fig., 9 tabl., 3 photos, 4 ann.
Ladouche B., Marchal J-Ch., Drfliger N., Lachassagne P., Lanini S., Le Strat P. (2005) Pompage
dessais sur le systme karstique des Cent Fonts (Cne de Causse de la Selle, Hrault), Prsentation
et interprtation des donnes recueillies, Rapport final, BRGM RP-54426-FR, 82ill., 45 tabl., 9
ann., 245pp.
Marchal, J.C., Perrochet, P., Lanini, S. Lachassagne, P. (2006, a) Analytical solutions to discharge
flow under varying drawdowns by superposition. Hydrogeology J., submitted.
Marchal, J-Ch., Drfliger, N., Lachassagne, P., Ladouche, B. (2006, b) A method for the interpret-
ation of long-duration pumping test in the drain of a mixed flow karst system, Water Resour. Res,
submitted.
Marsaud, B., (1997) A method for interpreting pumping tests in karst aquifers. Hydrogologie, 3,
3142.
Marsily, G. de., (1986) Quantitative Hydrogeology. Academic Press, 440 pp
Pinault, J.L., Plagnes, V., Aquilina, L., Bakalowicz, M. (2001) Inverse modeling of the hydrological
and the hydrochemical behavior of hydrosystems Characterization of karst system functioning.,
Water Resour. Res., Vol. 37, n
o
8, p. 21912204
Reynaud, A., Guglielmi, Y., Mudry, J. and Mangan, C., (1999) Hydrochemical approach to the alter-
ations of the recharge of a karst aquifer consecutive to a long pumping period; example taken from
Pinchinade Graben (Mouans-Sartoux, French Riviera). Ground Water, 37, 3, 414417.
CHAPTER 24
Multidisciplinary approach to a karstic region for
the use and protection of the water resource.
Application to the Causse de Sauveterre (south of
France)
Michel Lepiller
1
, Bernard Blavoux
2
, Stphane Brusset
3
, Laurent
Bruxelles
4
, Laurent Danneville
5
, Alain Mangin
6
& Pierre Marchet
7
1
PolytechOrlans Ecole Polytechnique de lUniversit dOrlans, Orlans Cedex, France
2
Universit dAvignon, Laboratoire dhydrogologie, rue Louis Pasteur Avignon, France
3
Laboratoire des Mcanismes et Transferts en Gologie, Universit Paul Sabatier IRD, avenue
Edouard Belin Toulouse, France
4
INRAP et UMR 8555 Centre dAnthropologie, ZA des Champs Pinsons, rue du ngoce,
Saint-Orens-de-Gameville, France
5
Parc Naturel Rgional des Grands Causses, Bd de lAyrolle, BP Millau Cedex, France
6
Laboratoire Souterrain du CNRS, Moulis, France
7
Agence de lEau Adour-Garonne, rue du Frtra Toulouse Cedex, France
ABSTRACT: A hydrogeological study was carried out between 2002 and 2004 for the Natural
Regional Parc of Grands Causses, France, on the western part of the Causse de Sauveterre (includ-
ing Causse de Sverac, Causse du Massegros, totalling 400km
2
with 23 communities and about
100 springs). The objectives of this study were to assess the water resources within major karst
systems and to characterize the sensitivity of the water resources to human activities. A multidisci-
plinary study of the karst hydrogeology was carried out (including geology, geomorphology, geo-
chemistry and hydrology) on 18 major springs that were equipped with gauging stations. Water
catchments were delineated based on geological, hydrogeological, hydrochemical and tracer tests.
The results will allow the implementation of the statutory protection perimeters for drinking water
resources.
1 INTRODUCTION
The Parc naturel regional des Grands Causses (Natural Regional Park of the Grands
Causses) endeavours to understand and look after its water resources as part of its mission
of protection, preservation and management of its natural heritage. After studying the
Nord Larzac and the Causse Rouge, the Park has begun a new wide-ranging hydrogeo-
logical study of the West of the Causse de Sauveterre, which is part of the long-term plan
to understand and protect. The water authority, Agence de lEau Adour-Garonne, the
Midi-Pyrnes region and the French government have given financial support to the two
main objectives of the study: general knowledge of the resource and obtaining the infor-
mation necessary to delineate the statutory catchment protection zones of some of the
springs in the studied area.
A multidisciplinary characterization of the structure and functioning of the karstic
aquifers has been carried out. This method (Marsaud, 1996) relies on:
(i) the identification of the geological and geomorphologic features, as well as those of
the main catchment areas from the hydrological study, the hydrochemical character-
ization and the tracer tests,
(ii) characterization of the karstic systems, in terms of functioning and resources, using
hydrological methods (notably the analysis of recession graphs), and
(iii) determination of the effects of human activities on the resource and of the protective
measures that need to be taken.
1.1 The study area
The study zone belongs to the Grands Causses in the southern part of the Hercynian
Massif Central. It is characterized by large tabular stretches of secondary limestone, cut
by narrow undulating valleys, which have been carved out by the rivers Tarn, Jonte and
Dourbie. These rivers cut the region into four main units: the Causse de Sauveterre, the
Causse Mjean, the Causse Noir and the Causse du Larzac (Fig. 1). The Causse de
Sauveterre is bordered by the Lot Valley to the north, 200 to 400m deep, and by the Tarn
gorges to the southeast, which it dominates from a height of 500m, over a distance of
60km. The Causse de Sauveterre (600km
2
) is crescent shaped. The western part of the
Causse (400km
2
) consists of the Causse de Svrac, the Causse du Massegros and its
Liassic fore-causses. The study carried out between 2002 and 2004 by the Parc des Grands
Causses will allow local communities to identify the protection zones for their water
sources.
The studied area contains approximately 100 springs (Fig. 2), 18 of which are equipped
with flow recorders. Selective measurements were carried out in wet and dry conditions at
the others.
1.2 The aquifers and their structure
The aquifers are divided into several karstic systems, which receive rainfall recharge
(unary systems, Marsaud, 1996) and have fixed head outlets along the edge of the study
zone represented by the rivers Aveyron, Tarn and Lot. The hydro-stratigraphy (Fig. 3) of
the Grands Causses includes two aquifers separated by Liassic marls. The lower unit rests
on the Triasic and an impermeable schist and granite bedrock. The lower aquifer is the
Liassic limestone and dolomite formations of the Hettangian-Sinemurian. Four of the
springs discharge from this aquifer. The impermeable supra-Liassic marl allows limited
hydraulic contact between the two aquifers wherever it thins or is locally faulted. The
upper aquifer is situated in the Aalenian to Portlandian limestone and dolomite formations.
It is drained by outlets at the contact with marls (perched karstic aquifer) and along the
main valleys.
Application to the Causse de Sauveterre (south of France) 319
The Causse de Sauveterre is divided into units by various types of faults (Fig. 4):
(i) steep reverse NW-SE trending faults belonging to the Pyrenean cycle: the Palanges
fault which creates a structural contact between the Hercynian basement and Liassic
layers, and the Vignes fault which puts into contact the Liassic and Dogger layers,
(ii) the Hercynain thrust, a south-facing thrust of the allochtonous Lvzou unit,
(iii) normal NW-SE to NNE-SSW trending faults linking the Dogger to the Permian-
Carboniferous in the Sermel basin.
Figure 1. Main units of the Grands Causses and the Parcs territory.
2 DETERMINATION OF THE CATCHMENT AREAS
2.1 Estimate of the surface area of the catchment zones
The water levels recorded at the hydrometric stations relate to the discharge from the main
springs. The flow rates of the secondary springs were measured in two field campaigns
reflecting high and low water levels. The capture zones were estimated for each spring
using the water budget method (Blavoux, 1991). In order to discount change in storage, the
budget was calculated for two years (20022004); evapotranspiration was calculated with
Figure 2. Location of the main hydraulic outlets.
Figure 3. Typical hydro-stratigraphy for the Grands Causses.
Turcs formula, using the observed rainfall and annual average temperature. The budget
indicated an annual flow rate of between 12 and 16l/s/km
2
.
2.2 Position of the catchment area
Hydrographic or topographic basins
A first analysis of the topographic catchment areas was made, based on the digital eleva-
tion model using the software HYDROKIT-STRATEGIS, i.e. no account was made of
possible cross catchment flow at this stage.
Two large basins, the Serre (37km
2
) and Cayrac (59km
2
) do not easily correlate with
the observed spring flows because the Tarn River also drains these basins. However, the
Tantayrou and Buzareingues basin springs in the Liassic strata have overlarge spring
flows, which imply some groundwater import form outside. Geochemical analyses con-
firm these conclusions.
For all the other springs, the topographic basins match the hydrogeological basins.
Tracer tests
More than 30 tracer tests were carried out with fluorescent dies, and these to help identify
the catchment areas where tracers were recovered.
Figure 4. Geology of the Causse de Svrac and the Causse du Massegros.
Figure 5. Main topographic basins.
Figure 6. Map of the catchment areas and tracer tests (Beldoire-Fontmaure and Rouveyrol
systems).
The positive results of many of the tests in the eastern part of the area suggest that water
is flowing from the Serre and the Cayrac basins towards the Tarn River.
Results of the geochemical tests
A sampling campaign was carried out in June 2003. The analyses from 90 spring sources
help characterize natural geological chemical markers (Bakalowicz, 1979; Mudry, 1987;
Plagnes, 1997).
The Ca/Mg molar ratio is an indicator of the percentage of dolomite in the system. If it
is between 1 and 2, it indicates that the ground is essentially dolomitic; when it is higher
than 8, it is essentially limestone, while two intermediate groups show the predominance
of limestone (between 5 and 7) and a mix of limestone and dolomite (between 2.5 and 4).
The dolomite is largely Liassic, apart from the southeastern end of the Causse du
Massegros, where Kimmeridgian dolomite crops out, and along the edge of the Causse de
Sauveterre, where there is brown Bajocian dolomite.
Silica concentrations greater than 12mg/l reflect the pre-Triassic silicates in the
Levezou massif to the south and to the west of the area and the superficial clay formations
transported in the depression from the outcrops of Bajocian shale limestone. (Brussels,
2001). Thus, a significant correlation was found between nitrates (agriculture) and silica
(clayed soils).
Monthly monitoring of the Tantayrou and Buzareingues springs in the Liassic basin,
shows that Buzareingues has a high bicarbonate concentration of 340mg/l, whereas
Tantayrou like the Dogger springs has a bicarbonate concentration of 281mg/l, similar to
that for Cayrac. If Buzareingues spring is supplied by the Hettangian dolomite, which out-
crops towards the West, Tantayrou is in confined conditions and could be supplied by
upwelling from the Dogger, the outcrops of which are to the east and north at the edge of
the Causse de Sauveterre (Fig. 7). The spring with the highest and unvarying temperature
is Tantayrou. Its temperature is approximately 1.8C higher than that of springs at the same
elevation, possibly reflecting deeper circulation (the proportion of deep groundwater esti-
mated at around 75% for Tantayrou and only 40% for Buzareingues).
Structural geologic features
The fractures within the limestone create its secondary permeability (Eraso, 1985). The spa-
tial distribution of the factures is anisotropic. A micro-tectonic analysis of available sections
was carried out to identify tectonic markers (faults, veins or stylolites). Their orientation was
measured revealing two groups, one nearly north-south, corresponding to vertical planes, the
other approximately east-west, corresponding to vertical or oblique planes (Fig. 8).
These form two preferred flow directions and have an influence on the spring locations.
The drainage planes of the springs along the edge of the Tarn River, to the East, are oriented
Figure 7. North-south trending geological cross-section near Tantayrou.
from west to east and are sub-vertical, allowing water to flow from the Causse de Sverac
and the Causse du Massegros to the Tarn River. The flow favoured by the Tarn gorges fol-
lows lithological discontinuities (limestone interspersed with marl) and the aquifer may be
multi-layered. The flow at the Fontmaure spring rises and falls with the Earth tides indicat-
ing a confined aquifer.
3 STRUCTURE AND FUNCTIONING OF THE AQUIFER
3.1 Results of the discharge study
The discharge of the 18 main springs was measured over at least two years (20022004).
The ratio of maximum to minimum daily flow indicates karstification (Mangin, 1971).
The lowest ratio is obtained for Cayrac (3) and thus shows a low level of karstification
and/or a high degree of regulation. The springs Mas de Lafon, Tantayrou and Esparses have
similar characteristics. However, the ratio is higher than 100 for the Verlenque (100),
Lestang, Serre, Roquaizou and Courtinaux (339) springs, indicating extreme karstification.
The rainfall-discharge relationship (Mangin, 1984) shows the system response, which
can be of two types: (i) for fast runoff, found in highly karstified aquifers, it is flashy,
(ii) for less karstified aquifers, it is wider with a lower maximum value (Fig. 9).
Analyses of the discharge from the 18 springs show a good response to rainfall, suggest-
ing high karstification, but also indicating some storage potential within the aquifer. There
are two groups, based on shape of the flow duration curve (Labat et al., 2002)(Table 1).
The first group (A or majority of A) draws from relatively well karstified, whereas the
second group (B) would be less so. The difference between the groups is caused by the
geo-morphological history where the second group is as karstified but some conduits are
now partially blocked.
3.2 Information obtained by tracer tests
The flow velocities from the residence-time distribution (Fig. 10) do not exceed 100 m/h.
Figure 8. Stereoplot of the drainage planes of the Causse de Sverac.
The highest velocities were measured from functional swallets (Lepiller et Mondain,
1986). The results are consistent for the Beldoire system (Fig. 6) although the hydrologi-
cal conditions varied (very low to medium water levels) and the injection points were dif-
ferent. This reflects the typical velocities indicating a drainage network which is
moderately well developed and contains water in storage.
4 ASSESSMENT OF THE GROUNDWATER RESOURCE
4.1 Estimate of the reserves
The calculated dynamic volumes (Mangin, 1970) show volumes greater than 1.5Mm
3
for Beldoire and Cayrac; volumes of between 0.5Mm
3
and 1.5Mm
3
for Bastide Lestang,
Mayrinhac, Sgala and Verlenque and smaller volumes, less than 0.5Mm
3
for the other springs.
In order to analyse the relationship of flow with infiltration the 2004 recession graphs
were interpreted using the classification of Mangin (1975), based on k and I, where k is the
ability of the system to stock rainwater and to return it. The greater this regulation capacity
is, the more k tends towards 1, which represents porous aquifers where the dynamic volume
is practically equal to the transit volume. For karstic systems, k is generally smaller than 0.5
and when k equals 0, the regulating capacity is negligible, and i is delayed infiltration.
A high value of i shows slow infiltration or the influence of an epikarstic aquifer. A low
value of i shows the rapid transit of the water towards the water-filled zone.
domain n
o
4 corresponds to karstic systems that are very karstified downstream, in
which the speleological networks are well developed; Courtinaux and Roquaizou
belong to this domain;
domain n
o
2 corresponds to systems which are more karstified upstream than down-
stream, with delayed water supply due to non karstic formations; the Duc, Sgala,
Serre and Lestang springs belong to this domain;
domain n
o
1 corresponds to complex, generally large, systems whose structure consists
of many smaller systems; Esparses, Mayrinhac and Buzareingues belong to this domain.
However, for three springs (Verlenque, Bastide and Cayrac), the coefficient k is higher
than 0.5. The springs are close to each other and their water is sourced from the northwest
of the Causse du Massegros. Their systems act as porous formations, with the coefficient
Figure 9. Examples of rainfall/discharge plots.
T
a
b
l
e

1
.
C
l
a
s
s
i
f
i
c
a
t
i
o
n

o
f

t
h
e

d
i
f
f
e
r
e
n
t

s
o
u
r
c
e
s

o
b
t
a
i
n
e
d

f
r
o
m

t
h
e

a
n
a
l
y
s
e
s
.
C
r
o
s
s

D
a
i
l
y

c
r
o
s
s

D
i
s
t
r
i
b
u
t
i
o
n

o
f

C
a
t
e
g
o
r
i
s
e
d

f
l
o
w

r
a
t
e
s
H
o
u
r
l
y

s
p
e
c
t
r
u
m
c
o
r
r
e
l
o
g
r
a
m

1
/
2

h
c
o
r
r
e
l
o
g
r
a
m
t
h
e

c
o
m
p
o
n
e
n
t
s
A

B

A

B

A

B

A

B

A

B

S
h
a
r
p

W
i
d
e

S
h
a
r
p

W
i
d
e

C
o
n
v
e
x
C
o
n
c
a
v
e
L
o
g
a
r
i
t
h
m
i
c

A
r
i
t
h
m
e
t
i
c

S
e
v
e
r
a
l

S
i
n
g
l
e
p
e
a
k
p
e
a
k
p
e
a
k
p
e
a
k
l
a
w
l
a
w
p
e
a
k
s
p
e
a
k
B
a
s
t
i
d
e
o
o
o
o
B
e
l
d
o
i
r
e
x
x
x
x
o
B
u
z
a
r
e
i
n
g
u
e
s
x
x
x
x
x
C
a
y
r
a
c
x
x
x
x
x
C
o
u
r
t
i
n
a
u
x
o
o
o
o
o
D
u
c
x
x
x
x
o
E
s
p
a
r
s
e
s
x
x
x
x
x
F
o
n
t
m
a
u
r
e
x
x
x
o
o
G
l
a
s
s
a
c
o
o
x
x
x
L
e
s
t
a
n
g
o
o
o
o
o
M
a
s

d
e

L
a
f
o
n
x
x
x
o
x
M
a
y
r
i
n
h
a
c
x
x
x
o
x
R
o
q
u
a
i
z
o
u
o
x
o
o
o
R
o
u
v
e
y
r
o
l
o
o
o
o
o
S
g
a
l
a
o
o
o
o
o
S
e
r
r
e
o
o
o
o
o
T
a
n
t
a
y
r
o
u
o
o
o
o
o
V
e
r
l
e
n
q
u
e
o
o
o
o
o
i increasing from Verlenque towards Cayrac. These results must be considered carefully as
they are only based on the 2004 data.
4.2 Anthropogenic influences
The chloride concentrations above 4mg/l, which is the natural value for effective rainfall,
reflect some human input. Chloride concentrations are, however, generally low, and only
rarely exceeds 12mg/l, with five samples above 25mg/l. These higher values were generally
near roads and/or towns. Over half of the samples (51/92) contain less than 6mg/l of NO
3
which is the threshold above which the impact of human activities is most noticeable, and in
just five springs it is greater than 25mg-NO
3
/l.
Figure 11. Position of the karstic systems in the Mangin (1975) classification.
Figure 10. Distribution of apparent velocities (m/h) obtained with the tracer tests.
The highest nitrate concentrations are found in the Liassic deposits in the Aveyron and
Serre valleys, which are the most urbanized and cultivated areas.
The correlation between chloride and nitrate and their variation with time (Figs. 12, 13
and 14), illustrates how the aquifer reacts to pollutants and their transport within the
aquifer (Mudry et al., 1995).
For example, in the case of chloride (Fig.13), some springs show constant and acceptable
contents over the whole hydrological cycle. Others show small but significant increases in
October, November and January during small rises in the water level, thus proving their
vulnerability. Conversely, two polluted springs, Courtinaux and Rouveyrol, experience a
decrease in concentration suggesting a dilution of a polluted flow permanently entering the
system. The same reactions are observed for the nitrate.
5 VULNERABILITY AND PROTECTIVE MEASURES
An inventory of potentially polluting activities has been made transport, agriculture
urban etc. The vulnerability map of the karst groundwaters depends on:
soil cover;
presence of preferential underground flow;
Figure 12. Chloride and nitrate correlation (in mg.l
1
).
presence of depressions, dolines and rock rills which facilitate rapid infiltration;
presence of swallets;
land use.
A vulnerability and hazard map was created for the eastern part of the study area
(Fig. 15).
Based on this map, suggestions for catchment protection zones were made. This
suggests that the catchments should be protected. The vulnerability classes range
from 0 to 11:
level 11 corresponds to a high vulnerability from swallets which offer a rapid infiltra-
tion route;
levels 3 to 10 are moderately vulnerable;
levels 0, 1 and 2 are weakly vulnerable but susceptible to diffuse pollution.
Figure 14. Variation in the nitrate content (in mg.l
1
NO
3
) from 2002 to 2003.
Figure 13. Variation in the chloride content (in mg.l
1
) from 2002 to 2003.
6 CONCLUSIONS
The surface areas of the catchment zones were obtained with the water budget method (calcu-
lated with Turcs formula) and measuring the outlet flow rates over the period 20022004.
They were then on the digital elevation model, and adjusted with information obtained from
tracer tests and groundwater chemistry. The catchment areas of the springs on the western side
correspond well with the topographic basins, whereas for those on the eastern side do not. This
may be due to underground flow from adjacent catchments, originally directed towards the
Aveyron and Serre rivers. Analyses of the discharge from 18 springs provides supporting
information on provenance. The hydrodynamic and hydrochemical behaviour have a clear
response to rainfall in a typically flashy karstic environment, although there is evidence also
of some storage. Geomorphology was used to assess groundwater vulnerability integrating all
the results regarding the functioning of the catchment areas and the inventory of potentially
polluting activities. This multi-disciplinary study (Parc Naturel Rgional, 2006), will assist
towards managing the water resources. The most important action that should now be taken is
the creation of statutory catchment protection areas.
Figure 15. Part of the vulnerability and hazard map.
REFERENCES
Bakalowicz M. (1979) Contribution de la gochimie des eaux la connaissance de laquifre kars-
tique et de la karstification (Contribution of geochemistry of water to the knowledge of karstic
aquifers and karstification). Doctorat es Sciences, University Pierre et Marie Curie, Paris, France.
Blavoux B., Mudry J., Puig JM. (1991/1992) Bilan, fonctionnement et protection du systme kars-
tique de la Fontaine de Vaucluse (sud-est de la France)(Water budget, functioning and protection of
the Fontaine de Vaucluse karst system (southeastern France), Geodinamica Acta, 5, 3, 153172.
Bruxelles L. (2001) Dpts et altrites du Larzac central: Causses de lHospitalet et de Campestre
(Aveyron, Gard, Hrault), Evolution morphognique, consquences gologiques et implications
pour lamnagement. PhD thesis, University of Provence, BRGM document n
o
304, Editions
BRGM, Orlans 2004.
Eraso A. (1985) Methodo de predicction de las direcciones principales de drenaje en el karst.
(Method of predicting the main drainage directions in karst) KOBIE (Serie Ciencias Naturales),
Bilbao, n
o
XV, 122 p.
Labat D., Abadou R., Mangin A. (2002) Analyse multirsolution croise de pluies et dbits de
sources karstiques, C.R.Geoscience, 334, 176195.
Lepiller M. et Mondain Ph. (1986), Les traages artificiels en hydrogologie karstique, Hydrogeologie,
BRGM, 1, 1986.
Mangin A. (1970) Contribution ltude daquifres karstiques partir de lanalyse des courbes de
dcrue et de tarissement (Contribution to karstic aquifers study from the decay and recession
curve analysis) Annales de Splologie, 25, 3, 581609.
Mangin A. (1971) Etude des dbits classs dexutoires karstiques portant sur un cycle hydrologique
(Discharge duration studies of karstic springs during a hydrological cycle) Annales de
Splologie, 26, 2, 283329.
Mangin A. (1975) Contribution ltude hydrodynamique des aquifres karstiques (Contribution to
the hydrodynamic study of karstic aquifers). Doctorat es Sciences, Dijon University; Annales de
Spelologie, 1974, 29, 3, 283332; 1974, 29, 4, 495601; 1975, 30, 1, 21124.
Mangin A. (1984) Pour une meilleure connaissance des systmes hydrologiques partir des analy-
ses corrlatoires et spectrales. J. of Hydrology, 67, 2543.
Marsaud B. (1996) Structure et fonctionnement de la zone noye des karsts partir des rsultats
exprimentaux (Structure and behavior of the saturated zone of karst aquifers from experimental
results) PhD Thesis, University Paris XI Orsay, BRGM document n
o
268, 1997, Editions BRGM.
Mudry J. (1987) Apport du traage physico-chimique naturel la connaissance hydrodynamique des
aquifres carbonats (Contribution of the physical and chemical natural tracers to the hydrody-
namic knowledge of carbonated aquifers). PhD Thesis, Franche Comt University, Besanon,
France.
Mudry J., Lastennet R., Puig JM., Blavoux B. (1995) Use of natural tracing to understand how an
aquifer works. Basic and applied hydrogeological research in French karstic areas, European
Commission, Action COST 65: Hydrogeological aspects of groundwater protection in karstic
areas, N. Crampon & M. Bakalowicz Editors, 2735.
Parc Naturel Rgional des Grands Causses (2006). Etude hydrogologique de la partie ouest du
Causse de Sauveterre. Rapport final dtude, 279 p., 8 cartes et annexes.
Plagnes V. (1997) Structure et fonctionnement des aquifres karstiques, caractrisation par la
gochimie des eaux (Structure and behavior of karstic aquifers, characterisation by hydrogeo-
chemistry) PhD. Thesis, Montpellier University; BRGM document n
o
293, 2000, Editions BRGM.
CHAPTER 25
Hydrogeological characterisation of the Oxfordian
limestone at the Bure URL
Martin Cruchaudet
1
, Jacques Delay
1
& Marc Distinguin
1,2
1
ANDRA Laboratoire de Recherche Souterrain de Meuse/Haute-Marne, BURE France
2
now AREVA NC, BP, VELIZY Cedex France
ABSTRACT: Andra, The French National Agency for Radioactive Waste Management has built
an Underground Research Laboratory (URL) for studying the feasibility of radioactive waste dis-
posal in Jurassic-age Callovo-Oxfordian argillaceous rock. One of the objectives of the many inves-
tigations performed since 1994, was to characterise the hydrogeological parameters of the
formations surrounding the argillaceous rock, especially the Oxfordian and Dogger limestone.
This article focuses on Hydrogeological investigations performed in the Oxfordian limestone at the
URL scale. These consist of:
Initial measurements in deep boreholes, including hydrogeological testing with and without pack-
ers, piezometric head measurements, geophysical logging and fluid conductivity logging;
Interference tests between the two access shafts to the laboratory and a network of four monitor-
ing boreholes.
1 INTRODUCTION
The French National Agency for Radioactive Waste Management is in charge of the inves-
tigation of potential reversible nuclear waste disposal in deep geological formations, using
underground research laboratories. Between 1994 and 1999, Andra performed initial
investigations in the south-west part of Meuse and the north-east part of Haute-Marne, two
dpartements of eastern France, to assess the feasibility of building a laboratory in the
Callovo-Oxfordian argillaceous rock, a layer located at a depth between 400 and 600m.
The objectives of the preliminary investigations were to verify the existence and the char-
acteristics of the host formation and to estimate the parameters of the geological formation
below and above it, particularly the hydrogeological parameters of the Oxfordian and
Dogger limestone. Knowledge of the hydrogeological properties of the Oxfordian lime-
stone was required for the safety assessment of an underground repository and to construct
groundwater flow identifying potential recharge and discharge areas. On 3 August 1999,
the French government authorised Andra to implement and operate the first French
Underground Research Laboratory (URL) at a site straddling the Meuse and the Haute-
Marne dpartements, near the village of Bure (Meuse, France, Figure 1).
From 1999 to 2005, the objectives of the URL were the in situ characterisation of the
physical and chemical properties of the Callovo-Oxfordian argillaceous rock (Delay and
Cruchaudet, 2004), and the characterisation of the physical, chemical and hydrogeological
properties of the rock above the Callovo-Oxfordian (Delay and Distinguin, 2004), the
Tithonian limestone, the Kimmeridgian clay and the Oxfordian limestone (Figure 2).
The Oxfordian limestone is 292m thick at the URL site, and is situated at a depth
between 125 and 417m in the main access shaft of the URL. The hydrogeological object-
ive for the Oxfordian limestone was to obtain undisturbed hydraulic head measurements
and to estimate hydrogeological properties for the rock (i.e. permeability and specific stor-
age coefficient). On a regional scale, the permeability for the Oxfordian limestone was
estimated between 5 10
10
and 3 10
8
m/s. These estimates were confirmed by measure-
ments obtained in three boreholes during the 19941999 survey: EST103 located on the
Figure 1. Location of the URL Meuse / Haute-Marne.
Figure 2. Simplified stratigraphy for the URL.
URL site, HTM102 located 2.8km south of the URL and MSE101 located 12.1km north
of the URL. The purpose of this paper is to give a detailed overview of the methodology
applied to characterise the Oxfordian limestone at the URL scale and to give some results.
2 GEOLOGICAL INVESTIGATIONS AND GLOBAL OVERVIEW OF THE
HYDROGEOLOGICAL PROGRAMME
The geological programme was based on a detailed 3D seismic campaign, which was
undertaken in a 4km
2
area centred around the URL. A network of boreholes drilled from
the surface to the Barrois, Kimmeridgian, Oxfordian and Callovo-Oxfordian formations
and detailed geological and geophysical logging was carried out. In five of these bore-
holes, specific measurements were made to obtain information about the Oxfordian lime-
stone (location on Figure 3): EST103 drilled during the 19941999 survey, EST201 and
EST203 drilled in 2000, EST204 and EST205 drilled in 2000 on the main shaft axis and
on the auxiliary shaft axis. These boreholes penetrate the full Oxfordian limestone
sequence.
The interpretation of all these measurements gave a precise representation of the geol-
ogy of the Oxfordian limestone in terms of lithology and stratigraphy. For hydrogeological
purposes, each sedimentary unit of at least 1m thick was defined in term of porosity,
lithology and horizontal extent at the URL scale. Seven main porous levels, corresponding
to zones of at least 20% of total porosity have been identified with geophysical logging.
These porous levels, named Hp1 to HP7, were found in all five boreholes (Figure 4)
described above. The total porosity was obtained from a magnetic resonance tool (CMR:
Combinable Magnetic Resonance).
The hydrogeological programme was based first on a detailed hydrogeological charac-
terisation in the boreholes EST201, EST203, EST204 and EST205, and then on an inter-
ference test between the two shafts and the four observation boreholes EST103, EST104,
EST201 and EST203. Data obtained in EST103 during the 19941999 survey are also
taken into account in this paper.
Hydrogeological characterisation of the Oxfordian limestone 335
Figure 3. Network of boreholes used to characterise the Oxfordian limestone on the URL.
3 CHARACTERISATION OF THE HYDRAULIC PROPERTIES IN BOREHOLES
3.1 Objectives, testing program and main characteristics of the Oxfordian limestone in
the boreholes
Hydrogeological characterisation was carried out in EST103, EST201, EST203, EST204
and EST205. Pumping, packer tests and a pressure monitoring were done in EST103
before 1999.
From the firsts results obtained in this borehole, the objectives for the four other bore-
holes were to verify these measurements and quantify the extent of these hydraulic prop-
erties at the URL scale with:
The measurement of the hydraulic heads before shaft sinking,
The characterisation of the overall transmissivity in the four boreholes,
The location of the inflow zones,
The measurement of the transmissivity and the hydraulic head for the main inflow
zones.
The methodology applied in these four boreholes used:
Test pumping associated with fluid conductivity logging,
Packer tests at the main inflow zones,
Long-term monitoring of the heads in the boreholes using multi-packers.

All the boreholes were air-drilled to avoid any contamination of the rock with drilling mud
as samples of the water formation in the Oxfordian were taken during drilling. In the five
boreholes (and EST104 located on the same platform as EST103), the elevation of the
ground surface is between 365m ASL (EST203) and 368m ASL (EST201). The thickness
of the Oxfordian limestone for the stratigraphic units C3b to K1 (Figure 4) is about 275m
Figure 4. Porous level location in the Oxfordian on URL site.
in these boreholes (0.5m). The depth for the top of the formation is between 125 and
130m and the depth for the bottom of the formation is between 399 and 405m.
3.2 Test pumping and fluid conductivity logging
Test pumping is classically used in hydrogeology to estimate the broad hydraulic proper-
ties of the rock (permeability, transmissivity). The test consists of an initial pressure recov-
ery followed by a pumping generally performed at a constant rate. This pumping phase
ideally lasts until the water level is more or less stabilised in the borehole. It is followed by
a pressure recovery phase. The analysis of both the pumping phase and the final pressure
recovery phase are classically performed with specific software which use the laws of
hydrodynamics and numerical inversion simulations. For the interpretations in EST103, a
GTFM simulator was used (see Pickens et al., 1987). For interpretations in EST201,
EST203, EST204 and EST205, a MULTISIM simulator was used (see Tauzin and Johns
(1997) for the description of the software, and Pickens et al. (1987) for the theoretical
model). Fluid conductivity logging (Tsang et al., 1990), performed in EST201, EST203,
EST204 and EST205, consisted of replacing the natural water filling the borehole with
water of contrasting electrical conductivity. The inflow of natural water to the hole is mon-
itored using repeated electrical conductivity logging. The electrical conductivity logging
shows peaks at the water inflow points. These peaks increase and move along the hole
depending on the inflow rates. This method allowed main inflow zones to be located and
was ideal for detecting low inflow rates which could not be detected with vertical flowmeter
logging (spinner). Furthermore, the hydraulic interpretation of the pumping tests associ-
ated with the analysis of fluid electrical conductivity logging in boreholes allowed the esti-
mation of inflow rates and transmissivity for each main inflow zone (Low et al., 1984).
Technically, the tools that were introduced in the borehole during this operation were first
a multi-parameter geochemical probe, containing a temperature and a conductivity sensor,
then a pump, set in the borehole at a depth between 130m and 150m and finally a pres-
sure sensor located above the pump to monitor the water level in the borehole. Table 1 indi-
cates the main characteristics for each pumping test with fluid conductivity logging in the
boreholes EST201, EST203, EST204 and EST205.
The main results obtained in EST201 and EST203 for production zones are synthesised
in Figure 5. The main inflow zones for these two boreholes are mainly located in the seven
porous levels identified from conductivity logging. In EST201, of 24 inflow zones identi-
fied, 19 were located in the porous levels, corresponding to a 91.6% of the overall produc-
tion. In EST203, for the same number of inflow zones identified, 19 were also located in
the porous levels, corresponding to 89.5% of the total production. The results show some
heterogeneity in terms of vertical distribution of the inflow zones: in EST201, the lower
part of the Oxfordian limestone (including Hp1 to Hp4) produces about 70% of the over-
all inflow whereas the same zone yields only 55% of the total inflow in EST203. The
results obtained in EST204 are similar to the results obtained in EST201: 90% of the
inflow zones located in porous levels with 75% of the overall flow coming from the lower
part of the Oxfordian limestone. Some heterogeneity occurs between these two boreholes
in terms of the number of the inflow zones (only 11 in EST204) and their exact location
and production rate. Similar results were obtained from EST203 and EST205: the latter
borehole with 47% of the inflow zones in the lower part of the Oxfordian limestone with
20 inflow zones located in the porous levels. Heterogeneity between the two boreholes is
Table 1. Characteristics of the pumping tests with fluid conductivity logging.
Borehole Tested zone Pumping characteristics Number of logs
EST201 Oxfordian to 410m depth Rate: 3.1l/min 9
diameter: 6
1
4 inches Duration: 33.5h
geological units*: K1 to C3a max. drawdown: 59m
EST203 Oxfordian to 410m depth Rate: 3.3l/min 10
diameter: 6
1
4 inches Duration: 42h
geological units*: K1 to C3a max. drawdown: 66m
EST204 Oxfordian and Callovo-Oxfordian Rate: 1.8l/min 8
to 508m depth diameter: Duration: 32h
6
1
4 inches geological max. drawdown: 52m
units*: K1 to C2a
EST205 Oxfordian and Callovo-Oxfordian Rate: 2.1l/min 9
to 422.2m depth diameter: Duration: 29h
6
1
4 inches max. drawdown: 66m
geological units*: K2d to C2d
* see Figure 4
-40
-20
0
20
40
60
80
100
120
140
160
180
200
220
E
l
e
v
a
t
i
o
n

(
m

A
S
L
)
2 3 4 5
Electric conductivity
(mS/cm at 25C)
0.0 0.2 0.4 0.6 0.8
0 20 40 60 80 100
Cumulated amount
of global inflow (%)
0 20 40 60 80 100
Cumulated amount
of global inflow (%)
0 0.2 0.4 0.6 0.8
2 3 4 5
Electric conductivity
(mS/cm at 25C)
logging n1
logging n3
logging n5
logging n7
logging n9
Total inflow : 3,3 L/min
Maximum drawdown : 66 m
L
o
w
e
r

g
r
o
u
p
=

4
5
%

i
n
f
l
o
w
U
p
p
e
r

g
r
o
u
p
=

5
5
%

i
n
f
l
o
w
0 0.2 0.4 0.6 0.8
Inflow zones
(L/min)
Total inflow : 3,2 L/min
maximum drawdown : 59 m
0 20 40 60 80 100 2 3 4 5
0 0.2 0.4 0.6 0.8
Inflow zones
(L/min) 0 20 40 60 80 100 2 3 4 5
U
p
p
e
r

g
r
o
u
p
=

3
0
%

i
n
f
l
o
w
L
o
w
e
r

g
r
o
u
p
=

7
0
%

i
n
f
l
o
w
Hp7
Hp6
Hp5
Hp4
Hp3b
Hp3a
Hp2
Hp1b
Hp1a
EST203
EST201
Porous
levels
O
x
f
o
r
d
i
a
n

l
i
m
e
s
t
o
n
e
L
o
w
.

K
i
m
/
U
p
.

O
x
f
L
o
w
.
K
i
m
K1
L2c
L2b
L2a
L1b
L1a
C3b
C3a
~
~
~ ~
Figure 5. Location of the inflow zones in EST201 and EST203 (by J.M. Joubert, Antea).
indicated by a different number of inflow zones (31 for EST205), and a major inflow zone
located between Hp6 and Hp7 in EST205, which mainly accounts for the fact that only
81% of the overall inflow is located in the porous horizons in this borehole.
In EST201, EST203, EST204 and EST205, permeability and transmissivity estimates
for the 275m thick Oxfordian limestone were obtained from the hydraulic interpretation of
the pumping tests (Table 2, from Antea, 2001). Permeability variation obtained from these
boreholes is less than one order of magnitude, ranging from 5,6 10
10
to 4,1 10
9
m/s.
These low values confirm the first measurements obtained in EST103 (best estimate:
2 10
9
m/s). Individual transmissivity estimates obtained for each of the 90 inflow zones
from the hydraulic interpretation of the fluid conductivity loggings associated with the test
pumping range from 10
9
to 2,5 10
7
m
2
/s (thickness of the inflow zones: 18m, except
for two zones of respectively 12 and 18m thick).
In each of the four boreholes drilled in 2000, the best estimate for the global transmis-
sivity obtained from test pumping interpretation is within or very closed to the range
obtained for the total of the individual permeabilities of each inflow zone obtained from
fluid conductivity logging. The two methods are thus giving similar results, highlighting
the consistency of the data and their interpretation.
3.3 Packer test results and comparison with test pumping and fluid
conductivity results
After the first characterisation phase using test pumping associated with fluid conductiv-
ity logging, four to five packer tests were made in EST201 and EST203 to specifically
investigate inflow zones in each of these boreholes obtain further results on their hydraulic
properties. The methodology consisted of isolating a selected zone in a borehole between
two packers and performing hydraulic tests in the interval. The setup used for the tests con-
sisted of a test string comprising two packers isolating the tested interval. Between the two
packers, the tubing is open to the borehole wall. Above this, a shut in valve is set inside the
tubing to isolate or connect the interval to the upper part of the tubing and to the surface.
Each test follows a precise procedure:
The testing tool is set in the borehole, the packers are inflated. The valve inside the cas-
ing is left open.
The test interval is isolated by closing the valve inside the casing. Pressure and temper-
ature recovery is monitored for up to half a day in the interval.
Table 2. Hydraulic properties obtained from global pumping and fluid conductivity logging.
Borehole Global Total transmissivity Permeability (m/s)
2
transmissivity (m
2
/s)
1
for the inflow zones
(m
2
/s)
2
EST201 6,1 10
7
7,44 10
7
1,36 10
6
2 10
9
3,6 10
9
EST203 6,2 10
7
3,1 10
7
8,6 10
7
1,1 10
9
3,1 10
9
EST204 6,9 10
7
3,7 10
7
1,7 10
6
7,7 10
10
3,4 10
9
EST205 5,8 10
7
2,2 10
7
6,8 10
7
5,6 10
10
1,7 10
9
1
results from global pumping interpretation
2
results from fluid conductivity logging interpretation
The water level in the casing above the valve is adjusted to the required pressure. This
pressure is either higher or lower than the stabilised pressure in the interval.
A pulse test is applied by opening and closing the valve in the casing for a few seconds.
The pressure recovery is monitored for up to one day.
Several pulse tests at different pressures are successively applied, each followed by a
pressure recovery.
At the end of the tests, the valve is opened, the packers deflated and the testing tool
retrieved from the borehole or moved to another testing section.
This kind of test is suitable for permeabilities 10
8
m/s and was thus well suited for the
Oxfordian limestone. Pulse tests are interpreted with semi-logarithmic plots or logarithmic
plots of the pressure derivative. Software using both direct and inverse simulation of the
pressure derivative has been used. The GTFM simulator (Pickens et al., 1987) was used to
interpret the tests in EST201 and EST203. Five packer tests were done in EST201 to test
separately Hp1, Hp2, Hp4, Hp5, Hp7, and in borehole EST203, four packer tests were car-
ried out to test separately Hp7, Hp6 and an inflow zone located just under this porous level,
Hp4, Hp3. Table 3 shows the results obtained from the packer tests in borehole EST201
and EST203 (Antea, 2001). Whilst the transmissivity determined from test pumping was
fairly consistent in all the boreholes drilled at the URL site, the results obtained from
packer tests show some variability for the porous levels transmissivity.
Pressure measurements between, below and above the two packers during packer tests
have shown the existence of vertical communication between the porous levels Hp1 to
Hp4 in borehole EST201. Transmissivity estimates derived from each packer test are not,
therefore, representative of the tested zone. Packer tests performed in EST203 and on Hp5
in EST201 showed no communication with porous levels located lower down in the
Oxfordian. Permeability values obtained from these packer tests are probably overesti-
mates in some cases as the extent of the tested zone cannot be defined accurately.
However, the transmissivities obtained from packer tests are consistent with transmissivi-
ties obtained from fluid conductivity loggings (Figure 6), which underlines the accuracy of
the estimates and the complementarity of the two methods.
3.4 Pressure monitoring in EST103, EST104, EST201 and EST203
The first pressure monitoring in the Oxfordian limestone at the URL was done in EST103
and EST104 with a pressure sensor (location on Figure 3). These two boreholes were
drilled and equipped before 2000. The monitored zone in EST103 corresponds to the 65m
horizon located at the bottom of the Oxfordian limestone, partly including the porous level
Hp1 (Figure 4). In EST104, the monitored zone corresponds to the 105m horizon located
just above the zone monitored in EST103, and including porous levels from the top of Hp1
to Hp4 (Figure 4). The hydraulic head measured in these two boreholes stabilised at a con-
stant level of 304.5m a.s.l., with no seasonal variation. This and the fact that the top of the
Oxfordian limestone is located at about 240m a.s.l. indicate that this rock formation is a
captive aquifer in the vicinity of the URL. In 2000, Andra decided to install multi-packer
devices (Delouvrier and Delay, 2004) with a pressure gauge dedicated to pressure meas-
urement for each inter-packer zone. These were set up in EST103 and EST104, EST201,
EST203 (Figure 7). There were two objectives. First, the multi-packer system enabled us
to measure the separate hydraulic head in up to five groups of porous levels (Figure 7).
Then this network of boreholes was used to monitor the effect of drainage during and after
the two shafts had been excavated nearby (see paragraph 4). The multipacker system
enabled the interference for each porous level or group of porous levels to be monitored
separately. In each borehole, monitoring zones have been located to isolate one or two
porous levels. Boreholes EST201 and EST203, specially drilled and equipped for the
interference test, have identical pressure measurement equipment with which to monitor
Table 3. Transmissivity estimates from packer tests in borehole EST201 and EST203.
Porous Results in EST201 Transmissivity for EST203 (m
2
/s)
level tested
Thickness of T (m
2
/s) Thickness of T (m
2
/s)
zone zone
Hp7 9.91 4,6 10
8
to 9,8 10
8
12.94 2,2 10
7
to 5,3 10
7
Hp6 12.94 8,0 10
8
to 1,4 10
7
Hp5 27.79 3,1 10
8
to 5,1 10
8

Hp4 9.91 2,0 10
7
to 2,2 10
7
12.94 3,6 10
8
to 5,2 10
8
Hp3 27.79 2,5 10
7
to 5,6 10
7
12.94 2,2 10
7
to 7,2 10
7
Hp2 27.79 9,2 10
8
to 1,5 10
7

Hp7
Hp6
Hp5
Hp4
Hp3b
Hp3a
Hp2
Hp1b
Hp1a
-20
0
20
40
60
80
100
120
140
160
180
200
220
sx
E
l
e
v
a
t
i
o
n
(
m
A
S
L
)
L
2
c
L
2
b
O
x
f
o
r
d
i
a
n

l
i
m
e
s
t
o
n
e
L
o
w
.
K
im
.
L
o
w
.
K
i m
.
/
u
p
.
O
x
f
S
t
r
a
t
i
g
r
a
p
h
y
L
i
t
h
o
l
o
g
y
T
K
2
0
0
0
-
0
4
T
K
2
0
0
0
-
0
3
T
K
2
0
0
0
-
0
1
T
K
2
0
0
0
-
0
5
T
K
2
0
0
0
-
0
5
T
K
2
0
0
0
-
0
4
T
K
2
0
0
0
-
0
2 T
K
2
0
0
0
-
0
3
T
K
2
0
0
0
-
0
2
T
K
2
0
0
0
-
0
2
L
1
b
C
3
b
L
1
a
EST203
-20
0
20
40
60
80
100
120
140
160
180
200
220
1.E-10 1.E-09 1.E-08 1.E-07 1.E-06
T(m/s)
E
le
v
a
ti
o
n

(
m
A
S
L
)
EST201
-20
0
20
40
60
80
100
120
140
160
180
200
220
1.E-09 1.E-08 1.E-07 1.E-06
T(m/s)
E
l
e
v
a
t
i
o
n
(
m
A
S
L
)
Legend
Porous level
Transmissivities
of water inflow zones
obtained from
fluid conductivity logging
Transmissivities from
packer tests
Total of transmissivities
of the water inflow zones
located in the intervals
tested by packer tests
range
for transmissivities
(sensitivity analysis)
range for transmissivity
upper value lower value
thickness of the tested zone
Figure 6. Transmissivities from packer tests and from fluid conductivity logging (by J.M. Joubert,
Antea).
the porous levels Hp1-Hp2, Hp3-Hp4, Hp5, Hp6 and Hp7 (Figure 7). As boreholes
EST103 and EST104 were already partly cased when the interference tests were designed,
only the porous levels Hp1 to Hp4 were monitored in these two boreholes.
The initial pressure monitoring before the shaft excavation began gave very similar
hydraulic head measurements in the four boreholes (504.5 to 505.5m ASL) and indicated
no vertical variations of hydraulic head in any of the boreholes.
4 INTERFERENCE TEST BETWEEN SHAFTS AND OBSERVATION
BOREHOLES
4.1 Experimental concept and devices
As the two access shafts were designed to drain the Oxfordian limestone, an interference
test was carried out at a larger scale by monitoring the outflow from the shafts and the
pressures in the network of observation boreholes (see Distinguin, 2005). The main shaft
and the auxiliary shaft have final diameters of 5 and 4m respectively. The main purposes
of the interference test were:
to determine the vertical connexions between the main inflow zones,
to evaluate a potential anisotropy amongst porous levels at the same scale,
to get an estimate of hydraulic parameters for the main inflow zones in the Oxfordian
limestone at the URL scale.
Figure 7. Pressure monitoring intervals location in observation boreholes.
A network of drains was positioned radially on the walls of the two shafts every 6m down
the shafts. The water collected from each porous level was directed to a flowmeter.
Furthermore, all the water collected from the Oxfordian limestone was connected to
another flowmeter. This allowed measurement of both the overall outflow from the
Oxfordian and the individual outflow from each porous level in each shaft. Complementary
data, which are not described in this paper, were gained from:
detailed geological mapping on the shaft walls, including identification of the extent
and orientation of fractures,
detailed sedimentary characterisation which also gave information to better understand

the structure of potential conductive levels and semi-permeable layers within the
Oxfordian.
24m long boreholes drilled every 12m in the main shaft. In ten of these boreholes,
constant head production tests or pulse tests were performed.
4.2 Measurements obtained in the shaft and the observation boreholes
The excavation of the Oxfordian limestone started on 11 November 2001 in the main shaft
and 29 March 2002 in the auxiliary shaft. The excavation lasted 909 days in the main shaft
and 748 days in the auxiliary shaft. Overall outflow measured just after the excavation of
the Oxfordian limestone was 9.7l/min in the main shaft and 8.4L/min in the auxiliary
shaft. In the main shaft, the four lower porous levels account for a 73.3% of the total
inflow, which is consistent with the results obtained from fluid conductivity logging in
EST204. In the auxiliary shaft, the same porous levels accounted for 57.2% of the total
inflow. This represented 10% more than was obtained in EST205 using fluid conductivity
logging. Pressure monitoring in borehole EST201 provides a good example of hydraulic
interference between shafts and monitoring boreholes. Figure 8 (Distinguin, 2005) shows
penetration versus time for the upper part the main shaft, with porous levels represented by
blue and black vertical bars corresponding to the depth and time the boreholes were drilled
during shaft sinking. The hydraulic head versus time is shown in the lower part of this fig-
ure. The first interference was observed when borehole PPA0012 was drilled across the
upper porous level Hp7. Drawdown was observed in the two upper intervals (correspond-
ing to Hp7 and Hp6) in borehole EST201, starting simultaneously on 26 November 2001.
These observations indicate that porous levels Hp7 and Hp6 are hydraulically connected.
During geological mapping in the main shaft, a vertical fracture from Hp7 to Hp6 was
observed on the wall and this may explain the observed vertical connectivity. These two
porous levels were also affected when Hp7 and Hp6 were excavated in the main shaft and
when borehole PPA0014 crossed Hp6. Further excavations in the main shaft have had no
effect but some effects observed later could be related to the auxiliary shaft excavation.
For interval three corresponding to porous level Hp5, no interference was observed until
borehole PPA0015 crossed Hp5 in the main shaft, indicating that this porous level is isol-
ated hydraulically from the two upper ones. Simultaneously, a slight decrease of hydraulic
head for the two deeper intervals in EST201 was observed, corresponding to porous levels
Hp1 to Hp4, indicating that there is a low hydraulic connectivity between Hp5 and the four
deeper porous levels. Further observation of pressures in borehole EST201 confirms that:
Hp7 and Hp6 are highly connected hydraulically with each other but not connected to
other porous levels,
Hp5 reacts on its own with a low connection to the four deeper porous levels,
Hp1 to Hp4 represent the third group of hydraulic response and are connected.
Measurements in EST103, EST104 and EST203 have given similar results in terms of
hydraulic connection between the porous levels.
4.3 Interference test interpretation
Three different modelling methods were carried out to interpret the measurements and
make some predictive simulations from the interference test for the Oxfordian limestone.
The two models performed by Antea and the BGR (Bundesanstalt fr Geowissenschaften
und Rohstoffe) are presented and compared below. The model run by Antea in 2004 with
a homogeneous mesh was first performed over an area of 1265km
2
centred on the URL.
The vertical mesh was then refined on a 0.13km
2
to take into account the multi-packer
design and the shaft sinking data. Specific storage coefficients were differentiated for each
rock level in the model. The analysis was based on hydraulic test interpretations with a
central well and radial flow. The Oxfordian limestone was divided into 17 levels consist-
ing either of porous levels or of matrix zones of lower permeability between the porous
levels. Hp5 was divided into three distinct levels to account for the shaft sinking stop from
May 2002 to June 2003. No vertical flow was taken into account in this simulation. Horizontal
3
0
/
1
0
/
0
1
2
7
/
0
2
/
0
2
2
7
/
0
6
/
0
2
2
5
/
1
0
/
0
2
2
2
/
0
2
/
0
3
2
2
/
0
6
/
0
3
2
0
/
1
0
/
0
3
1
7
/
0
2
/
0
4
1
6
/
0
6
/
0
4
1
4
/
1
0
/
0
4
1
1
/
0
2
/
0
5
1
1
/
0
6
/
0
5
0
9
/
1
0
/
0
5
0
6
/
0
2
/
0
6
0
6
/
0
6
/
0
6
0
4
/
1
0
/
0
6
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
H
y
d
r
a
u
l
i
c

h
e
a
d

(
m

a
b
o
v
e

s
e
a

l
e
v
e
l
)
550
500
450
400
350
300
250
200
150
100
50
0
500
450
400
350
300
250
200
150
100
50
0
D
e
p
t
h

(
m
)
EST201
interval 1: Hp1-Hp2
interval 2: Hp3-Hp4
interval 3: Hp5
interval 4: Hp6
interval 5: Hp7
Main shaft sinking
porous level:
Hp7
Hp6
Hp5
Hp4-Hp3
Hp2-Hp1
rock formation:
oxfordian limestone
callovo-oxfordian clay
borehole in the main shaft
P
P
A
0
0
1
2
P
P
A
0
0
1
5
P
P
A
0
0
1
3
P
P
A
0
0
1
4
P
P
A
0
0
1
7

a
n
d

P
P
A
0
0
1
8
P
P
A
0
0
1
6
P
P
A
0
0
1
9
P
P
A
0
0
2
0

a
n
d

P
P
A
0
0
2
1
2
7
/
1
1
/
0
1
1
3
/
0
2
/
0
2
2
4
/
0
3
/
0
2
2
2
/
0
4
/
0
2
2
5
/
0
5
/
0
3
2
7
/
0
7
/
0
3
1
2
/
0
9
/
0
3
0
7
/
1
1
/
0
3
3
0
/
1
1
/
0
3
Figure 8. Shaft sinking and hydraulic head measurements in borehole EST201.
permeability obtained from this model ranged from 5 10
10
and 3 10
8
m/s. Specific
storage coefficient varied between 10
6
to 1,5 10
6
m
1
. The model run by BGR in 2005
is based on the RockFlow code and uses a geological model for the Oxfordian obtained
from geological and geophysical data. Each level of clay, marl or limestone (Figure 9) was
accounted for in the model. Tectonic or sedimentary features were also taken into account
in the model, based on field observations: a vertical fracture in the main shaft axis con-
necting Hp7 and Hp6, vertical fractures connecting Hp5 and Hp4 to the east and west of
the URL site, vertical fractures connecting Hp6 and Hp5 to the west of the URL site, and
the lower part of the Oxfordian limestone (to Hp4) was divided into a western and an east-
ern domain. The boundary conditions correspond to a constant head on the model bound-
ary and a no flow zone below and above the Oxfordian limestone.
Figure 9. Location of lower permeability levels in the Oxfordian limestone (from BGR).
M
a
y
-
2
0
0
0
M
a
y
-
2
0
0
1
M
a
y
-
2
0
0
2
M
a
y
-
2
0
0
3
M
a
y
-
2
0
0
4
M
a
y
-
2
0
0
5
A
p
r
-
2
0
0
6
A
p
r
-
2
0
0
7
A
p
r
-
2
0
0
8
A
p
r
-
2
0
0
9
A
p
r
-
2
0
1
0
A
p
r
-
2
0
1
1
M
a
r
-
2
0
1
2
M
a
r
-
2
0
1
3
M
a
r
-
2
0
1
4
M
a
r
-
2
0
1
5
M
a
r
-
2
0
1
6
M
a
r
-
2
0
1
7
F
e
b
-
2
0
1
8
145
155
165
175
185
195
205
215
225
235
245
255
265
275
285
295
305
315
325
H
y
d
r
a
u
l
i
c

h
e
a
d

(
m

A
S
L
)
Hp7 measurement
Hp7 from BGR simulation
Hp7 from Antea simulation
Hp6 measurement
Hp5 measurement
Hp3-4 measurement
Hp3-4 from BGR simulation
Hp3-4 from Antea simulation
Hp1-2 measurements
Hp1-2 from BGR simulation
Hp1-2 from Antea simulation

M
D
/
e
s
t
2
0
1

s
im

L
T
.
g
r
f
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
D
r
a
w
d
o
w
n

(
m
)
Figure 10. Pressure measurements and simulations form the two models in EST201.
Horizontal permeability estimates for the porous levels range from 6,7 10
9
to
5,5 10
8
m/s. Permeability estimates for the zone between the porous levels range from
6,9 10
11
to 5 10
9
m/s. The very low permeability estimated for the layer between
Hp6 and hp5 (10
1 1
m/s) is consistent with the low hydraulic connectivity observed dur-
ing pressure monitoring. A homogeneous specific storage coefficient was obtained from
this model (3 10
6
m
1
). The range for permeability and specific storage coefficients
obtained by the two models is similar and are coherent with the estimates obtained in the
boreholes from test pumping. The main differences lie in the vertical distribution of the
permeabilities. Pressure simulations from the Antea model fit better with observed data
than pressure simulation from the BGR model in EST201 (Figure 10) and in EST203.
Estimates obtained from the Antea model are likely to be more representative for the ver-
tical and lateral variations of the hydraulic properties in the Oxfordian limestone.
5 CONCLUSIONS
From 1994 to 2006, Andra performed a very detailed hydrogeological characterisation of
the URL site. The overall permeability for this rock formation at the URL is between
5 10
10
and 5 10
9
m/s. This range is consistent with measurements obtained
12.1km north and 2.8km south of the URL. To the west, the Marne faults cross this rock
formation, where there are much higher permeabilities. 15km to the east, higher permea-
bility values were obtained which reflected lithological variations. Sophisticated method-
ologies like fluid conductivity logging or packer testing, which are not widely used in
hydrogeology, yielded interesting results for the detailed understanding of the hydrogeol-
ogy of this rock formation and proved to be efficient and complementary to more classi-
cal testing methods (e.g. test pumping, interference tests). Consequently water inflow
zones and their hydraulic properties have been identified at a multi-metric scale. The use
of specific devices for monitoring both the hydraulic head in the boreholes (multipacker
systems) and the outflow from the two access shafts to the URL (network of water collect-
ors and flowmeters) gave very accurate measurements, which were interpreted in terms of
hydraulic properties at a larger scale. The interference tests run since 2001 and the associ-
ated geological data have allowed a better understanding of the flow zones within the
Oxfordian limestone, and the horizontal extent of the inflow zones at the URL scale has
been demonstrated. The vertical connectivity or isolation of the inflow zones has also been
determined. Currently, measurements are still being taken in order to evaluate the zone of
influence caused by the drainage from the shafts in the Oxfordian limestone and estimate
the time needed to reach a steady state.
ACKNOWLEDGMENTS
The authors wish to thank:
The reviewers for their contribution towards the improvement of the manuscript,
The Antea team (including J.M. Joubert) who followed all the field work and made
some modelling,
Agns VINSOT (Andra) who was in charge of the scientific program in EST201,
EST203, EST204 and EST205 in 2000,
The BGR team who performed the modelling of the interference and provided Figure 9,
The Baker Oil Tool, Colenco, Hydroinvest and Intera companies who worked either on
performing the tests or on their interpretation.
REFERENCES
Antea (2001) Lot 2, forages de suivi des perturbations, rapport de synthse, Andra report n
o
B RP
0ANT 01010
Delay J, Distinguin M (2004) Hydrogeological Investigations in Deep Wells at the Meuse/Haute-
Marne Underground Research Laboratory. Proceedings EurEnGeo, Lige, vol. 104, pp 219225
Delay J, Cruchaudet M (2004) Hydraulic Monitoring of Low-Permeability Argillite at the
Meuse/Haute-Marne Underground Research Laboratory. Proceedings EurEnGeo, Lige, vol.104,
pp 341346
Delouvrier J, Delay J (2004) Multi-level Groundwater Pressure monitoring at the Meuse/Haute-
Marne Underground Research Laboratory, France. Proceedings EurEnGeo, Lige, vol.104,
pp 377384
Distinguin M (2005) Test hydrogologique dinterfrence lchelle du site. Exprimentation SPH.
Andra report n
o
D RP ADPE 050735
Gueutin P (2004) Quantification du drainage de lOxfordien par les puits et pizomtrie. Rapport de
DEA, Universit Pierre et Marie Curie, Andra
Low S, Kelley V, Vomvoris S (1984) Hydraulic Borehole Characterisation through the Application
of Moment Methods to Fluid Conductivity logs. Journal of Applied Geophysics, 31, 117131
Pickens JF, Grisak GE, Avis JD, Belanger DW (1987) Analysis and interpretation of borehole
hydraulic tests in deep boreholes: principles, model development and applications. Water Resour.
Res., 23, 13411375
Tauzin E, Johns RT (1997) A New Borehole Simulator for Well Test Analysis in Low-Permeability
Formations, paper presented at IAMG97, the annual conference of the international association
for mathematical geology, 2227 September 1997, Barcelona, Spain
Tsang CF, Hufschmied H, Hale FV (1990) Determination of fracture inflow parameters with bore-
hole fluid logging method, Water Resour. Res., 26, 561578
D. Geothermal aquifer systems
CHAPTER 26
Tracing high pH mineral waters ascribed to ultramafic
rocks (Central Portugal): Conceptual vs numerical
modelling
Jos M. Marques
1
, Maria J. Matias
1
, Paula M. Carreira
2
,
M. Rosrio Carvalho
3
, Fraser Goff
4
, Maria J. Basto
1
, Rui C. Grac
,
a
1
,
Mrio Andrade
1,2
, Lus Rocha
5
& Lus Aires-Barros
1
1
Instituto Superior Tcnico, Av. Rovisco Pais, Lisboa, Portugal
2
Instituto Tecnolgico e Nuclear, Sacavm, Portugal
3
Departamento de Geologia, Faculdade de Cincias, Universidade de Lisboa, Lisboa,
Portugal
4
Earth and Planetary Sciences Department, University of New Mexico, Albuquerque, USA
5
Junta de Freguesia, Av. da Libertao, Cabeo de Vide, Portugal
ABSTRACT: Cabeo de Vide mineral waters have a distinct chemical composition (high pH;
Na-Cl/Ca-OH type waters) and are associated with mafic/ultramafic rocks. Geochemical, isotopic
and numerical approaches have contributed to increased knowledge of the nature of these mineral
waters. Most of the local surface and shallow groundwater is of the Mg-HCO
3
type. Since both the
shallow Mg-HCO
3
and the mineral waters have similar
2
H and
18
O signatures there may be an evo-
lution from the HCO
3
-Mg waters towards the mineral waters. The reaction path simulations show
that the progressive evolution of the Ca-HCO
3
to Mg-HCO
3
waters can be attributed to the inter-
action of meteoric water with serpentinite. The sequential dissolution in CO
2
(g) closed system con-
ditions at depth leads to the precipitation of chrysotile, brucite and calcite, indicating that the water
is responsible for the serpentinization of fresh ultramafic rocks (dunites) present at depth.
1 INTRODUCTION
In the Cabeo de Vide region (Figure 1), spas must be considered to be one of the main
sources of local development. Thus, further knowledge of the local hydrogeology is
extremely important to achieve a sustainable use of this invisible georesource, which
could give rise to multiple new attractions to develop additional tourism. The mineral
springs of Cabeo de Vide have been well known since Roman times. In fact, the Romans
built a bath (dated 119 B.C.) for their legions in the study area. Cabeo de Vide mineral
waters, due to their particular chemical composition and their occurrence under complex
geologic and hydrogeologic conditions, have attracted the attention of scientists for a long
time (e.g. Portugal Ferreira and Mendona 1990; Costa et al., 1993). Their origin has been
a challenge for both geologists and hydrogeologists, although most studies have been
based mainly on results of detailed geological investigations.
Recently, within the scope of the TERMAVIDE R&D Project (funded by the local
Municipality) coupled geochemical, isotopic and hydrogeochemical modelling approaches
were designed to investigate the circulation paths of the waters and their correlation to local
surface, shallow and deeper groundwaters. The main objective of this paper is to summarise
the findings obtained during this Project, with special emphasis on hydrogeochemical and
isotopic investigations, comparing their results with recent hydrogeochemical modelling
using previous geochemical data.
2 GEOLOGICAL AND GEOMORPHOLOGICAL SETTINGS
The study region (Figure 1) is situated in the Central part of the Portuguese mainland, in
the Iberian Hercynian belt. The Lower Cambrian carbonate sequence was intruded and
metamorphosed by mafic and ultramafic rocks forming a NW-SE cumulate-type structure
of Ordovician age, which has been subjected to serpentinization rodingitization
processes (Costa et al., 1993). Cabeo de Vide mineral waters discharge at the intrusive
contact between the mafic/ultramafic rocks and the carbonate sequence, associated with
the main regional NNE-SSW trending fault (Figure 2). The ultramafic rocks range from
partially serpentinized dunites to serpentinites. The mafic rocks adjacent to the ultra-
mafics are mostly coarse- to fine-grained gabbros. In addition, several other distinct geo-
logical formations can be observed in the surroundings of Cabeo de Vide spas. The most
representative are: Precambrian metamorphised rocks (schists and graywackes), Cambrian
chloritized schists, quartzites and greywacke, and orthogneisses dated at 466 10Ma and
some hyperalkaline syenites.
The region extends from the Cabeo de Vide Spa (20km SW of Portalegre city)
towards the South Mamede Mountain ridge (5km NE of Portalegre city). This mountain
ridge (ca. 1027m a.s.l.) is the main geomorphologic feature of the Cabeo de Vide region.
The relief of the Cabeo de Vide area (260300m a.s.l.) tends to be rather flat. The
topography rises gradually from the Cabeo de Vide Spa at a distance of 20km from
Portalegre city (620m a.s.l.), after which the gradient becomes steeper, with a maximum
elevation at the crest of the South Mamede Mountain.
352 Aquifer Systems Management: Dracys legacy in a world or impending water stortage
Figure 1. Map of the Cabeo de Vide region.
3 SAMPLING PROCEDURES AND ANALYTICAL METHODS
Water samples were collected from streams, springs, dug wells and boreholes, for chemical
and isotopic analyses. Temperature (C), pH, electrical conductivity (S/cm) and redox
potential (mV) were determined in the field at the time of collection. Total alkalinity was
measured a few hours after collection. Major and minor elements in the waters were deter-
mined at Laboratrio de Mineralogia e Petrologia of Instituto Superior Tcnico (LAMP-
IST) / Portugal, using the following methods: atomic absorption spectrometry for Ca and
Mg; emission spectrometry for Na, K, Li, Rb and Cs; colorimetric methods for SiO
2
, Fe
total
,
F and Al; ion chromatography for SO
4
, NO
3
and Cl; potentiometry for alkalinity. Trace ele-
ment concentrations were determined at Los Alamos National Laboratory (LANL) / USA,
using the ICP-MS methodology. The
2
H and
18
O measurements (vs V-SMOW, Vienna
Standard Mean Ocean Water) were performed by mass spectrometry (SIRA 10VG ISO-
GAS) at the Instituto Tecnolgico e Nuclear (ITN Portugal) according to the analytical
methods of Epstein and Mayeda (1953) and Friedman (1953), with an accuracy of 1
for
2
H and 0.1 for
18
O. The
3
H content (reported in Tritium Units, TU) was also
determined at ITN, using electrolytic enrichment followed by liquid scintillation counting
method (standard deviation varies between 0.9 and 1.3 TU, depending on tritium activ-
ity of the water samples). The
13
C and
14
C determinations in waters were performed at the
Geochron Labs/USA by accelerator mass spectrometry (AMS). The
13
C values are reported
in vs V-PDB (Vienna Peedee Belemnite) standard, with an accuracy of 0.1.
14
C is given in pmC (per cent with respect to the so-called modern carbon).
Conceptual vs numerical modelling 353
Figure 2. Geologic map of the Cabeo de Vide region (courtesy Fernandes, J., personal communi-
cation). (1) serpentinized ultramafic rocks; (2) mafic and ultramafic rocks; (3) hornfels; (4) carbon-
ate rocks, displaying contact metamorphism; (5) Cambrian rocks: schists, quartzites and greywackes;
(6) Pre-Cambrian rocks: metamorphic schists and greywackes; (7) orthogneisses and (8) ortho-
gneisses and hyperalkaline syenites. Termas stands for Cabeo de Vide Spa. F stands for the main
regional NNE-SSW trending fault. The altitude parameter is given through the geodesic marks.
4.1 Water chemistry vs water-rock interaction
In the Cabeo de Vide region, the different geologic formations led to the development of
surface, shallow and groundwaters displaying different facies. Groundwaters discharging
within the Cambrian limestones are Ca-HCO
3
-type waters. Waters sampled in the
Precambrian/Cambrian slates and in the granitic rocks belong to the Na/Ca-HCO
3
and
Na-HCO
3
-type, respectively. The Mg-HCO
3
-type waters constitute most of the surface
(stream) and shallow groundwaters discharging from the serpentinites, suggesting that the
water chemistry is strongly host-rock dependent. In particular, serpentine (chrysotile) dis-
solution may explain the high Mg (Figure 3) and SiO
2
concentrations found in Mg-HCO
3
type waters (e.g. Barnes et al., 1967; Barnes and ONeil, 1969):
The main characteristics of Cabeo de Vide mineral waters can be summarized as follows:
i) Na-Cl/Ca-OH type, ii) very alkaline pH values (between 10.5 and 11.5), iii) low Mg and
SiO
2
concentrations (see Table 1), iv) rather low mineralization (dry residuum values up to
220mg/L) and v) the presence of reduced species of sulphur (HS
0.7mg/L). This type

of water has been well documented in the literature (e.g. Barnes and ONeil, 1969; Barnes
et al., 1967, Bruni et al., 2001). Some of these authors suggest that serpentinization of the
ultramafic rocks may be an important process contributing to the mineral water chemistry
through the reaction:
which may explain the low Mg (Figure 3) and SiO
2
concentrations found in the Cabeo de
Vide mineral waters. Usually, as stated by Hostetler et al. (1966), serpentinization is not a
2 3
2 4 2 2 3 2 5 4
Mg SiO H O Mg OH Mg Si O OH
forster
( ) ( )
( iite brucite chrysotile ) ( ) ( )
Mg Si O OH H Mg H SiO H O
chryso
4
0
3 2 5 4
2
4 2
6 3 2 ( )
(

ttile)
Figure 3. Mg (mg/L) vs pH plot of waters from different geological formations: () limestones;
() gabbros; () serpentinites; () contact gabbros/serpentinites; () contact gabbros/limestones.
() Cabeo de Vide stream. () Cabeo de Vide mineral waters. Adapted from Marques et al. (2004).
single episode in the history of ultramafic bodies, but has probably occurred at various
times and places for the same body of rock. The high Ca/Mg ratio of these waters may
indicate leaching of rodingitized rocks (Coleman, 1977; Costa et al, 1993). The rodingiti-
zation process is characterized by the addition of Ca and the removal of Si and some large
lithophile elements from the rocks, under relatively oxidizing conditions.
4.2 Isotope geochemistry
Environmental isotopes (
18
O and
2
H) were used to estimate the relative importance of
locally infiltrated meteoric waters in the recharge of the mineral aquifer system. The long-
term weighted mean value (
18
O and
2
H) of precipitation collected in the Portalegre mete-
orological station (597m a.s.l./approximately 15km NE of Cabeo de Vide) was used to
help the interpretations. The relationship between the concentrations of the environmentally
stable isotopes deuterium and oxygen-18 in all the water samples is similar to the relation-
ship in the global meteoric water line (GMWL:
2
H 8
18
O 10) defined by Craig
(1961), indicating i) that they are meteoric waters, which have not been subjected to surface
evaporation (being directly infiltrated into the ground) and ii) that there is no evidence of
water/rock interaction at high temperatures (Figure 4), consistent with the low issue tempera-
ture (19.5C) of the Cabeo de Vide mineral waters. However, the shift observed between
the GMWL and the isotopic composition of Cabeo de Vide mineral water can be due to the
analytical errors associated with the isotopic measurements (
2
H 1 ;
18
O 0.1
), although the possibility of isotopic fractionation between water and alteration / precip-
itation of mineral could be considered. Nevertheless, the formation of hydrous minerals
(e.g. clay minerals) and calcite precipitation can lead to depletion in the water isotopic com-
position. However, these types of minerals are enriched in
18
O but depleted in
2
H compared
to the water from which they precipitate at isotopic equilibrium (Chacko et al., 2001).
Therefore, Cabeo de Vide mineral waters (remaining water) should become progressively
depleted in
18
O and enriched in
2
H, during the precipitation of hydrous minerals. Cabeo de
Table 1. Physico-chemical characteristics of representative waters from the Cabeo de Vide region,
discharging from distinct geological formations. Concentrations in mg/L.
(L) (G) (Sp) (CV)
T(C) 19.4 18.2 19.2 19.6
Cond (S/cm) 724 785 740 646
pH 6.96 7.37 7.54 10.93
Na 14.3 9.53 9.56 49.0
K 0.96 0.35 0.44 5.04
Ca 100.0 41.9 38.7 22.5
Mg 26.5 77.3 72.2 0.16
HCO
3
469.2 540.3 542.9 n.d.
CO
3
0.24 0.70 1.06 6.9
SO
4
20.56 14.72 14.42 5.18
NO
3
19.48 12.33 9.07 6.22
Cl 16.68 14.44 9.08 32.14
SiO
2
24.6 74.3 72.8 5.50
OH n.d. n.d. n.d. 33.02
Notes: Groundwater issuing from (L) limestones (G) gabbros, (Sp) serpentinites. (CV) stands for Cabeo de Vide
mineral waters.
Vide mineral waters and the local Mg-HCO
3
waters (issuing from the serpeninized dunites)
have similar
2
H and
18
O values (around 27 and 4.5 , respectively), indicating a
common meteoric origin for these waters (Figure 4). Even the
2
H and
18
O contents of the
Cabeo de Vide stream waters (running along serpentinized dunites) are very similar to the
isotopic content of the Cabeo de Vide mineral waters. Considering the long-term mean
isotopic composition of precipitation at the Portalegre meteorological station
(
2
H 32.2;
18
O 5.44; (ITN 2002)) we can admit that the major source of
recharge of the Cabeo de Vide mineral waters appears to be limited to lower elevation sites
located at the local ultramafic/mafic outcrop (dunites, serpentinized dunites and gabbroic
rocks; see geological map in Figure 2).
4.3 Conceptual model
Since both Mg-HCO
3
waters and Cabeo de Vide mineral waters have similar isotopic
(
2
H and
18
O) signatures, we believe that the Mg-HCO
3
waters could evolve towards the
Cabeo de Vide mineral waters through interactions with the surrounding rocks. With this
water-rock interaction model, the local Mg-HCO
3
-type waters seem to be generated in an
initial step, under open CO
2
conditions, due to water-chrysotile interaction in a shallow
environment, whereas the Cabeo de Vide mineral waters are produced in a subsequent
step, under closed CO
2
conditions. In this second step, the increased pH values (due to
water-dunite interaction at depth) favour calcite precipitation (Figure 5).
In fact, the carbonate deposits occurring along veins in drill cores from the Cabeo de
Vide area are mainly calcite (Photo 1). Because most of the magnesium is retained in
chrysotile and vein brucite (Photo 2), Ca concentrations should increase proportionately in
the waters during the evolution of the HCO
3
-Mg-type waters towards the Cabeo de Vide
mineral water end member.
Assuming a common origin for the Mg-HCO
3
-type and the Cabeo de Vide mineral
waters, the higher Cl concentrations (accompanied by low
3
H values) found in the mineral
waters could be attributed to the increased water-rock interaction, as the result of a large
flow path and/or long residence time in the subsurface rocks (Figure 6).
The absence of
3
H determined in the Cabeo de Vide mineral waters (Figure 6) suggest
that recharge of the Cabeo de Vide mineral system corresponds to an age of more than
Figure 4.
2
H vs
18
O relations in water samples from Cabeo de Vide area. Adapted from
Marques et al. (2003).
60 years. Two samples of Cabeo de Vide mineral waters were collected for
14
C-age determi-
nations (AMS determinations; Geochron Laboratories, USA). The values obtained range
between 69.12 0.28pmC and 65.24 0.35pmC (AC5 and AC3 boreholes), indicating an
apparent groundwater age between 2970 40 and 3430 50 years BP, respectively.
Photo 2. Drillcore from borehole AC2 (Cabeo de Vide Spas) showing another example of serpen-
tinized dunite. The arrows indicate fibrous chrysotile and brucite along fractured zones.
Photo 1. Drillcore from borehole AC2 (Cabeo de Vide Spas) showing typical serpentinized
dunite. The arrows indicate the presence of veins filled with calcite.
Figure 5. Plot of HCO
3
CO
3
(mg/L) vs pH. Symbols as in Figure 3. Adapted from Marques
et al. (2004).
The study of carbon isotopes is more complex than that of O or H, due to the existence
of a number of different sources of carbon. In most groundwaters, only part of the bicar-
bonate carbon is recent and derived from biogenic CO
2
. The rest could be derived from
aquifer carbonate, which will normally be far older than the half-life of 5730 years and
will, therefore, have dead carbon (negligible
14
C content). Thus, the interpretation of
14
C
data has limitations but probably provides a limit on maximum age. The
13
C values of
22.9 and 18.0 of Cabeo de Vide mineral waters (boreholes AC3 and AC5 respec-
tively) suggest (i) an organic origin for the carbon in the Cabeo de Vide mineral waters,
and (ii) a negligible contribution from local carbonate rocks (
13
C values 1.48 ), sug-
gesting that
14
C-dating of mineral waters seems to provide quite reliable results.
4.4 Geochemical modelling
High-pH waters are well known to be associated with ultramafic rocks, namely serpen-
tinized rocks (e.g. Barnes et al., 1978, Drever, 1982, Bruni et al. 2002, among others). To
better understand the geochemical processes that most likely have caused the changes in
water composition, the water-rock mass transfer was simulated using chemical speciation,
mineral-solution equilibrium and reaction path modelling. The calculations were carried
out by means of the PHREEQC software code (Parkhurst and Appelo, 1999), referring to
the WATEQF thermodynamic database of 2005. As the water pH is the main parameter
affected during simulated aqueous changes, it is of interest to evaluate the saturation state
of minerals in the weathering environment, and elucidate the effect of pH on the saturation
state because early water-rock interaction is dominated by uptake of protons from the
water with a corresponding increase in pH (Gslason and Eugster, 1987a, b).
The groundwater from limestone (L, ) is at equilibrium with calcite (Figure 7) and
oversaturated in quartz and chalcedony but undersaturated with respect to amorphous sil-
ica (Figure 8) and magnesium carbonates. This observed oversaturation in silica minerals
is due to the limestone chemical composition, which is affected by contact metamorphism.
Groundwaters issuing from gabbros (G, ) are undersaturated with respect to the typical
minerals of mafic and ultramafic rocks (e.g. pyroxenes and forsterite, Figure 9) and
serpentinites (e.g. crhysotile, Figure 10). On the other hand, they are at equilibrium with
silica minerals and oversaturated in calcite as result of Ca-aluminosilicate dissolution. The
neutral Mg-HCO
3
waters from gabbro (G, ) and serpentinite (Sp, ) aquifers are
generally undersaturated with respect to the typical minerals of the ultramafic rocks
Figure 6. Plot of
3
H (TU) vs Cl (mg/L). Symbols as in Figure 3. Adapted from Marques et al. (2004).
(forsterite Figure 9) and serpentinite (chrysotile Figure 10), as well as to some solid
phases usually produced through interaction between these rocks and waters at relatively
low temperatures and pressures (brucite, sepiolite, and a variety of Mg-carbonates). These
waters are relatively close to saturation with Mg-montmorillonite, Mg-saponite, amor-
phous silica (Figure 8), and oversaturated in calcite as a result of Ca-aluminosilicate disso-
lution and a rise in the solution pH. In contrast, the Cabeo de Vide mineral waters (CV, ),
which are high-pH, Na-Cl/Ca-OH type waters, are generally strongly oversaturated with
chrysotile (Figure 10) and diopside, slightly undersaturated with respect to enstatite and
forsterite (Figure 9), and undersaturated in calcite, brucite and amorphous silica.
The progressive evolution from Ca-HCO
3
-type to Mg-HCO
3
-type waters can be attrib-
uted to low temperature interaction of meteoric waters with serpentinites, at open and/or
closed conditions with respect to CO
2
(g). To prove this hypothesis, the water-rock mass
transfer leading to these chemical changes in the aqueous phase was simulated through
progressive reaction of water from the limestones (L, as typically regional groundwater)
with serpentinites. The Cambrian limestones represent an important aquifer in the region
(the Monforte Alter do Cho carbonate aquifer) where the fractured carbonate rocks pro-
mote the development of Ca-HCO
3
waters. Since serpentinites are almost monomineralic
rocks, stoichiometric serpentine (Mg
3
Si
2
O
5
(OH)
4
chrysotile) was considered to be the
only solid phase under dissolution in the open system with respect to CO
2
(g). Calculations
were performed in titration mode, which means adding at each step of the reaction
-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
5 7 6 8 9 10 11 12
pH
L
o
g

S
I

C
a
l
c
i
t
e
(L)
(G)
(Sp)
(CV)
Equilib
Figure 7. Calcite water saturation index (SI)
versus the water pH.
-13
-11
-9
-7
-5
-3
-1
1
5 6 7 8 9 10 11 12
pH
L
o
g

S
I

S
i
O
2
(
a
)
(L)
(G)
(Sp)
(CV)
Equilib
Figure 8. Amorphous (a) silica water satura-
tion index (SI) versus the water pH.
-12
-8
-4
0
4
8
5 6 7 8 9 10 11 12
pH
L
o
g

S
I

F
o
r
s
t
e
r
i
t
e
(L)
(G)
(Sp)
(CV)
Equilib
-10
-8
-6
-4
-2
0
2
4
6
8
5 6 7 8 9 10 11 12
pH
L
o
g

S
I

C
h
r
y
s
o
t
i
l
e
(L)
(G)
(Sp)
(CV)
Equilib
Figure 9. Forsterite water saturation index (SI)
versus the water pH.
Figure 10. Chrysotile water saturation index
(SI) versus the water pH.
progress variable a corresponding amount of chrysotile, and only the chemical compos-
ition of the solid reaction was considered (i.e., both reaction kinetics and reactive surfaces
of solid phases were left out of consideration).
The major aqueous species increment during the reaction path, and the saturation of the
water varying with respect to forsterite, calcite, amorphous silica, brucite and magnesite is
represented in Figure 11(a) and (b), respectively, versus the pH of the water. During the
chrysotile dissolution reaction some changes are observed: (i) water pH ascent; (ii) increase
of aqueous Mg and silica concentrations; (iii) the water becomes oversaturated in calcite as a
result of pH rising; (iv) the water becomes close to saturation with respect to amorphous sil-
ica; (v) decreasing of the water undersaturation related to forsterite and chrysotile.
The very-high pH Na-OH water can only be achieved through the dissolution of ultra-
mafic and serpentinized rock minerals under closed system conditions with respect with
CO
2
(g). The water-rock mass transfer leading to these chemical changes in the aqueous
phase was simulated through reaction path modelling of water from serpentinites aquifer
(Sp) reacting with serpentinitic and dunitic rocks. The dunite dissolution simulation was
performed assuming that the olivine end-member forsterite (Mg
2
SiO
4
) is the main dis-
solved mineral. It was also assumed that forsterite dissolves stoichiometrically over a wide
range of solution pH (Gslason and Arnrsson, 1993) and that chrysotile, calcite, brucite
and amorphous silica can be formed at thermodynamic equilibrium.
The simulation was carried out in titration mode and in two steps: (1) dissolution of
chrysotile at open-system with a fixed PCO
2
(g) of 10
2
atm followed by chrysotile disso-
lution path in a closed-system and very reducing conditions (PO
2
10
50
atm); (2) dissolu-
tion of chrysotile in an open-system with a fixed PCO
2
(g) of 10
2
atm followed by forsterite
dissolution path at closed-system and very reducing conditions (PO
2
10
50
atm). The
reaction paths results are represented in Table 2 and Figures 12 and 13. Figure 12 shows (a)
the aqueous concentration variations of Ca
2
, Mg
2
, HCO
3
, and H
4
SiO
4
during the reactive
process of chrysotile dissolution, and (b) the saturation state of the water with respect to the
solid phases: forsterite, calcite, amorphous silica, brucite and magnesite. Figures 13(a) and
13(b) show the same parameters obtained for forsterite-water mass transfer, but the satur-
ation index for forsterite was substituted by the water saturation index for chrysotile.
The serpentinite dissolution does not determine a great increase in the pH and activity
of aqueous SiO
2
, since the water becomes saturated in chrysotile after dissolution of
(a)
0.E+00
1.E-03
2.E-03
3.E-03
4.E-03
5.E-03
6.E-03
7.E-03
8.E-03
9.E-03
1.E-02
0.E+00 4.E-04 8.E-04 1.E-03
Chrysotile reaction progress
(moles)
M
o
l
a
l
i
t
y
7.0 7.5 7.6 7.7
pH
Ca
+2
Mg
+2
H
4
SiO
4
HCO
3-
(b)
-12
-10
-8
-6
-4
-2
0
2
0.E+00 4.E-04 8.E-04 1.E-03
Chrysotile reaction progress
(moles)
L
o
g

S
I
1.0 3.0 5.0 7.0
pH
Forsterite
Calcite
SiO
2
(a)
Brucite
Magnesite
Figure 11. (a) Major aqueous species increment, and (b) the saturation index (SI) variations of
Ca-HCO
3
water with respect to forsterite, calcite and amorphous silica, brucite and magnesite dur-
ing the chrysotile reaction dissolution for an open system with respect to CO
2
(g).
(a) (b)
0.E+00
1.E-03
2.E-03
3.E-03
4.E-03
5.E-03
6.E-03
7.E-03
8.E-03
9.E-03
1.E-02
0.E+00 5.E-06 5.E-05 5.E-04 1.E-03
Chrysotile reaction progress (moles)
M
o
l
a
l
i
t
y
7.7 7.2 7.2 7.4 8.1
pH
Ca+2
Mg+2
H4SiO4
HCO3-
-9
-7
-5
-3
-1
1
3
5
0.E+00 5.E-06 5.E-05 5.E-04 1.E-03
Chrysotile reaction progress (moles)
L
o
g

S
I
7.7 7.2 7.2 7.4 8.1
pH
Forsterite
SiO2(a)
Brucite
Calcite
Magnesite
Figure 12. Chrysotile dissolution reaction path in close system to CO
2
(g): (a) Ionic water compo-
sition and pH variation; (b) mineral saturation index (SI) for some minerals and the water pH;
minerals are considered to form at thermodynamic equilibrium.
(a)
0.E+00
1.E-03
2.E-03
3.E-03
4.E-03
5.E-03
6.E-03
7.E-03
8.E-03
9.E-03
1.E-02
0.E+00 5.E-05 5.E-04 5.E-03 5.E-02 5.E-01
Forsterite reaction progress (moles)
M
o
l
a
l
i
t
y
7.7 7.2 7.4 7.9 10.8 10.8
pH
Ca+2
Mg+2
H4SiO4
HCO3-
(b)
-8
-7
-6
-5
-4
-3
-2
-1
0
1
0.E+00 5.E-05 5.E-04 5.E-03 5.E-02 5.E-01
Forsterite reaction progress (moles)
L
o
g

S
I
1.0 3.0 5.0 7.0 9.0 11.0
pH
Chrysotile
SiO2(a)
Brucite
Calcite
Magnesite
Figure 13. Forsterite dissolution reaction path in close system to CO
2
(g): (a) Ionic water compos-
ition and pH variation; (b) mineral saturation index (SI) for some minerals and the water pH; miner-
als are considered to form at thermodynamic equilibrium.
3 10
4
moles. During the mineral reaction the precipitation of calcite, brucite, magnesite
and amorphous silica was assumed to be in thermodynamic equilibrium. The general agree-
ment between theoretical paths and analytical data indicates that progressive interaction of
surface- and groundwaters with serpentinites leads to the formation of the Mg-HCO
3
type
waters, under open-system conditions with respect to CO
2(g)
(Table 2). The numerical mod-
elling points out: (i) the role of calcite precipitation during the closed-system dissolution in
depleting the aqueous solution in C and Ca species; (ii) the role of magnesite and brucite pre-
cipitation in the depleting dissolved Mg allowing the acquisition of a Na-OH composition;
iii) and the role of brucite formation on water pH. The low amounts of aqueous silica are the
result of the formation of amorphous silica and chrysoltile solid phases.
At their natural conditions, the Cabeo de Vide mineral waters are oversaturated in
chrysotile. Nevertheless, we have assumed thermodynamic equilibrium of the waters relative
to this mineral. There are limitations in thermochemical calculations of natural geological
processes, especially at low temperatures. Kinetics of mineral dissolution and mineral
precipitation at low temperature often is too slow for attaining water-rock equilibrium
within a reasonable time scale. The low silica content in these mineral waters does not cor-
roborate the hypothesis that groundwater might be at higher temperatures at depth, and the
low rate of chrysotile precipitation can explain the water oversatutarion.
The Cabeo de Vide mineral waters (high-pH, Na-Cl/Ca-OH type waters) are depleted in
Ca
2
, Mg
2
and SiO
2
. These geochemical signatures can only be obtained with sequential and
progressive interaction of meteoric waters with serpentinites and dunites, in a closed system
with respect to CO
2
. These signatures indicate that the Cabeo de Vide mineral waters would
be responsible for the serpentinization at low temperature conditions of the ultramafic rocks
present at depth. It should be stated that there is a general agreement between these theoretic-
al paths and the mineral assemblages observed in the cores from mineral water boreholes.
The results obtained from this study can be resolved as a hydrogeologic conceptual model
for the Cabeo de Vide mineral waters, which seem to be dominated by interactions between
locally recharged meteoric waters and ultramafic rocks. The local Mg-HCO
3
-type waters
(surface and shallow groundwaters) do indeed seem to evolve towards the Cabeo de Vide
mineral waters (high pH; Na-Cl/Ca-OH-type waters). Mg-HCO
3
-type waters are generated
under open CO
2
conditions, whereas the Na-Cl/Ca-OH-type waters are the result of water-
rock interactions under closed-system conditions. A schematic diagram (Figure 14) shows
the likely groundwater flow paths, the geological section and the main water-rock interaction
processes.
Mineropetrografic observations from drillcores provide strong evidence that consider-
able magnesium precipitation occurs when surface waters seep into the ground and react
progressively with the ultramafic rocks. Cabeo de Vide mineral waters are unsaturated
with respect to Mg-Olivine [forsterite Mg
2
SiO
4
] but supersaturated with serpentine
[chrysotile Mg
3
Si
2
O
5
(OH)
4
] indicating that the precipitation of Mg-rich minerals con-
trols aqueous magnesium concentrations. Serpentinization does not represent a single
episode in the history of Cabeo de Vide ultramafic body, but probably indicates many
Table 2. Theoretical chemical composition obtained by reaction path simulation of: (1) chrysotile
dissolution at open-system followed by chrysotile dissolution in a closed and reduced system;
(2) chrysotile dissolution at open-system followed by forsterite dissolution in a closed and reduced
system. (L) stands for the chemical composition of groundwaters issuing from the limestone; (CV)
stands for Cabeo de Vide mineral waters. The species concentrations are in mg/L.
(1) (L) Chrysotile Chrysotile (CV)
dissolution dissolution
pH 6.96 7.73 8.06 10.93
Ca 100 Open-system 11.34 Closed-system 11.74 22.5
Mg 26.5 PCO
2
(g) 104.05 PO
2
(g) 24.55 0.16
C total 92.35 10
2
atm 120.00 10
50
atm 41.64 1.38
SiO
2
24.6 104.60 104.60 5.50
(2) (L) Chrysotile Forsterite (CV)
dissolution dissolution
pH 6.96 7.73 10.76 10.93
Ca 100 Open-system 11.34 Closed-system 0.77 22.5
Mg 26.5 PCO
2
(g) 104.05 PO
2
(g) 1.50 0.16
C total 92.35 10
2
atm 120.00 10
50
atm 1.19 1.38
SiO
2
24.6 104.60 0.00 5.50
episodes at various times and various places for the same body of rock. Cabeo de Vide
mineral waters should be considered the fluid responsible for present-day serpentinization
of the local ultramafic body at depth. These hydrogeological signatures are being used by
the concessionaire of Cabeo de Vide Spas for the redefinition of well-head and aquifer
protection areas, drilling strategies, and future development plans.
ACKNOWLEDGEMENTS
This work has been supported by the Cabeo de Vide Municipality, under the Research
Contract TERMAVIDE / No. 693, and by the Centre of Petrology and Geochemistry of
Instituto Superior Tcnico, Technical University of Lisbon. F. Goff was supported by the
CO
2
-Sequestration Project of Los Alamos National Laboratory. Olivier Atteia and an
anonymous reviewer critically read an early draft of this manuscript and we gratefully
acknowledge their contribution. The authors would like to thank Nick S. Robins for edit-
ing the English.
REFERENCES
Barnes I, LaMarche VC Jr, Himmelberg GR (1967) Geochemical evidence of present-day serpen-
tinization. Science 56: 830832.
Figure 14. Hydrogeological conceptual model of Cabeo de Vide mineral waters, showing the evo-
lution of the local Mg-HCO
3
-type waters (generated under open CO
2
conditions) towards the
Cabeo de Vide mineral waters (high pH; Na-Cl/Ca-OH-type waters), as the result of water-dunite
interaction under closed-system conditions at depth.
Barnes I, ONeil JR (1969) The relationship between fluids in some fresh alpine-type ultramafics
and possible modern serpentinization, Western United States. Geol. Soc. America Bull. 80:
19471960.
Barnes I, ONeil JR, Trescases JJ (1978) Present-day serpentinization in New Caledonia, Oman and
Yugoslavia. Geochimica et Cosmochimica Acta 42: 144145.
Bruni J, Canepa M, Cipolli F, Marini L, Ottonello G, Vetuschi Zuccolini M, Chiodini G, Cioni R,
Longinelli, A (2001) Reactions governing the chemistry of waters interacting with serpentinites:
In: Cidu (eds) Water-Rock Interaction. Taylor & Francis / A.A. Balkema, London, pp 145148.
Bruni J, Canepa M, Chiodini G, Cioni R, Cipolli F, Longinelli A, Marini L, Ottonello G, Vetuschi
Zuccolini M (2002). Irreversible water-rock mass transfer accompanying the generation of the
neutral, Mg-HCO
3
and high-pH, Ca-OH spring waters of the Gnova province, Italy. Applied
Geochemistry 17: 455474.
Chacko T, Cole DR, Horita J (2001) Equilibrium oxygen, hydrogen and carbon isotope fractionation
factors applicable to geological systems: In: Valley JW, Cole DR (eds) Stable Isotopes
Geochemistry, Reviews in Mineralogy 43. Mineralog. Society America, Washington DC, pp 181.
Coleman RG (1977) Ophiolites. Springer-Verlag, Berlin, 229 pp.
Costa IR, Barriga F, Mata J, Munh JM (1993) Rodingitization and serpentinization processes in
Alter-do-Cho Massif (NE Alentejo): In: Noronha F, Marques M, Nogueira P (eds) Actas da IX
Semana de Geoqumica [Proceedings of the IX Geochemical Week]. Universidade do Porto.
Faculdade de Cincias. Museu e Laboratrio Mineralgico e Geolgico, pp 2731.
Craig H (1961) Isotopic variations in meteoric waters. Science 133: 17031703.
Drever JI (1982) The geochemistry of natural waters. Prentice Hall, Englewood Cliffs.
Epstein S, Mayeda T (1953) Variation of
18
O content of waters from natural sources. Geochimica
et Cosmochimica Acta 4: 21324.
Friedman I (1953) Deuterium content of natural waters and other substances. Geochimica
et Cosmochimica Acta 4: 89103.
Gslason SR, Arnrsson S (1993) Dissolution of primary basaltic minerals in natural waters: satur-
ation state and kinetics. Chemical Geology 105: 117135.
Gslason SR, Eugster HP (1987a) Meteoric water-basalt interactions, I. A laboratory study.
Geochimica et Cosmochimica Acta 51: 28272840.
Gslason SR, Eugster HP (1987b) Meteoric water-basalt interactions, II. A field study in N.E.
Iceland. Geochimica et Cosmochimica Acta 51: 28412855.
Hostetler PB, Coleman RG, Mumpton FA (1966) Brucite in alpine serpentinites. American
Mineralogist 51: 7598.
ITN [Instituto Tecnolgico e Nuclear] (2002) Precipitation data base. Portalegre meteorological sta-
tion (19882002).
Marques JM, Andrade M, Goff F, Basto MJ, Matias MJ, Graa RC, Carreira PM, Aires-Barros L,
Rocha L (2003) Origin and evolution of high pH mineral waters traced by isotope geochemistry
(S-Portugal): In: IAEA (eds) Book of Extended Synopses of the International Symposium on
Isotope Hydrology and Integrated Water Resources Management. IAEA-Vienna, pp 180181.
Marques JM, Matias MJ, Basto MJ, Graa RC, Aires-Barros L, Andrade M, Carreira PM, Goff F,
Rocha L (2004) Water-Rock interaction responsible for the origin of high pH mineral waters
(S. Portugal): In: Wanty RB, Seal II RR (eds) Water-Rock Interaction. Taylor & Francis /
A.A. Balkema, London, pp 293297.
Parkhurst DL, Appelo CAJ (1999) Users guide to PHREEQC (version 2)-A computer program for
speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations.
Water-Resources Investigations Report 994259. US. Geological Survey, Denver, Colorado.
Portugal Ferreira M, Mendona JL (1990) Enquadramento geolgico e hidrogeolgico das nascentes
de gua minero-medicinal de Cabeo de Vide [Geological and hydrogeological background of
Cabeo de Vide mineral waters]: In: Seco de Geologia Econmica e Aplicada (eds) Livro de
Homenagem a Carlos Romariz [Book of Homage to Carlos Romariz]. Departamento de Geologia,
Universidade de Lisboa, Lisboa, pp 151173.
CHAPTER 27
Fluid/mineral equilibrium calculations, isotopes and
geophysics as a multidisciplinary approach to the
characterization of Mono hydrothermal system
(NW-Portugal)
Paula M. Carreira
1
, Jos M. Marques
2
, M. Rosrio Carvalho
1
, Fernando
A. Monteiro Santos
3
, Hugo Matias
3
, Rafael Luzio
3
& Dina Nunes
1
1
Instituto Tecnolgico e Nuclear. Estrada Nacional n 10, Sacavm, Portugal
2
Instituto Superior Tcnico. Av. Rovisco Pais, Lisboa, Portugal
3
Centro de Geofsica da Universidade de Lisboa, Edifcio C, Lisboa, Portugal
ABSTRACT: Geochemical, environmental isotopes and geophysical data from thermal and cold
waters demonstrate the role of combined hydrogeological tools to improve knowledge of the concep-
tual circulation model of Mono hydrothermal system. Geophysical data show low resistivity zones
ascribed to the thermal waters circulation in a granitic environment. The isotope signatures indicate
that the aquifer recharge area is located to the south of the Mono Spas, up hill between 300 and
600m a.s.l. The waters issue at 48C, and are of Na-HCO
3
type. The mineral/solute equilibrium and
the geothermometric results show that Mono thermal waters are non-equilibrated due to the high
CO
2
content. The estimated deep temperatures derived from Na-K geothermometers give the most
reliable deep temperatures, in the range of 110 to 126C. Considering a mean geothermal gradient
of 30C/km at Mono region, the thermal aquifer is between 3.2 and 3.7km deep.
1 INTRODUCTION AND GEOLOGICAL SETTING
One of the main objectives of groundwater modelling is to determine the most likely work-
ing models to explain hydrogeological observations. With this goal, some integrated studies
(hydrogeochemical, isotopic and geophysical) were carried out at Mono region, North
of Portugal. An important aspect in the low-temperature geothermal system is the mainten-
ance of the chemical quality of the water, considering that Spas are very dependent on both
constant temperature and water quality. Therefore, the effects of mixing between the
hydrothermal system and the shallow cooler waters (or even with Minho River water) were
investigated. Furthermore, in the research region the intense land use for agriculture (vine-
yards) is a major pollution risk. With this objective an integrated interpretation of the results
was applied to update understanding of the Caldas de Mono low-temperature geothermal
system. Special emphasis was placed on the characterization of: (i) aquifer temperatures
of the Mono thermal waters from the chemical composition of the discharge fluids;
(ii) low resistivity zones to identify thermal waters flow paths; (iii) mixing processes
between shallow and deep groundwater systems and, (iv) altitude of the recharge areas of
the hydrothermal system. The better understanding of the aquifer structure will help in the
definition of strategic sites for drilling, bearing in mind that Mono Spas are expanding
their activities in response to increased demand. The boreholes located near the Minho
River enable the better understanding of the structure and geometric characteristics of the
hydrothermal system.
Three types of granitic rocks were identified based on their geometric relationships and
internal deformation (Fig. 1). According to Ribeiro and Moreira (1986) and Moreira and
Simes (1988) the rocks can be divided into: (i) sin-tectonic granites: with minute flakes of
muscovite and biotite, presenting metamorphic minerals, and strongly correlated with
migmatitic rocks; (ii) late-tectonic granites (frequently associated with the granodiorites):
with abundant biotite (muscovite is a secondary mineral). The degree of deformation sug-
gests an origin associated with the last Hercynian deformation phase (F3); (iii) post-tec-
tonic granites: usually characterized by the presence of mega crystals of potassium feldspar
and biotite. The lack of metamorphic minerals in these granites point to an age younger than
the last Hercynian deformation phase. Fluvial sandstones and conglomerates of Quaternary
age (sometimes with clay layers) constitute the most recent formations in the region, particu-
larly near the Caldas de Mono Spas and along the Minho River margins.
The main fracture systems in the region are represented by structural lineaments (strike-
slip faults), trending ENE-WSW, WNW-ESE, NNE-SSW and NNW-SSE, of late Hercynian
age, and still active during the Meso-Cenozoic. The geological studies carried out in Mono
region have indicated that the fracture system ENE-WSW (left strike-slip fault) is respon-
sible for the morphology of the Minho River valley. The thermal borehole waters and
Figure 1. Geological sketch map of region (after Pedrosa, 1999). The location of water sampling
sites is presented (for both thermal and shallow cold dilute water systems).
springs are located along the ENE-WSW structures, the groundwater flowpath being con-
trolled by this fracture system.
2 SAMPLING AND METHODS
Two fieldwork campaigns were carried out at Mono region, the first during February
2002 and the second one in February 2003. Groundwater samples were collected from
boreholes (thermal system) and springs (located at different altitude sites) representing the
shallow cold dilute systems (local circulation). Isotopic determinations (
2
H,
13
C,
18
O,
3
H and
14
C) and chemical analysis (major and minor species) were carried out. During
these campaigns two water samples were collected from Minho River for isotopic and
chemical analysis.
The environmental isotopes (
2
H,
13
C,
18
O,
3
H and
14
C) were measured at Instituto
Tecnolgico e Nuclear (Portugal). The
18
O and
2
H determinations were carried out using a
mass spectrometer SIRA 10 VG-ISOGAS. Epstein and Mayeda (1953) and Friedman
(1953) have described the analytical methods used in the determinations of oxygen-18 and
deuterium respectively. The results are reported in notation in reported to V-SMOW
with the accuracy of the measurements are 1 for
2
H and 0.1 for
18
O.
The
3
H was determined for all water samples using an electrolytic enrichment method
and measured by liquid scintilation counting (PACKARD TRI-CARB 2000 CA/LL). The
detection limit of this equipment is 0.5 TU. The associated error to the measurements
varies with the tritium concentration in the samples, although usually is around 0.6 TU.
The analytical method is describe in I.A.E.A. (1976).
The
14
C measurements were carried out on the Total Dissolved Inorganic Carbon (TDIC)
extracted in the field as BaCO
3
in a pH environment higher than 9.0 (I.A.E.A. 1981). The
barium carbonate obtained in situ is transformed into benzene in the laboratory and the
counting rates of the
14
C are obtained using a liquid scintillation counter (PACKARD TRI-
CARB 4530). The errors associated to this method vary with the amount of carbon avail-
able in each sample, and are greatest for the lowest
14
C content expressed as a percentage
of modern carbon (pmC).
During the benzene synthesis a CO
2
gas sample is collected for
13
C determinations (by
mass spectrometer). The isotopic composition is reported to V-PDB in . The associated
error to this measurement is 0.1.
Temperature (C), electrical conductivity (E.Cond., S/cm) and pH were measured in situ.
The water samples collected for chemical analysis were stored in two polyethylene bottles
(1 L), one acidified with HCl for analyses of Ca, Mg, Na, K, Li, Al and SiO
2
, the other was
reserved unacidified for the F, Cl, SO
4
and NO
3
determinations. The chemical parameters
were determined at Laboratrio de Mineralogia e Petrologia of Instituto Superior Tcnico
(Portugal), using the following methods: atomic absorption spectrometry for Ca and Mg;
emission spectrometry for Na, K and Li; colorimetric methods for Al, F and SiO
2
(the sam-
ples were specifically treated ultrafiltration before Al measurements); ion chromatogra-
phy for SO
4
, NO
3
and Cl; potentiometry for alkalinity, here referred to as HCO
3
.
Geophysical studies have been performed near Caldas de Mono Spas. Resistivity sur-
veys were carried out with the main goal to detect low resistivity zones that may be con-
nected to the thermal circulation system. These investigations were developed by the Instituto
de Cincias da Terra e do Espao (Portugal).
Fluid/mineral equilibrium calculations, isotopes and geophysics as a multidisciplinary 367
3 GEOPHYSICAL APPROACH
Resistivity surveys have been performed near Caldas de Mono Spas to detect low resist-
ivity zones that represent the thermal water circulation system. The survey comprised two
dipole-dipole lines that have been designed to cross the main fault system, tentatively asso-
ciated to the Mono hydrogeological system (Fig. 2A). Fig. 2B shows the field apparent
resistivity pseudo-section corresponding to the dipole-dipole line #2, carried out southwards
of Caldas de Mono (dipole length of 10m). The total length of this profile is 720m. The
field data was inverted using the RES2DINV code (Loke and Barker 1996). The resistiv-
ity model is presented in Fig. 2.
The main characteristics of this model are (i) overburden with a thickness ranging from
6 to 20m and resistivity ranging from 200 to 3000ohm-m; (ii) a more homogeneous resist-
ivity distribution (600ohm-m in average) at depths greater than 20m; and (iii) the pres-
ence of three relatively low-resistivity (100 to 300ohm-m) anomalous zones crossing all
the section at 100m, 260300m and 400m coordinates.
Previous geological studies that have been performed in the region highlight the strike
slip regional faults that trend NNE-SSW and NW-SE to E-W, and which basically control
all the important hydrogeological systems in northern Portugal (Pedrosa, 1999). The high
resistivity zones, in the overburden, are related to non-altered granite, which possess a few
fracture zones that act as conductors within the profile. The low-resistivity zone in the
western part of the profile (coordinate 100m) is well correlated with a mapped fault (Fig.
2A) that runs NNE-SSW in the Mono village direction. Taking into account the resistiv-
ity contrast (from 600 to 60100ohm-m) this fractured zone has to be infilled with high
conductivity material, probably clay and mineralized water. Therefore, the low-conductivity
zones, at coordinates 260300m and 400m, have been interpreted as fracture zones filled
with clay and mineralized water. The same criteria was applied to the resistivity model
obtained from line #1, which reveals low resistivity zones that correlate with faults marked
F2 and F3 in (Fig. 2A). These fault systems cross the area towards the Caldas de Mono
Spas. The models suggest that those fault systems should go deep into the granitic massif
as predicted by geological studies.
Figure 2. (A) Location of the two dipole-dipole lines plotted on the topographic map of Caldas de
Mono area (1: 25000); (B) Field apparent resistivity pseudo-section corresponding to the dipole-
dipole line #1, carried out southwards of Caldas de Mono; (C) Field apparent resistivity pseudo-
section corresponding to the dipole-dipole line #2, carried out southwards of Caldas de Mono.
4 ISOTOPIC APPROACH OF CALDAS DE MONO GROUNDWATER
SYSTEMS
4.1 Stable isotopes features
In general, the distribution of
18
O and
2
H mimics the topography of the continents; moun-
tains chains are marked by more negative values. This feature is called altitude effect
(Dansgaard, 1964; Rozanski et al., 1982; 1992; 1993). The lowering of temperature with
increasing elevation in mountain regions usually leads to enhanced condensation and, there-
fore, to a progressive depletion in heavy isotopes in rain with altitude. This altitude effect
has been used in numerous hydrological studies to identify the recharge areas and to inves-
tigate the origin and interconnection of water bodies. The vertical isotope gradient varies
between 0.15 and 0.50 100m
1
for oxygen-18 and about 1 to 4 100 m
1
for
deuterium (Aragus-Aragus et al., 2000; Gonfiantini et al., 2001).
Using the isotopic composition of the shallow cold groundwater samples the equation
of the Local Meteoric Water line (Fig. 3A) is
2
H 10.17
18
O 20.19 (the isotopic com-
position of the water samples are presented in Tables 1 and 2). In the same diagram is the
isotopic composition of Minho River and the Global Meteoric Water Line (GMWL). The
isotopic gap found between the two river samples should be attributed to the different sam-
pling periods (end of summer season October 1999 and middle of wet season February
2003). In the Caldas de Mono area the isotopic gradient obtained for
18
O (altitude
effect ) was 0.18 per 100m of altitude (Fig. 3B). This isotopic gradient was estimated
using the discharge altitude of the springs (cold dilute groundwater systems) and not the
infiltration altitude, in view of the fact that these are local circulation systems, with small
circulation paths.
Based on the isotopic gradient of the region and the isotopic signatures of the thermal
waters, the recharge altitude of Caldas de Mono hydrothermal system was estimated.
The values obtained range between 300m and 600m with a mean altitude around 400m
a.s.l. (at the South of Caldas de Mono). The recharge altitude values obtained at Caldas
de Mono low-temperature geothermal system are in agreement to those proposed by
Lima (2001), on the basis of the precipitation record.
One of the main objectives of this study was the identification of possible mixture between
Minho river water with the hydrothermal system. The isotopic composition can be con-
sidered as a fingerprint of each system. Minho river depleted isotopic composition (February,
50
45
40
35
30
25
20
8,0 7,0 6,0 5,0 4,0 3,0
18
O vs VSMOW (
o
/
oo
)
2
H

v
s

V
S
M
O
W

(
o
/
o
o
)
Local Meteoric Water Line
2
H = 10.17
18
O + 20.19
r = 0.86
Global Meteoric Water Line
2
H = 8
18
O + 10
A
6,0
5,5
5,0
4,5
4,0
3,5
0 100 200 300 400 500 600
Altitude (m a.s.l.)
1
8
O

v
s

V
S
M
O
W

(
o
/
o
o
)
18
O = 0.0018 Alt - 4.526
r = 0.881
AC1
AC2
B
Figure 3. (A)
2
H vs
18
O for groundwater samples from Caldas de Mono area; (B) Relation of
18
O values as a function of the altitude of sampling sites. The arrows indicate the estimation of the
mean altitude of the hydrothermal system (boreholes AC1 and AC2). The different symbols stand
for: () thermal borehole waters; () shallow cold dilute waters; () Minho river water.
2003) indicates that most of the flow contributions derive from high elevations. According
to Loureiro and Machado (1986), the source of the Minho River is located about 750m
a.s.l. on Meira Mountain (NW-Spain) receiving an important recharge contribution of the
Sil River also in NW Spain (Jistreda Mountain Cantbric Mountain Range) about 1500m
a.s.l.. The Minho River geographical and topographical features are well marked in the
isotopic composition variation, a depletion of about 1.5 in oxygen-18 (see Table 1). In a
first approach this isotopic fingerprint in Minho river water suggests that the contribution
of this superficial water body to Mono hydrothermal system should be extremely small
or even non-existent.
4.2 Dating Caldas de Mono thermomineral groundwater system
Within the environmental radioisotopes, tritium cannot be detected in waters with more
than approximately 5060 years due to its short half-life () (
3
H: 12.32 years: in Lucas
and Unterwager, 2000). Among the radioactive isotopes with a half-life higher than 10
3
years, carbon-14 ( 5730 years: in Mook, 2000) represents the most important tool in
groundwater dating.
Tritium concentrations were measured in all water samples (Table 1 and Table 2), as a
qualitative indicator of age since it is important to mention that the tritium being an envir-
onmental radioactive isotope originated in the high levels of the atmosphere has also an
anthropogenic origin associated with the release to the atmosphere from the test phase for
hydrogen bombs. Tritium in precipitation has been monitored by IAEA/WMO since the
bomb tests in late 1952 (Gonfiantini et al., 1990; Rozanski et al., 1991), these data enable
the establishment of timescales for transport of water, e.g., groundwater, which does not
contain tritium, must have infiltrated as rain before 1952, since the natural concentrations
have already decayed below the detection limit.
The tritium concentrations measured in Mono shallow cold dilute groundwater sys-
tems indicate a local recharge and a rather short residence time (Table 1). When these
groundwater content is compared with the precipitation tritium record measured at Porto
meteorological station (Serra do Pilar, approximately 100km south from Caldas de Mono),
similar values are observed. The precipitation data present a mean arithmetic value of 4.5
TU (monthly record from 1988 to 2000, ITN data base in: Carreira et al., 2003). Furthermore,
it is important to notice that the isotopic records of the Portuguese Network Isotopes in
Table 1. Data from shallow cold dilute groundwater samples from Caldas de Mono area
discharge altitude, issue temperature, electrical conductivity and isotopic composition.
Samples Sampling Altitude Temp. Elec.Cond.
18
O
2
H
3
H
date (m) (C) (S/cm) () () (TU)
NS1 Oct-99* 140 16.0 162 5.01 27.7 5.2 1.0
Feb-02 14.4 114 4.66 27.3 3.2 0.6
Feb-03 12.8 123 4.81 26.5 2.6 0.6
NS2 Oct-99* 260 13.3 62 4.88 31.5 4.8 1.1
Feb-03 11.7 80 5.19 33.6 2.1 0.6
NS3 Feb-03 550 13.1 39 5.50 36.8 2.4 0.5
River Minho Oct-99* 17.5 114 5.10 30.0 4.5 0.9
Feb-03 10.4 112 6.87 48.2 2.3 0.6
Note: * Data from Nascimento (2000).
Precipitation reported (monthly arithmetic average of the weighted means of tritium in
modern precipitation over Portugal from 19882000) have found tritium concentrations
varying between 4.5 TU and 5.3 TU in coastal and in continental regions, respectively
(Carreira et al., 2003).
The
3
H content found in the shallow cold dilute groundwater systems range between 2.1
TU and 5.2 TU, reflecting local recharge and short residence times. However, the absence
of
3
H in the thermal borehole waters indicates a relatively long residence time, at least
greater than 40 years (Table 2), if the tritium input is of the same order of magnitude as
found in Porto meteorological station (4.5 TU mean arithmetic weight value).
Fig. 4 shows a rather similar isotopic composition (
18
O and tritium) in the thermal
waters collected from boreholes AC1 and AC2. In a first approach, the isotopic deviation
in tritium content and depletion in oxygen-18 (AC1/ Oct. 99) could represent mixing with
other water bodies (e.g. Minho River waters), more depleted and younger than the waters
from the thermal system.
Table 2. Data from Mono hydrothermal system issue temperature, electrical conductivity, pH
and isotopic composition (boreholes AC1 and AC2).
Date Temp. E.Cond. pH
18
O
2
H
3
H (TU)
13
C
14
C (pmC)
(C) (S/cm) () () ()
AC1 Oct-99* 48.1 633 7.31 6.54 45.2 4.1 0.6
Feb-02 52.0 670 7.05 5.19 31.6 0 0.6
Feb-03 47.8 619 6.72 4.77 33.9 0 0.6 7.06 7.43 0.34
AC2 Oct-99* 45.0 693 6.98 5.20 35.2 0 1.0
Feb-02 46.5 741 7.06 5.25 30.9 0 0.6
Feb-03 42.2 793 6.55 5.22 30.0 0 0.6 6.25 4.82 1.00
Note: *Data from Nascimento (2000).
Figure 4. (A)
18
O() vs
3
H (TU); (B) Temperature (C) vs
3
H (TU); (C) C1
(mg/L) vs
3
H (TU)
for the water samples collected in 1999 in 2002 and during 2003 fieldwork campaigns. Symbols as
in Fig. 3.
However, this mixing hypothesis is not enhanced in Fig. 4, either in the isotopic compos-
ition of all hydrothermal samples. Based on the
3
H content measured in the shallow aquifer,
Minho River or precipitation data, no mixing is apparently occurring in the hydrothermal
system or the contribution to other water bodies should be extremely small. The observed
isotopic gap in
18
O values between the two river samples could be explained by seasonal
variation (October and February represent the end of the dry and rainy season, respectively).
The
13
C determinations were performed at Mono hydrothermal system (TDIC), giv-
ing values of 7.06 in AC1 and 6.25 in AC2 boreholes, suggesting an origin for the
CO
2
that can be related with mixture between atmospheric CO
2
, decay of organic matter,
dissolution of limestone, deep-seated (upper mantle) CO
2
or metamorphism of carbonates
producing CO
2
with slightly positive
13
C values (Truesdell and Hulston, 1980; Bergfeld
et al., 2001). However, limestone dissolution is unlikely, because no carbonate matrix is
found in the system. Based on the geology and tectonic structure of the region the more
reasonable source of carbon would be deep CO
2
(upper mantle origin).
Carbon-14 determinations were also carried out on the thermal waters, boreholes AC1
and AC2. It is important to mention that this radioisotope is present in the atmosphere,
soil, aquifer matrix, etc. In fact, radiocarbon dating of groundwater in the simplest form
assumes that: the
14
C moves with the water molecules along the flow path with a hydraulic
gradient and the only mechanism enable to change the carbon-14 content within the ground-
water systems is radioactive decay [A A
o
e
t; where A stands for the measured activity

of
14
C in the sample; A
o
initial
14
C activity; the decay constant for
14
C and t stands for
time]. The time can be re-arranged to provide an estimation of the groundwater age
[t
1
ln (A/A
o
)]. In order to apply these equations, it is important to estimate the ini-
tial concentration of
14
C (A
o
) at the time when the recharge occurs. Problems can arise
when the measurements are performed in the Total Dissolved Inorganic Carbon (TDIC) as
a consequence of the different incomes of carbon to the water system since the dissolved
inorganic carbon in the surface is not conservative and potentially interacts with soil and
aquifer carbonate matrix and organic matter, leading to a dilution of the initial
14
C content,
and this results in an overestimation of the groundwater age.
At Caldas de Mono Spas the low
14
C content, determined in the Total Dissolved
Inorganic Carbon (TDIC) in AC1 (4.82 1.00pmC) and in AC2 (7.43 0.34pmC) bore-
hole waters, together with the absence of tritium support the hypothesis of a long residence
time for the thermal waters. A close system model was used to calculate the apparent car-
bon-14 age of the thermal waters.
The carbon-14 age calculations was based upon the assumption that the initial activity
of the dating material is 100% modern CO
2
activity (100pmC); for the
13
C of the reser-
voir was adopted the value of 3 1 (assuming the main origin is deep CO
2
(Berfeld
et al., 2001); the
13
C for the CO
2
in the soil zone is 25 2; the fractionation factor
HCO
3
is 8 0.5, and the error associated to the measurements of the TDIC is 0.5. The
chosen mathematical model, therefore, includes mixture of soil CO
2
with deep CO
2
since
in the region no other source is known to explain the high bicarbonate content. The appar-
ent groundwater age obtained is 11.39 2.68ka BP for AC1 and 13.13 3.59ka BP for
AC2 borehole waters.
The radiometric age obtained in the hydrothermal system does not reflect the true
groundwater age, since the most dissolved carbon has its origin in the upper mantle (
14
C
free), and the age reflects its source: dissolved carbon from deep CO
2
(upper mantle). Using
simple mass balance calculations for a two-component system, an estimation can be made
of the maximum amount of upper mantle CO
2
at Mono hydrothermal system.
13
C val-
ues of upper mantle CO
2
was assumed to be 3 (Bergfeld et al., 2001) and for the shal-
low aquifer
13
CO
2
22.9 (Marques et al., 2000). For a two-component system carbon
isotope values indicate that the fraction of upper mantle CO
2
in Mono could be as high
as 7983%.
5 WATER CHEMISTRY APPROACH
Groundwater samples were collected from the two boreholes (AC1 and AC2) ascribed to the
Caldas de Mono Spas, from the shallow cold dilute groundwater systems and from Minho
River. The mean temperature of Caldas de Mono thermal waters (at the wellhead) is
around 48C, while the shallow cold dilute groundwater systems issue at temperatures
around 14C. A gap in the degree of mineralization is observed between these two water sys-
tems. The thermal waters are characterized by dry residuum between 420 to 470mg/l, while
the shallow cold groundwater systems have dry residuum values varying between 24 up to
120mg/l. Both hydrogeological systems present the same HCO
3
-Na facies. The Human
impact in the shallow cold groundwater systems is enhanced by the increase in NO
3
, Cl
and
SO
4
2
concentrations. By contrast, the thermal water samples fall in a tight cluster of data
points. Chloride is a conservative element often used to identify mixing processes in thermal
fluids (Marini and Susangkyono, 1999; Marques et al., 2003). Therefore, geochemical data
(Ca, Na, K, NO
3
, SO
4
and HCO
3
) were plotted against chloride content (Fig. 5).
Two clusters of water samples can be observed, one representing the hydrothermal system
and another representing the shallow cold dilute groundwaters and Minho river water. In the
diagram SO
4
2
versus Cl
the dispersion of the water samples from the shallow cold dilute
system is well marked. The highest NO
3
2
and SO
4
2
concentrations should be associated to
the geographic location of Bica dos Milagres spring. This spring is located at low altitude
within an area of intensely cultivated land. However it is important to emphasize that theses
two parameters (NO
3
2
and SO
4
2
) are bellow the recommended maximum values in drink-
ing waters for human use. It seems that no mixing process seems to occur between the ther-
mal waters and the local shallow/surface waters (including Minho river water).
6 GEOTHERMOMETRIC APPROACH
The aquifer temperatures of Caldas de Mono thermal waters have been estimated with
aqueous geothermometers (Table 3). These geothermometers are all based on the assump-
tion that specific temperature dependent mineral/solute equilibrium is attained in the
reservoir. The deep temperatures range from a minimum of 76C (Na/Li geothermometer
from Fouillac and Michard, 1981) to a maximum of 135C using the quartz geothermome-
ter (Fournier, 1977). This range of values may be due to lack of equilibration between the
respective solutes and hydrothermal minerals as result of deep CO
2
gas input.
Reed and Spycher (1984) have proposed that the state of equilibrium between water and
the many hydrothermal minerals as a function of temperature can provide the best estimate
of reservoir temperature. Waters that are far from equilibrium with hydrothermal minerals
will not show convergence of mineral equilibrium at any specific temperature. Reed and
Spycher (1984) consider their procedure a valuable tool to distinguish between equilibrated
geothermal waters and waters that have departed from equilibrium due to boiling or mix-
ing with shallow colder waters.
The reservoir temperature was predicted on the basis of the temperature dependent
equilibrium between many mineral phases and the water. The WATCH program version 2.0
Table 3. Summary of results from aqueous geothermometric calculations applied to Mono ther-
mal waters. Geothermometers used are from: (1) Fournier (1977); (2) Fournier and Truesdell (1973);
(3) Arnrsson et al. (1983); (4) Kharaka and Mariner (1989); (5) Fouillac and Michard (1981).
Aqueous Geothermometers
Well/Spring Date Calcedony Calcedony Quartz- Quartz- Na/K Na/K Mg/Li Na/Li
(1) (1) H 0 S 0 (2) (3) (4) (5)
(1) (1)
Mono-AC1
Oct-99 108 106 131 135 102 121 110 80

Mono-AC1
Feb-02 97 97 122 125 107 126 109 81

Mono-AC1
Feb-03 103 102 127 131 97 116 139 151

Mono-AC2
Oct-99 104 103 128 131 90 110 109 78

Mono-AC2
Feb-02 91 91 118 120 102 121 108 77

Mono-AC2
Feb-03 97 97 123 125 101 121 138 135

Sr
a
Sade
Oct-99 107 106 130 134 100 120 109 76

Note: (
) stands for borehole waters; (
) stands for spring waters.

0
5
10
15
20
25
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/l)
C
a
2
+

(
m
g
/
l
)
A
0
1
2
3
4
5
6
7
8
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/l)
K
+

(
m
g
/
l
)
C
0
2
4
6
8
10
12
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/l)
S
O
4
2
-

(
m
g
/
l
)
E
0
25
50
75
100
125
150
175
200
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/l)
N
a
+

(
m
g
/
l
)
B
0
3
6
9
12
15
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/l)
N
O
3
-
(
m
g
/
l
)
D
1
10
100
1000
0 5 10 15 20 25 30 35 40 45 50
Cl
-
(mg/L)
H
C
O
3
-

(
m
g
/
L
)
F
Figure 5. (A) Ca
2
vs Cl
; (B) Na
vs Cl
; (C) K
vs Cl
; (D) NO
3
2
vs Cl
; (E) SO
4
2
vs Cl
and
(F) HCO
3
vs Cl diagrams for the water samples collected in 1999 (Nascimento 2000), 2002 and
2003 fieldwork campaigns. Symbols as in Fig. 3.
(Arnrsson and Bjarnason, 1994) was used to calculate the deep-water composition of
Mono thermal aquifer and the state of equilibrium between the water and the minerals
as a function of temperature (Fig. 6). The range of possible intersection temperatures is
enormous, if one considers all the minerals for which thermodynamic data are available in
the programme. The minerals likely to be present in the aquifer matrix have been considered.
For most of the minerals, the deep equilibrium temperatures range from 105 to 130C (Fig.
6). Nevertheless, the estimated temperature for the deep water of the Sra
a
Sade thermal
spring is lower (Fig. 6(C)) because of the lack of aqueous aluminium data and, probably,
because of silica re-equilibration during upflow.
The mineral/solute equilibrium and the geothermometric results obtained through the
WATCH programme show that Mono thermal waters are non-equilibrated waters due to
the high value of dissolved CO
2
. As the geothermometric temperatures based on the silica
minerals equilibrium are strongly dependent on the fluid pH, the estimated deep tempera-
tures from the application of the conventional Na-K geothermometers give the most reli-
able deep temperatures, in the range of 110 to 126C (Arnrsson et al., 1983).
The deep water composition of the Mono hydrothermal system was calculated using
the WATCH programme and considering: (i) 120C as deep temperature; (ii) conductive
cooling in the upflow; (iii) that the total CO
2
analysed is already present in reservoir water
(not updated during upflow). The Mono hydrothermal water supplying the AC1 and AC2
Figure 6. The saturation state of Mono thermal waters for selected minerals versus temperature:
(A) water from AC1 well; (B) water from AC2 well; (C) water from N Sra. Sade spring. The legend
represented in (A) is equivalent for (B) and (C).
wells has pH around 6.706.85, with PCO
2
of 0.2090.402 bar-abs and enthalpy close to
500kJ/kg. The deep pH obtained for the spring water is higher (pH of 7.63) probably due
to the lost of CO
2(g)
during upflow.
In the aquifer, at 120C, the water is slightly oversaturated in calcite, fluorite, Mg-mont-
morillonite, Ca-montmorillonite, muscovite, quartz, and talc (Fig. 7), which mean an
excess of Ca, K, and Mg for the used reference equilibrium temperatures. The waters are
undersaturated with respect to albite, chalcedony, Na-montmorillonite and amorphous sil-
ica. This can be explained by ionic exchange reactions or mixing of waters with different
geochemical signatures. Assuming the aquifer temperature of 120C and a mean geother-
mal gradient of 30C/km at Mono region (Duque et al., 1998), the Caldas de Mono
thermal aquifer is situated between 3.2 and 3.7km at depth.
7 CONCLUSIONS
The regional geomorphology of the region favours a conceptual circulation model for
Caldas de Mono hydrothermal system where the recharge area is located at South of
Caldas de Mono, between 300 and 600m a.s.l. (based on
18
O values). The ground-
water flow paths are associated with the NNE-SSW fault systems and their NW-SE associ-
ated systems. The geophysical models suggest that the tectonic systems may be deep and
filled with mineralized water. The low
14
C (TDIC) content (between 4.82 1.00pmC and
7.43 0.34pmC) of Caldas de Mono thermal waters together with the absence of
3
H
support the hypothesis of a long circulation path through the subsurface rocks. This
hypothesis is corroborated by the minimum depth reached by the Caldas de Mono ther-
mal water system (3.23.7km) estimated by the geothermometric approach, considering
that the thermal aquifer temperature is at 120C. The
13
C determinations give values in
the range of 7 to 6, indicating a complex origin for the CO
2
in these waters (mix-
ture between atmospheric CO
2
, decay of organic matter and mainly by upper mantle CO
2
).
Geophysical, isotopic and geochemical studies performed at Caldas de Mono region
have increase knowledge on the interaction between local shallow cold groundwater sys-
tems and the low-temperature geothermal waters. The better understanding of the Mono
hydrothermal aquifer will help in the definition of drilling strategies, and prevent physico-
chemical fluid alterations during further aquifer exploitation.
Figure 7. Water saturation state for some minerals, at the reservoir temperature of 120C.
ACKNOWLEDGEMENTS
This work is being funded by FCT under the Project POCTI/CTA/45159/200 DISGAS
Dissolved gases in subsurface hydrology CO
2
-rich thermomineral waters (N-Portugal).
J.L.Michelot and an anonymous reviewer critically read an early draft of this manuscript
and we gratefully acknowledge their contribution. The authors would like to thank Nick
Robins for editing the English.
REFERENCES
Aragus-Aragus L, Froehlich K, Rozanski K (2000) Deuterium and oxygen-18 isotope compos-
ition of precipitation and atmospheric moisture. Hydrological Processes, 14, 13411355
Arnrsson S, Gunnlaugsson E, Svavarson, H (1983) The chemistry of geothermal waters in Iceland. III.
Chemical geothermometry in geothermal investigations. Geochim Cosmochim Acta, 47, 567577
Arnrsson, S, Bjarnason JO (1994) WATCH. Version 2.1. Icelandic Water Chemistry Group,
Iceland, 7 p.
Bergfeld D, Goff F, Janik CJ (2001) Carbon isotope systematics and CO
2
sources in the Geysers-
Clear Lake region, northern California, USA. Geothermics, 30, 303331
Carreira PM, Barbosa T, Valrio P, Arajo MF (2003) Teores em trtio nas guas de precipitao em
Portugal continental: variabilidade e factores condicionantes. Resumos do IV Congresso Ibrico
de Geoqumica. XIII Semana de Geoqumica, 353355
Duque R, Monteiro Santos FA, Mendes-Victor LA (1998) Heat flow and deep temperatures in the
Chaves Geothermal system, northern Portugal. Geothermics, 27, 1, 7587
Dansgaard W (1964) Stable isotopes in precipitation. Tellus XVI, 4, 436468
Epstein S, Mayeda T (1953) Variations of
18
O content of waters from natural sources. Geochimica
Cosmochimica Acta, 4, 213224
Fouillac C, Michard G (1981) Sodium/lithium ratios in water applied to geothermometry of geother-
mal reservoirs. Geothermics, 10, 5570
Fournier RO (1977) Chemical geothermometers and mixing models for geothermal systems.
Geothermics, 5, 4150
Fournier RO, Truesdell AH (1973) An empirical Na-K-Ca geothermometer for natural waters.
Geochim Cosmochim Acta, 37, 12551275
Friedman I (1953) Deuterium content of natural waters and other substances. Geochimica et
Cosmochimica Acta, 4, 89103
Gonfiantini R, Aragus-Aragus L, Rozanski K (1990) Tritium in precipitation: comparison between
the years of high tritium (196367) and recent years. Bulletin of Radiation Protection, 13, 16
Gonfiantini R, Roche MA, Olivry JC, Fontes JCh, Zuppi GM (2001) The altitude effect on the iso-
topic composition of tropical rains. Chemical Geology, 181, 147167
I.A.E.A. (1976) Procedure and technique critique for tritium enrichment by electrolysis at IAEA
laboratory. Technical Procedure n19, International Atomic Energy Agency, Vienna.
I.A.E.A. (1981) Sampling of Water for
14
C Analysis. International Atomic Energy Agency. Isotope
Hydrology Laboratory. Vienna, 1981: 7 pp
Kharaka, YK, Mariner, RH (1989) Chemical geothermometers and their application to Formation
Water from Sedimentary Basins, in Thermal History of Sedimentary Basins Methods and Case
Histories. Ed. Naeser, ND, McCulloh TM, 99117
Lima AS (2001) Hidrogeologia de Terrenos Granticos. Minho Portugal. PhD Thesis. Universidade
do Minho: 451 pp
Loke MH, Barker RD (1996) Rapid least-squares inversion of apparent resistivity pseudosections by
a quasi-Newton method. Geophys. Prosp., 44, 131152
Loureiro JJM, Machado MLR (1986) Bacia hidrogrfica do rio Minho. Monografias Hidrolgicas
dos Principais Cursos de gua de Portugal Continental. D.G.R.A.H., Lisboa, 227
Lucas LL, Unterweger MP (2000) Comprehensive review and critical evaluation of the half-life of
tritium. J. Res. Natl. Inst. Technol., 105, 541549
Marini L, Susangkyono AE (1999) Fluid geochemistry of Ambon Island (Indonesia). Geothermics,
28, 189204
Marques JM, Carreira PM, Aires-Barros L, Graa RC (2000) Nature and role of CO
2
in some hot
and cold HCO
3
/Na/CO
2
-rich Portuguese mineral waters: a review and reinterpretation.
Environmental Geology, 40, 12, 5363
Marques JM, Espinha Marques J, Carreira PM, Graa RC, Aires-Barros L, Carvalho JM, Chamin
HI, Borges FS (2003) Geothermal fluids circulation at Caldas do Moledo area, Northern Portugal:
geochemical and isotopic signatures. Geofluids, 3, 189201
Mook WG (2000) Environmental isotopes in hydrological cycle. Principles and applications. IHP-V,
Technical Documents in Hydrology (UNESCO/IAEA) N 39, Vol. I., 280 pp
Moreira A, Simes M (1988) Carta Geolgica de Portugal na escala 1:50000. Notcia Explicativa da
Folha 1-D (Arcos de Valdevez). Servios Geolgicos de Portugal.
Nascimento IB (2000) Contribuio para o Conhecimento das guas Subterrneas da Regio de
Mono. MsC Thesis. Instituto Superior Tcnico. Universidade Tcnica de Lisboa. 92 pp
Pedrosa MY (1999) Carta hidrogeolgica de Portugal na escala 1:200000. Notcia explicativa da
folha 1. Instituto Geolgico e Mineiro Departamento de Hidrogeologia. 70 pp
Reed M, Spycher N (1984) Calculation of pH and mineral equilibria in hydrothermal waters with
application to geothermometry and studies of boiling and dilution. Geochim. Cosmochim Acta
48, 14791492
Ribeiro ML, Moreira A (1986) Carta Geolgica de Portugal na escala 1:50000. Notcia Explicativa
da Folha 1-B (Mono). Servios Geolgicos de Portugal. 46 pp
Rozanski K, Sonntag C, Munnich KO (1982) Factors controlling stable isotope composition of
European precipitation. Tellus 34: 142150
Rozanski K, Gonfiantini R, Aragus-Aragus L (1991) Tritium in the global atmosphere: Distribution
patterns and recent trends. Journal of Physics G: Nuclear Particle Physics, 17, 55235536
Rozanski K, Aragus-Aragus L, Gonfiantini R (1992) Relation between long-term trends of
oxygen-18 isotope composition of precipitation and climate. Science, 258, 981985
Rozanski K, Aragus-Aragus L, Gonfiantini R (1993) Isotopic patterns in modern global precipi-
tation. American Geophysical Union, Geophysical Monograph 78, Climate Change in Continental
Isotopic Records. 136
Truesdell AH, Hulston JR (1980) Isotopic evidence on environments of geothermal systems. In:
Fritz P, Fontes JCh (ed) Handbook of Environmental Isotope Geochemistry. The Terrestrial
Environment 1, 179226
CHAPTER 28
Predictive model for the management of thermal resources
Alain Dupuy
1
, Ronan le Fanic
2
& Pierre Pouchan
1
1
EGID University of Bordeaux, Pessac, France
2
Nestl Waters M.T., Vittel, France
ABSTRACT: The exploitation and protection of thermal water requires an understanding of the local
thermal systems. It is important to know a sustainable abstraction rate that will not impact the quality
of the resource. In the area around Dax (SW France), there are several hot springs. The groundwater
comes from the Senonian fissured/karstified dolomitic aquifer, which has a sub-vertical dip as it lies
on the flank of a salt diapir. In order to create a management tool for the water resource, nested models
were developed to simulate the local impact of the withdrawals from all of the thermal wells. The over-
all capture zone was mapped using the flow paths induced by the wells. The models can be used as pre-
dictive tools to simulate future impact on the resource and to manage risk in terms of quantity and
quality.
1 INTRODUCTION
Within the global framework of sustainable development, the protection of water resources,
both quantitatively and qualitatively, is one of the key objectives of applied research in
hydrogeology. Hydrotherapy depends on the preservation of the original quality of the
resource. In France, thermal activities are subject to strict regulations concerning the water
resource quality. The Academy of Medicine and the Medical Faculties have, for several
years, required quality control of thermal products. Professionals who bottle thermo-mineral
water for consumption (still water) or use the thermal water for therapeutic purposes (medi-
cal hydrotherapy) must protect, by careful exploitation and management, the qualitative
stability of the water and guard against all possible risks.
Thermo-mineral water often results from specific, and often very complex, geological
and hydrogeological configurations. This is the main reason why the management of thermal
aquifers and springs is very difficult. This is particularly true in the southwest France where
many thermal spas exploit deep groundwater resources for medical hydrotherapy and con-
sumption. The Adour-Garonne basin is the primary French thermal basin with 35 spas, or
30% of all the French spas. In Aquitaine, hydrotherapy is an activity with a high tourist
potential and significant economic impact. The province of Aquitaine, receives over
90000 patients per year, which is the second highest in France. The Landes administrative
region and the area around Dax attract the greatest numbers of patients close to 76000
in 2003. About 5.5 million m
3
of water from deep Palaeocene and Senonian aquifers are
used annually for hydrotherapy in the Landes. Around 12000 jobs are generated directly or
indirectly by hydrotherapy activities in the region. The Dax Saint-Paul-ls-Dax hydrother-
mal system has been the object of descriptive studies, but the hydraulics of the system, the
origin of the water, and possible derogation effects due to intensive exploitation have not
been studied in depth. The various stakeholders in the local economy are aware of the dif-
ficulties involved in managing the resource. Until now, each one has done what they con-
sidered necessary to guarantee the quality of the water. In 2003, the Aquitaine Region and
the local spa owners commissioned a study to investigate the Dax hydrogeological system
in order to develop the proper management and protection of the resource.
2 GEOLOGICAL SETTING
The study area is located in the Adour River basin in the south of the Aquitaine Basin. It is
a part of the Pyrenean foreland and the geological history of this zone is intimately linked
to that of the mountain range. This part of the Aquitaine Basin is characterized by a suc-
cession of folds and small basins parallel to the Pyrenean foreland. It represents a transi-
tion zone between the Pyrenean domain, which is strongly affected by tectonics, and the
northern domain, which is a normally active margin (Rocher et al., 1999). In the Adour
River basin, halokinesis was particularly active. It seems to have started during the Upper
Jurassic and must have continued until the end of the Oligocene. Petroleum investigations
(BRGM et al., 1974) have revealed numerous salt structures, some of them salt domes like
the Dax diapir, others just injected faults (Figure 1). The diapirs are the result of Triassic
fault movements in the bedrock (Dupouy-Camet, 1952). In the area around Dax, the
Triassic salt diapir is sealed by alluvial deposits from the Adour River and, therefore, does
not mark the local topography. This salt dome is about 4.5km long east-west by 3km
north-south. At the beginning of the 20th century, a cap rock of halite was exploited at a
depth of 60m covered by clays (1030m) and alluvial fans (020m).
The Dax hydrothermal zone is contained in a dolomite formation, of Upper Cretaceous
age, which has been uplifted to a sub-vertical position along the northern side of the diapiric
structure (Figure 2). The contact between the aquifer and the salt mass occurs along a rec-
tilinear fault called the thermal fault. The Upper Senonian thermal aquifer is sheared and
Figure 1. General location, structure and isobaths of the Upper Cretaceous formation.
fractured at various scales, but most of the fractures are inter-connected. The fractures pro-
vide the aquifer with significant secondary porosity. The fracture anisotropy controls the
emergence zones of the thermal springs and diffuse contact with the Adour River alluvium.
3 ORIGIN OF THERMAL WATER
The lithology of the thermal aquifer is fine-grained chalky limestone, with fine grained-
texture and bioclasts (Cerepi et al., 2003; Dupouy-Camet 1952). The late dolomitization
phase and the karstification have greatly improved the hydrodynamic properties of the
thermal dolomite, especially on the flanks of the diapir. Moreover, the intense halokinesis
induced major fracturing of the thermal layer. The thermal aquifer contains large discontinu-
ities. Many fractures have been opened by dissolution, generating the complex system which
supports the emergences (Pouchan, 1995). The effective width of the dolomite aquifer is
less than 100m; its length has been estimated at 1600 meters using geophysical data.
Historically, the natural discharge of the thermal springs was approximately 5000m
3
/day.
Total mineral content of the water is around 1g/l and the water is of the sulphate-
calcium-chloride-sodium type (Dupouy-Camet, 1952; Pouchan et al., 1991). The temperature
of the water as it reaches the surface at Fontaine Chaude is 62C. Currently, the thermal
resource is exploited in Dax by 6 wells: Boulogne, Baignots, Fontaine Chaude and Place
de la Course, feeding the municipal thermal distribution network for connected spas, and
Elvina and St Christophe which supply the thermal baths at the private spa, Thermes
Adour. The total flow rate during the high season is about 300m
3
h
1
. The natural hot
springs have disappeared in response to the intensive pumping of the thermal water.
Predictive model for the management of thermal resources 381
Figure 2. Simplified cross-section of the northern flank of the Dax diapir (Pouchan, 1995, modified).
The zone of emergence in Dax is characterized by the presence of two dolomitic bodies
(Figure 3) with similar facies. One is dated to the Palaeocene and the other Upper Senonian
(Pouchan, 1979). Only the colour of the dolomite makes it possible to differentiate them. The
Senonian dolomite is white whereas the Palaeocene dolomite is pink, due to iron precipita-
tion during the emersion of the Palaeocene formations. There used to be small outcrops
along the Adour River but these are now obscured by urbanization.
4 HYDROGEOLOGICAL AND THERMAL SETTINGS
Given the head variations time lag between Fontaine Chaude and SPDX under pumping
conditions, a large hydraulic conductivity in the fractures had to be assigned in order to
model appropriate hydraulic heads. The transient flow analysis showed that porous matrix
properties alone cannot explain the observed discharge values, especially in the Fontaine
Chaude area. Analyses of pumping tests of the thermal wells also yielded a high value for
hydraulic conductivity (Le Fanic, 2005). The thermal parameters used in the model are
those resulting from drill-core measurements, which allow an accurate regional geother-
mal gradient of 2.7C/100m to be fitted. The Dax Saint-Paul-ls-Dax system is charac-
terized by the rise, along its northern borders, of water with a temperature exceeding 62C.
The local water emergence at this temperature causes a positive anomaly in the geothermal
field due to a very large thermal outflow. The limited geographical extent of the hydrother-
mal system, however, requires some clarification of the various modes of heat transfer
(Dupuy, 2001). The major synclinal zones have weak geothermal gradients. The negative
anomalies correspond to zones where the heat flux is overexploited compared to the total
average geothermal gradient because of the quantity of heat evacuated by the thermal
springs and the diapir. The discharge of all of the thermal wells is approximately 300m
3
/h.
This means that, in addition to heat conduction, the heat is also transported by the water,
i.e., both by advection and conduction. Heat transfer in the Dax thermal complex is mainly
due to advection in the fractured part of the Senonian thermal layer and to conduction in
the salt diapir, in the aquiclude layers and in the low zones of the Upper Senonian thermal
layer.
Figure 3. Simplified cross-section of the position of Senonian and Palaeocene springs (Le Fanic,
2005).
5 MODEL SETTINGS
The boundaries of a large-scale 3D model are shown in Figure 4. The total surface area is
96.4km
2
. It extends from the Go1 well in the Northeast to the Dax thermal wells in the South.
The geological data available in the sector are integrated into the model. The eastern
parts of the Sebastopol and Magescq salt domes have been included in the modelled zone.
Several seismic profiles provide the depths for the most important layers in the major syn-
clinal zones. To the south, the diapir at Dax is located between the thermal wells of Dax
and the southern limit of the study zone. The bedrock is fixed at a depth of 5000m. All of
the layers from the Quaternary to the Trias are taken into account with fourteen geological
formations and three aquifers (alluvial, Oligocene and thermal Senonian). The grid is
directly generated by FEFLOW (Diersch, 2002) and each well in the zone corresponds
to a node of the grid. The principal faults are integrated into the grid. The final large-scale
3-D model includes 27968 elements per layer (363584 elements for the fourteen layers).
On average, the triangular elements are 120 meters wide. Grid refinement is applied in
order to study thermal behaviour in the aquifer. Two overlapping zones of refinement were
created in the sub-vertical part of the thermal dolomitic body. The first corresponds to a
refinement to a grid of about 40m, the second, representing the Dax zone of emergences,
to a grid of about 8m.
The first simulations were carried out by applying a homogeneous permeability (K) to each
layer. This equivalent permeability (matrix and fissures) is used to represent the thermal
fissured/karstified aquifer (Debieche et al., 2002). The aquiclude layers have a permeability
of 1 10
9
m/s. Dolomites of the Upper Cretaceous have a permeability K of 2.3 10
6
m/s
(Moro, 1999). The Oligocene, Palaeocene and Upper Cretaceous aquifers are the only layers
whose permeabilities were fitted in a steady-state regime. The south-eastern zone, correspon-
ding to the Dax diapir, is assigned a permeability of 1 10
11
m/s. For the Upper Cretaceous
Figure 4. 3-D view of the Senonian layer of the large-scale model (Le Fanic, 2005).
aquifer, the range of K values used is between 10
5
m/s in the low zones and 8 10
3
m/s in
the fissured/karstified part on the northern side of the diapir at Dax. The specific storage coeffi-
cient S
S
is set at 4.2 10
6
m
1
in the fissured zone and at 7.7 10
5
m
1
in the other zones.
The value of K in the sub-vertical zone is greater than those obtained by the interpretation of
the pumping tests (4 10
4
m/s). Simulations of the local geothermal gradient were carried
out using constant values of thermal conductivity.
6 RESULTS
6.1 Natural state
The first results provided by the large-scale 3D model under steady-state conditions were
head distributions. By taking into account all the geological layers that make up the
hydrothermal system, it was possible to describe the local geothermal gradient. The geo-
metry and the nature of the various formations have a significant role in the simulation of
the temperature fields. Under transient-state conditions, the multi-layered structure of the
model enables variations of the observed piezometric heads to be restored as functions of
the pumping rates. The hydrodynamic/thermal coupling makes it possible to reproduce the
local geothermal gradient and, relatively precisely, the observed temperatures. The temper-
ature field is a key parameter to the simulation of the flow dynamics of the thermal aquifer.
By including all the geological layers of the hydrothermal structure, it was possible to con-
struct the first coupled model of the Oligocene aquifer, which is the main drinking-water
resource in the area. The piezometric heads and temperatures are thus reproduced by con-
sidering the entire multi-layered aquifer system. For the thermal aquifer, the use of equiva-
lent permeabilities in the fractured zone gives satisfactory results in transient conditions.
From this large-scale model, a specific nested sub-model was constructed. The flow dynam-
ics were simulated considering only the effects of temperature on the hydraulic heads. The
aim was to study the relationship that may exist in the Dax area between ascending thermal
water fluxes and those of the cold Adour River alluvial aquifer. This was done to prove or
invalidate the existence of a potential risk of degradation of the thermal water quality. In the
outcrop area of the thermal dolomitic body, there were four layers representing the aquifers:
the Quaternary layer, a marl cover corresponding to the Eocene aquiclude, the Upper
Cretaceous (Senonian) thermal aquifer and the salt dome. The local model covers an area of
1.4km
2
. It extends 1 600m eastward and 900m northward. The grid was done with trian-
gular elements with 8m sides. Under natural conditions (not influenced by pumping), the
simulated piezometric heads of the thermal aquifer are always higher than those of the allu-
vial aquifer (water table) thus preventing any pollution of or mixing with the mineral water.
Without pumping (as in the 19th century), there was a series of natural springs (about thirty
natural hot springs), the main one being the Fontaine Chaude spring (Figure 5). The exis-
tence of these springs demonstrates that the pressure in the thermal aquifer was higher than
that of the overlying alluvial aquifer. During this period, many inhabitants of Dax had
warm water wells in their gardens (Daguin 1948; Dupouy-Camet 1952).
All these springs and warm wells have disappeared following the commissioning of
the major abstraction wells. Since the 1960s, pumping has created a cone of depression
depriving the natural warm water springs from their source. Nowadays, there are no longer
any natural artesian phenomena.
6.2 Influenced state
Under pumping conditions, the piezometric heads of the thermal aquifer have fallen below
those in the alluvial aquifer (3.5m a.s.l.), especially near the thermal wells. There might,
therefore, be a potential risk of degradation of mineral water quality under intensive opera-
ting conditions (Figure 6). In the central part of the thermal aquifer, between Baignots and
Fontaine Chaude (FC in Figure 6), the simulated heads show a zone with higher values
than in the other parts.
In order to implement improved management of the thermal resource, it is important to
have a tool making it possible to identify: (i) preferential mixing areas, (ii) risk zones, (iii)
maximum pumping rates without deteriorating the quality of the resource.
Risk zones were defined where the local thickness of the sediments deposited at the
top of the thermal dolomitic body does not exceed 10m. They correspond to the historical
zones of the springs (Figure 5): the Splendid and the Fontaine Chaude as well as an area of
100m to the East of Baignots where there were springs before pumping began (the Est,
Seris, and Demi Lune springs) (Counilh, 1985).
Before the period of intensive pumping, the potential risk zones corresponded to emer-
gences, but since then the role of these zones has changed. They are now either a potential
discharge system or entrance points. These sectors seem to be the most sensitive part of the
actual Dax thermal zone. With the hydrogeologic model, a systematic study of the flow
paths of the water fluxes was carried out to assess the risk zones. These zones now have
relatively modest surface area and are all on the northern side of the thermal dolomitic
body (Figure 7). Because of the high permeabilities that characterize the thermal aquifer
in its upper part, they present a potential risk of contamination. Regarding the protection
of the thermal resource, therefore, major construction work, or work affecting the ground
of any other nature, could affect the integrity of the cover and should be avoided. This
study of the emergence zone makes it possible to describe the risk zones of the upper part
of the thermal aquifer. They correspond mostly to the high points of the aquifer, which
were the natural outlets before pumping began. These risk zones are very sensitive.
Figure 5. Position map and historical list of springs (natural and wells) in the Dax area (Le Fanic,
2005).
7 MANAGING TOOL
To quantify the potential risk of disturbance of the thermal resource, the capture zones of
active thermal wells were simulated by a particle tracking method. Figure 8 illustrates
some of these catchments, in particular those of the Fontaine Chaude (FC), Place de la
Course (PC) and Elvina wells. This result is confirmed by the physicochemical data from
three wells with a similar history.
Figure 6. Simulation of piezometric heads of the thermal aquifer under influenced conditions (Le
Fanic, 2005).
Figure 7. Potential risk zones of contamination of the thermal resource (Le Fanic, 2005).
This study also demonstrated the importance, for better thermal resource management,
of identifying any interference that may exist between the abstraction wells. The model is
a useful tool to assess the impact of a proposed well or to improve on-site maintenance of
an existing well. This nested model is adapted to simulate phenomena of significant mag-
nitude and rapidity in the highly permeable fractured aquifer.
8 CONCLUSION
The numerical model can be used as a local management tool for the thermal water
resource in order to develop protection policies. This type of modelling could be general-
ized and applied to major hydrothermal and mineral springs elsewhere. The objectives are
to manage the groundwater resource and preserve its quantity and quality. The modelling
will assist in answering questions from the Academy of Medicine and the Medical
Faculties about the preservation of the original qualities of the resource.
ACKNOWLEDGMENTS
The authors would like to thank the Conseil Rgional dAquitaine for the financial support
of this project, Gaz de France, Thermes Adour and the Commune de Saint-Paul-ls-Dax
for access to their data and wells.
REFERENCES
BRGM, ELF RE, ESSO REP, SNPA (1974) Atlas gologique du Bassin dAquitaine, 27 pl.
Cerepi A, Barde J-P and Labat N (2003) High-resolution characterization and integrated study of a
reservoir formation: the danian carbonate platform in the Aquitaine Basin (France), Marine and
Petroleum Geology, 20, 11611183.
Figure 8. Simulated capture areas of all the present thermal wells (Le Fanic, 2005).
Counilh F (1985) Captages, temprature et rsistivit des eaux minrales de Dax, PhD Thesis,
Universit Bordeaux 2, France, 128 pp.
Daguin A (1948) LAquitaine occidentale, 230 pp.
Debieche T H, Guglielmi Y and Mudry J (2002) Modeling the hydraulical behavior of a fissured-
karstic aquifer in exploitation conditions, J. of Hydrology, 257, 247255.
Diersch H J G (2002) FEFLOW Reference Manual, 277 pp.
Dupouy-Camet J (1952) Recherches structurales sur les accidents triasiques du sud-ouest de
lAquitaine, Bulletin de la carte gologique, 233, XLIX: 249531.
Dupuy A (2001) Global modelling of the behaviour of a complex geothermal system: example of
Dax area, S-W of France Heuristic model, New approaches characterizing groundwater flow, in
New approaches characterizing groundwater flow, Proceedings of the XXXI IAH congress
Munich, Seiler and Wohnlich (eds) Balkema, 2: pp 931934.
Le Fanic R (2005) Hydrogologie dun systme thermal et modlisation couple hydrodynamique
thermique en vue de la gestion de la ressource: Application au systme de Dax Saint-Paul-ls-
Dax, PhD Thesis, Universit de Bordeaux 3, France, 285 pp.
Moro S (1999) Siougos: interprtation de lessai de moyenne dure de Gourbera 1, Rapport, Gaz de
France, m.DPS-SMO G 54895SM, 9 p.
Pouchan P (1979) Le gisement hydrothermal de Dax, Bordeaux Mdical, 15, 3, 10091010.
Pouchan P, Canellas J and Nguyen Ba C (1991) Aspect de lhydrothermalisme landais, Journal
Franais dHydrologie, 22, 1, 916.
Pouchan P (1995) Lmergence thermominrale, La Houille Blanche, 23, 4650.
Rocher M, Lacombe O, Angelier J, Deffontaines B and Verdier F (1999) Cenozoic folding and fault-
ing in the South Aquitaine Basin (France): insights from combined structural and paleostress
analyses, J. of Structural Analyses, 22, 627645.
E. Aquifer contamination studies
CHAPTER 29
Urban hydrogeology of the Paranhos sector, Porto city
(NW Portugal): A geoenvironmental perspective
Maria Jos Afonso
1,2
, Jos M. Marques
3
, Laura Guimares
4
, Isilda
Costa
2,1
, Jos Teixeira
2
, Clara Seabra
1
, Fernando T. Rocha
2
, Lcia
Guilhermino
4,5
& Helder I. Chamin
1,2
1
Dep. Eng. Geotcnica, ISEP, R. Dr. A. Bernardino Almeida, Porto, Portugal
2
Centro de Minerais Industriais e Argilas, Univ. Aveiro, Aveiro, Portugal
3
Dep. Eng. Minas e Georrecursos, IST, Lisboa, Portugal
4
CIIMAR/CIMAR-LA, Laboratrio de Ecotoxicologia, Porto, Portugal
5
ICBAS, Dep. Estudos de Populaes, Lab. Ecotoxicologia, Univ. Porto, Porto, Portugal
ABSTRACT: This study highlights the importance of urban hydrogeological mapping, under a
geoenvironmental perspective, as well as the vital significance of an inventory of potential surface con-
tamination sources, as useful tools for understanding the potential contamination of the Porto City
groundwater systems (NW Portugal, Iberian Massif). Paranhos spring horizontal galleries catchworks
(Arca dgua sector; ca. 1km extension and a 20m of depth) was one of the main water supplies to
Porto city for more than five centuries and, nowadays, these waters are still appropriate for irrigation
uses. Therefore, the following steps were performed: i) for the characterisation of the underground area,
mainly granitic rocks, subsurface geological and hydrogeological mapping were carried out; ii) hydro-
geological, hydrogeochemical and ecotoxicological techniques were used to describe the groundwa-
ter; iii) potential sources of groundwater contamination located in the surrounding areas of the galleries
catchworks were registered and mapped.
1 INTRODUCTION
Urbanisation has increasingly intensified in the last few decades and now some 50% of the
worlds population lives in urban centres (Foster, 1996, 1997, Chilton, 1997). Urban geo-
logy is an interdisciplinary field in geo- and socio-economic sciences addressing Earth-
related problems in urbanised areas. Apart from the more traditional disciplines such as
geological mapping, structural geology and geomorphology, some interdisciplinary fields
such as hydrogeology, engineering geology, and environmental geology play a dominant
role in urban geosciences. Geology exercises critical control over most phases of water use
and supply so that, without adequate geological information, planning regarding water man-
agement cannot be carried out effectively and may even, in extreme cases, be bad planning
(Legget, 1973).
Since water is an integral part of the environment, the hydrogeology of cities can be
seen as a vital key issue in all successful urban planning. A city cannot exist without water
and obtaining it in sufficient quantity and good quality, both for domestic and industrial
use, is a primary civic service. Urbanisation can impact the hydrological cycle deeply and
at different scales. The urban subsurface is a network of pipes, conduits and other struc-
tures that alter the natural hydraulic conductivity of the ground. These infrastructures were
built to facilitate transportation, drainage, sewage disposal and the water supply system for
the population. Aquifers characterised by the presence of abundant, but vulnerable
groundwater, are in many cases located below major cities, where the poor knowledge of
aquifer characteristics, uncontrolled exploitation and indiscriminate effluent and waste
disposal practices contribute to groundwater resources degradation (e.g., Foster, 1996,
Custodio, 1997, Morris et al., 1997, Foster et al., 1999, Chilton, 1999, Lerner, 2004,
Zaadnoordijk et al., 2004).
Groundwater quality is threatened mainly by human activities, although harmful sub-
stances are sometimes introduced by natural processes. Sustainable groundwater manage-
ment must be based not only on prevention of the overexploitation of groundwater
resources but also on prevention of contamination because the latter, unlike treatment at
the point of use, protects the whole resource. The sources and causes of groundwater con-
tamination are numerous and as diverse as human activities (e.g., Fetter, 1993, Vrba and
Zaporozec, 1994, Eaton and Zaporozec, 1997, Zaporozec and Miller, 2000, Zaporozec,
2004). With the growing number of potential contamination sources, attempts to classify
them have also increased. Agencies and scientists in the field of groundwater contam-
ination have often followed their own categorisation. A large number of source classifica-
tion methods have been developed and sub-classifications have also been proposed
(Zaporozec, 2004).
In contrast to topographic and geographic maps, which are in general acceptance, the
value of thematic maps (hydrogeological maps included) is often disputed. Hydrogeological
maps can be very powerful tools for conveying information, particularly for planning and
management as well as for education and public information (Struckmeier and Margat,
1995). Hydrogeological maps which give a picture of groundwater-related information
have been around for many years. Examples of such maps are plentiful (e.g., UNESCO,
1983, Struckmeier et al., 1986, Struckmeier, 1989, Struckmeier and Margat, 1995, NRC,
1997). The tendency for hydrogeological maps is to have a large central base map showing
some hydrogeological variables. On such maps the results of surveys of potential
subsurface contaminant load can be superimposed, to facilitate the assessment of ground-
water pollution hazards. Well-executed fieldwork is the key to obtaining reliable and
good quality data needed for an understanding of the basic geology and hydrogeology of the
site and of the types of contaminants and their behaviour in the subsurface media
(Zaporozec, 2004).
In this work, pilot hydrogeochemical and ecotoxicological analyses conducted to assess
the quality of Paranhos spring galleries catchworks demonstrated the need to ascertain the
potential sources of contamination of this groundwater. Hence, using a GIS technique, an
inventory of surface contamination sources of this aquifer was performed, based on both
the character of the source and its characterisation by origin. This methodology proved
valuable in deepening our understanding of the susceptibility of Porto City (NW Portugal,
Iberian Massif) groundwater systems. In addition, it can provide guidelines for the
planning and management of water resources exploitation in an equitable, sustainable and
ethical manner.
2 GEOLOGICAL AND HYDROGEOLOGICAL SETTING
Porto city is the capital of the second biggest urban area in Portugal, with an area of ca. 42km
2
and a population of 260,000 inhabitants. Porto is located on the banks of the Douro River,
in northern Portugal, which is one of the oldest cities in Europe. The history of Porto City
dates back at least to the 6th century, since the days of Suevians. The conquest of the so-called
Portucale, the previous designation for the Porto area, in 868 A.D. is considered a major
event of the most ancient history of Porto. However, as of 868 A.D. it became the centre
of the movement of Christians re-conquest of Iberian Peninsula (Oliveira Marques, 1972).
Porto City became an important conurbation since the 12th century and has been developed
on granitic hill slopes of the Douro River.
The Porto urban area is located in a complex geotectonical domain of the Iberian Massif
on the so-called Ossa-Morena Zone and Central-Iberian Zone boundary (Ribeiro et al.,
1990) along the western border of the PortoCoimbraTomar shear zone (Chamin, 2000,
Chamin et al., 2003). Arajo et al. (2003) pointed out that the geomorphologic configu-
ration of this region consists of a littoral platform characterised by a quite regular plana-
tion surface dipping gently to the west, culminating at around 120m a.m.s.l., and in the
east bounded by a series of hill ridges (250300m a.m.s.l. on the top). The regional fracture
network defines some preferential orientations: NNW-SSE is dominant, and more discreetly,
NE-SW; while the predominant dip of the discontinuities is vertical to sub-vertical. The
regional geotectonic framework of Porto urban area (e.g., Sharpe, 1849, Barata, 1910,
Carrngton da Costa, 1938, 1958, Almeida, 2001, Chamin et al., 2003, and references
therein) comprises a crystalline fissured basement complex of highly deformed and over-
thrusted Late Proterozoic/Palaeozoic metasedimentary rocks and granites.
The crystalline bedrock of the study area consists mainly of granitic rocks, representing
the so-called Porto granite facies (Almeida, 2001): two-micas, medium- to coarse-grained,
deformed, greyish in colour, changing to yellowish when weathered. The granitic facies in
this area is, generally, weathered to different grades, from fresh-rock to residual soil, show-
ing highly variable conditions, resulting in arenisation and kaolinisation, which may reach
depths of more than 100m (e.g., Begonha, 2001, Gaj et al., 2003). The local fracture net-
work is intense and defines some preferential orientations: NNE-SSW to NE-SW, namely
N2030E, with a dip of 7080 towards NW. Several fault sets were also recognised
with a mean orientation of NE-SW, N-S and a sub-horizontal set.
The regional hydrogeological units described in Porto metropolitan area are presented
in Figure 1 and Table 1, facing their future use as important tools for the sustainable manage-
ment of local water resources. The negative impact on the quality of groundwater resources
can require considerable time before being detected, since the response time of ground-
water systems is the longest of all components of the urban hydrological cycle. Thus, in these
areas, hydrogeological data acquisition namely morphotectonic and geological, is, generally,
the key to predicting the possible negative impacts of surface watergroundwater interaction.
The intensity of the impacts usually depends on the susceptibility to pollution of underlying
aquifers and is directly connected with the type and stage of urban development. For more
than five centuries, the water supply of Porto city was possible through fountains which
were fed by numerous springs (e.g., Bourbon e Noronha, 1885, Ferreira da Silva, 1889,
Fontes, 1908, Carteado Mena, 1908, Carrngton da Costa, 1938, Afonso, 2003, Afonso
et al., 2006). Several underground galleries were excavated throughout the last five centuries
Urban hydrogeology of the Paranhos sector, Porto city (NW Portugal) 393
to gather the water from these springs. Paranhos spring collection chambers constituted
one of the main galleries.
3 METHODOLOGY
In order to assess and understand the potential anthropogenic and natural contamination of
the Porto City groundwater systems, namely Paranhos spring horizontal galleries catch-
works, this study followed a geoenvironmental approach, where hydrogeological mapping
Figure 1. Regional hydrogeological setting from Porto metropolitan area (geological background
adapted from Pereira et al., 1989, Chamin, 2000 and Chamin et al., 2003).
T
a
b
l
e

1
.
R
e
g
i
o
n
a
l

h
y
d
r
o
g
e
o
l
o
g
i
c
a
l

u
n
i
t
s

a
n
d

r
e
l
a
t
e
d

f
e
a
t
u
r
e
s

i
n

t
h
e

P
o
r
t
o

m
e
t
r
o
p
o
l
i
t
a
n

a
r
e
a

(
s
e
e

f
i
g
u
r
e

1
)
.
R
e
g
i
o
n
a
l

H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
H
Y
D
R
O
G
E
O
L
O
G
I
C
A
L

F
E
A
T
U
R
E
S
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l
U
n
i
t
s
C
o
n
n
e
c
t
i
v
i
t
y

t
o

t
h
e

T
y
p
e

o
f

f
l
o
w
W
e
a
t
h
e
r
i
n
g
M
o
r
e

s
u
i
t
a
b
l
e

G
r
o
u
p
s
d
r
a
i
n
a
g
e

n
e
t
w
o
r
k
e
x
p
l
o
i
t
a
t
i
o
n

s
t
r
u
c
t
u
r
e
s
w
i
t
h
w
i
t
h
o
u
t
p
o
s
s
i
b
l
e
p
o
r
o
u
s

f
i
s
s
u
r
e
d

l
o
w

h
i
g
h

c
l
a
y
e
y
s
a
n
d
y
d
u
g
-
w
e
l
l
s
,

b
o
r
e
h
o
l
e
s
m
e
d
i
u
m
m
e
d
i
u
m
t
h
i
c
k
n
e
s
s
t
h
i
c
k
n
e
s
s
g
a
l
l
e
r
i
e
s

a
n
d

s
p
r
i
n
g
s
S
e
d
i
m
e
n
t
a
r
y

S
a
n
d
s

a
n
d

a
l
l
u
v
i
u
m
x
x
n
.
a
.
n
.
a
.
n
.
a
.
n
.
a
.
x
c
o
v
e
r
s
a
n
d
s
t
o
n
e
s

a
n
d

x
x
n
.
a
.
n
.
a
.
n
.
a
.
n
.
a
.
x
c
o
n
g
l
o
m
e
r
a
t
e
s
M
e
t
a
s
e
d
i
m
e
n
t
a
r
y
Q
u
a
r
t
z
-
p
y
l
l
i
t
e
s
,

x
x
x
x
x
x
r
o
c
k
s
m
i
c
a
s
c
h
i
s
t
s

a
n
d

b
l
a
c
k

s
h
a
l
e
s
Q
u
a
r
t
z
i
t
e
s

a
n
d

x
x
x
x
x
x
s
l
a
t
e
s
S
c
h
i
s
t
s
,

x
x
x
x
x
g
r
a
y
w
a
c
k
e
s

a
n
d

m
e
t
a
c
o
n
g
l
o
m
e
r
a
t
e
s
G
r
a
n
i
t
i
c

r
o
c
k
s
G
r
a
n
i
t
e
,

m
e
d
i
u
m

x
x
x
x
x
t
o

c
o
a
r
s
e

g
r
a
i
n
e
d
,

w
i
t
h

K
-
f
e
l
d
s
p
a
r

M
e
g
a
c
r
y
s
t
a
l
s
G
r
a
n
i
t
e
,

m
e
d
i
u
m

x
x
x
x
x
t
o

f
i
n
e

g
r
a
i
n
e
d
,

e
s
s
e
n
t
i
a
l
l
y

b
i
o
t
i
t
i
c
G
n
e
i
s
s
e
s

a
n
d

x
x
x
x
x
x
m
i
g
m
a
t
i
t
e
s
n
.
a
.
n
o
t

a
p
p
l
i
c
a
b
l
e
.
and an inventory of surface potential contamination sources were used as fundamental
tools. For this purpose, the following steps were taken:
i) Subsurface geological/geotechnical and hydrogeological mapping (scale: 1/1.000)
enabled characterisation of the Paranhos underground area (Arca dgua sector), using
methodologies and techniques proposed by CFCFF (1996) and Assaad et al. (2004).
ii) Description of groundwater quality, by using hydrogeochemical and ecotoxicological
(Cooney, 1995, OECD, 2000) analyses. Hydrogeological fieldwork campaigns were per-
formed and in situ determinations included: temperature (C), pH and electrical con-
ductivity (S/cm). The hydrochemical analyses were performed at Centro de Estudos
de guas / ISEP (Porto, Portugal). Sixteen sampling sites were established for hydro-
geochemical and ecotoxicological evaluation. All water samples were analysed for major
element concentrations. For the ecotoxicological evaluation, standard acute bioassays
with Daphnia magna (OECD, 2000) were carried out. Experiments were performed
according to the methodology described in Guilhermino et al. (1996). In brief, all
bioassays were carried out with animals from a single clone (clone A, sensu Baird et
al., 1989) and initiated with third to fifth brood neonates (24h old). The test medium
was ASTM hard water (ASTM, 1980) without any organic additive; the organisms were
not fed during the tests. The photoperiod was 16h L:8h D and the temperature was
20 1C. Twenty animals were used per treatment, in groups of 5 per 100ml of test
solution in a glass flask. In each bioassay, five test solutions of groundwater, with a con-
centration factor equal or lower than 2 between them, plus a control solution (100% dilu-
tion water) were prepared by dilution of the original sample with ASTM hard water
(ASTM, 1980). For collected samples showing pH values outside the range 79, two tests
were simultaneously performed: one with pH correction and another without pH correc-
tion. Each test had a duration of 48h. The measured effect was death recognised by
immobilisation of the organisms for 15 seconds after stimulation by a bright light.
Temperature and photoperiod were as described above. In all test vessels, pH, tempera-
ture, and conductivity were measured at the start of the test, and after 24 and 48h.
iii) Regarding groundwater degradation conditions, potential sources of groundwater con-
tamination occurring in the surrounding areas of the galleries catchworks were registered
and mapped. This approach was further refined through the application of geographical
information system (GIS) tools. A GIS is a powerful computer-based tool for integrating
and analysing data obtained from a wide range of sources (Zaporozec, 2004).
4 RESULTS
4.1 Geological and hydrogeological mapping
The characterisation of the underground area (ca. 1km long and at 20m depth b.g.l.),
mainly comprising granitic rocks, was done by means of subsurface geological/geotechni-
cal and hydrogeological mapping (scale: 1: 1.000). For the structural geology and geotech-
nical characterisation, the scanline sampling technique of discontinuities has been applied
to the deformed granitic rock-mass faces from the Paranhos spring galleries. Surface and
subsurface geological fieldwork surveys allowed to identify major tectonic accidents respon-
sible for groundwater circulation paths and to assess lithological and structural heterogeneities
(see figure 2).
F
i
g
u
r
e

2
.
H
y
d
r
o
g
e
o
l
o
g
i
c
a
l

m
a
p
p
i
n
g

o
f

t
h
e

P
a
r
a
n
h
o
s

s
p
r
i
n
g

g
a
l
l
e
r
i
e
s

c
a
t
c
h
w
o
r
k
s
,

A
r
c
a

d
g
u
a

s
e
c
t
o
r
.

T
h
e

s
t
r
u
c
t
u
r
a
l

d
i
a
g
r
a
m

(
I
n
s
e
t

r
i
g
h
t
)

s
h
o
w
s

t
h
e

m
a
i
n
d
i
s
c
o
n
t
i
n
u
i
t
i
e
s

s
e
t
s

o
r
i
e
n
t
a
t
i
o
n

(
n
3
4
4
)
.
4.2 Hydrogeochemical and ecotoxicological assessment
From a hydrogeochemical point of view, most of the sampled groundwaters presented near-
neutral pH values (median 6.8), the lowest rate being 5.4 and the highest 7.9. Regarding
the electrical conductivity, most of the samples presented medium electrical conductivities
(median 434S/cm), with a minimum value of 405S/cm and a maximum of 530S/cm.
Concerning chemical signatures, the studied groundwaters show a wide range; Ca
2
fol-
lowed by Na
dominate cations and anions are dominated by SO

4
2
and Cl
. By observing
the Piper diagram of Figure 3, one can conclude that most of the groundwaters are mixed
SO
4
2
/Cl
Ca
2
/Na
waters. Besides major elements, all water samples have been analysed
for minor constituents, such as nitrate. According to the Portuguese and European legisla-
tions, the majority of these samples exceed the maximum admissible value (50mg/L) for
this element. The SAR (sodium adsorption ratio) plot of groundwater chemistries showed
that these waters have a medium salinity and low sodium concentrations. Concerning the
results of the ecotoxicological evaluation of Paranhos groundwater, within all test vessels pH,
temperature, and conductivity variation was always lower than 1 unit. Mortality recorded
in control solutions was less than 5% (Figure 4, upper left). No mortality was recorded at
Figure 3. Piper diagram for water samples collected from Paranhos spring galleries catchworks,
Arca dgua sector (Porto city).
S
u
r
v
i
v
a
l

(
%
)
Concentration
Figure 4. Results obtained for four of the acute bioassays performed to assess the ecotoxicological
quality of Paranhos groundwater, Arca dgua sector (Porto city).
the lowest concentrations of groundwater in any of the tests, whereas occasional lethality,
always around or below 10%, was recorded at concentrations greater than 25% groundwater
(Figure 4). The low mortality observed in the tested groundwater concentrations usually
occurred in the tests without pH correction, and was no longer observed in the corresponding
pH corrected tests (Figure 4, upper and lower right). These results are in good agreement
both with the low pH values registered at some sample sites and the high sensitivity exhib-
ited by Daphnia magna to acidic environments (Peters 1987).
4.3 Contamination source inventory
The results led us to inquire which potential sources of contamination could explain the
high nitrate and sulphate contents of this groundwater. In order to assess existing or poten-
tial impacts of human activities on groundwater, we proceeded to prepare an inventory of
all existing and potential sources of contamination their location, type, characteristics,
and estimated magnitude of impact on groundwater in the surrounding area of Paranhos
spring galleries catchworks. The inventory of contamination sources included three major
phases: i) recording of elements of inventory design; ii) inventory implementation; and iii)
evaluation and analysis of the elements of the groundwater contamination inventory. An
assessment of both the existing and potential sources of contamination and the spatial
extent of the present groundwater contamination is needed before considering how to
monitor and prevent future groundwater quality problems. In this paper, a classification of
the contamination sources based on character (Table 2), i.e. point, line and diffuse sources,
was used, although the source categorisation by origin was also applied (Zaporozec 2004).
On the map of Figure 5, the contamination sources and the areas of existing contamination
were labelled according to whether they were rated high, moderate, or low. A letter indi-
cates the rating of the potential contamination source: H high, M moderate, and L
low. All identified contamination sources are summarised in Table 2 and Figure 5.
5 DISCUSSION
The hydrogeochemical analyses performed in this work have shown that this groundwater is
nitrate and sulphate-enriched, and thus unsuitable for human consumption. Nevertheless,
the SAR plot of groundwater chemistry permitted to conclude that these waters induce a
medium salinity and a low sodium hazard for irrigation purposes. In contrast to hydro-
geochemical analyses, the ecotoxicological assessment provides a means, both of appraisal of
Table 2. Groundwater potential contamination sources around Paranhos spring galleries catch-
works, Arca dgua sector.
Category Source Usual character: Inventory Potential groundwater
type point (P); line numbers contamination load:
(L); diffuse (D) H (high), M (moderate),
L (low)
Agriculture Public garden D 7 M-H
with pond
Public garden D 11 M
Urbanisation On-site sanitation P 9, 33 L
Hospital P 15, 16 M
Ventilation shaft P 18, 19, 23,
30, 31, 32
Spring galleries P 24, 25, 26, 28,
entrance 29, 34, 35
Industry Garage P, L 1, 2, 3, 5, 6, 8, H
12, 14, 22
Metallurgy P 10 M
Petrol station P, L 27 H
Cable wire factory P 13 M
Water Production well P 20 M-H
mismanagement Abandoned well P 21
Miscellaneous Riding school D 4 M-H
Artificial lake D 17
Watercourse L 36
canalized
Spring galleries L 37 H
F
i
g
u
r
e

5
.
S
k
e
t
c
h

o
f

g
r
o
u
n
d
w
a
t
e
r

p
o
t
e
n
t
i
a
l

c
o
n
t
a
m
i
n
a
t
i
o
n

s
o
u
r
c
e
s

a
r
o
u
n
d

P
a
r
a
n
h
o
s

s
p
r
i
n
g

g
a
l
l
e
r
i
e
s

c
a
t
c
h
w
o
r
k
s
,

A
r
c
a

d
g
u
a

s
e
c
t
o
r

(
t
o
p
o
g
r
a
p
h
i
c

c
o
n
t
o
u
r
i
n
t
e
r
v
a
l
1
0
m
)
.
environmental effects resulting from complex mixtures of pollutants, and extrapolation
between chemical concentrations and likely biological effects. Within this context, it is
worth noting that the European Council Directive on the protection of groundwater against
pollution and deterioration (PE-CONS 3658/06) recommends that ecotoxicology knowl-
edge should be taken into account for the evaluation of the chemical status of groundwater.
Standard acute bioassays are test methods used for the toxicity assessment and monitoring
of groundwater and its impacts on associated aquatic ecosystems (e.g., EPA, 1993,
Connon et al., 2003, Dewhurst et al., 2002). These tests provide rapid information on the
degree of pollution present in the groundwater and the potential effect of that pollution on
indicator or surrogate organisms. Typically, their result is the determination of the median
lethal concentration (LC50), i.e., the concentration that causes a mortality of 50% of the
tested organisms (Cooney, 1995) in the specific conditions of the bioassay. In spite of the
high nitrate and sulphate concentrations found, no considerable mortality was observed in
any of the acute bioassays performed, suggesting a low toxicity potential for this ground-
water. The results obtained so far do not discard, however, the possible occurrence of more
subtle effects on organisms, such as changes in growth and/or reproduction, or seasonal
patterns of toxicity, possibly associated with seasonal alterations and/or human activities.
These aspects will be investigated in the next steps of the study. These earlier pilot studies
raised concern over what potential contamination sources could contribute to explain the
results obtained so far. To answer this question we developed a potential contamination
source inventory, which provided a useful basis for planning the quality assessment of the
Paranhos spring galleries catchworks, especially when cross-linked with: i) the regional
and local hydrogeological environments, ii) the structural geology features and iii) the
geotechnical fractured rock mass behaviour. In agreement with the adopted methodology,
most of the contamination sources were mapped as being point sources; nevertheless some
of them were considered to be diffuse and line sources. Results of the contamination
source inventory indicated that the greatest potential problems are the leaking under-
ground storage tanks, spills of hazardous materials (mostly concentrated along main roads
and streets), garages, ventilation shafts, and latrines. Moreover, several abandoned domes-
tic wells are assumed to exist within areas now served by municipal water systems.
Individual on-site septic-tank systems were not catalogued. The most difficult part of the
inventory was the identification of abandoned wells. As the public water systems were
built and their areas expanded, the domestic wells were often abandoned without being
proper closure and sealing. In the case of subsurface contaminant load, the largest and
most sophisticated facilities are not necessarily the ones which generate the largest subsur-
face contaminant load and the highest groundwater contamination risk. This is because
their chemical handling and effluent disposal practices are more carefully controlled and
monitored. Equal or greater concern is associated with small services, and activities,
because they are widely disseminated, often use considerable quantities of potentially
toxic groundwater contaminants, and their effluent disposal practices may not be subject to
strict control. Regarding the high concentrations of sulphate and nitrate found in these
groundwaters they can be explained by the interaction of surface water and groundwater.
In addition, the origin of sulphates might be attributed to air pollution (Afonso et al., 2007)
atmospheric gaseous SO
2
and particulate matter and to detergents (Foster, 1997).
Specifically, geoenvironmental and ecotoxicological studies are required to assess poten-
tial variations in water composition and toxicity associated with seasonal changes in cli-
mate and human activities.
The delineation of the potential problematic areas was not intended to suggest that these
were the only areas where potential groundwater quality problems may occur. All contami-
nation sources listed in the inventory have to be considered as having a potential to contam-
inate groundwater. The map of potential problematic areas was drawn up only as a planning
tool to rank the possible problems and to indicate which areas should be addressed first in
a regional water quality management plan. To avoid misinterpretation, it is important to stress
that the developed method is qualitative and should be used as a planning and screening tool
and that it cannot replace site-specific, detailed investigations. Indeed, the local approach
methodology described above is regarded as the first step in a groundwater contamination
risk assessment, and is intended to prioritise areas, but not to substitute for systematic field
inspection and monitoring. In pursuit of these objectives, the inventory of potential contami-
nation sources presented here will now be used to plan additional hydrogeochemical and
ecotoxicological field studies required to assess variations in water composition and toxi-
city resulting from seasonal changes in climate and human activities.
Multidisciplinary geoengineering approaches probably offer the best starting point for
reliable groundwater and surface water studies and for assessment of geospatial parameters
variability, such as, lithological heterogeneity, structural geology features, hydrogeology and
hydrobiology of a specific site. Thus, combined consistent hydrogeological and ecotoxico-
logical local data as well as GIS technology, offer promises of a better understanding of the
hydrological functioning of large urban areas.
ACKNOWLEDGMENTS
This study was performed within the scope of the GROUNDURBAN Urban groundwater
and environmental management in the Northwest Portugal R&D Project granted by the
Portuguese Foundation for Science and Technology (FCT) and FEDER EU funds, contract
POCTI/CTE-GIN/59081/2004. Special acknowledgements are due to V. Beleza (Centro de
Estudos de guas, ISEP, Porto) for his support on the hydrochemical analyses. This paper
is a LABCARGA I&D (Laboratrio de Cartografia e Geologia Aplicada, ISEP) contribu-
tion. We acknowledge Ccile Garnier, an anonymous referee and Nick Robins for construc-
tive reviews that helped to improve the clarity of the manuscript.
REFERENCES
Afonso MJ (2003) Hidrogeologia de rochas granticas da regio do Porto (NW de Portugal)
[Hydrogeology of granitic rocks from Porto region (NW Portugal)]. Cadernos Laboratrio Xeolxico
de Laxe, Corua, 28,173192.
Afonso MJ, Chamin HI, Gomes A, Fonseca PE, Marques JM, Guimares L, Guilhermino L,
Teixeira J, Martins Carvalho J, Rocha FT (2006) Urban hydrogeomorphology and geology of the
Porto Metropolitan Area (NW Portugal). Pre-Congress Proceedings, The 10th IAEG International
Congress Engineering geology for tomorrows cities, Nottingham, U.K. The Geological Society
of London. 19 pp. (CD-Rom)
Afonso MJ, Chamin HI, Carvalho JM, Marques JM, Gomes A, Arajo MA, Fonseca PE, Teixeira J,
Marques da Silva MA, Rocha FT (2007) Urban groundwater resources: a case study of Porto met-
ropolitan area (Iberian Massif, NW Portugal). In: Howard KWF (ed). Urban Groundwater: Meeting
the Challenge. International Association of Hydrogeologists Selected Papers, Taylor & Francis
Group, London, 8, 271287.
Almeida A (2001) Caracterizao geoqumica e geocronolgica do granito de duas micas sintec-
tnico do Porto (NW de Portugal) [Geochemical and geochronological characterization of the
syntectonic two-mica granite of Porto (NW Portugal)]. In: Proceedings III
er
Congreso Ibrico de
Geoqumica. Instituto Tecnolgico de Aragn, Zaragoza, pp. 311315.
Arajo MA, Gomes A, Chamin HI, Fonseca PE, Gama Pereira LC, Pinto de Jesus A (2003)
Geomorfologia e geologia regional do sector de Porto-Espinho (W de Portugal): implicaes mor-
foestruturais na cobertura sedimentar Cenozica [Regional geomorphology and geology from the
PortoEspinho sector (W Portugal): morphostructural implications to cenozoic sedimentary
cover]. Cadernos Laboratrio Xeolxico de Laxe, A Corua, 28, 79105.
Assaad FA, LaMoreaux PE, Hughes TH, Wangfang Z, Jordan H (2004) Field methods for geologists
and hydrogeologists. Springer-Verlag. 420 pp.
ASTM American Society for Testing and Materials (1980) Standard practice for conducting acute
toxicity tests with fishes, macroinvertebrates and amphibians. Report E 79080. American
Society for Testing and Materials, Philadelphia.
Baird DJ, Soares, AMVM, Girling A, Barber I, Bradley M, Calow P (1989) The long-term mainte-
nance of Daphnia magna Straus for use in ecotoxicity tests: problems and prospects. In: Lokke H,
Tyle H, Bro-Rasmussen F (eds), Proceedings First European Conference on Ecotoxicology,
Lyngby, pp. 144148.
Barata JMP (1910) Contribuio para o estudo das rochas do Porto [Contribution to the study of the
Porto rocks]. Faculdade de Philosophia Natural, Coimbra, 59 pp.
Begonha A (2001) Meteorizao do granito e deteriorao da pedra em monumentos e edifcios da
cidade do Porto [Granite weathering and stone deterioration in monuments and buildings of the
Porto city]. Coleco monografias, FEUP Edies, Porto, 2, 1445.
Bourbon e Noronha T (1885) As aguas do Porto [Porto waters]. Dissertao inaugural apresentada
Escola Mdico-Cirrgica do Porto. Typ. Occidental, Porto, 89 pp.
Carteado Mena J (1908) Contribuio para o estudo da Hygiene do Porto: analyse sanitaria do seu
abastecimento em gua potvel. III. Estudo sobre os poos do Porto [Contributions to the hygiene
of the Porto city: sanitary analysis for the potable water-supply. I. Study about Porto dug-wells].
Dissertao inaugural apresentada Escola Mdico-Cirrgica do Porto. Typ. Encyclopedia
Portugueza, Porto, 270 pp.
Carrngton da Costa J (1938) O Prto: geografia-geologia [Porto: geography-geology]. Nova
Monografia do Porto, pp. 332.
Carrngton da Costa J (1958) A geologia da regio portuense e os seus problemas [Porto region geo-
logy and their problems]. Boletim da Academia Cincias Lisboa, 30, 3658.
Chamin HI (2000) Estratigrafia e estrutura da faixa metamrfica de Espinho-Albergaria-a-Velha
(Zona de Ossa-Morena): implicaes geodinmicas [Stratigraphy and structure of the Espinho-
Albergaria-a-Velha metamorphic belt (Ossa-Morena Zone): geodynamic implications].
Faculdade de Cincias da Universidade do Porto, 497 pp. (PhD Thesis)
Chamin HI, Gama Pereira LC, Fonseca PE, Noronha F, Lemos de Sousa MJ (2003)
Tectonoestratigrafia da faixa de cisalhamento de PortoAlbergaria-a-VelhaCoimbraTomar,
entre as Zonas Centro-Ibrica e de Ossa-Morena (Macio Ibrico, W de Portugal)
[Tectonostratigraphy of the PortoAlbergaria-a-VelhaCoimbraTomar Shear Zone between
Central-Iberian and Ossa-Morena Zones (Iberian Massif, W Portugal)]. Cadernos Laboratrio
Xeolxico de Laxe, A Corua 28, 3778.
Chilton J (1997) Groundwater in the urban environment: selected city profiles. A. A. Balkema.
342 pp.
Chilton J (1999) Groundwater in the urban environment: problems, process and management. A. A.
Balkema. 682 pp.
CFCFF Committee on Fracture Characterization and Fluid Flow (1996) Rock fractures and fluid
flow: contemporary understanding and applications. National Research Council, National Academy
Press, 568 pp.
Connon R, Dewhurst RE, Crane M, Callaghan A (2003) Haem peroxidase activity in Daphnia
magna: A biomarker for sub-lethal toxicity assessments of kerosene-contaminated groundwater.
Ecotoxicology, 12, 5, 387395.
Cooney JD (1995) Freshwater tests. In: Rand GM (ed), Aquatic toxicology: effects, environmental
fata, and risk assessment (2nd edition), London, pp. 71102.
Custodio E (1997) Groundwater quantity and quality changes related to land and water management
around urban areas: blessings and misfortunes. In: Chilton J et al. (eds.) Proceedings of the 27th
IAH Congress on Groundwater in the urban environment: Problems, processes and management.
Balkema, Rotterdam, 1, 1122.
Dewhurst RE, Wheeler JR, Chummun KS, Mather JD, Callaghan A, Crane M (2002) The compari-
son of rapid bioassays for the assessment of urban groundwater quality. Chemosphere, 47, 5,
547554.
Eaton TT, Zaporozec A (1997) Evaluation of groundwater vulnerability in an urbanizing area. In:
Chilton, J. et al. (eds.), Groundwater in the Urban Environment, Volume 1, Problems, processes
and management. A. A. Balkema, Rotterdam, pp. 57782.
EPA Environmental Protection Agency (1993) Methods for measuring the acute toxicity of efflu-
ents and receiving waters to freshwater and marine organisms, 4th edition, Office of Research and
Development, Washington, DC, EPA/600/4-90/027F.
Ferreira da Silva AJ (1889) Contribuies para a hygiene da cidade do Porto [Contributions to the
hygiene of the Porto city]. Typ. Antnio Jos da Silva Teixeira, Porto, 263 pp.
Fetter CW (1993) Contaminant hydrogeology. Macmillan Publishing Comp., New York, NY, USA.
500 pp.
Fontes A (1908) Contribuio para a Hygiene do Porto: Analyse sanitaria do seu abastecimento em
gua potvel. I. Estudo dos Mananciais de Paranhos e Salgueiros [Contributions to the hygiene of
the Porto city: sanitary analysis for the potable water-supply. I. Study of the Paranhos and
Salgueiros springs]. Dissertao inaugural apresentada Escola Mdico-Cirrgica do Porto. Typ.
Encyclopedia Portugueza, Porto, 172 pp.
Foster S (1996) Groundwater quality concerns in rapidly-developing cities. In: Guswa JH (ed.),
Hydrology and hydrogeology of urban and urbanizing areas. American Institute of Hydrology, St.
Paul. pp. MIU12MIU26.
Foster S (1997) The urban environment. Evaluation of hydrological changes and their consequences.
In: Proceedings of an International Symposium on Isotope Techniques in the Study of Past and
Current Environmental Changes in the Hydrosphere and the Atmosphere. International Atomic
Energy Agency, Vienna, pp. 321338.
Foster S, Morris B, Lawrence A, Chilton J (1999) Groundwater impacts and issues in developing
cities: an introductory review. In: Chilton J (eds.), Proceedings of the 27th IAH Congress
on Groundwater in the urban environment: Selected cities profiles. Balkema, Rotterdam, 21,
316.
Gaj F, Guglielmetti V, Grasso P, Giacomin G (2003) Experience on Porto: EPB follow-up. Tunnels &
Tunnelling International, pp. 1518.
Guilhermino L, Lopes MC, Carvalho AP, Soares AMVM (1996) Inhibition of acetylcholinesterase
activity as effect criterion in acute tests with juvenile Daphnia magna. Chemosphere, 32, 4,
727738.
Legget RF (1973) Cities and geology. McGraw-Hill, New York, 624 pp.
Lerner D (2004) Urban groundwater pollution. A. A. Balkema, 277 pp.
Morris BL, Lawrence AR, Foster SD (1997) Sustainable groundwater management for fast-growing
cities: mission achievable or mission impossible?. In: Chilton J et al. (eds.), Proceedings of the
27th IAH Congress on Groundwater in the Urban Environment: Problems, processes and manage-
ment. Balkema, Rotterdam, 21, 5566.
NRC National Research Council (1997) Valuing ground water: economic concepts and approaches.
Nat. Academy Press, Washington, DC, USA, 189 pp.
OECD Organisation for Economic Co-operation and Development (2000). OECD Guidelines for
testing of chemicals Revised protocol for updating Guideline 202 Daphnia sp. Acute
Immobilisation Test.
Oliveira Marques AH de (1972) History of Portugal. Vol. 1: from Lusitania to Empire. Columbia
University Press, New York, 507 pp.
PE-CONS 3658/06 (2006) The European Parliament and the Council of the European Union,
Legislative Acts and Other Instruments: directive of the European Parliament and of the Council
on the protection of groundwater against pollution and deterioration. 42 pp.
Pereira E, Ribeiro A, Carvalho GS, Noronha F, Ferreira N, Monteiro JH (1989) Carta Geolgica de
Portugal, escala 1/200.000. Folha 1 [Geological Map of Portugal, scale 1/200.000, Sheet 1].
Servios Geolgicos de Portugal, Lisbon.
Peters RH (1987) Daphnia culture. In: Peters RH, de Bernardi R (eds), Daphnia. Memorie
Dellistituto Italiano di Idrobiologia Dott. Marco de Marchi, 45, 483495.
Ribeiro A, Quesada C, Dallmeyer RD (1990) Geodynamic evolution of the Iberian Massif. In:
Dallmeyer RD and Martnez-Garca E (eds.) Pre-Mesozoic Geology of Iberia, Berlin, Heidelberg.
Springer-Verlag. pp. 397410.
Sharpe D (1849) On the Geology of the neighbourhood of OPorto, including the Silurian coal and
slates of Vallongo. Quart. Journ. Geol. Soc. London Proceedings, 5, 142153.
Struckmeier WF (1989) Types and uses of hydrogeological maps. IAH, Mem. Intern. Symp. on
Hydrogeological Maps as Tools for Economic and Social Development, Hannover, pp. 1730.
Struckmeier WF, Margat J (1995) Hydrogeological maps: a guide and standard legend. IAH, Intern.
Contr. to Hydrogeology, Heise, Hannover, 17, 1177.
Struckmeier WF, Engelen GB, Galitzin MS, Shakchnova RK (1986) Methods of representation of
water data. In: Engelen GB and Jones GP (eds.), Developments in the Analysis of Groundwater
Flow Systems, IAHS, 163, 4763.
UNESCO (1983) International legend for hydrogeological maps. Revised Edition, Unesco, Paris.
Vrba J, Zaporozec A (1994) Guidebook on mapping groundwater vulnerability. Verlag Heinz Heise,
Hannover, Germany, International Contributions to Hydrogeology, Vol. 16. Heise, Hannover.
Zaadnoordijk WJ, van den Brink C, van den Akker C, Chambers J (2004) Values and functions of
groundwater under cities. In: Lerner DN (ed.), Urban groundwater pollution, A. A. Balkema,
Lisse, 24, 128.
Zaporozec A (ed.) (2004) Groundwater contamination inventory: a methodological guide with a
model legend for groundwater contamination inventory and risk maps. IHP-VI, series on ground-
water, 2. UNESCO, Paris, France, 160 pp.
Zaporozec A, Miller JC (2000) Groundwater pollution. UNESCO, Paris, France, 24 pp.
CHAPTER 30
Modelling of chlorinated solvent transport and natural
attenuation in groundwater
Fabrice Quiot
1
, Claire Rollin
1
, Olivier Bour
1
, Salvador Jordana
2
,
Eduardo Ruiz
2
, Jordi Guimer
2
, Jean Schwartz
3
& Patrick Goblet
4
1
INERIS, Parc Technologique ALATA, Verneuil-en-Halatte, France
2
ENVIROS, Spain S.L., Passeig de Rub, Valldoreix, Spain
3
ANTEA, Direction Technique, avenue Claude Guillemin, Orlans Cedex France
4
Ecole des Mines de Paris (ENSMP), Centre dInformatique Gologique, Rue Saint Honor,
Fontainebleau, France
ABSTRACT: The objective of this study is to test different modelling approaches and the reliability of
the codes used in water pollution risk assessment for a site contaminated by chlorinated solvents. The
final goal is to establish, within the framework of the TRANSPOL research programme, guidelines that
may be used as scientific support for contaminated aquifer management. The models developed by four
teams (ANTEA, ENSMP, ENVIROS and INERIS) to simulate contamination of groundwater by per-
chloroethene were compared (calibrated results, methodology, hypotheses ) to each other. The total
amount of perchloroethene discharged into the aquifer was unknown and a simplified conceptual model
was considered. An initial synthesis of the results shows two main difficulties regarding the estimate of
source concentration and the simulation of natural attenuation phenomena (sorption/degradation). A
more accurate diagnosis is needed in order to reduce the uncertainties of the model parameters.
1 CONTEXT AND OBJECT
When risk assessment approaches are applied to contaminated sites, it is extremely important
to know the fate of pollutants in the groundwater and soil. The acquisition of this knowledge
is based on extensive investigation, available data from monitoring of pollutants in time and
space and numerical modelling of pollutant transport in the environment. In practice, results
may vary greatly among users, according to the available data and the aims of the risk assess-
ment. The lack of a reliable methodology may prevent high quality expertise. In this context,
INERIS (the French National Institute for Industrial Environment and Risks, www.ineris.fr)
has developed a research program, TRANSPOL, to promote better practice of the use of trans-
port models for different groups of pollutants. The chosen way to reach this objective is based
on the comparison of the methodologies used by private consultancies, universities and
research centres. This article presents the results of a case study, named Real Case 3, concern-
ing chlorinated solvent fate, transport and natural attenuation in groundwater. The aim of this
study is to develop a natural attenuation model to predict the fate and transport of chlorinated
solvents and their degradation products in saturated groundwater systems.
Figure 1. Site details and well locations (UTM coordinates).
Figure 2. South North cross-section of the site.
2 SITE DESCRIPTION
A contaminated alluvial aquifer in northern Spain (Figure 1) has been studied by
ENVIROS since 2000. Figure 2 presents a conceptual cross-section model of the site. The
site has a length of 1000m along the direction of groundwater flow (S-N) and is 600m
wide. A river crosses this site (the Ter River). The aquifer thickness is 16m including a 6m
thick vadose zone. The sedimentary materials are unconsolidated and very heterogeneous
(from clay to rolling cobbles), and of varied composition (from limestone and sandstone to
metamorphic and igneous rocks). Perchloroethene (PCE) is the primary solvent contami-
nant that was originally disposed of at this industrial site. However, concentrations of chlo-
rinated solvent degradation products such as trichloroethene (TCE), cis-dichloroethene
and trans-dichloroethene (cis-DCE, trans-DCE) were measured in the monitoring wells
located downstream from the source location. The observation well network is made up of
19 piezometers. Six of these boreholes are pumping wells (with pumping rates ranging
between 1100 to 70000m
3
/y in PL2 the pumping rate is unknown). There are three
potential recharge zones (Figure 1, irrigation zones are in green). The regional average nat-
ural net infiltration is approximately 150mm/y.
3 CONCEPTUAL MODEL
As a first step, a workshop was held with all modellers (Cahier des charges n1, Rollin &
Bour, 2002), to identify the processes contributing to the behaviour of the pollutant, to
select the hydrodynamic and transport parameters, and the sources appearing to be the
most likely, as would be done in a proper risk assessment. The conclusions of this work-
shop were given in a second Cahier des Charges (Rollin, 2003). This presents the con-
ceptual model, initial conditions, calibration data (heads and concentration values), which
are presented below. Some parameter (K, K
d
) values and a range of variations were
given based on other site investigations or from the literature.
3.1 Conceptual model for groundwater flow
The boundary conditions are summarised in Figure 3.
Modelling of chlorinated solvent transport and natural attenuation in groundwater 409
Figure 3. Modelled domain and boundary conditions.
Flow conditions are assumed to be at steady state. The groundwater hydraulic head val-
ues at the model boundaries were based on observed data (December 2003). The aquifer is
modelled as a single, continuous, unconfined unit with spatially variable transmissivity
values. The different transmissivity zones used in the model were based on pumping tests
and the calibration of head values (hydraulic conductivity, Table 3).
3.2 Conceptual model for contaminant transport
The source of PCE is modelled as an injection of mass directly into the saturated zone. The
injection is located inside a 30m-diameter circle centred in the middle of an industrial
building. The total amount of PCE discharged into the aquifer is an unknown quantity, as
is often the case on contaminated sites. The time simulation is 10 years:
initial time t
0
(12 December 1995) : PCE concentration is zero;
t
1
(15 November 2000): the plume is stabilised (measured concentrations -16 November
2000- were used to calibrate transport parameters);
t
0
to t
1
: the injection rate is constant during the first 5 years (R1inkg/d);
t
1
to t
2
: the injection rate starts decreasing slightly from the beginning of the 6th year
until the end of the 10th year.
High concentrations of PCE and chlorinated solvent degradation products such as
trichloroethene, cis-dichloroethene and trans-dichloroethene were observed in the moni-
toring wells. Reductive dechlorination of chlorinated solvents is, therefore, suspected. The
kinetics of all the degradation reactions (Figure 4) is assumed to be first-order kinetics.
PCE TCE DCE VC (Vinyl Chloride)
Figure 4. Anaerobic degradation pathway of chlorinated solvents.
Sorption processes could be taken into account. It was suggested that the K
d
approach
be used. Values of the transport parameter range are between different literature values
and/or estimated values obtained by the concentration calibration process:
porosity;
dispersivity;
molecular diffusion;
sorption constants (K
d
);
kinetic reaction rates of dechlorination ().
4 RESULTS AND COMPARISON
The models inter-comparison methodology is based on a comparison between:
flow (head) and transport (concentrations) calibration results;
calibrated parameter values;
hypotheses respected;
natural attenuation phenomena considered (particularly sorption and degradation).
Note that the main goal is not to decide which is the best code but to evaluate the methodo-
logy, and the hypotheses, etc. used by the modeller. Four teams were involved in the
benchmarking process. Different numerical codes, commercial or developed internally,
were used (Table 1). All the transport codes are able to consider sorption and degradation
(natural attenuation phenomenon).
4.1 Groundwater flow simulations
The relationship between the aquifer and the river was not taken into account by all the
teams. The groundwater flow codes were used to simulate the groundwater head distribu-
tion. The calibration results were assessed by using the differences between calculated and
observed heads in 16 piezometers. Errors in computed head values are listed in Table 2.
Table 3 presents the hydraulic conductivity values calibrated in the models.
Table 1. Teams and codes used in the inter-comparison benchmark.
ANTEA cole des Mines ENVIROS INERIS
de Paris
3D Flow MARTHE FD METIS FE TRANSIN FE MODFLOW 2000
Modelling V. 6.3 V1 (March 2004) V. IV FD
(Dec. 2002) and V. GMS 5.0
MODFLOW 2000 (May 2004)
FD V. GMS 4.0
3D MARTHE FD Metamodel kriging TRANSIN FE RT3D TVD
Transport V. 6.3 and V. IV V. GMS 5.0
Modelling (Dec. 2002) RT3D TVD V. (May 2004)
GMS 4.0
FE : finite elements; FD : finite differences; TVD : total variation decrease ( Zheng et al. 2001).
Table 2. Hydrodynamic calibration Error summary.
ANTEA ENSMP ENVIROS INERIS
Mean error (m) 0.04 0.14 0.09 0.04
Mean Abs. Error (m) 0.05 0.06 0.04 0.03
Root mean Sq. error (m) 0.002 0.004 0.002 0.001
Table 3. Hydrodynamic calibration Hydraulic conductivity.
Hydraulic conductivity (m/d) 109 and 150 150 73 350, 200
and 70
Recharge (mm/y) 146 irrigation 150 uniform Not defined 150 irrigation
zones zones
Pumping rate well PL2 (m
3
/d) 5 0.1 0 10
The hydraulic conductivity ranges from 70 to 350m/d. The final calibrated values fall
within the ranges suggested in the specifications (results of investigations: 60 to 600m/d).
An anisotropy factor (K
x
24m/d and K
y
32m/d) was tested by ENSMP and calibration
results were better for piezometers near the source.
4.2 Contaminant transport simulations
The models reproduced successfully the general trends of the contamination plumes but
the time of peak arrival and the concentration at the observation points may vary accord-
ing to the user and the different features of the numerical codes. Results show that with
such a simple conceptual model (homogeneous, isotropic and with a steady flow with
steady boundary conditions) it is impossible to reproduce the observed oscillating concen-
trations. Thus the concentration calibration objective was to fit to the order of magnitude.
The transport parameters used in the models are summarised in Table 4. Transport proper-
ties were estimated based on literature values (for similar types of geologic materials) and
calibration (using measured concentrations).
The differences in source volume are linked to the choice of surface and by the conside-
red injection rate (30 to 650g/d). The variations of the extent of the source and the injec-
tion rate are different and influence the range of variation of PCE (TCE, DCE and VC)
concentration in the models. Table 5 shows the differences between observed and calcu-
lated concentrations (PCE, TCE and DCE) in two piezometers located downstream of the
source. It appears that the variations are considerable between different teams.
Table 4. Parameters selected for transport calibration.
PCE Injection rate 0.65 0.17 0.17 0.03
R1(kg/d)
Source surface 707 100 91 25
(m
2
)
Effective porosity 10 30 10 20 and 30
n
e
(%)
Dynamic
L
30 22 10 20
dispersivity (m)
T
3 8.5 8 4
Molecular diffusion Not defined PCE, TCE, PCE, TCE, Not defined
(m
2
/d) DCE and DCE and
CV 1 10
4
CV 1 10
6
Bulk density of aquifer 1700 1600 Not defined Not defined
(kg/m
3
)
Sorption PCE 1.4 10
4
9.4 10
5
Not defined Not defined
coefficient K
d
TCE 5.8 10
5
9.4 10
5
(m
3
/kg)
DCE 5.2 10
5
9.4 10
5
CV Not defined 9.4 10
5
First order PCE 0.15 0.95 1.1 1.1
degradation TCE 5.11 0.04 21.68 0.73
constant DCE 0.26 0.18 6.35 0.18
(y
1
) CV Not defined 0.11 10.95 1.1
Figures 5, 6 and 7 present PCE, TCE and DCE results for six piezometers distributed
from the source downstream at 1800 days of simulation (calibration date).
The highest PCE concentrations were calculated by ANTEA. They consider higher
PCE injection rate and source concentration. This hypothesis may be justified by the fact
that PCE concentrations are very high in the aquifer near the source (but the solubility value
is not reached in observation wells, the maximum is 29mg/l in S-2). ANTEA made the
choice initially to calibrate the concentrations with piezometers near the source. Three others
models give similar results for PCE except near the source. Indeed, PCE concentrations in
S-2 and S-3 are lower for INERIS than ENSMP and ENVIROS. INERIS considered a
smaller source concentration and calibrated concentrations principally on piezometers
downstream from the source. This assumption may be justified by the fact that piezo-
meters downstream from the source integrate multiple small source contributions.
Deviations of modelled TCE and DCE concentrations compared to the measured values are
mainly due to sorption and degradation conditions. Table 4 summarises the assumed values of
sorption and degradation coefficients. Each team gave a different calibrated value. The four
teams considered degradation of chlorinated solvents but values of the first-order degradation
constant were very different. For example in the ENVIROS and INERIS models, the first
order PCE degradation constant is about 0.003d
1
(constant half-life degradation : 230d), it
Table 5. Differences between measured and calculated concentrations.
C measured C calculated (g/l) ANTEA ENSMP ENVIROS INERIS Measure (g/l)
PR-3 t1 PCE (g/l) 2383.1 201.2 480.3 0.6 135.0
TCE (g/l) 68.9 209.5 25.6 56.9 0.1
DCE (g/l) 123.6 47.5 6.6 39.2 50.1
PL t1 PCE (g/l) 1261.3 242.4 173.8 274.4 290.0
TCE (g/l) 65.5 237.9 2.7 31.4 2.2
DCE (gl) 476.8 0.5 6.6 24.2 8.3
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
6500
7000
P
C
E

c
o
n
c
e
n
t
r
a
t
i
o
n
s

t
1

(
g
/
l
)
S-2 S-3 PL PR-3 S-5 BV
ANTEA
ENSMP
ENVIROS
INERIS
Measure
(16/11/2000)
N S
Figure 5. PCE concentrations (g/l) at t
1
.
is about 0.0026d
1
for ENSMP (267d) but only 0.0004d
1
for ANTEA (1733d). Close to
the source area, observed concentrations of DCE are higher than those of TCE. Thus in the
developed models (expect for ENSMP) the degradation factor of TCE is faster than that of
DCE. Only two teams (ANTEA and ENSMP) considered sorption of chlorinated solvents.
The retardation factors considered were 3.4 (PCE), 2 (TCE) and 1.9 (DCE) for ANTEA and
1.5 (PCE, TCE and DCE) for ENSMP. Figures 8 and 9 present PCE time/space results for two
piezometers distributed downstream of the source.
0
50
100
150
200
250
300
350
400
T
C
E

c
o
n
c
e
n
t
r
a
t
i
o
n
s

t
1

(
g
/
l
)
ANTEA
ENSMP
ENVIROS
INERIS
Measure
(16/11/2000)
S N
Figure 6. TCE concentrations (g/l) at t
1
.
0
50
100
150
200
250
300
350
400
450
500
550
600
D
C
E

c
o
n
c
e
n
t
r
a
t
i
o
n
s

t
1

(
g
/
l
)
ANTEA
ENSMP
ENVIROS
INERIS
Measure
(16/11/2000)
S N
Figure 7. DCE concentrations (g/l) at t
1
.
As shown in Figure 10, ANTEA favours PCE sorption (with high PCE injection) and
ENVIROS PCE degradation (with no sorption). For ENVIROS, no retardation due to
sorption processes was considered. The values of PCE mass coming out (through wells or
boundaries) and remaining in the model were thus lower than ANTEAs. The high degrad-
ation allows calibration of the concentrations despite a high injection rate.
Figure 9. PCE concentrations (g/l) calculated for PL.
0
50
100
150
200
250
300
350
400
0 500 1000 1500 2000 2500 3000 3500
Time (d)
P
C
E

c
o
n
c
e
n
t
r
a
t
i
o
n
s

(
g
/
L
)
ANTEA
ENSMP
ENVIROS
INERIS
Measure
Figure 8. PCE concentrations (g/l) calculated for Edilkamin.
0
200
400
600
800
1000
1200
1400
1600
1800
0 500 1000 1500 2000 2500 3000 3500
Time (d)
P
C
E

c
o
n
c
e
n
t
r
a
t
i
o
n
s

(
g
/
L
)

ANTEA
ENSMP
ENVIROS
INERIS
Measure
5 CONCLUSION
Multi-species (PCE, TCE and DCE) models were used to model transport (homogeneous
and isotropic) and natural attenuation processes occurring at a chlorinated solvent release
site. Different codes were used by four modelling teams (ANTEA, ENSMP, ENVIROS
and INERIS). The model inter-comparison methodology was based on comparisons of
flow and transport calibration results, calibrated parameter values, respected hypotheses
and natural attenuation phenomena considered (particularly sorption and degradation).
Results show the difficulty to estimate :
source concentration;
sorption and degradation.
A more accurate diagnosis is needed in order to reduce the uncertainty of model para-
meters and improve the conceptual model. Thus, more accurate data were acquired during
the second semester of 2005, concerning:
the hydraulic conductivity;
the presents distribution of pollutants;
the location and rate of the source term;
the degradation and sorption rate of each substance.
Mass out (limits)
1%
Mass remaining in
the model (sorption
or migration)
53%
Mass out (wells)
25%
Mass degraded
21%
Mass out (limits)
0,1%
Mass remaining in
the model (sorption
or migration)
16%
Mass out (wells)
10,9%
Mass degraded
73%
ANTEA: high sorption
ENVIROS: high degradation
Figure 10. Comparison between ANTEA and ENVIROS PCE mass balance (kg) (0 to 1800 days).
These new data will allow a second phase of inter-comparisons to be undertaken to evalu-
ate the contribution of these parameters to the new calibration results. TRANSPOL results
are presented on the website : http://transpol.ineris.fr/
ACKNOWLEDGEMENTS
We thank the participants of the TRANSPOL programme for their contributions to the
research (ANTEA, ENSMP and ENVIROS) as well as the French Ministry of the
Environment (Ministre de lEcologie et du Dveloppement Durable) and of Sustainable
Development for its financial support.
REFERENCES
Castro MC, Goblet P (2003) Calibration of regional groundwater flow models: Working toward a
better understanding of site-specific systems. Water Resour. Res., 39 (6) : p 11721172
Clement TP, Sun Y, Hooker BS, Petersen JN (1998) Modeling Multispecies Reactive Transport in
Ground Water. Ground Water Monitoring & Remediation, 18(2) : p 7992
Clement TP, Johnson CD, Sun Y, Klecka GM, Bartlett C (2000) Natural Attenuation of chlorinated
solvent compounds: model development and field-scale application at the Dover Site. J. of
Contaminant Hydrology, 42(24) : p 113140
Dan A, Goblet P (2004) Programme TRANSPOL II (INERIS 2003). ENSMP-CIG-LHM/RD/04/
18, 23 p
Dan A, Jamet P (2005) Pollutant source-term characterization and transport parameter estimation
with metamodel-based methods: a chlorinated solvent case study. In Water resources management
III, Ecology and the Environment volume 80. Cunha MDC and Brebbia CA, p 227236
Poirot N, Schwartz J (2004) Impact dun dversement de composs chlors sur la qualit des eaux
souterraines TRANSPOL II Cas Rel n3. ANTEA- A33548/A, 18 p
Quiot F (2004) Modlisation hydrogologique dune nappe alluviale pollue par des solvants chlors
TRANSPOL Cas Rel n3. INERIS-DRC-04-55503/DESP-R01, 38 p
Rollin C, Bour O (2002) TRANSPOL II Cahier des charges pour la ralisation dun modle
numrique (CdC n1). Cas rel n3: impact dun dversement de composs chlors sur la qualit
des eaux souterraines INERIS-DRC-02-25294/DESP-R18, 8 p
Rollin C (2003) TRANSPOL II Cahier des charges pour la ralisation dun modle numrique (CdC
n2). Cas rel n3: impact dun dversement de composs chlors sur la qualit des eaux souter-
raines INERIS-DRC-03-46524/DESP-R01, 21 p
Ruiz E, Jordana S, Guimer J (2004) Simulations of the impact of a spill of chlorinated compounds
on groundwater quality TRANSPOL II CR-3. ENVIROS R-2225-r1, 43 p
Zheng C, Hill MC, Hsieh PA (2001) MODFLOW-2000 The U.S. Geological Survey modular
ground-water model User guide to the LMT6 package, the linkage with MT3DMS for multi-
species mass transport modeling: U.S. Geological Survey Open-File Report 0182, 43 p
CHAPTER 31
Variable water saturation impact on sorption/desorption
characteristics of multiple species heavy metals from
gold mining
Aksara Putthividhya
1
& Srilert Chotpantarat
2
1
Department of Water Resources Engineering, Faculty of Engineering, Chulalongkorn University,
Pathumwan, Bangkok
2
National Research Centre for Environmental and Hazardous Waste Management,
Pathumwan, Bangkok
ABSTRACT: Gold mining is a growth industry, producing substantial amounts of waste that can
impact the environment. It is important to understand the mechanisms of releasing metal elements
from these mines and their wastes in order to predict migration of these contaminants and design a
mitigation strategy for pollution control and prevention in the local groundwater system. The study
focuses on the impacts of variable water saturation on migration and adsorption/desorption charac-
teristics of multiple heavy metal species derived from mine water leachate entering the groundwater
environment. A series of batch and column tests was conducted to evaluate the effects of pH on the
sorption/desorption characteristics of metals and to estimate the sorption/desorption isotherms of
the multiple heavy metal species. The results from the column experiments were then used as input
parameters in HYDRUS-2D model to simulate the migration of heavy metals through variably satu-
rated porous media. The computer simulations revealed that the migration of the mixed contamin-
ated plume was governed mainly by the mechanisms prevailing in the unsaturated zone, rather than
the hydraulic head gradient in the saturated zone. In addition, decreasing water content resulted in
higher sorption, promoted the late arrival of water fronts, and therefore could extend the contamin-
ants lifetime in the system. Drier soil tended to retard transport of metals particularly in the shallow
unsaturated zone, leading to a possible pathway for the contaminants to get back into the system via
plant root uptake. This study provides a basis for understanding processes governing the transport
and migration of heavy metals in groundwater, which could lead to a more robust planning strategy
for contaminant controls and prevention in the future.
1 INTRODUCTION AND BACKGROUND
The presence of heavy metals in surface and groundwater has been a major concern to com-
munities and municipalities. Concerns over contamination of groundwater and surface water
by heavy metals from previously abandoned waste disposal sites and some currently operat-
ing sites, including mining, have generated a programme to remediate contaminated soils.
Metals can cause significant damage to the environment and human health as a result of their
mobilities and solubilities. The ubiquitous nature of heavy metals, their toxicity even in trace
quantities, their tendency to bioaccumulate in the food chain, and the stricter environmental
regulations related to heavy metal discharges make it necessary to develop schemes for the
removal of heavy metals from both wastewaters and landfill leachates. The selection of the
most appropriate soil and sediment remediation method depends on the site characteristics,
concentration, type of pollutants to be removed, fate and transport of metals, and the end use
of the contaminant medium.
Metals typically become groundwater problems under the following situations: (1)
activities associated with the plating industry (Ma et al., 1994), where a wide variety of
metals are present at high concentrations in soluble forms; (2) sites with releases of
radionuclides, due to unique health risks and the use of chelating and complexing agents,
can have significant impact at very low concentrations; and (3) metals and high levels of
inorganic Total Dissolved Solids (TDS) are associated with leaks from sanitary, solid and
hazardous waste landfill, and mining. Many studies have been conducted on the crystal
structure and chemistry of synthetic and mineral apatites (Chen et al., 1997, Wright, 1990,
McConnell, 1973) and many toxic heavy metals have been investigated (Ma et al., 1994;
Minamisawa et al., 2004, Um and Papelis, 2004, Schneider and Rubio, 1999). Additionally,
many studies have studied the sorption of metals onto hydroxides, especially iron hydrox-
ides (Hayes and Leckie, 1986, Dzombak and Morel, 1990), aluminium hydroxides
(Benyahna and Garnier, 1999, Bowers 1983), silicon hydroxides (Meng and Letterman,
1993), and kaolinite (Dalang et al., 1984). However, research related to studying multiple
component toxic heavy metals in the unsaturated zone where there is variable degree of
water saturation is rare.
This paper focuses on multiple component toxic heavy metals as contaminants in
groundwater. Groundwater can be contaminated with metals directly by infiltration of
leachate from mine tailings and other mining wastes. A variety of reactions, which influ-
ence the speciation and mobility of metal contaminants, can occur. These reactions include,
acid/base, precipitation/dissolution, oxidation/reduction, sorption/desorption, and ion
exchange. In general, precipitation, sorption, and ion exchange can retard the movement of
metals in groundwater. The rate and extent of these reactions depend in factors such as pH,
Eh, complexation with other dissolved constituents, sorption and ion exchange capacity of
the geological materials, and organic matter content. Soil moisture contents as well as
groundwater flow characteristics also significantly influence the transport of metal contam-
inants. It is important to understand the mechanisms and processes controlling the release
of these heavy metals into the environment from mine wastes in order to predict or estimate
the migration of these toxic contaminants and initiate a groundwater protection strategy as
well as remediation plan for pollution control and prevention.
This paper focuses on the equilibrium-state chemistry for the four possible species of
heavy metals released from gold mining leachate: lead (Pb), cadmium (Cd), nickel (Ni),
and zinc (Zn). In this study, the impact of leachate from mine tailing is being investigated
at Akara mine in Thailand. Akara Mining Limited has developed the Chatree Gold Project,
which is located 280km north of Bangkok. This project involves the mining and process-
ing of gold ore during a fourteen-year period. During the mining phase, waste rocks have
been extracted and used in the construction of the tailings storage facility. During ore pro-
cessing, 750,000 dry tons per year of tailings were generated and deposited in the tailing
storage facility. These mine tailing was left as particulate suspension (i.e., fine grained
sediment-water slurry) after the valuable metals had been extracted. However, there are
shallow municipal wells located within the nearby villages, where local people rely on
groundwater for their supply. After preliminary investigation, these shallow wells probably
intersect the shallow, unconfined, alluvial, clayey sand aquifers at depths between 1.57m
beneath land surface, and the groundwater table is approximately 6m below the ground
surface. Surface soil was examined, and it was found that the soil pH ranged from 3 to 7,
i.e. the soils were acidic to neutral. The metal levels were considered to be relatively low
for a mineralized area (comparing to other mining sites around Thailand and elsewhere).
The primary objective of this study was to investigate the effect of environmental cond-
itions such as soil moisture contents on migration of multi-species heavy metals from min-
ing leachate into natural subsurface environments. For this purpose, the study was divided
into three parts. Firstly, real tailings soil samples from Akara mine were used in a series of
batch and column experiments to investigate the sorption and desorption characteristics of
the multi-species metals, respectively. Secondly, a series of batch experiments were under-
taken to evaluate the sorption isotherms for each heavy metal under various pH conditions.
Finally, the transport and migration of the multi-species under various soil moisture con-
tents was investigated using a 2D numerical model (HYDRUS-2D) by employing the sorp-
tion isotherms obtained from previous batch equilibrium experiments as inputs.
The amount of heavy metals in the tailing at Akara Mine was initially measured by collect-
ing soil samples at the tailing storage facility at different depths using a hand auger and
clean plastic containers. The soil samples were collected from the tailings storage facility
at 13 different locations distributed across the area and at depths from the soil surface
down to around 2 m. The samples were secured separately in clean plastic bags and trans-
ported back to the laboratory. Each sample was then air-dried and sieved through sieve
number 200 prior to heavy metal and pH analysis.
In order to analyze for total heavy metal contents in the tailings, EPA method 351 was
employed in which the contaminant concentrations are most commonly expressed as total
dissolved metals in mass concentrations or in molar concentrations. Our tailing soil samples
were batch tested under three pH conditions (i.e., pH 4, 7, and 10) in order to estimate the
equilibrium time in each state. For this, the tailing soil samples were digested in 10ml con-
centrated nitric acid for 10 minutes in a microwave digesting oven. The sample and acid were
then placed in a fluorocarbon (PFA or TFM) microwave vessel. The vessel was capped and
heated in the microwave again. After cooling down, the vessel contents were filtered and cen-
trifuged (or allowed to settle). The supernatant was diluted with 50ml buffer solution, and
then shaken at 180rpm. The slurry was then filtered through GF/C filter paper and the aque-
ous phase was analyzed via ICP-OES or ICP-AES (inductively coupled plasma optical emis-
sion spectroscopy). Several buffer solutions were tested in order to select the most efficient
one under these conditions. The results are shown and discussed later in this paper.
To estimate desorption of heavy metals, samples were again tested under 3pH conditions
(i.e., at pH 4, 7, and 10). Triplicate tailing samples were adjusted using 50ml of the appro-
priate buffer. The samples then were centrifuged at 180rpm and time to reach local equilib-
rium was measured under each condition. After that, the slurry was filtered through GF/C
filter paper (Whatmann) before analyzing for metals using ICP-OES or ICP-AES. In order
to determine metal sorption isotherm parameters of the soil laterite aquifer materials found
Impact on sorption/desorption characteristics of multiple species heavy metals 421
at the site, the samples were collected, transported back to the laboratory, and kept at 4C
prior to use. Table 1 shows the main soil characteristics of the porous medium from the
laterite layer.
Zn and Pb were selected as the representatives of the multi-species heavy metals found
in the leachate because of their relatively high concentrations found at the site and because
of their toxicity to human and the environment. A series of sorption batch experiments
were conducted under 2 different pH conditions (i.e., pH 4 and 7). For this, 5ppm of Zn
and Pb each was added as the initial aqueous metal concentration in the system with 1g of
the laterite sample. The samples were then shaken on a platform shaker for 120 hours or
until local equilibrium had been reached. Changes in sorbed metal concentration in the
system were measured periodically. This was done at a pH higher than 7 but a significant
precipitation was observed and the results are not included here.
3 RESULTS
The results of the overall regional groundwater analysis in the Akara Mine area revealed
that several elements exceeded the standards shown in Table 2. There was concern about
the concentrations of metals that may be present in the sediment as a solid phase, instead
of the dissolved phase. Therefore, several tailing soil samples were collected and analyzed
for metal concentrations. The results from multi-element analysis of 36 tailing soil sam-
ples collected from the tailings storage facility are presenting in Table 3.
Adsorption and desorption experiments of heavy metals from the tailing samples were
undertaken under different pH conditions as explained above. Therefore, several buffers
were used in order to control the pH and to ensure that it was consistent throughout the
experiments. The efficiency of the pH buffers and the optimal amount of buffer were
tested by measuring the pH in the system for a significant period of time. The results are
plotted in Figure 1 as shown below.
The results in Figure 1 indicate that NaAc at 0.14M was sufficient to control the pH of
the system at 4 throughout the entire experiment. In addition, the system reached an equi-
librium condition rather quickly, indicating that there was no sign of mass transfer limita-
tion between solid and aqueous phase. Similarly, PIPEs buffer at concentration of 0.124 M
or more was sufficient to control the pH at 7 with no sign of mass transfer limitation. CAPs
buffer at 0.0075M was used at a higher pH of 10 with no mass transfer limitation. It is
probably a better idea to select the proper buffer based on the concentration basis, i.e., the
least amount of buffer, the better, in order to avoid any possible metal precipitation or any
surface complexation. The effects of the buffer on metal precipitation and complexation
are currently under investigation.
Table 1. Soil characteristics of the laterite layer.
pH-H
2
O 5.4
Organic matter (%) 0.13
Clay fraction (%) 50.2
Sand fraction (%) 30.8
Silt fraction (%) 19.0
CEC (cmol
c
/kg) 8.2
Bulk density(g/cm
3
) 1.10
Table 2. Regional estimate of groundwater at Akara Mine.
Elements and Location Concentration (ppm)
Mn
Piezometer 690 0.510.75
UD 0.8526.7
ST5 0.510.66
Fe
ST2 0.311.90
ST3 0.204.50
Cd
DC 0.026.10
UD 0.022.35
SO
4
2
UD 2352120
Table 3. Chemical analysis of tailing soil samples at Akara Mine.
Parameter Concentration (mg/g) Minimum Maximum Concentration (mg/g)
Aluminium 10.82 7.2813.67
Arsenic ND ND
Barium 0.15 0.040.29
Bismuth ND ND
Boron 0.147 0.040.29
Cadmium ND ND
Calcium 41.80 24.2151.76
Chromium ND ND
Cobalt ND ND
Copper 0.07 0.020.09
Iron 19.71 11.4122.90
Lead 0.04 00.13
Magnesium 4.87 3.865.36
Manganese 2.49 1.672.84
Mercury 0.001 00.005
Nickel 0.007 0.0050.027
Potassium 2.53 2.372.79
Silver ND ND
Sodium 0.41 0.330.45
Strontium 0.04 0.020.05
Zinc 0.17 0.050.28
Note: ND Not Detected.
Batch desorption experiments (leaching tests) of several heavy metal species from the
tailing soil were conducted to estimate the amount of heavy metals that could be released
from the sorbed phase into the aqueous phase and potentially contaminate groundwater
resources as well as surface water resources in the nearby area. Additionally, the effect of
pH was also determined on the release of heavy metals from the tailing soil as a threat to
the environment. The results are illustrated in Figure 2 below.
The desorption characteristics of Mn are pH dependent (Figure 2). The pH of the solu-
tion greatly affected the metal desorption mechanism from mine tailings as less Mn was
discovered in the aqueous phase when the pH was increased. Additionally, the aqueous
metal concentrations (i.e., y-axis in Figure 2) reached a plateau level after 48, 24, and less
than 24 hours for pH 4, 7, and 10, respectively, indicating that the system reached local
equilibrium conditions faster with increasing pH.
Heavy metal desorption characteristics from the tailing soils are also species dependent.
Figure 3 compares the desorption characteristics of Mn, Fe, Cu, and Pb, for the same pH
condition and buffer type. The results suggested that Mn desorption was highest among
the four metal species tested, while Cu and Pb could desorb relatively less than Fe and Mn.
Moreover, Pb took as long as 312 hours to reach equilibrium, which was the longest
Figure 1. Efficiency and optimal amount of different buffers operating under 3pH conditions.
among the four metal species tested, whereas Mn and Fe reached equilibrium at approx-
imately the same time (i.e. 36 hours).
Sorption batch experiments of Zn and Pb with laterite soil were undertaken at pH 4 and
7. The results showing time to reach equilibrium are presented in Figure 4. It is important
to note here that data at higher pH for Pb in the sorption experiment could not be obtained
due to the very low solubility of Pb, resulting in substantial precipitation of Pb. Therefore,
results for Pb at pH 4 only are shown in Figure 4. The y-axis represents sorbed normalized
metal concentration on soil materials, whereas the x-axis is the sampling time. The results
show that while the reaction time increased from 2 to 24 hours, the total Zn sorption over
this period increased from 60% to 80% under pH 7. Also, the sorption capacity of Zn onto
the laterite soil increased with increasing pH from 4 to 7, based on the higher normalized
concentration of Zn observed in the system to around 80%. Additionally, under more
acidic conditions, Zn sorption capacity seemed to reach its equilibrium slower.
The sorption characteristics of Zn and Pb at pH 4 can be compared in Figure 4. Firstly,
both Zn and Pb sorption characteristics followed a similar trend. Secondly, the plots indicate
that Pb at the same pH conditions sorbed onto the laterite soil matrix better than Zn, based
on the maximum sorbed Pb concentration of 40% after approximately 96 hours. Moreover,
the system composed of Pb ions achieved local equilibrium faster (approximately after
Figure 2. Effects of pH on desorption and time to reach equilibrium for manganese (Mn).
24 hours) than that in the system with Zn and adsorption of Zn did not further increase after
approximately 48 hours.
Adsorption isotherm parameters obtained from batch experiments were employed as an
initial input parameters for modelling the transport and migration behaviour of multiple
heavy metal components in the unsaturated zone. Since the HYDRUS-2D model cannot
account for a variation in pH, the simulations were undertaken under constant pH condi-
tions. The conceptual model (Figure 5) was constructed and several scenarios generated by
computer simulations for estimating metal concentration from mine leachate that could
contaminate the groundwater environment underneath the tailings storage facility. The
model was designed with the boundary conditions shown in Figure 5. The initial focus was
to estimate the maximum metal concentration level (i.e., Thai drinking water standard
shown in Table 4) and the time for heavy metals to travel to the groundwater table under
various water saturation conditions. Once the heavy metals desorbed from the tailing soil,
they moved through pores as rainwater infiltrated through the soil. Metals in the tailing
Figure 3. Heavy metal desorption characteristics of Akara Mine tailing soil at pH 4.
Figure 4. Zn and Pb sorption characteristics at pH 4 and 7.
Transport through
variable water
saturation
Constant head
Constant Flux
Groundwater table
Tailing waste
n
o

f
l
o
w

b
o
u
n
d
a
r
y
f
o
r

l
i
q
u
i
d

a
n
d

c
h
e
m
i
c
a
l
s

n
o

f
l
o
w

b
o
u
n
d
a
r
y
f
o
r

l
i
q
u
i
d

a
n
d

c
h
e
m
i
c
a
l
s

Figure 5. Conceptual model and domain for numerical simulation.
Table 4. Some important input parameters for HYDRUS-2D simulations.
Dispersivity, cm 60
Residual Water Content 0.092
Saturated Water Content 0.456
Bulk Density, g/cm
3
1.10
Saturated Hydraulic Conductivity, cm/day 12.72
Initial Metal Concentration, mg/l 20 for Pb
35 for Zn
Maximum Concentration Level 0.01 for Pb
(i.e., Thai Drinking Water Standard), mg/l 5 for Zn
soil could dissolve in the percolating water and move vertically down to the water table.
Metal contaminant fluxes were assumed to be steady and were estimated from two years
of pumping rates recorded from the seepage wells under the tailings storage facility. The
final estimated Darcy flux between the years 2004 and 2005 was 0.011mm/day with the
maximum was 0.17mm/day. Some important input parameters for HYDRUS-2D simula-
tions are shown in Table 4. The modelled water retention curve generated in HYDRUS-2D
is shown in Figure 6.
The computer simulation using HYDRUS-2D revealed the results from the variation in
water saturation on the migration of multiple component metals in the unsaturated zone
(Figure 7). Figure 7 shows the moisture content in the soil (y-axis) versus the time to reach
the required metal concentrations (x-axis) at different depths from the surface. Since the
water table in the Akara Mine area was approximately 6m below the ground surface, the
simulation was done at different depths at 1m increments (i.e., from 1m to 5m below the
surface). The results show that the transport and migration of multiple component heavy
metals in the unsaturated zone are dependant on soil water content. For Zn at pH 4 at a
depth of 1m below the ground surface, for example, and by increasing the soil water con-
tent by 25% (from around 0.3 to more than 0.4), it took two orders of magnitude longer for
Zn to pass through the soil matrix to reach a certain concentration at a certain depth in the
environment.
When comparing different metal species (i.e., Zn and Pb), the simulation results indi-
cated that Zn travelled faster in the laterite soil than Pb, leading to a much shorter time
required to reach a specified metal concentration at a specific depth in the subsurface. This
result supported the observation from batch experiments described earlier since the sorp-
tion capacity of Pb was higher than that of Zn.
Water fronts at each depth under various soil moisture content conditions are plotted in
Figure 8. This shows the final water content (y-axis) as a function of time (x-axis) under
various initial soil moisture contents. The results in Figure 8 demonstrate that increasing
the initial soil moisture content led to the early arrival of the water front at each depth.
Furthermore, at the same initial soil water saturation, the time required to reach a specified
final soil moisture content in the unsaturated zone depended mainly on the location in the
vertical direction of the domain. Taking the system with an initial soil moisture content of
50% of its maximum value as an example, the results shown in Figure 8 demonstrate that
it would take 600 days and 400 days at depth 3m and 2m from the ground surface, respect-
ively, in order for the final water content in the soil to get up to the value of 0.3. Moreover,
Figure 6. Modelled water retention curve.
the water front profile at 6m depth did not really respond to the change in the initial water
content as shown by the flat water front profiles simulated by the model.
4 DISCUSSIONS AND CONCLUSIONS
In the Akara mining process, the oxidation stage produces acidic water which allows the
release of heavy metals, including lead (Pb), copper (Cu), iron (Fe), manganese (Mn), and
Figure 7. Effects of variable soil moisture content on the transport and migration of multiple com-
ponent heavy metals in the unsaturated zone.
zinc (Zn), from the mineralized zone. The mechanism for their release is due to the infil-
tration of fresh water from the surface to the tailing pit, creating an oxidised environment,
which causes the release. Once the release occurs, water containing heavy metals migrates
further downward, and eventually can contaminate the underlying groundwater. The eval-
uation looked at the desorption of heavy metal components from the tailing soils collected
from the Akara Mine area caused by infiltrating water. This focused on the effects of pH
on the desorption of heavy metals.
The desorption study (Figures 2 and 3) are extremely important, particularly for con-
taminated soil, in order to predict the fate and mobility of contaminants and to develop
sound and cost-effective remediation strategies. The results (Figure 2) demonstrate that
decreasing the pH led to a significant release of heavy metals into the environment, sug-
gesting that under acidic mine drainage conditions, such as those found at the Akara Mine,
there would be more mobility of multi-component heavy metals. This observation may be
explained by the competition between metal anions and protons in the aqueous solution at
low pH conditions, resulting in smaller fractions of heavy metals being sorbed onto adsor-
bent surfaces. In addition, the presence of inorganic anions such as carbonate, phosphate,
and sulphide in the soil water can influence the soils ability to fix metals chemically
(Diatta, 2002). These anions can form relatively insoluble complexes with metal ions and
cause the metals to desorb and/or precipitate in their presence.
The adsorption parameters of multiple component heavy metals onto laterite soil and
the sorption isotherm are extremely important in assessing the mobility of dissolved and
Figure 8. Effects of variable soil moisture content on transport and migration of multiple compon-
ent heavy metals in the unsaturated zone.
sorbed chemicals, including heavy metals, in soil environments. Analysis of the adsorption
data (Figure 4) using the Freundlich adsorption model, i.e. that the pollutant uptake per
unit weight of adsorbent (soil matrix) to the equilibrium adsorbate (heavy metals) concen-
tration in the bulk phase, demonstrates that the sorption isotherms of Zn and Pb fit the
Freundlich adsorption model (Sposito, 1981), Equation 1:
(1)
Where K
F
and n are Freundlich constants and commonly employed for modelling adsorp-
tion characteristics of adsorbates in the environment.
After applying the Freundlich isotherm model in the system with multiple heavy metal
components, adsorption parameters K
F
and n were calculated under different pH condi-
tions (Table 5). For the purpose of comparison between the model and the observed data,
the correlation coefficient R
2
for each case is also listed.
The relatively high correlation coefficient R
2
values in most cases observed from this
study indicate that the transport and migration behaviour of heavy metals in the unsatu-
rated zone is governed mainly by their sorption capacity and/or other immobilizing mech-
anisms such as precipitation. The Freundlich isotherm parameters tabulated in Table 3
were the outcome of well-established linear relationships of log Q versus log C
e
(C
e
: equi-
librium concentration and Q: amount of sorbed phase concentration).
From the adsorption and desorption studies, it was observed that desorption was a more
difficult process than adsorption and that not all of the adsorbate (i.e., heavy metal ions)
was desorbed, i.e., the reactions appear to be partially irreversible. Such apparent irre-
versibility may result in the adsorption and desorption isotherms corresponding to the for-
ward and backward reactions not being coincident (Verburg and Baveye, 1994). The
time-dependent desorption behaviour from laterite soil in this case could be ascribed to
heavy metals being trapped in soil micropores, which resulted in a tortuous and slow
release process (Sparks, 1995). This interesting observation stimulates the need for con-
ducting further studies at macroscopic scale to determine the mechanisms for sorption-
desorption phenomena as macroscopic measurements and models done in this work reveal
some important information about sorption/desorption phenomena only, although no
mechanistic or molecular information can be explained.
The analysis of water front profiles at each depth under various initial soil moisture con-
tent conditions (Figures 7 and 8) revealed that the wetter the soil the greater the heavy metal
mobility. The simulation of water fronts at 6m depth from the ground surface did not illus-
trate the effects of the initial soil water contents at all due to the presence of the water table
at around 6m below ground level. The presence of the groundwater table would partially
Q K C
F e
n
1
Table 5. Freundlich adsorption isotherm parameters for Zn and Pb under different pH conditions.
Heavy metals pH K
F
n R
2
Pb 3.88 0.07 20.96 2.93 1.20 0.06 0.99
Zn 4.01 0.05 1.78 0.06 1.81 0.06 0.77
Zn 7.09 0.04 142.80 22.64 0.90 0.05 0.99
saturate the system, and soil moisture contents would reach its saturated condition rather
quickly no matter what the initial soil moisture contents were. More importantly, the
results in Figure 8 indicated that the initial soil moisture content affected the water front
profile of the system more significantly at the shallower depths.
The aqueous speciation of the target heavy metals as a function of pH was also studied
by employing MINEQLversion 4.5 (Schecher and McAvoy, 2001) because the hydroly-
sis behaviour of metal ions may influence their sorption behaviour. The systems were com-
posed of initial metal ions at various concentrations (20mg/l, 35mg/l, 40mg/l, and
55mg/l for Pb, Zn, Ni, and Mn, respectively) and the ionic strength of the system was kept
constant at 0.05M. The results are presented in Figure 9 below.
The results in Figure 9 indicated that under operating conditions between pH 4 and 7,
metal ions dominated the system. For Pb, at pH below 6, Pb
2
and PbNO
3
were the only

two significant species. The latter was generated from the buffer solution used to control
the pH of the system. Between pH 6 and 7, some solid precipitation occurred in the form
of Pb(OH)
2
. As the Pb concentration in the aqueous phase was reduced due to the solid for-
mation process within this pH range, the sorbed concentration of Pb in the system could
be overestimated. However, all the metal analysis was done on ICP-OES in which the total
metal concentration was measured instead of the ionic form. Therefore, the errors pertain-
ing to the effect of solid formation/precipitation in the system were likely to be minimal.
Figure 7 shows that HYDRUS-2D model was run for a relatively long time because natu-
rally occurring metals travel slowly in subsurface environment as they tend to sorb more onto
the soil matrix surfaces. The correlations shown reflect the effects of the water content on the
time to reach the maximum concentration level for each metal. The steeper slopes of the plots
between final water contents versus time to reach a certain metal concentration with increas-
ing depth also show that the initial soil moisture content does affect the mobility of multi-
species heavy metals in the shallow unsaturated zone. This shallow zone is characterized by
Pb
Ni
Zn
Mn
Figure 9. Aqueous speciation of the target heavy metals as a function of pH.
biological and vegetation root uptake activities, and these may yield a potential for toxic
heavy metals to be removed from the subsurface environment via biodegradation and
biosorption (Minamisawa et al., 2004). However, the return of heavy metals to the surface
via vegetation may raise alarm because of the potential accumulation of heavy metals in agri-
cultural products. This study thus provides a basis for understanding the possible application
of bioremediation and phytoremediation to remove multiple component heavy metals in the
subsurface environment; there has been interest in the use of natural plants in treating heavy
metal polluted effluents during the past two decades (Schneider and Rubio, 1999).
REFERENCES
Benyahna L., and J-M Garnier (1999) Effect of Salicylic Acid upon Trace-Metal Sorption (Cd
II
, Zn
II
,
Co
II
, and Mn
II
) onto Alumina, Silica, and Kaolinite as a Function of pH. Environmental Science
and Technology, 33, 13981407.
Bowers, A.R. Adsorption Characteristics of Various Heavy Metals at the Oxide-Solution Interface:
Effect of Complex Formation. Ph.D. Thesis, University of Delaware, 1983.
Chen, X., J.V. Wright, J.L. Conca, and L.M. Peurrung (1997) Effects of pH on Heavy Metal Sorption
on Mineral Apatite. Environmental Science and Technology, 31, 624631.
Dalang, F, J. Buffle, W. Haerdi (1984) Study of the Influence of Fulvic Substances on the Adsorption
of Copper (II) Ions at the Kaolinite Surface. Environmental Science and Technology, 18,135141.
Diatta, J.B. (2002) Evaluation of adsorption parameters and charge densities of some selected soils:
application to lead. Journal of Polish Agricultural Universities, 5, 4050.
Dzombak, D.A., Morel, FMM Surface Complexation Modeling Hydrous Ferric Oxide; Wiley-
Interscience: New York, 1990.
Hayes, K.F., and J.O. Leckie In Geochemical Processes at Mineral Surfaces, Davis, JA, Hayes, KF,
Eds; ACS Sym. Ser. 323; American Chemical Society: Washington, DC, 1986; pp 114141.
Ma, Q.Y., S.J. Tralna, TJ Logan, and JA Ryan (1994) Effects of Aqueous Al, Cd, Cu, Fe(II), Ni, and Zn
on Pb Immobilization by Hydroxyapatite. Environmental Science and Technology, 28, 12191228.
McConnell, D. Apatites: Its Crystal Chemistry, Mineralogy, Utilization, and Geologic and Biologic
Occurrences; Springer-Verlag: Berlin, 1973.
Meng, X., and R.D. Letterman (1993) Modeling Ion Adsorption on Aluminum Hydroxide-Modified
Silica. Environmental Science and Technology, 27, 19241929.
Minamisawa M., H. Minamisawa, S. Yoshida, and N. Takai (2004) Adsorption Behavior of Heavy
Metals in Biomaterials. Journal of Agricultural and Food Chemistry, 52, 56065611.
Schecher, W.D., and D.C. McAvoy (2001) MINEQL: A Chemical Equilibrium Modeling System.
Version 4.5 for windows. Environmental Research Software, Hallowell, Maine.
Schneider, I.A.H., and J. Rubio (1999) Sorption of Heavy Metal Ions by the Nonliving Biomass of
Freshwater Macrophytes. Environmental Science and Technology, 33, 22132217.
Simunek, J., M. Sejna, and M. Th. van Genuchten (1999) The HYDRUS-2D Software Package for
Simulating Two-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably
Saturated Media. Version 2.0, IGWMC TPS 53, International Ground Water Modeling Center,
Colorado School of Mines, Golden, Colorado, 251pp.
Sparks, D.L. (1995) Environmental Soil Chemistry. Academic Press, New York.
Sposito, G. (1981) The Thermodynamics of Soil Solutions. Oxford University Press, Oxford.
Um, W, and C Papelis (2004) Metal Ion Sorption and Desorption on Zeolitized Tuffs from the
Nevada Test Site. Environmental Science and Technology, 38, 496502.
Verburg, K., and P. Baveye (1994) Hysteresis in the Binary Exchange of Cations on 2:1 Clay
Minerals: A Critical Review. Clays Clay Miner, 42, 207220.
Wright, J. In Skeletal Biomineralization: Patterns, Processes, and Evolutionary Trends. Volume I;
Carter, J.G., Ed.; Van Nostrand Reinhold: New York, 1990, pp 445459.
CHAPTER 32
Aquifer protection: A combined study of microstructure
and transport properties in clay-sand mixtures
Paolo Macini, Ezio Mesini & Rossella Viola
University of Bologna, DICMA, Viale del Risorgimento, Bologna, Italy
The study focuses on the analysis of the effects of swell-shrinkage potentials on the micro-
structure of bentonite-sand mixtures, to gain insight into changes with varying hydration
states, performed by field emission scanning electron microscopy (FESEM) and Nuclear
Magnetic Resonance (NMR). In addition, permeability experiments on similar artificial
soils were undertaken in oedometric conditions, by means of a constant head permeame-
ter, to validate the suitability of the soil for the construction of liners in landfills, for the
purpose of aquifer protection. FESEM observations indicate microstructural changes with
dehydration that involve reorientation of the tactoids and closure of micropores. As to per-
meability tests on the mixtures, the results showed that the addition of very small percent-
ages of clay induces variations of more than three orders of magnitude. Following the
results, a discussion highlights the suitability of the Revil and Cathles (1999) model, using
NMR and permeability tests.
1 INTRODUCTION
Clay-sand mixtures are receiving increasing attention as technical buffer materials in the
construction of hydraulic caps and barriers because of their low permeability. Engineered
barriers are constructed containment systems, used to isolate hazardous sites from their
surroundings and to protect aquifers. They include subsurface barriers (denoted as vertical
barriers) and caps. Subsurface barriers are vertical cut-offs that prevent the horizontal
migration of the groundwater. Vertical barriers, typically used to control sources of haz-
ardous waste, are soil-clay, soil-cement-clay, cement-clay, sheet pile (steel or high density
polyethylene), and clay barriers. Impervious caps prevent the downward migration of sur-
face runoff and rain. They are built of clay or geosynthetic materials. Initially, engineered
subsurface barriers were the major component of passive containment systems to prevent
migration of contaminated groundwater from hazardous waste sites. Caps were often
added to completely isolate such sites. This paper deals with the feasibility of clay-sand
mixtures to meet the requirements concerning low hydraulic transport. Moreover, it is
worth noting that, quite recently, engineers have begun to be concerned not only about
the hydraulic transport of contaminants, but also about the diffusion of contaminants
through vertical barriers. This issue is crucial for the long term, i.e. for the integrity of
vertical barriers. New technologies are emerging to increase the sorption capacity of ver-
tical barriers, primarily through the use of additives in the backfill materials (EPA, 1998).
As far as clay-sand mixtures are concerned, clay fabric content, microstructure, and
grain-size distribution determine the porosity and the permeability of porous media, as
well as the evolution of mesopore space forming between sand grains at low clay content and
at low hydration state. The spatial organization of the clay particles consisting of colloid-size
stacks of clay platelets (tactoids) and the spacing between individual platelets are functions
of the hydration state. Clay platelets are organized in a complex fashion resulting in a hier-
archy of structures and pores that affect retention of water and volume changes. Large changes
in volume and in pore space, with changes in the hydration state of the soil, make it diffi-
cult to predict and model hydraulic functions (Tuller and Or, 2003).
Laminar swelling alone cannot explain the observed volume changes and the amount of
water retained in the clay fabric. Evidence from Scanning Electron Microscope (SEM)
observations clearly shows that the clay fabric can arrange itself in an open micropore
structure due to face-to-edge bonds of tactoids (Tessier, 1990). Thus, when dealing with
pure clay, it is important to consider both the spaces between the platelets and the microp-
ores. Considering other textural components such as sand or silt grains, another important
pore-space feature has to be incorporated, the so-called mesopores. Mesopores are voids
that form between sand or silt grains; they are larger than both the interlaminar spaces and
typical micropores in a network of tactoids. These mesopores form the backbone of the
so-called textural pore size of soils. Effects of different clay contents on porosity and
permeability have to be taken into account in clay soils. At high sand contents, grains
commonly form a primary skeleton with distinct mesopore spaces between individual
grains.
Data from geophysical surveys for the petroleum industry (Revil and Cathles, 1999)
show a clear trend of reduction in porosity of the clay-sand mixture towards a critical clay
content where the void space formed by the coarse texture is completely filled by the clay
fabric. Subsequent increase in the clay content increases the distance between adjacent
sand grains that float in the clay matrix. Measurements obtained for clay-sand mixtures
reveal a critical clay content of about 3540% by volume at which the clay-sand skeletal
porosity is completely filled with clay matrix.
Clay-sand microstructure can be characterised by techniques such as field emission
scanning electron microscopy (FESEM) and, recently, Environmental SEM (ESEM). In
order to visualize pore space geometry and spatial phase arrangement in clay-sand mix-
tures as a function of the hydration state, microscopy studies with samples varying in clay
content were made by means of FESEM and ESEM. FESEM observations clearly indicate
dramatic microstructural changes with dehydration that involve reorientation of the tac-
toids, and closure of micropores. In a previous paper (Viola et al., 2005) formation of
mesopores between sand grains and the clay fabric have been revealed by ESEM.
Together with SEM, NMR Relaxometry (MRR) of water
1
H nuclei was introduced in this
study, as it is a powerful tool for the investigation of the pore space structure in high
Surface-to-Volume ratio (S/V) systems. In addition, permeability experiments on similar arti-
ficial soils were undertaken in oedometric conditions, by means of a constant-head permea-
meter, to validate the suitability of the soil for the construction of liners in landfills, for the
purpose of aquifer protection (Macini and Brighenti, 1993, 1994). Following the results, a
discussion highlights the suitability of the Revil and Cathles (1999) model, using NMR and
permeability tests.
2 THEORY
The clay fabric is an assembly of colloid-size platelets (tactoids) characterized by a large
surface area associated with the laminar nature of clay particles, and by a negative lattice
charge balanced by exchangeable cations on their surfaces (Quirk, 1986). Additionally,
Tessier (1990) and Quirk (1986) have shown that tactoid orientation and other micro struc-
tural features of active clays vary with the type of clay and exchangeable cations. In add-
ition to the effects of drying (or wetting), electrolyte type and concentration are key
components in determining the extent of the diffuse double layer and thus affecting the clay
fabric microstructure. The spacing between the platelets increases with increasing chemical
potential (less negative) due to the interaction of charged surfaces and associated diffuse
double layers (Low, 1980; Warkentin et al., 1957).
Changes in interlaminar spacing, chemical potential, electrolyte concentration and
charge density, cause the shrink-swell phenomenon to occur (Fig. 1). The theoretical basis
for modelling interactions between charged surfaces in the presence of electrolyte is pro-
vided by the DLVO theory (Derjaguin and Landau, 1941; Verwey and Overbeek, 1948)
using the disjoining pressure as the basic thermodynamic property (or changes in free
energy per unit surface area per film thickness).
NMR relaxometry of water
1
H nuclei was introduced to investigate pore-space structure
in high surface-to-volume ratio (S/V) systems. Under specific experimental conditions, the
signal amplitude is proportional to the number of nuclear spins at a given value of the mag-
netic field. The detection of water
1
H nuclei trapped in a porous system allows porosity to be
Aquifer protection: A combined study of microstructure and transport properties 437
Figure 1. Chemical potential vs lamellar spacing. The spacing between the platelet increases with
increasing chemical potential (Low, 1980; Warkentin et al., 1957).
calculated. Moreover, structural characteristics of the pore space itself can be determined,
relying on the so-called surface effects. NMR relaxation times of confined water may be
many orders of magnitude shorter than the times in bulk water, and the distribution of the
relaxation times may be several decades wide. These effects can be related to properties of
the porous medium such as local S/V ratio distributions. The relaxation rates and pore
characteristics are related by the simple relationship 1/T
1,2
1,2
S/V 1/T
1,2bulk
, where
1/T
1
and 1/T
2
are the longitudinal and transverse relaxation rates (the reciprocals of the
relaxation times), respectively, and is the so-called surface relaxivity. In real porous
media, pores are not isolated, and NMR relaxation data are sums of exponentially decaying
components, reflecting local S/V ratios, combined with some unavoidable measurement
noise. Experimental data may be turned into pore-size distributions (Borgia et al., 1998)
by the so-called inversion of the decay data.
Traditional SEM is a recognized tool for the structural characterization of materials.
Bombarding sample surfaces with electron beams and detecting those that are emitted or
backscattered yield maximum resolutions of about 1nm and the intricate details of the
material structure to be studied. To minimize attenuation of the electron beam due to water
or gas molecules, observations are carried out under high vacuum. This creates difficulties for
the observation of hydrated media such as partially saturated soils or plant tissue due to alter-
a-tion of their natural structures. To overcome this problem, a number of techniques such as
freeze-drying (Smart and Tovey, 1982), critical point drying (Crang, 1988), resin impregna-
tion (Kim et al., 1995), or cryofixation (Chenu and Tessier, 1995) have been applied to
preserve sample structure at a given hydration state. In this paper, the freeze-drying technique
has been employed.
The samples analysed in this study were obtained by mixing Wyoming Na-bentonite and
Ottawa foundry sand in different amounts. A concentration of 60% bentonite and 40%
sand by weight was used, as illustrated in Fig. 2. Each sample was manually mixed for 30
minutes to ensure that clay particles were evenly distributed. The mixture was then poured
into a cylinder with filter paper glued to the bottom and placed on top of a porous stone
contained in a vessel. Following saturation, the samples were removed from the vessel and
stored in a refrigerator at 10C. The gravimetric water content of the saturated sample was
determined for a small sub-sample by oven-drying. To obtain a set of samples with vary-
ing water content for the FESEM observations, the saturated sample was divided into sub-
samples and slowly dried to the desired water content.
It is essential that the samples are dried prior to FESEM analysis, while maintaining the
sample (micro-) structure associated with a specific hydration state. The freeze-drying
approach was used to preserve the sample structure.
During the freeze-drying process, solid water (ice) is directly transferred to the vapour
phase. Sub-samples were sealed in small plastic containers and submerged in a vessel con-
taining isopentane, surrounded by liquid N
2
in an outer container. If immersed directly into
liquid N
2
, samples will be surrounded by a gas layer that has an insulating effect, thereby
decreasing the freezing rate. Slow freezing rates might lead to ice crystal growth and
potential structural damage. To avoid crystal growth, samples must be rapidly frozen at
temperatures below 130C. After freezing, samples are immediately transferred to a
vacuum oven, where the temperature is slowly increased to above 0C while a vacuum
below 6 kPa is maintained. This allows ice to thaw, but since there is not enough pressure
for a liquid to form, the ice is directly transferred into the vapour phase.
Concerning MRR experiments, specimens of natural dry bentonite were tested in vials
of 6mm height and 7mm diameter. A few drops of brine (deionised water and NaCl 2%
by weight) were poured into the vial tube, and MRR measurements were made after 24h.
MRR exploits the interaction of the nuclei of a sample with magnetic fields. A strong
magnetic field is applied to polarize the nuclear magnetic moments; time-dependent radio-
frequency (rf) irradiations are used to stimulate the spectroscopic response. After rf irradi-
ation, the decay of the nuclear magnetization of the
1
H nuclei can be obtained on the monitor.
The frequency L with which the atomic nuclei respond is called the Larmor frequency. It is
related to the strength of the magnetic field, B
o
, at the site of the nucleus by the fundamental
equation LB
o
, where is the gyromagnetic ratio of the observed nucleus.
1
H longitu-
dinal and transverse relaxation curves were obtained at 25C and 20MHz.
Permeability tests were carried out on artificial samples, prepared by mixing natural
sand and bentonite. An alluvial sand and an industrial bentonite were used. In particular,
bentonite-sand mixtures with concentrations of 0, 2, 4, 6, 8, 10% of bentonite (by weight)
were used.
Permeability was measured by means of a high-precision cylindrical permeameter in
oedometric conditions. Dry samples were placed inside the sealed sample-holder (diam-
eter 50mm and height about 30mm), de-aerated for 3 hours and then saturated with
deionised water. After saturation, a load cycle from 0 to 2MPa and back was repeated five
times on each sample in order to obtain the same standard of compaction. All measure-
ments were carried out with a constant vertical stress of 2MPa at room temperature. Two
systematic sets of measurements were run. The first one concerned samples with 0, 2, 4,
6, 8, 10% of bentonite by weight and saturated. These samples were then flushed with
NaCl brine (about 3 pore volume) having an increasing NaCl concentration of 50,000,
100,000, 200,000ppm (5%, 10%, 20% by weight) respectively. For each concentration,
including deionised water, the permeability of the sample was measured. The second set
of measurements concerned samples with 6, 8 and 10% of bentonite, saturated with a solu-
tion of 200,000ppm NaCl (20% NaCl by weight). These samples were then flushed with a
saline solution having a decreasing NaCl concentration of 100,000, 50,000 and 0ppm
(10%, 5% and 0%) by weight, respectively.
Figure 2. FESEM micrographs (magnification 10000x) depicting orientation and spatial
arrangement of clay tactoids at different hydration states. (a) Open micropore structure with face-to-
edge bonding of tactoids at saturation; (b) and (c) micropore closure and unidirectional alignment of
tactoids with decreasing water content (Viola et al., 2005).
4 RESULTS
Fig. 2 depicts a sequence of FESEM micrographs showing the bentonite fabric of a bentonite-
sand mixture (60% bentonite, 40% sand by weight) at different gravimetric water contents
(2.68kg/kg, 1.37kg/kg, 0.84kg/kg from (a) to (c)). At a magnification of 10000x, we can
clearly identify the arrangement of clay tactoids (stacks of individual lamellae). In the
saturated stage (Fig. 2a) tactoids form an open micropore structure with face-to-edge
bonding of clay layers. As the degree of saturation decreases (Figs.2b and 2c), we
observe micropore closures and a more directional alignment of tactoids in a face-to-face
orientation.
Fig. 3 shows the pore-size distributions obtained by NMR for natural dry bentonite satur-
ated with both deionised water and NaCl brine. Although the curves cover the same range,
from 5 to 20ms, the peak is delayed with deionised water, indicating that the swelling effect
with deionised water turns develops larger local V/S ratios experienced by the
1
H mol-
ecules. This means that the water uptake inside clay layers is exchanged with the water that
is outside rapidly, so that the average diffusion cell becomes wider.
The first set of permeability tests confirmed that in all samples, for a fixed bentonite
percentage, the permeability increases with increase of water salinity (Fig. 4). Fig. 5 shows
permeability against bentonite content (at constant water salinity). Several measurements
were made with different water salinity. The observed trend indicates that the addition of
very small percentages of clay induces variations in permeability of more than 3 orders of
magnitude.
During the experiments, the evolution of permeability with time of contact with the
saline solution was also examined. In most of the cases, a reasonable equilibrium was
attained within 1015 hours. For example, Fig. 6 shows the behaviour of the 8% bentonite
sample flushed with a brine at 50,000ppm NaCl.
The second set of measurements initially showed a slow decrease in permeability with
the decrease in brine concentration, and then a fast decrease in response to deionised water
flow. This suggests that the micro-texture of the sample, i.e. the spatial arrangement of clay
Figure 3. Pore size distributions of 100% clay with deionised water and NaCl. (Viola, 2005).
Figure 5. Permeability vs bentonite content: 0ppm NaCl, 50,000ppm NaCl, 100,000ppm NaCl,
200,000ppm NaCl.
Figure 6. Permeability against time in the 8% bentonite sample flushed with brine 100,000ppm NaCl.
Figure 4. Permeability vs water salinity: 0%, 2%, 4%, 6%, 8%, 10% bentonite from top to bottom.
particles, varies more quickly in response to small reductions of water salinity towards
deionised water (Fig. 7).
5 DISCUSSION
The results serve to highlight the influence of deionised and saline water on clay and clay-
sand mixtures. By means of FESEM analysis, it can be seen that deionised water has the
effect of enhancing an open structure in a clay fabric, in terms of edge-to-face arrange-
ment of tactoids. The DLVO theory and the hydration force explain the swell-shrink phe-
nomenon with particular emphasis on the separation of clay surfaces.
NMR experiments distinguished two different microstructures that arise when a clay
surface comes into contact with either deionised water or NaCl brine, i.e. pore sizes seem
to be larger in the case of deionised water than in the case of NaCl brine. It is worth stress-
ing that the relationship between local S/V values and pore dimensions depends on the
pore structure, which can be very complex. However, it is important to bear in mind that
in relaxation measurements the size tends to correspond to pore dimensions, and does
not include reference to the size of the pore throats.
Only if one assumes that the bigger the pores, the higher the permeability, one might
expect higher values of permeability in clay in association with deionised water. But it is
well known that deionised water destabilizes clay minerals, and the permeability measured
with deionised water is underestimated (Moore et al., 1982). It is likely that clay particles
float into the pore space, plug the pore throats, and block the flow. On the other hand, at
high salinities, cations from the electrolyte are directly adsorbed in the Stern layer, and this
adsorption stabilizes the clay mineral surface (Rowlands et al., 1997). Thus, the paradox
may be explained by exploiting the fact that NMR reflects the pore dimensions, whereas
transport properties are controlled by the throats. For 100% clay it is likely that, although
deionised water is responsible for bigger pores than water with NaCl, the throats are
blocked and the permeability is lower (Fig. 8).
Figure 7. Permeability against water salinity in samples initially saturated with saline solution
(200,000ppm NaCl): 6%, 8%, 10% bentonite.
Finally, permeability tests on clay-sand mixtures confirm that small variations of water
salinity do not generate as significant a permeability reduction as that observed between brine
and deionised water. For the sake of clarity, at the beginning it was shown how permeability
decreases when moving from 2% to 10% of bentonite in the mixtures, both flushed with
deionised water and NaCl brine. This trend correlates with the first part of the curve depicted
in the model drawn by Revil and Cathles, valid for brine (Fig. 9). This model further indicates
that when the clay content is higher than that which can be accommodated within the voids of
the coarser fraction, the permeability is primarily controlled by the clay content alone.
A deeper analysis of the experimental results, in terms of porous structures in the mix-
ture, allows us to explain, at each different bentonite percentage, the higher values of
permeability with NaCl water than with deionised water. When other textural components
are added, e.g. sand grains, mesopores form between them. Mesopores are larger than both
the interlaminar spaces and micropores in clay fabric. Thus, since clay swelling is less
evident with brine than with deionised water (NMR and clay model), more space is avail-
able and the permeability is higher.
REFERENCES
Borgia GC, Brown RJS, Fantazzini P (1998) Uniform-Penalty Inversion of Multiexponential Decay
Data, J Magn Reson, 132, 6577
Figure 8. Modelled effects of deionised water (left) and NaCl (right) on the microstructure of
100% clay.
Figure 9. Modelled effects of deionised water (left) and NaCl (right) on a clay-sand mixture up to
10% of clay.
Brighenti G, Macini P (1993) Permeability of clayey soils to liquid pollutants. In: Geology and
Confinement of Toxic Wastes (ed. by M. Barrs, B. Come and M. Arnould) (proc. Geoconfine 93,
Symposium, Montpellier, June 1993) vol.1, 1520. Balkema, Rotterdam
Brighenti G, Macini P (1994) Effectiveness of clayey layers as protective barriers for aquifers
against pollution. In: Future Groundwater Resources at risk (proc. Helsinki Conference, June
1994) IAHS Publ. no. 222, 409416
Chenu C, Tessier D (1995) Low temperature scanning electron microscopy of clay and organic
constituents and their relevance to soil microstructures. Scanning Microscopy, 9, 4, 9891010
Churaev NV, Sobolev VD (1995) Prediction of wetting conditions on the basis of disjoining pressure
isotherm. Computations Colloid J, 57, 6, 843850
Crang RFE (1988) Artifacts in specimen preparation for scanning electron microscopy. In: Artifacts
in biological microscopy (ed. by R.F.E Crang and K.L.K Klomparen) New York, Plenum Press
Derjaguin BV, Churaev NV, Muller VM (1987). Surface Forces. Plenum, New York.
Derjaguin BV, Landau L (1941) Acta Physicochim URSS 14, 633662
EPA United States Protection Agency (1998) Evaluation of Subsurface Engineered Barriers at Waste
Sites. EPA Report 542-R-98-005: www.clu-in.com
Israelachvili JN (1992) Intermolecular and Surface Forces, 2nd ed. Academic, San Diego
Kim JW, Peacor DR, Tessier D, and Elsass F (1995) A technique for maintaining texture and perman-
ent expansion of smectite interlayers for TEM observations. Clays and Clay Minerals, 43, 1,
5157
Low PF (1980) The swelling of clay. II. Montmorillonites. Soil Sci Soc Am J, 44, 4, 667676
Moore DE, Morrow CA and Byerlee JD (1982) Use of swelling clays to reduce permeability and its
potential application to nuclear waste repository sealing, Geophys Res Lett, 9, 10091012
Murray RS, Quirk JP (1990) Interparticle forces in relation to the stability of soil aggregates. In: Soil
Colloids and their Associations in Aggregates, (ed. by De Boodt, M.F., Hayes, M.H.B., Herbillon,
A.), Plenum Press, New York
Paunov VN, Dimova RI, Kralchevsky PA, Broze G, Mehreteab A (1996). The hydration repulsion
between charged surfaces as an interplay of volume exclusion and dielectric saturation effects.
J Colloid Interf Sci, 182, 239248
Quirk JP (1986) Soil permeability in relation to sodicity and salinity. Phil Trans R Soc Lond A
316:297317
Revil A, Cathles LM III (1999) Permeability of shaly sand. Water Resour Res, 35, 3, 651662
Rowlands WN, OBrien RW, Hunter PJ and Patrick V (1997) Surface properties of aluminium
hydroxide at high salt concentration, J Colloid Interface Sc., 188, 325335
Smart P and Tovey NK (1982) Electron microscopy of soils and sediments: techniques. Oxford,
Clarendon Press
Tessier D (1990) Behaviour and microstructure of clay minerals. In: Soil colloids and their associ-
ations in aggregates; (ed. by De Boodt, M.F., Hayes, M.H.B., A. Herbillon), New York, Plenum
Press
Tuller M, Or D (2003) Hydraulic functions for swelling soils: pore scale considerations. J of
Hydrology, 272, 5071
Verwey EJW, Overbeek JTG (1948) Theory of the stability of Lyophobic Colloids: the Interactions
of Soil Particles having an Electric Double Layer, Elsevier, New York
Viola R (2005) A critical analysis of pore space structure and transport properties in consolidated
and unconsolidated porous media by NMR relaxometry. PhD thesis, University of Bologna, Italy
Viola R, Tuller M, Or D, Drasdis J (2005) Microstructure of clay-sand mixture at different hydration
states. In: Advanced Experimental Unsaturated Soil Mechanics (ed. by Tarantino, A., Romero, E.,
Cui Y.J.) (Proc. Experus 2005, Trento, June 2005). 437442. Balkema, Rotterdam
Warkentin BP, Bolt GH, Miller RD (1957) Swelling pressure of montmorillonite. Soil Sci Soc Am
Proc., 21, 495497
CHAPTER 33
Modelling aquifer salinity in the Potash Basin (Alsace)
Marie-Luce Noyer
1
& Philippe Elsass
2
1
BRGM service EAU, Orleans cedex, France
2
BRGM Alsace, Parc Club des Tanneries, Tanneries cedex, France
ABSTRACT: Saline groundwater pollution originating from tailing dumps at the Mines de Potasse
dAlsace (MDPA) affects a large part of the Haut-Rhin Dpartement. Even if the upper layer of the
underlying aquifer shows a general tendency to improve following remediation works implemented
by the MDPA with the financial partnership of Agence de leau Rhin-Meuse (AERM), investigation
still has to be carried out on the salinity and salinity trends of the deeper aquifer layers. It is, therefore,
important to be able to use a numerical model to simulate the evolution of salinity in both the deep
and the shallow layers. The objectives of the model are to predict the long-term evolution of aquifer
salinity and to host test pumping scenarios, which will help to design an optimum programme of aquifer
quality restoration. This paper describes the different steps in the construction and calibration of the
model and the main results which were obtained.
1 INTRODUCTION
Groundwater in the Quaternary sediments of the central part of the southern Upper Rhine
Graben is strongly affected by chloride pollution as a consequence of the long time mining for
potash from Oligocene salt deposits. The two main areas of extensive mining are the Alsace
Potash Basin on the French side of the Rhine and the Heitersheim and Buggingen area on the
German side. Mining started in 1910 in the Potash Basin North of Mulhouse (France) and in
1927 near Buggingen (Germany). It ended in 1976 on the German side but was operated until
2002 on the French side. In this sensitive area, different hydrodynamic models have been
implemented in the past either in France or in Germany or through a cross border collab-
oration; for instance within the framework of the European programmes LIFE or INTER-
REG II (ANTEA 1996, BRGM 1996, BURGEAP-IMF 1996). BRGM has been involved
in two main studies over the last 10 years to investigate the mass transport modelling of the
saline pollution:
Modelling of the Potash Basin pollution using a 3-layer model built in 1998 (Noyer et al.,
1998a, b). This model is called the prototype model hereafter.
In collaboration with German partners, cross border modelling of the pollution in the
area extending from Fessenheim (France) to Breisach (Germany) near the Rhine river;
in this area, the pollution is mainly due to past infiltration from the former brine storage
basins on the Fessenheim island (Lang et al., 2005).
These two models concern two distinct areas in the Rhine valley each with different sources
of pollution; moreover, the models use different numerical codes and are based on differ-
ent calibration hypotheses and methodologies.
The present paper describes a third study performed by BRGM on the same area as the
prototype model (Potash Basin and downstream). This project was started in 2003 with the
financial partnership of Rgion Alsace, AERM and MDPA, following the completion of the
prototype study. This prototype model had been built on a state of the art basis, with a uniform
500m grid inherited from the former regional studies under the LIFE programme. In spite of
the innovative results that it brought, the grid of this model was too coarse to be used as a man-
agement tool. It was, therefore, necessary to build a new model with a finer grid. Moreover,
the conditions retained for the transient state calibration of the prototype were not optimal
(hydrodynamic calibration on a 5 year period only, transport calibration on 20 years but at a
yearly time step); the temporal discretization also had to be refined in the new study.
2 HYDROGEOLOGICAL CONTEXT
The model area covers the sector of the Alsace aquifer from Mulhouse upstream to Colmar
downstream, and from the Vosges slopes in the West to the Rhine River, which is taken into
account in the modelling, in the east (Fig.1). Geologically, the whole area belongs to the Rhine
plain with Quaternary alluvial deposits resting on a substratum of Oligocene age marls. The
alluvial deposits consist of a mixture of sands, pebbles and gravels locally divided by clay
layers of varying extension and thickness. The area can be subdivided into two parts of
unequal hydrogeological importance:
East of the Ill river, the thickness of the aquifer increases regularly from south to north,
from around 20m North of Mulhouse to more than 200m near Neuf-Brisach, with a
rise in the substratum along the Ill river, North of Ensisheim, due to the presence of salt
diapirs such as the Meyenheim crest and the Hettenschlag dome.
West of the Ill river, the aquifer thickness decreases regularly from 80m approximately
near the Ill in direction of the Vosges mountains; it amounts to 20 to 30m in the sector of
the Potash Basin, located north-west of Mulhouse and is characterized by the complex
topography of the Tertiary substratum which is carved by a succession of palaeovalleys.
The complex hydrographic network includes the Rhine River along with the Grand Canal
dAlsace, the Fecht, the Ill and its tributaries: Lauch, Thur, Vieille-Thur and Doller, the main
canals (Vauban, Widensolen, Hardt, Rhne-au-Rhin, Huningue) as well as a network of
drainage ditches. Inputs to the groundwater model consist in the following components:
recharge from rainfall, infiltrations from the surface network and inflow coming from the hills
located at the southern and western limits of the model (Vosges, Sundgau). Outputs cor-
respond to the withdrawals for drinking and industrial water in the whole model area plus the
withdrawals from the remediation wells installed in the vicinity of the tailing dumps in the
Potash Basin area.
3 SALINITY CONTEXT
Potash has been mined since the beginning of the 20th century in the Potash Basin located
in the south-western part of the model area (Fig. 2). The by-products of exploitation, mainly
sodium chloride, have been deposited on tailing dumps on the permeable alluvium or dis-
charged into the rivers (the Thur and the Vieille-Thur) until 1975; after this date, the brines
were fed into the pipe system called saumoduc, created in 1934 to transport and discharge
the salt into the Rhine River at Fessenheim. The mine dumps were fed by deposits of salty
residues from the ore processing until the end of the 1950s and the salinity of the aquifer
results from the dissolution processes of the salt contained in the dumps by rainfall recharge
since the beginning of potash mining. The pollution from the mine dumps results in two distinct
saline plumes in groundwater with the following configuration since the beginning of the
1990s: the Eastern plume (originating mainly from the dumps named Joseph Else, Amlie,
Anna, Fernand) is initially directed west to east then presents a sharp azimuth variation
towards the north-east and becomes parallel to the western plume coming from the dumps
at Marie-Louise, Alex and Rodolphe. This characterization of the source of groundwater
salinity presents two zones of interest regarding data analysis and groundwater modelling:
the Potash Basin zone, which includes the tailing dumps as main sources of saline
infiltration, with high local concentrations (up to 50g/l in the vicinity of the dumps);
Modelling aquifer salinity in the Potash Basin (Alsace) 447
Figure 1. Morphological map of the area.
the plume zone downstream in which the main processes are related to the transport
with its advective, diffusive and dispersive components, due to the heterogeneity of the
velocity field. In this zone, salt concentrations are much lower. The only other salinity
sources present in this zone are of much lower amplitude and are due to the infiltration
from the surface network into groundwater.
Within the Potash Basin, the fifteen principal mine dumps representing the sources of pollu-
tion are organized in seven sectors; their detailed periodic survey is the subject of an annual
synthesis report on salinity prepared by BRGM. Today all these tailing dumps either have
already been remediated or are in the process of remediation: the most saline dumps are
treated by accelerated dissolution with collection of the resulting brine by a surrounding
drainage ditch and by fixation wells, while the less saline dumps are covered with a grass-
seeded impervious cap. To complete the review of the known sources of salinity, it is neces-
sary to add a last source of lower amplitude located upstream of the Potash Basin (south-west
of the model area). At the outlet of the Thur valley, groundwater is contaminated by the
Ochsenfeld industrial dumps. The by-products of the industrial processes generate, among
other elements, sodium chloride. The remediation wells, set up in 1974, limit their impact to
a narrow plume (200mg/l chloride), which joins the plumes issuing from the Potash Basin
mine dumps. This pollution should cease in 2005 following the installation of a watertight
barrier. All these sources of salinity which are known and taken into account in the model-
ling are well identified anthropogenic sources. It is not excluded that there may also be nat-
ural sources such as brine rising up through faults or issuing from the dissolution of salt
domes, but these have not yet been clearly identified.
Figure 2. Observed salinity in 2004 in layer 2.
4 CONSTRUCTION OF THE REFINED MODEL
The hydrodynamic and transport software package used in this study is the MARTHE code,
developed by BRGM (Thiry, 1995). The construction of the new refined model was based
on the re-use of data from the prototype model wherever possible, plus the analysis and
integration of new knowledge obtained since 1998. The main process was as follows:
4.1 Complete redefinition of the geometry of the alluvial layers
The coarse Quaternary alluvial deposits of the Rhine plain were deposited by the Rhine River,
bringing clastic material of alpine origin, while the tributaries from the Vosges and the Black
Forest brought eroded material from ancient mountain ranges. The distinction between the
areas with Vosges material with the alpine material is fundamental because the Rhine alluvial
material (of alpine origin) is well rounded and sorted with little fine-grade material, while the
Vosges deposits are not so well sorted and are less clean (i.e. contain more silt and clay). A
multi-layer conceptual model had been established for the prototype hydrodynamic model
carried out by the BRGM in 1997 (Noyer et al., 1998a,b). It was based on the cross-border
work undertaken in collaboration with the teams of the Bade-Wurtemberg Geological Survey.
The 1998 conceptual model considered three layers inside the alpine alluvia; this subdivision
was confirmed by drilling which was carried out within the framework of the INTERREG II
project Reconnaissance transfrontalire de laquifre profond dans la bande rhnane entre
Fessenheim et Breisach (Regierungsprsidium Freiburg 2002). This three layer conceptual
model was thus retained in the present study for the alpine alluvium.
The conceptual model had to be modified for the Vosges alluvium. The drilling carried out
since 1998 in the Potash Basin downstream of the mine dumps Amlie and Rodolphe in par-
ticular, had highlighted the presence of a basal sequence in the Vosges sequence consisting
of coarse-grained sand and gravel. It is relatively permeable, and is sometimes isolated
under a clay layer, which had previously been mistaken for the substratum. In addition to
the redefinition of the alluvium, the substrata and topography were updated. The topography
of the prototype model based on the Digital Elevation Model from the Institut Gographique
National has been updated for the Potash Basin with data communicated by the MDPA to
take into account mining subsidence; these data were topographic contours measured in
December 2003. As mining ceased in September 2002 it is likely that the topography of
the MDPA in December 2003 will be the final topography of the area.
The conceptual model has been greatly modified since the 1998 prototype version. The
three layers of the model after discretization at the 500m mesh size are presented in Figure
3. In each layer the thickness varies widely, increasing from south-west towards the north-
east from a sometimes very low value (of the order of one metre) to approximately 65m
for layer 1, 175m for layer 2 and 37m for layer 3. None of these layers is completely con-
tinuous throughout the whole of the study area. One of the assets of the MARTHE code is
that it can manage the local absence of layers as well as the numerical problems likely to
occur by mesh dewatering in the south-east part of the model where the layers are thinnest.
4.2 A horizontal discretization locally refined with a grid of 125m
The horizontal discretization at a 500m mesh size of the prototype model was sufficient
to represent regional piezometry and even the flows in the rivers, but not to simulate the
transport of brines, in particular in the Potash Basin where the distances between the dumps
and the remediation wells vary between 100m and 1km. It is thus clear that in certain cases,
a 500m mesh contains the dumps and one or several wells; consequently, the prototype
model can only simulate an average effect in this mesh resulting from the infiltration
under the dump and the withdrawals from the wells; obviously, this average effect can be
rather different from the really observed phenomena. In addition, one important characteris-
tic of the salinity field is the relative narrowness of the saline plumes induced by the trans-
port; thus, the western plume is only about 500m wide for several kilometres of its length
and this is also the size of the prototype mesh; the strong transverse gradients observed in
salinity cannot, therefore, be correctly reproduced. For the construction of the refined model,
the choice was made to use the nested grid option of the MARTHE code in order to refine
the horizontal discretization in the whole Potash Basin where the mesh surface has been
divided by 16 as compared to the prototype model (new mesh size is 125m). In this way, all
the data are treated in a much more precise way in this sensitive zone; in particular, it can
determine the impact of infiltration under the dumps and of abstraction from the fixation
wells which are now in distinct meshes; as already mentioned.
Figure 3. Model geometry extension and thickness of each layer.
4.3 Updating of all the data necessary for hydrodynamic simulations
Previous studies have shown that the Alsace alluvial aquifer is in hydraulic contact with the
hydrographic network (Rhine, rivers, canals, drains). The dynamic interactions between the
aquifer and the surface waters must be carefully taken into account in the modelling.
Consequently, it is necessary to collect the data specific to this interaction:
data characterizing the hydrographic network for each mesh crossed by an element of the
surface waters: length and width of river section, water height, thickness and permeabil-
ity of the river bed, roughness of the bed; the last three parameters are usually unknown
and must be tested during the calibration process;
time series data for monthly discharge since 1975 up to 2004 at 10 gauging stations
obtained by interrogation of the HYDRO database; among the 10 stations, five are used
to characterize the discharge at the entry points of the rivers in the model (Didenheim on
the Ill, Masevaux on the Doller, Willer on the Thur, Ungersheim on the Vieille-Thur and
Guebwiller on the Lauch River respectively); the other five stations are used as control
points for the calibration (Staffelfelden on the Thur, Reiningue on the Doller, Ensisheim,
Oberhergheim and Colmar on the Ill);
time series data for monthly discharges of the Rhine during the same period at the Kembs
station (entry point of the Rhine in the model) and some information on the canals pro-
vided by the Service de la Navigation de Strasbourg and by the Direction Dpartementale
de lAgriculture et de la Fort (DDAF) du Haut-Rhin;
available data on the network of shallow drainage ditches.
The calculation grid (groundwater and surface network) is presented in Figure 4 along with
the position of the gauging stations (the mesh is not shown in the 125m grid area). The other
required data are withdrawal data, hydro-climatic data for recharge calculation and at least
one piezometric map and time series data for piezometry at different observation points.
Piezometric data: for steady state calibration, the average situation for June 2002 was
used. The observed piezometry was established using data from the APRONA and
MDPA piezometric networks. A series of heads measured on about 20 piezometers dis-
tributed in the whole model area was used to calibrate the model in the transient state over
27 years (from January 1978 to December 2004). These time series were supplied by
APRONA or extracted from the BRGM data bank (Banque de Donnes du Sous-Sol).
Hydro-climatic data: long series of temperature, insolation duration and rainfall data
acquired from the Mto-France database at 4 climatic stations (Meyenheim, Baldersheim,
Mulhouse, Burnhaupt) between 1960 and 2004, at a monthly time step.
Withdrawals: annual withdrawals since 1975 for drinking and industrial water in the whole
model area plus, in the Potash Basin, the withdrawals from the remediation wells installed
in the vicinity of the mine dumps. This includes all the annual withdrawals except those
for irrigation, which are poorly known and otherwise neglected. Following the redefi-
nition of the model geometry, these withdrawals were then distributed in each layer
according to the position of the borehole screens. The WinMarthe pre-processor enables
these data to be imported to the input format of the MARTHE software and to sum them
up by mesh and layer. These values can vary significantly from one year to another in
particular in the Potash Basin because of the progressive installation and decommission-
ing of the remediation wells.
4.4 A better definition of the salinity and its evolution under the effect of various
remediation operations
BRGM (Service Rgional dAlsace) monitors the groundwater salinity, collects chloride
concentrations from various locations of the aquifer and surface waters and publishes annual
reports. For these reports, it establishes the annual map of salinity distribution. For the trans-
port simulation, two salinity situations were discretized, layer by layer, using the observed
data for 1991 as an initial condition for the transport calibration and for the 2004 in order to
check the results at the end of the calibration period. Salinity data in the rivers come from
only a few gauging stations equipped for recording chloride concentrations and are located
on the main rivers: Thur, Vieille-Thur, Ill, Lauch, Doller. The other required data for the
transport modelling are related to the dumps (surface, treatment history and time series of
infiltration, planned evolution) and the remediation wells (time series of concentration and
discharge, planned evolution); these data were provided by the MDPA for the calibration
period (19912004). The infiltration data under the dumps are estimated by the MDPA on
the basis of the study known as the Graillat study (Graillat and Brunck, 1980) and progres-
sively adjusted according to the observations made during operations on the dumps and to
the evolution of their surface of influence during their rehabilitation. The infiltration data
were provided by the MDPA expressed as tonnes of NaCl per year for each dump; they were
transformed into kg of Cl
per month per m

2
; MARTHE is then able to read these values
directly at each time step of the calculation.
Figure 4. Aquifer and surface network discretisation with locations of gauging stations and mine
dumps.
5 MODEL CALIBRATION
The model calibration was carried out in three phases with iterations between phases:
Hydrodynamic calibration in steady sate regime on the June 2002 situation.
Hydrodynamic calibration in transient state regime over 27 hydrological cycles (January

1978 to December 2004) at a monthly time step, taking into account the dynamic inter-
actions with the hydrographic network (Rhine, rivers, canals, drains); 5 gauge stations
and approximately 20 observation wells were used as control points. It must be noted that
the hydrographic network is only partly included in the model area; it is, therefore, neces-
sary to explicitly define source points for the different rivers and canals at their entry
into the model: for the rivers, the discharges at 5 gauge stations close to the river entry
points are used (Didenheim on the Ill, Masevaux on the Doller, Willer on the Thur,
Ungersheim on the Vieille-Thur and Guebwiller on the Lauch respectively); for the
Rhine, the discharge at the Kembs station is used; for the canals, no gauge station is avail-
able but the discharges at the canal entry points have been estimated and calibrated.
Mass transport calibration at a monthly time step from January 1992 to December 2004
with control on the salinity maps of 2004 and on the evolutions in 19 observation wells
with one or more screened levels, distributed in the Potash Basin and the east and west
plumes, and in 17 remediation wells located downstream from the mine dumps.
For the calibration, the following options were carried out:
Integrated recharge calculation: definition of 5 hydroclimatic zones based on a map of

long term average annual rainfall from AERM and on long series of meteorological
data on 5 stations. For each zone, the BRGM lumped hydrological model GARDENIA
(Thiry, 2003) was used for rainfall/discharge calculations with 14 to 30 years of meas-
ured rainfall data and calculated potential evapotranspiration data (PET), to provide
estimates of the calibration parameters. These estimates are read by the hydrodynamic
code MARTHE, which is then able to calculate the balance rainfall/PET and to distrib-
ute the surplus of rainfall between infiltration to the water table and runoff to the sur-
face network for each zone and each time step.
Groundwater/surface water interactions: a dynamic treatment of the interactions was

carried out based on the relationship between water height and the flow in the rivers
according to the Manning-Strickler law. It was calibrated on the river discharge meas-
ured at five gauging stations.
Density effects: on the extension of the salt plumes, salinity varies from almost 50g/l under
the tailing dumps to less than 200mg/l at the downstream end of the plumes; the calcula-
tions previously carried out with the prototype model indicated that the density effects can
affect the velocity field only in the immediate vicinity of the dumps, and isodensity trans-
port driven only by the pressure gradient is dominant downstream. Moreover, in order
to perform a detailed simulation of the density effect in the vicinity of the dumps, it would
be necessary to set up an inner model which would be based on the 3 layer geometry of the
present model with a refined vertical discretization of each layer into several sub-layers.
Obviously this fine vertical discretization is not necessary over the whole extent of the
present model. In order to avoid unnecessarily complicating the already time-consuming
calculations, the decision was taken to simulate the density effects under the dumps by dis-
tributing the infiltration in the 3 layers of the model. This distribution is calibrated on the
chloride concentration observed in the remediation wells downstream from the dumps and
on the concentration maps observed during 2004 in each layer.
6 RESULTS
In Figures 5 and 6, two illustrations of the hydrodynamic calibration in the transient regime
are presented, the first one on an observation well located near Ensisheim, the second one
on a gauging station at Staffelfelden, both in the Potash Basin; the locations of these two
points are shown in Figure 4. After an adjustment period on some of the hydrological cycles
required because the initial condition (steady state simulation on the June 2002 situation) is
Observation well 4132X0193 Ensisheim 41C
210
212
214
216
218
220
222
1978 1981 1985 1993 1989 1997 2001
Years
P
i
e
z
o
m
e
t
r
i
c

H
e
a
d

(
m
)
Observed Simulated
Figure 5. Example of head calibration on an observation well at the output of the Potash Basin.
THUR river flow at Staffelfelden station
0
5
10
15
20
25
30
35
F
l
o
w

(
m
3
/
s
)
1978 1981 1985 1993 1989 1997 2001
Years
Observed Simulated
Figure 6. Example of discharge calibration on the Thur River inside the Potash Basin.
different from the observed situation at the beginning of the series (January 1978), it can be
observed that the simulated hydrograph fit is quite satisfactory.
Also the fitting of the simulated discharge of the Thur River at Staffelfelden is good on
the whole simulation period, especially for the low water periods.
The results on the other observation wells and gauging stations are similar for the cali-
bration period and can be seen in Noyer and Elsass (2006). In Figures 7 and 8, two illus-
trations of the mass transport calibration are presented. Figure 7 presents the concentration
evolution on an observation well with multiple screening: 1/15M and 2/25M tapping layer 2,
4/38M tapping layer 3. The simulated concentration curves are labelled according to the
tapped layer.
Figure 8 presents the concentration evolution on a decontamination well downstream from
the dump Amlie in an area of high saline density. The well does not tap layer 3, and at the
end of the observed curve, a sudden increase in measured concentration can be observed.
This is due to the start of the accelerated dissolution treatment of the dump; this feature has
not been taken into account in the modelling because of the complexity of the process.
Moreover, the observed and simulated salinity maps at the end of the calibration period
(December 2004) are presented in Figure 9. They show an overall satisfactory agreement.
Observation well 4131X0294 Nonnenbruch Wittenheim
0
2000
4000
6000
8000
10000
12000
14000
1991 1993 1995 1997 1999 2001 2003
Years
C
h
l
o
r
i
d
e

C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
g
/
L
)
1/15M layer 2
2/25M layer 2
4/38M layer 3
Simulated layer 2
Simulated layer 3
Figure 7. Example of transport calibration on a multiple observation well in the Potash Basin.
Fixation well 4131x0213 Amlie 1 P1
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
C
h
l
o
r
i
d
e

C
o
n
c
e
n
t
r
a
t
i
o
n

(
m
g
/
L
)
Observed
Simulated layer1
Simulated layer2
1991 1993 1995 1997 1999 2001 2003
Years
Figure 8. Example of transport calibration on a fixation well near dump Amlie.
Figure 9. Observed and simulated salinity maps in 2004.
7 CONCLUSIONS
In spite of some inaccuracies in the immediate vicinity of the tailing dumps where the dens-
ity effects are very significant, the resolution at the end of the calibration period of the
observed maps and series is acceptable. After the completion of the calibration process, vari-
ous exploitation scenarios were carried out in order to predict the future evolution of the
salinity in the study area, given various assumptions on the likely course of remediation to be
adopted by the MDPA.
These scenarios have shown that in 2014 the downstream part of the plumes will be
largely decontaminated in the upper layer but not in the deep layers. In 2027, the east plume
will be cleaned-up within the deep layers too but not the west plume. In the Potash Basin
itself, the upper layer will be cleaned up by 2027 but concentrations up to 2g/l may still be
present in the deep layers.
The site specific study integrating regional flow and transport modelling for the simu-
lation of aquifer remediation is in line with the monitoring of the impact of former mining
activities, a new public service, recently entrusted to BRGM.
ACKNOWLEDGEMENTS
This study was funded by BRGM (Research Direction) and by the following partners :
Rgion Alsace, Agence de leau Rhin-Meuse and Mines de Potasse dAlsace. The authors
want to thank the two anonymous reviewers for their useful suggestions and comments.
REFERENCES
ANTEA (1996) Projet LIFE. Modlisation des eaux souterraines de la plaine dAlsace. Secteurs
Mulhouse-Colmar zone 4 et Ble-Mulhouse zone 5. Rap. A07533.
BRGM (1996) Projet LIFE. Modlisation des secteurs franais en rgime permanent pour les basses
eaux. Rap. R 39068.
BURGEAP-IMF Strasbourg (1996) Rgion ALSACE Projet LIFE Modlisation des secteurs de
Colmar et de Mulhouse/Ble en rgime permanent pour les Hautes Eaux et les Moyennes Eaux.
Elsass P, Rau S (1995) Coupes hydrogologiques Strasbourg-Offenburg. Notice. Initiative commu-
nautaire INTERREG, Cartographie hydrogologique du Rhin suprieur. Rgion Alsace,
Strasbourg, avril 1995.
Graillat A, Brunck R (1980) Direction interdpartementale de lindustrie rgion Alsace. Les terrils
du bassin potassique. Etude en vue de leur neutralisation vis vis de la pollution saline des eaux
phratiques et de leur insertion dans le paysage. Novembre 1980.
Lang U, Gudera T, Elsass P, Wirsing G (2005) Numerical modelling of chloride propagation in the
Quaternary aquifer of the southern Upper Rhine Graben. International Journal of Earth Sciences,
94, 550564.
LfU (1996) Projet de dmonstration pour la protection et la gestion de la nappe franco-germano-
suisse du Rhin suprieur. Demonstrationsvorhaben zum Schutz und zur Bewirtschaftung des
Grundwassers des deutsch-franzsisch-schweizerischen Oberrheingrabens. LIFE
Abschlussbericht, Karlsruhe.
MDPA (1991) Sol et sous-sol. Les affaissements miniers. Potasse magazine N101, novembre 1991,
pp 46.
Noyer ML, Menjoz A, Elsass P, Thiry D, Martin JC (1998a) Modle de la nappe dAlsace. Projet de
dveloppement 312. Rapport final. Rapport BRGM R 40335.
Noyer ML, Menjoz A, Thiry D, Elsass P, Martin JC (1998b) Modlisations monocouche et multi-
couche de la nappe dAlsace dans la zone du Bassin Potassique. in AIH Journe technique du
Comit franais de lAIH 5me Strasbourg France 27/11/1998, pp 3338.
Noyer ML, Elsass P (2006) Modlisation de la salure profonde au droit et en aval du Bassin
Potassique. Rapport BRGM RP54389-FR.
Regierungsprsidium Freiburg (2002) Reconnaissance transfrontalire de laquifre profond dans la
bande rhnane entre Fessenheim et Breisach. Rapport final projet Interreg II.
Thiry D (1995) Modlisation 3D du transport de masse avec le logiciel MARTHE version 5.4.
Rapport BRGM R 38149 DR/HYT 95.
Thiry D (2003) Logiciel GARDNIA version 6.0. Guide dutilisation. Rapport BRGM
RP52832-FR.
CHAPTER 34
A multidisciplinary approach for assessing the risk of
seawater intrusion in coastal aquifers: The case of the
Roussillon Basin (France)
Bertrand Aunay
1,2
, Nathalie Drfliger
1
, Cdric Duvail
1,3
, Frdric Grelot
4
,
Paul Le Strat
1
, Marielle Montginoul
4
& Jean-Daniel Rinaudo
1
1
BRGM EAU/RMD, Montpellier, France
2
Hydrosciences Montpellier, University of Montpellier II, France
3
Laboratoire Dynamique de la Lithosphre, University of Montpellier II, France
4
Cemagref, UMR G-EAU, Montpellier, France
ABSTRACT: Littoral zones are important places of economic development, often linked to inten-
sive water needs. Geological, hydrogeological and socio-economic studies have been carried out as
part of a multidisciplinary approach on a coastal Mediterranean aquifer the Roussillon in order
to design effective water management strategies in areas sensitive to seawater intrusion risks.
Geology provides the geometry and architecture of the different aquifers, hydrogeology assesses the
seawater intrusion risk while the socio-economic study includes consulting stakeholders with the
aim of suggesting water management and policy options acceptable to the majority of the popula-
tion. This paper highlights the economic interests at stake, the diversity of viewpoints expressed by
stakeholders and the political dimensions of the issue, which are likely to be encountered for all sim-
ilar situations on both sides of the Mediterranean Sea.
1 INTRODUCTION
In many Mediterranean coastal areas, agriculture, drinking water supply, tourism and
industry strongly depend on groundwater resources. As a result of significant economic
development over the last three decades along the Mediterranean coast (growth in tourism,
population migration and agricultural intensification), abstraction from coastal aquifers
has greatly increased. This has often resulted in a significant decline in water tables that
has, depending on the local geological context, increased the risk of seawater intrusion
(Nixon et al., 2003, Plan Bleu, 2005). In a number of locations (in Spain, Italy and Greece
for instance), although action is clearly needed to avoid any irreversible degradation of
groundwater resources, the status quo often prevails and pumping fever does not seem to
be affected by the threat of salt water intrusion (Aguillera-Klink et al., 2000). As in many
other Tragedy of the Commons (Hardin, 1968), this is often explained by the intensity
of the interests at stake and the lack of appropriate institutions. We argue here that another
key factor explains the absence of social response. Stakeholders, who are increasingly
involved in water planning and management decisions, often do not share the same view
of the nature and the intensity of the risk of salt water intrusion. Since they do not agree
with the nature of the problem, its origin and its intensity, there can be no consensus on
how it should be solved (Rinaudo and Garin, 2005). Clearly, constructing a common under-
standing of the risk of sea water intrusion is a prerequisite to defining a sustainable water
management strategy. However, assessing this risk is a difficult exercise for Mediterranean
coastal aquifers, which are generally complex multi-layered systems. It requires extensive
knowledge of the geology and the hydraulic conditions of the aquifer, both onshore and off-
shore. The first part of this paper illustrates, through a French case study, how the risk of
seawater intrusion can be assessed using reservoir geology methods applied onshore
(through genetic stratigraphy) and offshore (through seismic stratigraphy) (Cross et al.,
1993, Homewood et al., 2002). The method applied as well as some of the results could
be used elsewhere around the Mediterranean. In its second part, the paper illustrates that, as
long as this risk is not assessed precisely by scientific experts, stakeholders may have very
different perceptions of the risk and may promote very different groundwater management
strategies.
2 PRESENTATION OF THE CASE STUDY AREA
The case study area selected for carrying out this multidisciplinary research is the Roussillon
Basin, located along the southernmost part of the French Mediterranean coast, near to the
Spanish border. This 700km
2
sedimentary basin is bordered by the foothills of the Pyrenean
Mountains to the south, the Corbires karstic region to the north and the Mediterranean Sea
to the east (Fig. 1). The groundwater resources comprise several Pliocene confined aquifers,
which are overlain by a Quaternary deposit aquifer and embedded within a Miocene struc-
tured margin (Fig. 1). These aquifers are intensively used for drinking water purposes, for
tourism related activities along the coast and by agriculture for irrigating orchards and veg-
etables crops. Since the superficial aquifer was increasingly affected by diffuse pollution
(nitrates and pesticides), the Pliocene aquifers have been increasingly exploited, in particu-
lar by municipalities but also by large vegetable producers. The resulting decline in the
water tables, which has been observed over the last 20 to 30 years, is expected to continue
as the population keeps growing and the farming sector progressively abandons ancient
surface canal irrigation systems in favour of new drilled irrigation wells.
Although high chloride concentrations have been reported in several wells located along
the coast, it has not yet been proven that seawater intrusion actually takes place in the cur-
rent situation. In particular, uncertainties remain concerning the relation between the
Quaternary aquifer (which may be brackish in certain locations) and the Pliocene aquifers
(Fig. 2). Also, experts had no information before this research was initiated on the
geology offshore, which determines if sea water intrusion can or cannot take place in the
Pliocene aquifers (Fig. 2). The mathematical model that was developed in the early 1990s
(Auroux et al., 1992) has the same caveat. In a context of sustained growth in groundwa-
ter exploitation, policy makers decided there was an urgent need to improve our knowl-
edge of the structure and functioning of the Pliocene aquifers and to identify possible
strategies to mitigate the risk of groundwater deterioration. The following section presents
an attempt to answer this demand.
A multidisciplinary approach for assessing the risk of seawater intrusion in coastal aquifers 461
Figure 1. Location of the Roussillon Basin.
Figure 2. Cross section of the traditional conceptual model of the Roussillon Basin (Marchal et al.,
1976).
3 ASSESSING THE RISK OF SEAWATER INTRUSION IN THE
ROUSSILLON BASIN
In complex multi-layer aquifers, assessing the risk of sea water intrusion requires a detailed
characterisation of aquifer boundary conditions and interconnections between aquifer layers.
This detailed characterisation requires a good understanding of regional geological history
(section (a) below) before applying classical hydrogeological methods (section (b)). These
two items of information combined lead to the development of a conceptual model (section
(c)) that can be used to assess the risk of sea water intrusion.
3.1 Geological history of the case study area
During the Late Miocene, a marine regression followed the partial drying of the Mediter-
ranean Sea, (Messinian salinity crisis, approximately 5.8Ma). A major erosion phase
induced a depression of the fluvial networks and the creation of deep canyons that have
been identified in the Roussillon (Clauzon et al., 1987). During the Pliocene (5.3Ma), the
opening of the Gibraltar Straits allowed the return of seawater, which penetrated into the
deep valleys and transformed the Roussillon Basin into a ria (fluvial canyon drowned by
marine transgression). All the onshore Pliocene sediments were deposited within 1.7 m.y.
(i.e. 75cm/1000 years, Clauzon et al., 1987). The Quaternary deposits were due to inter-
glacial transgression and regression phases. The last transgression (sea level rise) resulted
in some coastal constructions and a lagoon area from Leucate to Argels. Today, only the
Salses-Leucate and Canet lagoons remain.
The first part of this study consisted in developing a detailed lithological and geometric
geological model of this basin, on the basis of reservoir geology methods (Duvail et al.,
2005): (i) onshore, through genetic stratigraphy performed with the BRGM public geolog-
ical database (more than 500 wells, with about 120 logs), the interpretation of well logs and
cross-checking with the outcrops, (ii) offshore, through seismic stratigraphy (existing oil
company sections and high-resolution seismic profiles from the University of Perpignan).
The result is a detailed mapping of the elevation of the top and bottom of the aquifers, the
aquitards and aquicludes along with their lithology, their interrelationship with other
aquifers (especially karstic, whose structure is also influenced by sea level variations).
Lithological and geometrical knowledge of the aquifers makes it possible, in particular, to
determine if coastal aquifers are directly or indirectly connected to the Mediterranean Sea.
3.2 Hydrogeology of the case study area
The second step of the analysis consisted in determining the link between the geology and
the hydrogeology of the studied area (Fig. 3). Several small-size aquifers make up each
aquifer layer, each one having different hydrogeological properties. For instance, marine
Pliocene sediments are composed of stacks of five different sedimentary prisms (sand
and clay).
The Quaternary aquifer is mainly composed of alluvium (flood plain or conglomer-
ate) and coastal deposits, the geometry of which is linked to Pliocene-Quaternary glacio-
eustatic variations. The unconfined Quaternary aquifer lies along the main rivers and the
coastline. It is mainly exploited by farmers, private individuals and campsite owners.
The water quality of the Quaternary upper aquifer is poor near the coast due to a high chlo-
ride concentration (the electrical conductivity can reach 5000 S/cm at St-Laurent-de-la-
Salanque, 5km from the coastline).
The Pliocene aquifers (continental and marine), whose characteristics are linked to the
Messinian event, comprise fluvial deposits and marine sands and clays. Thirty years ago
there were many artesian wells tapping these aquifers but, nowadays these wells are no
flow naturally. The confined Pliocene aquifers are mainly exploited for drinking water
supply and by farmers. Generally, both water quality and the hydraulic head of the
Pliocene aquifers increase with the depth of the exploiting well. Even under the offshore
bar, the water is still of good quality: chloride concentrations are lower than 40mg/l.
Locally, some interconnections exist between the Quaternary and Pliocene aquifers, mainly
due to two reasons: (i) leakage from existing wells and (ii) incision of the Pliocene aquifers
by Quaternary channels. These interconnections tend to deteriorate the water quality of the
Pliocene aquifers.
Fluvial sands channels of deposited in a deltaic environmenent constitute the Continental
Pliocene aquifer. This water in this aquifer is only excessively mineralised in the northern
part of the basin, along the lagoon and even more so along the coast at Barcars. The conti-
nental Pliocene aquifer is very productive, especially in the Salanque Plain. Because of its
Figure 3. Geological (1) hydrogeological (2) cross sections of the Roussillon Basin. Vertical
leakage linked to hydraulic head inversion is illustrated in the hydrogeological cross section. Vertical
leakage takes place through a more sandy continental lignite layer or through defective boreholes.
Scale and location are not presented because these theoretical cross sections only illustrate the
hydrogeological problem.
geological formation (stream channel and flood plain alternations), this aquifer is considered
to be a heterogeneous aquifer: the permeability tensor and salt water interface position could
vary widely in three dimensions. Usually, layers of lignite with plant remains alternating with
plastic marsh clays are found beneath the Continental Pliocene aquifer. This non-continuous
impervious layer separates the continental Pliocene aquifer from the sandy marine Pliocene
aquifer.
Sandy marine Pliocene sediments correspond to the prograding deltaic shore face.
Theoretically, this homogeneous aquifer is not connected to the sea and it is preserved
from seawater intrusion. However, some areas contain excessive mineralization. The ori-
gin of this contamination is vertical leakage from the partially contaminated Quaternary
aquifer to the Pliocene aquifers, due to defective boreholes (Marchal and Chery, 1995).
New boreholes drilled next to the old ones show a strong decrease in water salt content.
Without any extraction, the hydraulic head of the sandy marine Pliocene aquifer is higher
than that of the continental Pliocene aquifer.
Improved understanding of the hydrogeological model elaborated from reservoir geology
methods clears up the uncertainties of the traditional model concerning aquifer interconnec-
tions and extensions. Our investigations suggest that a strong link exists between the upper
and lower layers, although the piezometric levels are apparently different because of a pro-
nounced inertia of the Roussillon Pliocene aquifers (Aunay, forthcoming). They also indicate
that, whereas the Quaternary aquifer is vulnerable to seawater intrusion, the Pliocene aquifers
are most probably protected by a five kilometre long impervious layer (clay on Fig. 3).
3.3 Increasing salt water intrusion risk
Based on electrical conductivity measurements and chemical analyses (Fig. 4), seawater
intrusion is well recognized in the Salanque Plain a few kilometres from the coast relating
to the unconfined superficial Quaternary aquifer. In the South Roussillon Basin, seawater
intrusion is restricted to within one kilometre from the coastline.
Near the coast and since exploitation began, the water level in the Pliocene aquifers has
dropped to 0m regularly during the year (Fig. 5). Moreover, the range in annual variation
increases year after year. Although abstractions are almost constant in the Quaternary
aquifer, there has been a threefold increase for the Pliocene aquifers since 1975 (Accord-
cadre, 2003). A drop in groundwater levels means that leakage direction and vertical flow
are reversed: leakage was upward before exploitation, whereas leakage is now downward.
Even if no direct connections exist between the sea and the Pliocene aquifers, the risk of
seawater intrusion is real because of vertical flow through the contaminated superficial
Quaternary aquifer. In the same way, chloride concentration increases locally in a few
observation points located in the Pliocene aquifers. At present, these increases are geo-
graphically limited, but they are also continuous for each observed point.
Today, the situation is serious but not totally irreversible. Nevertheless, after field obser-
vations, field measurements and geological studies, some quantitative hydrogeological
studies are necessary in order to determine the most suitable method for exploiting the
Roussillon aquifers. These quantitative hydrogeological studies began with variable-density
flow and solute transport simulations (Aunay, forthcoming). In these simulations, because
good geological information is available, interconnection and aquifer geometry constitute
key parameters. In the same way as geological information, knowledge of the hydraulic con-
ductivity distribution comes from conventional well tests and from geostatistical stress
analyses of the geological model of the Roussillon Plain. The aim of the modelling is to
determine, amongst other things, the time of occurrence and the velocity of vertical leakage.
Such studies enable the risk of seawater intrusion into the aquifers to be determined in
different coastal areas. The saltwater intrusion risk may be assessed for the present time and
for various water management scenarios. For example, in the Salanque Plain, because of a
specific geology, the origin of which is attributed to differential subsidence, the seawater
intrusion risk is more pronounced than in the south of the Roussillon Basin. Regarding
hydrogeological results, the water management recommendations will be different for each
aquifer. These recommendations, resulting from geological and hydrogeological models,
need to be discussed with the stakeholders to determine the best ways of ensuring sustain-
able water management.
Figure 4. Coastal water quality in September 2004. Values of electrical conductivity from uncon-
fined superficial aquifer and chloride concentration from confined aquifers. In the north of
Roussillon, the Salanque Plain coastal area is more vulnerable to a decrease in water quality because
of the presence of the Salses-Leucate lagoon, high hydraulic conductivity and a large number of wells
including defective boreholes. In the southern Roussillon, water quality decreases only in the uncon-
fined superficial aquifer in a few places close to the sea (campsites with superficial boreholes).
4 IDENTIFYING WATER MANAGEMENT SCENARIOS THROUGH
CONSULTATIONS WITH STAKEHOLDERS
Because stakeholders have different technical backgrounds, information levels, interests at
stake and strategic objectives, they have diverging opinions on the strategy that should be
selected for ensuring a sustainable management of the Pliocene coastal aquifer. Taking all
these viewpoints into consideration and debating them publicly is a prerequisite for the
acceptability of the final decision and successful implementation of the selected policy
option. One option for eliciting the diversity of stakeholders viewpoints consists of con-
ducting a stakeholder analysis (Wellard, 1997) using semi-directive individual interviews.
These interviews are preceded by a questionnaire addressing the following issues with (i)
current water management practices (opinions, concerns, etc.); (ii) perception of decrease
in water quality linked to saltwater intrusion (origin, location, interaction between ground-
water and surface water, etc.); (iii) perception of groundwater use by different economic
sectors (households, municipalities, agriculture and tourism); (iv) existing and potential
water use conflicts; perception of the need for reinforcing regulations and economic tools
(tax system, quotas, restrictive measures for drilling new boreholes, fines, etc.). This sec-
tion presents the result of such a consultation conducted in the Roussillon case study area,
where 35 representatives from public territorial bodies (2), municipalities (9), drinking
water utilities (7), farmers organisations (4), private companies (2), regional research
departments and local experts in hydrogeology (7) and government agencies (4) were
interviewed. A qualitative summary of the findings is presented below.
4.1 Stakeholders perception of the risk of seawater intrusion
The results of this consultation highlight the fact that stakeholders do not share the same view
of the risk of seawater intrusion, its significance in the current situation, its origin and its pos-
sible consequences in the future. Overall, stakeholders views of the aquifer correspond to the
Figure 5. Piezometric level and chloride concentration as a function of time. All these observations
points are located in the Pliocene aquifers near the sea.
traditional conceptual model presented in the previous sections (Fig. 2). Although all of the
stakeholders concerned acknowledge that the water table has been declining for several
decades, some of them do not perceive this as a serious threat. Some of them argue that the
drop in the water table is not likely to generate seawater intrusion, since there could be an
impervious layer between the sea and the Pliocene aquifers preventing seawater inflow.
Similarly, most of the stakeholders are aware that high chloride concentrations have been
found in several boreholes along the coast, but their opinions on the origin of the contamina-
tion differ. For some stakeholders, the presence of chloride confirms that intrusion actually
takes place, whereas others assert that the chlorides found stem from the brackish water con-
tained in the superficial Quaternary deposits through abandoned boreholes. Given the lack of
scientific evidence available, the construction of a common perception of the aquifer does not
seem possible. As a result, stakeholders are likely to remain entrenched in their beliefs and
viewpoints and no real debate can take place as long as these key technical questions remain
unsolved by scientific and technical experts and passed on to the public.
4.2 Alternative water management strategies advocated
Since they do not agree on the nature and the intensity of the problem, the groundwater
management strategies they advocate significantly differ. A first group of players has
called for an immediate and drastic reduction in groundwater abstraction. Others are
opposed to this decision and simply recommend that the monitoring of chloride concentra-
tions is upgraded and the control of existing boreholes reinforced, both in terms of moni-
toring water abstraction and the compliance of boreholes with construction quality
standards (to reduce the risk of contamination of deep layers with superficial contami-
nants). Similarly, although all stakeholders agree that the quantitative pressure exerted on
the aquifer will continue to increase over the coming years, mainly due to sustained popu-
lation growth, they disagree on the policy that should be developed at a regional level to
meet the mounting water demand. The following three main strategies are advocated,
reflecting the diversity of interests at stake.
The first strategy is based on the assumption that the satisfaction of increasing drinking
water supply should be considered as a priority objective, overruling all other objectives
agriculture development in particular. The Pliocene aquifer should only be used for drink-
ing water purposes and its use for irrigation in agriculture restricted. This would require
developing additional surface water supply for agriculture use: three reservoirs (Caramany,
Vina and Villeneuve de la Raho) could be used as a substitute for groundwater. Inter-basin
transfer has also been suggested as a possible solution that could be implemented, with the
construction of an aqueduct supplying raw water taken from the Rhne River. The success
of this strategy would, however, require that public authorities are able to identify all exist-
ing boreholes (most of them are not officially registered), that appropriate regulations are
developed to allow the closure of existing wells (and possibly financial compensation allo-
cated) and that local politicians are willing to support this decision.
A second strategy would consist of preserving all existing agricultural and drinking
water supply boreholes and allocating them water use quotas corresponding to their cur-
rent exploitation level (or slightly below if hydrogeological studies confirm that the total
abstraction has to be reduced). New drinking water demand would be satisfied through the
development of non-conventional resources (desalination) or inter-basin transfers (impor-
tation of water from the nearby Corbires karstic aquifer, an aqueduct bringing water from
the Rhne River, etc.). The successful implementation of this strategy would require car-
rying out a census of existing boreholes, registering them, installing meters and monitor-
ing the implementation of the quota system.
An alternative to the two previous strategies would consist in implementing water
demand management actions targeting both the agricultural and drinking water sectors.
Irrigation efficiency could significantly be increased for a large number of farms who still
use basin or furrow irrigation techniques (Salanque region). Significant water savings
could also be achieved in the drinking water sector through various actions, such as reduc-
ing leakages in mains distribution systems, households and public services, installing
water saving devices such as double button toilet flushes etc., providing economic incen-
tives for reducing per capita water use in households through increased water pricing,
reducing irrigation of public gardens, developing waste water recycling systems for irri-
gating greens on golf courses, public gardens, etc.
4.3 Socio- political implications
The implementation of any of the strategies described above would, however, be difficult due
to political and economic constraints. For instance, a reduction in groundwater use by the
farming sector is a very sensitive political issue; farmers are strongly opposed to any restric-
tions in their access to the aquifer. They have only accepted, after long negotiations, that a cen-
sus of agricultural wells is carried out. Water meters remain very rare, even though this has
been required by law since 1997 and they are generally opposed to the idea of substituting
groundwater with surface water resources, since this alternative would probably be more
costly, its supply less flexible and provide water of lower quality. Many stakeholders are also
opposed to the idea that the drinking water supply should be considered as a priority as long
as no water saving efforts have been made in this sector. Between 25 and 50% of losses occur
in drinking water networks and huge volumes of water pumped from the Pliocene aquifers are
used for watering lawns or golf courses and filling swimming pools. Some stakeholders also
point out that many households have drilled private wells to access the Pliocene aquifers
(Perpignan region) and that these should be regulated as a priority. Building a political consen-
sus will require a significant negotiation effort and the water management policy that may be
adopted by all the parties will probably be a compromise of the three options described above.
5 CONCLUSION
Based on a French case study, this paper shows that assessing the risk of sea water intru-
sion in complex multi-layer systems is technically and sociologically complex. It also
shows that this complexity favours the emergence of different and sometimes opposing
viewpoints among stakeholders, who, because they do not share the same diagnosis of the
problem, are not likely to agree on the groundwater management solutions to implement.
Major sources of uncertainties concerning the intensity of the risk of sea water intrusion
are highlighted and a method is proposed for eliminating them. This method consists of
conducting a high-resolution geological characterisation of sedimentary prisms resulting
in a precise geometry of the aquifers and aquicludes. This knowledge establishes the rela-
tionships existing between the different aquifers and their possible connection to the sea.
It contributes to establishing the origin of salinity of contaminated boreholes or aquifers in
such a complex multi-layered system. Indeed, it is possible to determine whether direct or
indirect (defective boreholes) connections with the sea exist. Understanding of the geolog-
ical framework, hydrogeological field observation measurements and modelling allow the
degree of vulnerability to seawater intrusion to be determined.
The results of the stakeholder consultation presented in this study also highlight that,
whilst constructing a common understanding of the functioning of the aquifer is a prereq-
uisite to collective action, it is far from being the end of the process. Differences in stake-
holders viewpoints also reflect diverging economic interests. In the Roussillon as in many
other places in the world, proponents of demand based management options are opposed
to stakeholders calling for the mobilisation of additional resources, either through dam or
desalination plant construction (Giordana and Montginoul, 2006).
ACKNOWLEDGEMENTS
This study forms part of Bertrand Aunays and Cdric Duvails PhD work at the University
of Montpellier II (France), funded by the regional government of Languedoc-Roussillon
(CPER Syscolag Program) and by the BRGM in the framework of the COM-
PLEXAQUI project (EAU15). This work also benefited from financial support from the
French Ministry for Ecology and Sustainable Development.
REFERENCES
Accord Cadre pour la dfinition dun programme global de protection et de gestion concerte des
ressources en eau de lensemble aquifre multicouche plio-quaternaire de la plaine du Roussillon
Connaissance des eaux souterraines de de la plaine du Roussillon (2003) Water management
institutions of Roussillon Basin report. Regional council of PyrnesOrientales (CG66),
Perpignan, France, 47 p.
Aguilera-Klink F, Perez-Moriana E and Sanchez-Garcia J (2000) The social construction of
scarcity. The case of water in Tenerife (Canary Islands). Ecological Economics, 34, 2, 233245.
Aunay B (forthcoming) Importance of sequence stratigraphy knowledge for hydrogeological
modeling accuracy applied to coastal water resources management. University of Montpellier II,
PhD dissertation, (in French) University of Montpellier II.
Auroux F, Noyer ML, Marchal JP and Martin JC (1992) Management mathematical model of the
Roussillon. -in French- BRGM, Montpellier, France, R34 981 LRO 4S 92. 31 p.
Clauzon G, Aguilar J-P and Michaux J (1987) Pliocene Roussillon basin (PyrnesOrientales,
France): geodynamical evolution of a Mediterranean ria linked to the Messinian salinity crisis. -
in French- C. R. Acad. Sc. Paris 304 II,11, 585590.
Cross TA, Baker MR, Chapin MA, Clark MS, Garner MH, Hason MS, Lessenger MA, Little LD, Mc
Donough KJ, Sonnenfeld MD, Valasek DW, Williams MR and Witter DN (1993) Application of
High-Resolution Sequence Stratigraphy to Reservoir Analysis. In Eschard R and Doligez B eds,
Subsurface Reservoir Characterization from Outcrop Observations. Technip, Paris: 1133.
Duvail C, Gorini C, Lofi J, Le Strat P, Clauzon G and Dos Reis AT (2005) Correlation between
onshore and offshore Pliocene-Quaternary systems tracts below the Roussillon Basin (eastern
Pyrenees, France). Marine and Petroleum Geology 22, 67, 747756.
Giordana G, and Montginoul M (in press, corrected proof) Policy instruments to fight against sea-
water intrusion in coastal aquifers: an overview. Life and Environment.
Hardin G (1968) The Tragedy of the Commons. Science 162, 12431248.
Homewood P, Mauraud P and Lafont F (2002) Best practice in Sequence Stratigraphy for
Explorationists and Reservoir Engineers. Vol. Mem. 25. Total ed. Pau. 81p.
Marchal J-P, Plegat R, Benech C, Biscaldi R, Bourgeois M, Juncy G, Ramil F and Sauvel C (1976)
Roussillon Plain. -in French- Hydrogeological map of France, BRGM, Orlans.
Marchal J-P and Chery L (1995) Marine contamination of the Roussillon Plio-quaternary aquifer
(Pyrnes-Orientales) -in french- BRGM report, BRGM R38604, Orlans, France 85 p.
Nixon S, Trent Z, Marcuello C and Lallana C (2003) Europes water: an indicator-based assess-
ment European Environment Agency report, 1/2003.
Plan-Bleu (2005) A sustainable future for the Mediterranean. The Blue Plans environment and
development outlook Benoit G and Comeau A (eds), Earthscan, London Streling VA. 432 p.
Rinaudo J-D and Garin P (2005) The benefits of combining lay and expert knowledge for water
management planning at the watershed level. Water Policy, 7, 279293.
Wellard K (1997) Stakeholder methodologies in natural resource management: a review of princi-
ples, contexts, experiences and opportunities Agricultural Systems 55, 2, 173193.
CHAPTER 35
Cost-Benefit analysis of the alternative scenarios for
reducing diffuse nitrogen pollution in the Seine River basin
Christophe Viavattene
1
, Jean-Marie Monget
2
, Emmanuel Ledoux
3
&
Bernard Barraqu
4
1
ENPC, LATTS, Marne-la-Valle
2
ENSMP, Centre de Gosciences, Rue Claude Daunesse BP Sophia Antipolis
3
ENSMP, Centre de Gosciences, Fontainebleau
4
ENPC, LATTS, Marne-la-Valle
ABSTRACT: The MODSUR-STICS-NEWSAM model simulates the distribution of nitrate in
groundwater over three aquifers of the Seine River basin. Used in a prospective evaluation, the future
trend of water quality has been assessed under two scenarios, Current status quo and Agro-
environmental measures. A cost-benefit analysis of the results is used in order to compare each scenario.
The farmers costs are determined from direct measures costs and from the crop yield variations
computed by the model. The evaluation of benefits is limited to drinking water for domestic con-
sumers. The corresponding damage is computed for the whole basin by applying the systematic
curative treatment of water wells according to Rauchers method.
1 INTRODUCTION
The French water laws of 1964 and 1992 provide the basis for an integrated nationwide
water management system decentralized among six hydrographic districts. Adopted on
December 22, 2000, the European Water Framework Directive (WFD) does not funda-
mentally revise this type of basin organization but it brings a new environmental emphasis
at the scale of smaller hydrographic entities called water bodies. Based on the DPSIR
(Drive-Pressure-State-Impact-Response) framework (Karakos et al., 2003), member states
need to characterize all river basins by assessing the different human activities and their
present and future effects on the water system (Driver and Pressure). This requires the phys-
ical, chemical and biological characteristics of the resource at the scale of the water body
(State) to be identified and also the effect on environmental and societal issues (Impact).
In 2009, the next step will mark the adoption of a river basin management plan including a
programme of measures to reach good chemical and biological water status (Response).
Among the objectives for the improvement of water quality status is the proper reduction
of nitrate in the groundwater bodies in the Seine River basin where nitrate concentrations
in groundwater have risen constantly for several decades. The contamination source is dif-
fuse pollution generated by widely established farming practices throughout the watershed.
Agro-environmental measures have been promoted in the area for several years in order to
reduce the nitrogen load from agriculture. However, due to the poor uptake of these meas-
ures by the farmers and the extremely long response time of the nitrate concentration in the
groundwater, the capacity of these measures in reaching a good water status by 2015 can
be questioned. Use of modelling is essential in order to assess the future state of ground-
water resources given the complexity of the overall system. The MODSUR-STICS-
NEWSAM agro-hydrology numerical model was used in order to simulate the distribution
of nitrate in groundwater over the three main aquifers of the Seine River basin. Proper
assessment of Driving, Pressures and States factors, leads to a socio-economic analysis
following WFD recommendations.
In the Seine River basin, the main impact of groundwater nitrate contamination con-
cerns the domestic water production. The MODSUR-STICS-NEWSAM spatially distrib-
uted model has been used to simulate the distribution of nitrate in groundwater over three
aquifers of the Seine River basin. Two different methods have followed in order to assess the
damage of diffuse pollution to potable groundwater under two scenarios: Current status quo
and Agro-environmental measures. In order to lead to a cost-benefits analysis, the cost
of preventive measures was also further assessed.
2 THE TRENDS IN GROUNDWATER NITRATE CONCENTRATIONS
IN THE SEINE BASIN
The French Ministry of Environment created a national database for groundwater quality
monitoring called the Observatoire National de la Qualit des Eaux Souterraines in
1984 (ONQES). This database has been maintained by the Service Gologique National
(National Geological Survey) inside the Bureau de Recherches Gologiques et Minires
(BRGM). It is a composite of different data sources which were set up either locally, region-
ally or at the scale of the whole river basins (Fig. 1). It is essentially recording quality
measurements made from fresh groundwater wells. Management basin organizations such
the Agence de lEau Seine-Normandie (AESN) have followed the same scheme for their
own surveillance network database.
Over the geographic area covered by the Seine River basin, 6500 water boreholes are
stored in the ONQES database for a surveying period ranging from 1972 to 1995. But,
interruptions in the observation programme and the relative scarcity of sampling after 1990,
limit reliable nitrate concentration statistics for the whole basin to the 19751988 period
over a subset of 3297 homogeneous time-sequences for sampled pumping wells. To complete
the study up to the year 2000, another database, SISE-Eaux, from the Direction Rgionale
des Affaires Sanitaires et Sociales (Departement of Social and Sanitary Services) was used.
However, the gradual decrease in the total number of boreholes monitored over time since
the 1990s introduces a bias, as the drinking water wells that are monitored are not neces-
sarily representative of the regional quality of the aquifer.
The computed median of the nitrate concentration distribution (Fig. 2) displays a strik-
ingly linear increase for all three aquifers. This sustained increase rate is estimated at
0.6mgNO
3
/l per year. The year 2000 median value is about 35mgNO
3
/l for Chalk and
Oligocene aquifers and about 25mgNO
3
/l for the Eocene aquifer. Most of the observed
data are under the threshold value of 50mg NO
3
/l defining the good status of groundwa-
ter bodies for the nitrate parameter, but they could probably reach this limit in a few
decades if present loading conditions are sustained. Thus the French water agency of the
Cost-benefit analysis of the alternative scenarios for reducing diffuse nitrogen pollution 473
Figure 1. Geographic distribution of the 6500 groundwater boreholes available over the Seine
River basin in the ONQES national database.
Figure 2. Observed evolution of nitrate concentration in water wells over the three main aquifers
of the Seine River basin.
Seine River basin estimates that 39 groundwater bodies among the 46 covering the district
are at risk of not meeting the good chemical status for both the nitrate and pesticide
parameters in 2015 (AESN, 2004).
3 BASIN WIDE GROUNDWATER NITRATE CONCENTRATION MODELLING
In the framework of the PIREN-Seine research program, the MODSUR-STICS-NEWSAM
model has been developed to simulate the impact on the groundwater resources of
diffuse nitrogen pollution originating from agricultural practices extending over the
whole Seine River basin. Following the WFD guidelines and in concordance with the AESN
basin agency recommendations, the evolution of the overall quality of groundwater
water resources has been assessed for an agriculture and land use baseline scenario up
to 2015.
The modelling (Rousset et al., 2004) is composed of a cascade of coupled distributed
models: MODSUR for hydrology, STICS for water and nitrogen balance of the soil-crop
system (Schnebelen, 2004) and NEWSAM (Ledoux et al., 1990) for groundwater flow and
transport (Fig. 3). This cascade allows for the simulation of the transfer of nitrate from the
Figure 3. Functional diagram of the MODSUR- STICS-NEWSAM modelling system.
upper soil down to the three main top aquifers system of the Seine basin: Oligocene, Eocene
and Chalk. Daily agronomic and climatic data have been used as input in order to compute
the estimated yearly nitrate concentration for each water system on the basis of a kilometre
uniform grid. For the definition of the spatial variability of agronomic input, the watershed
has been divided into 12 areas known for their homogeneity in farm practices. This seg-
mentation results from the overall aggregation of 150 initial regional units called: Petites
Rgions Agricoles or PRA (Mignolet et al., 2004). The simulation is based on the repetition
over time of the agriculture practices observed during the period 19902000 and the cli-
mate conditions between 1970 and 1990. The 19902000 climate data have not been used
as it is affected by climate change (GIEC, 2001). The resulting agriculture baseline sce-
nario can, therefore, be considered as the result of a Business as usual policy originating
from the 19902000 period.
Model calibration has been undertaken using the annual evolution of the median of the
statistical distribution of nitrate from the existing measurements of nitrogen concentration
for the period 19701990 in more than 3000 water wells. The observed and computed
median has been compared by a least-squares fitting method. Excellent results are obtained,
with an error variance below 1mgNO
3
/l for each aquifer. The model is, therefore, a good
estimator of the overall evolution of the nitrate concentration.
Any predictive exercise must be validated and the uncertainties examined. In order to
estimate the local reliability of the models results, deviation from the yearly mean value
between observed and computed concentrations have been calculated for each borehole.
The resulting statistical distribution of the deviations for the Oligocene aquifer (Fig. 4) can
be modelled by a Gaussian distribution with a 28mgNO
3
/l deviation whatever the year.
The same results have been observed for the two other aquifers.
One may postulate that the Gaussian distribution of modelling discrepancies will be
stable over future times. It can be used as an uncertainty model for the nitrate levels com-
puted for any model cell under the hypothesis of spatial uniformity. This probability distri-
bution can also be used in order to compute the risk (or probability) of exceeding the
50 and 100mg/l thresholds over typical milestones for any particular model cell.
Simulation results are managed using a standard GIS (Geographic Information System),
Figure 4. Observed deviations between measured nitrate concentrations in water wells and
MODSUR-STICS-NEWSAM simulation for the Oligocene aquifer.
thus providing spatial aggregation by either WFD compliant groundwater bodies or
municipal boundaries.
Within the European WFD guidelines, baseline scenarios are meant as being Present
status quo (reference) hypotheses used in a forward extrapolation of present pressures on
the environment as well as management practices. Their real purpose is to conduct a risk
analysis of non-compliance and evaluate the probability of economic alternatives in order
to reach the good water status for particular water bodies. In this context, the production
of concentration maps at given time horizons (here, 2015) is seldom the appropriate end
product for decision makers. In risk management studies it is often preferable to manipu-
late entities to which probabilities can be attached and eventually combined with monetary
concepts such the Net Present Value (NPV) (Pollio, 1999) or the various economic values
attached to the environment (Pearson, 2000). In order to fulfil these demands, uncertainty
probability modelling enables threshold probability maps to be prepared which display the
probability that a certain regulatory index level be exceeded. Fig. 5 provides such an exam-
ple, where MODSUR-STICS-NEWSAM results at 2015 horizon have been used in order
to evaluate the probability of exceeding a 50mgNO
3
/l regulatory level in the water wells
drawing water out of the Oligocene of the Seine River basin.
4 EVALUATION OF BASIN WIDE NITRATES REDUCTION MEASURES
AND COST IMPLICATIONS FOR THE FARMER
The WFD requires the achievement of good groundwater status and to that end provides
for the monitoring of groundwater bodies as well as measures to protect and restore
groundwater. But is it economically viable for society to pay to reduce groundwater nitrate
concentrations by investing in programmes that result in the increased adoption of best
management practices (Yadav, 1998)? In order to answer this question, the MODSUR-
STICS-NEWSAM model has been used in order to compare different policy alternatives
and their respective costs.
Figure 5. Municipality based probability map of exceeding the level of 50mgNO3/l in the ground-
water wells for the Oligocene aquifer of the Seine River basin for the year 2015.
In France, groundwater protection measures are mainly based on the control of fertilisers
input and the use of nitrogen fixing crops (CIPAN in French). New agricultural scenarios
including these two practices have been simulated with the STICS model as an input to the
NEWSAM hydrological model in order to evaluate their effect at a regional level. The
Business as usual hypothesis (scenario A) has been contrasted with a change in agricul-
tural practices applied after 2005 which is: a reduction of 20% of N-input and use of
CIPAN (scenario B: Agro-environmental measures). The third hypothesis (scenario C)
assesses how the natural system would react if all human impact would cease after 2005.
It is also assessing the nitrate discharge capacity of the aquifer and the ideal chemical
status that could be reached.
The following graphs (Fig. 6) show the evolution of the calculated median of nitrate
concentrations modelled by NEWSAM under the different scenarios for each of the main
aquifers of the Seine River basin. Over a short-time span, effects of preventive measures
show little impact on the nitrate concentration evolution at basin scale, especially for deeper
parts of the Chalk aquifer for which no trend reversal can be observed but merely a stabi-
lization of the present level.
In order to compare the different scenarios, the cost of the agro-environmental measures
needs to be assessed in order to compare the cost on farmers to change practices.
With preventive measures, the decrease in crop yield implies a loss of income for agri-
cultural activities. AESN (Agence de lEau Seine-Normandie) estimated at around 60
/ha the average cost of preventive actions (Guillemin and Roux, 1992). Costs of CIPAN
have been evaluated at around 75 /ha by one source (Martin, 1996) or between 97 and
188 /ha by another (Mollard, 1997). By reducing the amount of fertiliser needed for the
next annual crop, CIPAN could also induce savings of 2 /ha for high-yield cereals. In this
study, the farmers added costs are evaluated from direct measures costs at 75 /ha for
CIPAN and from the crops yield value variation computed by the STICS agronomy model.
Depending upon the type of crops and soils, most of the computed costs lay between 50
and 200 /ha. The total amount for a systematic enforcement of preventive measures over
the whole basin is estimated at 645 millions , the main part coming from loss of income
(527 millions ).
5 ASSESSMENT OF THE BENEFITS FOR THE DRINKING WATER DOMESTIC
CONSUMER USING R.L. RAUCHER METHOD
A direct approach to the evaluation of the economic value of groundwater quality is to
consider only the costs that users have to bear if groundwater quality deteriorates. The
underlying concept is that these costs would no longer have to be paid if groundwater qual-
ity could be restored. In this sense, the benefits of groundwater protection take the form of
avoided damage costs (Grlach, 2003).
Groundwater contamination generates different types of damage which can be evaluated
as the remediation costs (Rinaudo et al., 2005) met by water users who have to deal with
ecological damage or even the loss of recreational value, when groundwater contamination
has an impact on surface ecosystems (rivers, wetland, coast). This study focuses on the
estimation of the damage associated with the domestic water supply. In this case, the esti-
mated benefits can be interpreted as a lower bound estimate of the value of groundwater
protection. This approach has already been applied in the case of contamination of an
Figure 6. Evolution of the median of MODSUR-STICS-NEWSAM simulated nitrate concentra-
tions in the three main aquifers of the Seine River Basin for various scenarios.
aquifer by toxic materials (Raucher, 1983). The author explained that the expected social
benefits of protection are defined by the change in the expected damage E(D) associated
with the contamination as being expressed by:
E(D) p[q*Cr (1q)*Cu]
Where:
p : probability that contamination will occur (0 p 1)
q : probability that contamination would be detected before polluted water was
used (0 q 1)
Cr : expense of the most economically efficient response to the contamination
incident (Cr 0)
Cu : cost incurred if contaminated water were used in the same manner as prior
to the incident
The concept of Raucher needs to be adapted (Barraqu et al., 2005) in the case of the con-
tamination of wells by nitrate for domestic water supply. The probability (q) can be con-
sidered as equal to 1, so that the expected damage is expressed by:
E(D) p*Cr
The p value is computed from model results after aggregation at municipality level. It rep-
resents the risk of exceeding the 50mg/l maximum admissible nitrate concentration. In
order to compute the expected damage for each local groundwater managing unit (also
called Unit de Gestion et dExploitation or UGE, in France) the assumption was made
that, for each borehole logged in the ONQES database, a treatment would be systemati-
cally set up in case of nitrate concentration exceeding 50mg/l. For each borehole the cor-
responding Cr term (see equation 1, above) can thus be determined as:
Cr Cost of water treatment * number of water consumers
* water consumption per habitant
The actual determination of Cr is based on the following values: a cost of raw water treat-
ment of 0.22 per m
3
(AESN-DEPEEP-PEP, 2002); a water consumption per inhabitant
of 150l per person per day and the likely number of water consumers in 2015 based on an
extrapolation of the 1998-population census (population of the each municipalities linked
to the particular UGE) using a population growth rate (source: INSEE census) calcu-
lated at bassin de vie scale for the period 19901999 (Source: INSEE). The bassin de
vie is a spatial cluster of municipalities representing the smallest territory where people
have access to the same social equipments and employment facilities. The Seine River
basin counts 341 bassin de vie regrouping 8118 individual municipalities. This approach
does not, however, take in account the fact that treated water can be blended with raw water
from a different well. The loss of water in the network is also discarded.
Expected damage (Fig. 7) has been computed following the previous hypotheses for
each of 1684 UGE of the Seine River basin (6.5 Millions inhabitants) situated outside
the Paris urban area (Viavattene, 2006):
The resulting benefits of the preventive measures (Table 1) increase rapidly from the
starting date of 2005 and reach a level of 714000 by 2015. However, when compared
with the total costs borne by the farmers, these benefits remain low. This type of result
could be criticized by arguing that farmers costs have been overestimated because preven-
tive measures have been considered to be indiscriminately applied over the whole river
basin area. It is obvious that a more locally adapted application of such measures could lead
to a similar efficiency at lower cost. On the other hand, the benefits are probably undervalued,
as the whole range of damage caused by nitrate pollution was not taken into account.
Moreover the global approach used here does not realistically reflect the actual behaviour
of the UGEs and the full range of individual costs as damage was only considered through
the probabilistic expected value.
6 CONCLUSION
On the Seine River Basin the MODSUR-STICS-NEWSAM has been shown to be a good
estimator of the evolution of nitrate concentrations and was used to assess future trends in
groundwater nitrate contamination under three contrasting scenarios. The local reliability
of the results is limited, probably linked to an oversimplification of agricultural input data.
New simulations with inputs defined at the smaller scale (Petites Rgions Agricoles or
PRA) are presently under way and reflect this hypothesis. Instead of using direct point
concentrations provided by the model it is more efficient to use an index of the pollution
impact based on the probability of exceeding a defined nitrate concentration level. This
index is computed on the base of the statistical distribution of model deviations.
On the base of this index, the economic impact of the nitrate on domestic water production
has been assessed by two methods. The Raucher method allows for the calculation of the
expected damage for the whole basin which is found to be equal to 19 millions in 2005,
Figure 7. Evolution of the expected damage in the Seine River basin for Business as usual and
Agro-environmental measures scenarios.
Table 1. Evolution of the benefits for the period 20052015.
Year 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Benefits 5 27 62 163 229 291 345 422 518 617 714
(Thousand e)
reaching 27 millions in 2050 under a Business as usual scenario. With preventives
measures the expected damage is reduced but the resulting benefits on a short-time hori-
zon are insignificant compared to the costs incurred by the farmer. Although this is only a
first step approach to the economic evaluation of protection measures, the results obtained
appear promising. Improvements on the description of the physical and technical environ-
ment need, however, to be pursued.
REFERENCES
Agence de lEau Seine-Normandie, AESN (2004) Etat des lieux: bassin Seine et cours deau ctiers
normands, 172 pp
Agence de lEau Seine-Normandie, AESN-DEPEE-PEP (2002) Analyse conomique de la rcupration
des cots: premiers chiffres, premires hypothses, mise en discussion. Agence de lEau Seine
Normandie Direction des Etudes, de la Prospective et de lEvaluation environnementales Ple
Evaluation et Prospective, 51 pp
Barraqu B, Ledoux E, Monget J-M, Viavattene C (2005) Assessing the impact of future nitrogen
diffused pollution scenarios on the ground water resources of the Seine River basin. Second
International Workshop on Implementing Economic Analysis in the Water Framework Directive,
Paris, 1718 Feb. 2005, 18 pp
Groupe dExperts Intergouvernemental sur lEvolution du Climat, GIEC (2001) Bilan 2001 des
changements climatiques: les lments scientifiques. Rapport du groupe de travail I du Groupe
dExperts Intergouvernemental sur lEvolution du Climat, 97 pp
Grlach B, Interwies E (2003) Economic Assessment of Groundwater Protection: A survey of the
Literature. Ecologic, Berlin, 64 pp
Guillemin C, Roux JC (1992) Pollutions des eaux souterraines en France : bilan des connaissances,
impacts et moyens de prvention. Manuels et Mthodes n23, BRGM, 254 pp
Karakos A, Skoulikaris X, Monget J-M (2003) The broadcasting on Internet of water DPSIR indicators.
Experiment on the Nestos Delta, Greece. Global Nest: The Int. J., 5, 2, 8187
Ledoux E, Sauvagnac S, Rivera A (1990) A Compatible Single-Phase/Two-Phase Numerical Model: 1.
Modeling the Transient Salt-Water/Fresh-Water Interface Motion. Ground Water, 28, 1, 79
Martin Y (1996) Rapport sur la gestion durable des eaux souterraines. Paris, Ministre de lIndustrie,
de la Poste et des Tlcommunications, Conseil gnral des Mines, 29 pp
Mignolet C, Schott C, Benot M (2004) Spatial dynamics of agricultural practices on a basin territory:
a retrospective study to implement models simulating nitrate flow. The case of the Seine basin.
Agronomie 24, pp 219236
Mollard A (1997) Solutions curatives ou prventives la pollution diffuse de leau? Une approche
de la soutenabilit de la ressource en eau. Natures Sciences et Socit, 5, 3, 521
Pearson CS (2000) Economics and the global environment. Cambridge University Press, 583 pp
Pollio G (1999) International project analysis and financing, Mac Millan Press, 235 pp
Raucher RL (1983) A conceptual framework for measuring the benefits of groundwater protection.
Water Resources Research, 19, 2, 320326
Rousset F, Habets F, Gomez E, Le Moigne P, Morel S, Noilhan J, Ledoux, E (2004) Hydrometeorolog-
ical modeling of the Seine basin using the SAFRAN-ISBA-MODCOU system. Journal of
Geophysical Research, 109 (D14105)
Schnebelen N, Nicoullaud B, Hocine Bourennanec H, Couturierc A, Verbequed B, Revalierd C,
Bruande A, Ledoux E (2004) The STICS model to predict nitrate leaching following agricultural
practices. Agronomie, 24, 423435
Viavattene C (2006) Exploitation socio-conomique de la modlisation souterraine du transfert des
nitrates lchelle du basin de la Seine. PhD thesis (in French), Ecole Nationale Suprieure des
Mines de Paris, France
Yadav SN, Wall DB (1998) Benefit-cost analysis of best management practices implemented to con-
trol nitrate contamination of groundwater. Water Resources Research, 34, 3, 497504
CHAPTER 36
Integrating economic and groundwater models for
developing long-term nitrate concentration scenarios in
a large aquifer
Nina Graveline
1
, Jean-Daniel Rinaudo
1
, Volker Segger
2
, Hendrik
Lambrecht
3
, Markus Casper
3
, Philippe Elsass
1
, Jost Grimm-Strele
3
,
Thomas Gudera
3
, Remi Koller
4
, & Paul Van Dijk
4
1
BRGM, 1039, rue de Pinville, 34000 Montpellier, France
2
Landesanstalt zur Entwicklung der Landwirtschaft-Schwabische Gmnd, Germany
3
Landesanstalt fr Umwelt, Messungen und Naturschurz Baden-Wrttemberg, Karlsruhe,
Germany
4
Association pour la Relance Agricole en Alsace, Schiltigheim, France
ABSTRACT: This paper presents a multi-disciplinary approach aimed at assessing groundwater
nitrate in a transboundary aquifer lying between Germany and France. Developed as part of the
MONIT EU-InterReg project, the methodology consists of integrating different models representing
economical, physical, chemical and hydrogeological processes as well as their interactions. The aim
of this paper is to present the work of the economic working group who developed a model aimed at
representing the evolution of agricultural land by modelling the farm production choices and their
practices. Simulations are run at a 2015 time horizon to assess the impact of three contrasting scen-
arios (baseline scenario, and two contrasted scenarios derived from the International Panel
Climate Change work). The scenarios include various assumptions on several driving forces such as
the CAP reform, the risk of corn rootworm proliferation, energy prices, water policy, and the devel-
opment of biofuels. Results suggest that the evolution of cropping patterns will be significant
between now and 2015, and this will have consequences on both nitrate leaching and the implemen-
tation of groundwater protection measures. The long-term consequences of nitrate contamination of
groundwater are also simulated using the chain of models described in the paper.
1 INTRODUCTION
During the late 1980s, the concerns expressed by the public in Europe about steadily
increasing nitrate concentrations in drinking water resources triggered a policy debate on
agricultural non-point source pollution. This led to the Nitrate Directive (91/676/EEC) in
1991, followed by a first wave of action by Member States in the farming sector. A com-
munication of the European Commission (European Commission, 2002) which reviews
the measures implemented between 1991 and 2000 showed that these measures have not
been sufficient to reverse the trend and to achieve the targeted nitrate concentration of
50mg/l in all declared vulnerable areas. This statement particularly applies to ground-
water bodies, a number of which still show increasing nitrate concentration trends and
are increasingly concerned by diffuse pesticide pollution.
This debate has been revived with the publication of the Water Framework Directive
(2000/60/EC), which requires Member States to be able to maintain or restore good
chemical status for all water bodies by 2015. Specific pollution abatement measures have
to be defined and implemented by 2009 in order to reach nitrate concentrations lower than
50mg/l in all groundwater bodies (an objective recently clarified with the publication of
the daughter groundwater directive). Member States are allowed to extend the period to
attain good chemical status until 2027 at the latest under certain conditions. This delay
must be anticipated and notified to the European Commission in the River Basin
Management Plan which has to be published by 2009.
At the river basin district level (or sub-district level), where groundwater protection
measures have to be defined, a key issue that policy makers and water planners need to
investigate is the expected future evolution of groundwater quality. This requires under-
standing and modelling of the dynamics of groundwater quality. The approach presented
in this paper recognises that water quality evolution is not only determined by biophysical
drivers (such as climate, groundwater flows, etc.) but also by economic drivers, which con-
siderably influence the activities generating the pollution. This paper presents the result of
an economic study aimed at (i) identifying the driving forces likely to influence future
agriculture practices (e.g. crop choices, use of fertilisers), (ii) assessing this impact in quan-
titative terms using micro-economic farm models and (iii) estimating the impact of changes
in farm practices on nitrate contamination and groundwater quality. The study was con-
ducted as part of the trans-boundary MONIT InterReg project, which has as its main
objective the development of an integrated modelling platform for simulating future evo-
lution of nitrate concentration in the upper Rhine valley aquifer. This modelling platform,
which has been described elsewhere (see Casper et al., 2005; LUBW, 2006) comprises a
soil-plant model (simulation of nitrate transfer in the unsaturated zone), a nitrogen balance
model (simulating nitrogen infiltration in the aquifer), a hydrogeological model simulat-
ing water and nitrogen flows in the aquifer and an economic model (simulating farmers
decisions in terms of crop choices). The present paper focuses on the farm economic
model. It presents the results of two years of collaborative research between economists,
policy makers and stakeholders conducted as part of the Socio-economic Working
Group of the InterReg III MONIT project.
2 NITRATE GROUNDWATER POLLUTION IN THE UPPER RHINE VALLEY
The Upper Rhine valley aquifer extends over 4200km
2
and lies within both Germany and
France. With a reserve of approximately 45 billions m
3
of water, it is one of the largest
freshwater reserves in Europe. The water supply of more than three million inhabitants of
the Alsace (France), Baden (Germany) and Basel (Switzerland) regions directly depend on
this resource. Approximately 300 millions m
3
are extracted every year for drinking water,
45% being used in France, 35% in Germany and 20% in Switzerland. Groundwater also
fulfils 50% of industrial water needs (approximately 300 millions m
3
pumped every year).
Since the 1970s, the groundwater has increasingly been affected by diffuse nitrate and
pesticide pollution, mainly due to agriculture intensification. The nitrate pollution problem
is particularly acute on both sides of the Rhine. While the nitrate concentrations were lower
than 50mg/l in the entire aquifer in the early 1970s, 15% of the 1100 monitored points had
a nitrate concentration exceeding 50mg/l in 1997. The European guide value of 25mg/l
was exceeded in 36% of the monitored points (Rgion Alsace, 2000). Groundwater shows
high nitrate concentrations in areas where intensive crops (corn or vine) are cultivated. The
area given over to corn and vine is increasing whereas the area occupied by grassland
steadily declines. Between the two last agricultural censuses (1988 and 2000), the area
under vine has increased by 13%, the area under corn by 60% whereas area under wheat
and grassland decreased by respectively 25% and 13%. This trend is likely to be progres-
sively reversed after the implementation of the reform of the Common Agricultural Policy
(in 2006 in Germany and 2013 in France).
Nitrate contamination of groundwater causes significant economic damage at a regional
level. It has lead to the closure of an increasing number of drinking water wells, it con-
tributes to the decline of consumer trust in tap water and a rise in bottled water consump-
tion, and it creates technical and economic constraints for industry. In the Alsace region,
the total cost due to nitrate contamination between 1988 and 2002 has been estimated by
BRGM at 26 millions for drinking water utilities (investment), 160 millions for house-
holds (bottled water purchase, and the installation of filtering devices at home). The cost
borne by the industrial sector could not be assessed but the example of one large brewery,
which spent over 10 millions in ten years for the construction and operation of a nitrate
removal plant reveals the significance of the cost (Rinaudo et al., 2004). This trend is,
however, likely to be significantly altered in the future in response to changes in European
agriculture and environmental policy, agricultural market, energy prices and the natural
environment.
3 METHODOLOGY: INTEGRATION OF ECONOMICS IN
THE CHAIN OF MODELS
The integrated modelling platform comprises four main models (LUBW, 2006). The first
model (economic farm optimisation model) simulates crop choices and fertilizer use prac-
tices for all the farms in the area. The results of this model are used as input to a nitrogen
balance model (STOFFBILANZ), which calculates nitrate leaching in the unsaturated
zone. The parameters of this nitrogen balance model are adjusted using a soil-plant model
(STICS), which is developed and calibrated for different crops and types of soils. The
results of the nitrate balance model are then fed into a groundwater flow model, which
simulates groundwater flows and piezometric head as well as nitrogen transport within the
aquifer (combined use of MODFLOW and MT3D).
The economic optimisation models on which this paper focuses aim at simulating farm-
ers behaviour in terms of crop choice and cropping practice (e.g. fertiliser use, intercrop soil
management practices, etc.) for different economic conditions. Models assume that farmers
select the combination of crops which maximises their income under a set of technical, regu-
latory and economic constraints (Hazell and Norton, 1986). They simulate crop choices, input
consumption (fertiliser, labour, energy) and farm income for different input parameter values
(agricultural prices and subsidies, regulatory constraints, changes in the price of input such as
energy, fertiliser, labour, minimum set aside constraint, etc). In their current version, the
models incorporate constraints related to crop rotations, labour availability, production quotas
Developing long-term nitrate concentration scenarios in a large aquifer 485
(sugar beet and milk quotas), manure storage and management (for livestock oriented farms).
Gross margin values and technical coefficients are based on figures reported by farmers, cross
checked with standard values used by agricultural experts and published by professional agri-
cultural organisations. The economic linear programming models representing the production
choices of each of these representative farms are calibrated at the farm level by comparing the
simulated with the current cropping patterns. They are implemented using a mathematical
solver and a simulation engine developed using Visual Basic, which allows repeated simula-
tions for a range of input parameter values.
The diversity of farms in the region was taken into account, based on a farm typology
developed as part of the project (Graveline et al., 2005). The typology was based on a
farming system analysis conducted by the regional Agriculture Chamber (Chambre
dAgriculture dAlsace, 2003) and additional investigations and farm statistical classifica-
tion conducted by LEL in Germany as part of the project. Twelve main farm types, re-
presenting respectively 84% and 69% of the cultivable area in Alsace and Baden regions were
selected. For each type, a real farm was selected and a microeconomic model developed,
based on information collected through interviews. The models are calibrated for these 12
farms before being adapted to represent a statistical average farm for each group. The eco-
nomic models are then used in simulations and the results obtained at the farm level are
extrapolated to the entire farm type. Results are spatially distributed at the level of small
agricultural regions (13 in total) and the distribution at the grid cell level (used by the
nitrate balance model) is assumed homogeneous. The simulation results are scrutinized by
a group of experts ultimately leading to the validation of the models, after several rounds
of parameter adjustment and validation.
The group of experts is also used to identify future trends for economic parameters
determining farm production choices. Several scenarios are constructed. For each scenario,
the economic models are used to simulate (i) cropping patterns (area under each crop);
(ii) total gross margin; (iii) input use (in particular fertilizer, energy, water). Simulated
cropping patterns are then transmitted to the nitrate balance model and groundwater model
for assessing the environmental consequences of the scenarios.
4 DEVELOPMENT OF PROSPECTIVE SCENARIOS
The first step of the methodology used to develop the scenarios is based on the earlier
approach by Alcamo et al. (1996), and consists of identifying driving forces (or drivers)
likely to influence future production choices (cropping patterns) and farming practices.
Twenty-six driving forces were identified by the French and German expert group and
gathered into six categories:
(i) agricultural market and policy changes following the reform of the Common
Agricultural Policy (CAP), World Trade Organisation (WTO) negotiations, European
enlargement, the development of a demand for products of high quality, etc.;
(ii) other markets and economic policy evolution (land, labour and energy markets),
(iii) the evolution of environmental policies (farming practices, cross-compliance, second
pillar of the CAP, implementation of the Water Framework Directive, etc.);
(iv) the evolution of the natural environment, in particular climate change (impact on
crops and nitrate migration processes in soils) and proliferation of a corn parasite;
(v) technological change (selected varieties, simplified farming practices, GMOs);
(vi) the evolution of social demand for organic farming and quality labels;
(vii) the evolution of socio-economical characteristics of the agricultural population
(demography, concentration of farms, capital level etc.).
For each of the 26 drivers, experts assessed and described qualitatively the most likely
trend and uncertainty of each. Whereas some of the changes are considered as almost cer-
tain, others will not necessarily occur before 2015 (e.g. climate change). A prioritisation of
factors according to intensity of impact and uncertainty leads to the identification of six
key driving forces that will be taken into consideration in the models for the 2015 simula-
tions. The six driving forces are the following:
The 2003 CAP reform: the main feature of the reform consists in decoupling financial
subsidies from the nature and quantity of production, in order to make agriculture more
responsive to market changes. Decoupling consists of replacing the former system of
subsidies per hectare (or per animal capita) by a single payment per hectare (calculated
on past cropping patterns). France keeps a 25% coupling while Germany has opted for
a complete decoupling (100%). This reform is likely to induce major changes in crop-
ping patterns of the upper Rhine valley (MAAPAR, 2004; Deutscher Bauernverband,
2003).
The corn rootworm: in 2003, this parasite caused major damage to corn fields in the
south of the upper Rhine valley (Hardt and Sundgau). In a 10000 ha safety area, corn
has now been reduced by 70% due to both regulatory measures and self-limitation by
farmers. The propagation of the parasite around Paris in France, in Eastern Europe and
in the United States suggests that the parasite could spread in the upper Rhine valley,
possibly resulting in a drastic reduction of corn cropping patterns.
Energy price: if the oil price trend is confirmed, the profitability of energy-consuming
crops could decrease, in particular crops requiring important mechanical operations,
irrigation and drying. Fertilizer prices, which are strongly correlated to the price of
energy, could also rise. This could induce farmers to reduce yield objectives and nitro-
gen supply, and, as a consequence, reduce nitrate emissions.
The development of bioenergy crops: the price of energy could also lead to the develop-
ment of increased biofuel and bioenergy demand. Two types of industries could emerge:
the first one, in Baden only, consists of industrial production of biogas from corn; the
second consists in producing raw vegetable oil from rape, either at farm level (auto-con-
sumption) or at industrial scale. At farm level, farmers can produce their own fuel and
use oilcakes as fodder for animals. The development of these crops will largely depend
on economic policies (taxes and subsidies), legislation and the local development of the
necessary transformation industry.
Price and taxes for irrigation water: in Alsace, where the area under irrigation is signifi-
cant, largely because of more favourable taxes compared with Baden Wrttemberg
(regulations, tax exemption and Wasserpfennig in Baden). But this tax could be
implemented (debate on the water law) within the implementation of the Water
Framework Directive that stipulates that Member States have to put incentive tariffs on
water. This would lead to a reduction of the mean profitability of irrigated corn and veg-
etables and could favour the development of alternative crops.
European enlargement may imply an increase of seasonal labour costs in Germany,
where it is nowadays relatively low because of the possibility of hiring workers from
Eastern Europe (Poland mainly) with no regulatory minimum salary (which is not the
case in France where a minimum salary exists). The recent accession of these countries
to the European Union will certainly lead, until 2015, to a progressive increase in salary
levels. This will have repercussions on the seasonal labour market in Germany. Gross
margins of vegetable crops will in consequence be reduced, perhaps leading to a reduc-
tion in those areas, which are today severely contaminated by nitrates.
The six driving forces described above correspond to parameters in the economical models
that can be modified in order to carry out the simulations. Table 1 describes this corres-
pondence between driving forces and changeable parameters from models.
Given that different trend assumptions can be made for each of the six driving forces, the
total number of scenarios resulting from a combination exercise can be very significant. In
order to reduce the number of scenarios to be simulated, three contrasting scenarios have
been developed. Each scenario corresponds to a combination of assumptions which are
internally consistent. A baseline scenario, representing the most plausible evolution of the
economic, regulatory and natural environment, has been constructed. Two other scenarios
have been built on more extreme assumptions in order to cover a larger spectrum of possi-
bilities. These two scenarios have largely been inspired by the global emission scenarios
proposed by the Intergovernmental Panel on Climate Change (IPCC, 2000). The underlying
assumptions have been defined both on literature searches and by the expert group.
The baseline scenario assumes the following changes in agriculture driving forces. The
corn rootworm, which has been detected since 2003 in the area, extends over large areas,
forcing farmers to increase crop rotations (area under corn cannot exceed 50% of the cul-
tivable area). Energy prices (vegetable fuel-oil) are supposed to increase by 6% per year on
average (2015 prices are twice those of 2003) and it is assumed that no financial compen-
sation mechanism will be implemented by national governments. As a result of energy
Table 1. Correspondance between driving forces and parameters of economical models.
Driving forces Parameter of economical models modified Concerned countries
to simulate impact of driving forces
CAP reform Gross margins produced per area unity (/ha) France and Germany
Upkeep cost for non productive land (set aside) France and Germany
Corn rootworm Rotation constraints represented by a lower France and Germany
limit for area under corn (in %)
Energy price Farm gasoil price and other energy sources France and Germany
appearing in the technical itinerary (field
works, irrigation and drying)
Mineral fertilizer prices France and Germany
European Cost of seasonal labour force Germany
enlargement
Water price and taxes Price and taxes on consumed water France and Germany
Development of Possibility to produce raw vegetable oil France and Germany
biofuels and from rape and to produce agricultural
bioenergies fuel-oil
Possibility of selling the corn production Germany
(silage corn practices) to a biogas plant
price growth, the price of fertilizer will increase by 1.5% per year. Due to European enlarge-
ment, temporary labour force cost increases (66% in twelve years), reducing the prof-
itability of vine, fruit and vegetable crops (in particular in Germany where a foreign labour
force is significantly used for these crops). In France, it is assumed that the reform of the
Water Act establishes a new water abstraction tax of 0.025 per cubic meter. In Germany,
the tax called Wasserpfennig is maintained at its 2003 level (0.05 /m
3
). In France as in
Germany, farmers are allowed to produce bio-diesel (rape) on farm for their own use, but
not for sale. The biogas industry will develop in Germany only with the possibility of sell-
ing the product locally.
The second scenario, called A1 with reference to IPCC emission scenario A1, depicts
a more liberal future. With this vision, agriculture development aims at maximising com-
petitiveness in markets which tend to function without protectionist barriers and with a min-
imum of environmental constraints: taxes on water abstraction are suppressed in Germany
and not introduced in France. The liberalisation trend leads France to fully implement the
decoupling principle promoted by the Common Agriculture Reform, which results in a
drastic change in crop gross margins and crop profitability. Energy price increases are com-
pensated by a fiscal stabilisation mechanism and do not exceed respectively 40% and 68%
in Germany and France. Significant technical means are mobilised by government agencies
to fight against the corn rootworm (pesticides sprayed from helicopters). And the biofuel
industry develops, representing new market opportunities for farmers.
The third scenario, inspired by the B2 IPCC emission scenario, corresponds to a vision
of the future where agriculture evolves under the double pressure of increasing input
prices (energy, fertiliser) and more stringent environmental constraints. Water abstraction
taxes are established at the level of the baseline scenario. A tax on fertilisers is introduced
at the level of 0.15 /kg in France and 0,26 /kg in Germany. Due to high energy price and
active government support to the bio-fuel industry, crops used for bio-fuel production
represent a very attractive market. The proliferation of the corn rootworm compels farm-
ers to reduce the area under corn. And financial support is granted to fruit and vegetable
farms to invest in machinery and compensate for the increase of temporary labour costs.
If the first scenario reflects the more likely evolution of driving forces within the limits
of todays knowledge according the expert group, alternative scenarios follow less accepted
logic. The scenarios help to evaluate the relative effect (with comparison with the 2003 ref-
erence and baseline scenario) of these types of global policy and economic evolutions in
nitrate contamination. The B2 scenario is clearly an environmental orientation (corn root-
worm is not eliminated, water protection policies are created, absence of exemption tax
even when energy prices are very high induces a reduction in consumption) while the A1
scenario is more liberal and not environmental because an exemption on the energy price
is made and no environmental policy is implemented.
5 SIMULATION RESULTS
Simulated change in cropping patterns

With the baseline scenarios, farmers diversify their cropping patterns and replace 26% of
cultivated crops in the 2003 reference situation (2003). The change is even more important
with the B2 scenario (41% of the land occupation will change) due to stronger assump-
tions than in the baseline scenario. Conversely, the A1 scenario induces a change of only
T
a
b
l
e

2
.
S
y
n
t
h
e
s
i
s

o
f

t
h
e

a
s
s
u
m
p
t
i
o
n
s

m
a
d
e

f
o
r

t
h
e

t
h
r
e
e

s
c
e
n
a
r
i
o
s

i
n

F
r
a
n
c
e

a
n
d

G
e
r
m
a
n
y
1
.
1
S
o
u
r
c
e
s

u
s
e
d

f
o
r

d
e
v
e
l
o
p
i
n
g

t
h
e

s
c
e
n
a
r
i
o
s

a
r
e

t
h
e

f
o
l
l
o
w
i
n
g
:

*

a
c
c
o
r
d
i
n
g

t
o

e
x
p
e
r
t

a
d
v
i
c
e

o
r

c
o
m
p
r
o
m
i
s
e

f
r
o
m

t
h
e

w
o
r
k
i
n
g

g
r
o
u
p
;

(
a
)

L
E
L

e
t

S
t
a
t
i
s
t
i
s
c
h
e
s

B
u
n
d
e
s
a
m
t
;

(
b
)

L
a
F
i
s
c
a
l
i
t

d
e

l
E
n
e
r
g
i
e

-
D
G
E
M
P
-
M
i
n
i
s
t
r
e

d
e

l
E
c
o
n
o
m
i
e

d
e
s

F
i
n
a
n
c
e
s

e
t

d
e

l
I
n
d
u
s
t
r
i
e
;

(
c

)

b
e
t
w
e
e
n

2
0
0
3

a
n
d

2
0
0
6

a
n

i
n
c
r
e
a
s
e

o
f

a
l
r
e
a
d
y

5
0
%
,

t
h
e

b
a
s
i
s

f
o
r

t
h
e

w
o
r
k
i
n
g

g
r
o
u
p
a
s
s
u
m
a
p
t
i
o
n

;

(
d
)

L
E
L

:

L
a
n
d
e
s
a
n
s
t
a
l
t

z
u
r

E
n
t
w
i
c
k
l
u
n
g

d
e
r

L
a
n
d
w
i
r
t
s
c
h
a
f
t

(
S
c
h
w
a
b
i
s
c
h

G
m
n
d
)
;

(
e

)

f
r
o
m

t
h
e

u
n
i
t

p
r
i
c
e

p
e
r

f
e
r
t
i
l
i
z
e
r

t
y
p
e

o
f

t
h
e

B
u
r
e
a
u

d
e

l
a

C
o
n
j
o
n
c
t
u
r
e

d
u
S
C
E
E
S

a
n
d

t
h
e

F
e
r
t
i
-
m
i
e
u
x

s
t
u
d
y

A
R
A
A
;

(
f
)

I
N
S
E
E
;

(
g
)

c
o
n
s
e
n
s
u
a
l

v
a
l
u
e
s

f
r
o
m

U
N
I
F
A

2
0
0
5

a
n
d

w
w
w
.
e
c
o
n
o
l
o
g
i
e
.
f
r
;

(
h
)

d
r
a
f
t

F
r
e
n
c
h

w
a
t
e
r

l
a
w

(
2
0
0
5
)
.

I
t

i
s

a
s
s
u
m
e
d

t
h
a
t
p
r
o
d
u
c
t
i
o
n
s

p
r
i
c
e
s

a
n
d

y
i
e
l
d
s

r
e
m
a
i
n

c
o
n
s
t
a
n
t

u
n
l
e
s
s

a

1
,
5
%

p
e
r

y
e
a
r

y
i
e
l
d

i
n
c
r
e
a
s
e

f
o
r

c
o
r
n
.
17% of agricultural land occupation. The cropping pattern corresponding to the three scen-
arios and the 2003 reference year are described in Figure 1.
In the baseline scenario, the area under corn decreases in the whole study zone (24%
of the total agricultural land) to the benefit of cereals (wheat and a little barley) and,
slightly, to rape which is transformed in biofuel on the farm. Corn monoculture is no longer
practiced and two- to three-year rotation becomes more frequently used, (corn/wheat/
rape) and is generalised in scenario B2. This is due to the expansion of the corn rootworm
and to the increase in the fuel oil price and, to a lesser extent, to increased fertilizer prices.
For a few farm types, fertilizer tax contributes to this evolution (wheat replaces corn in a
greater proportion). Livestock (mainly milk production) and crops with high added values
(e.g. sugar beet) are not modified, except for market garden crops, which disappear in
Rye
Barley
Summer
cereals
Silage
corn
Corn
Wheat
Tobacco
Vegetables
Potato
Oleaginous/
Leguminous
Rape
Fallow
Sugar
beet
Reference scenario (2003)
Rye
Barley
Summer
cereals
Silage
corn
Corn
Wheat
Tobacco
Vegetables
Potato
Oleaginous/
Leguminous
Rape
Fallow
Sugar
beet
Tendency scenario (2015)
Rye
Barley
Summer
cereals
Sugar
beet Fallow
Rape
Oleaginous/
Leguminous
Potato
Vegetables
Tobacco
Wheat
Corn
Silage
corn
Scenario B2 (2015)
Rye
Barley
Summer
cereals
Silage
corn
Corn
Wheat
Tobacco
Vegetables
Potato
Oleaginous/
Leguminous
Rape
Fallow
Sugar
beet
Scenario A1 (2015)
Figure 1. Distribution of total cultivated area per crop simulated with economic models.
response to the increased labour price (not in scenario B2) and tobacco, which disappears
partially (due to the CAP reform).
Although the evolution is similar in Alsace and Baden, impacts are stronger in Alsace in
terms of cropping patterns (32% in Alsace against 11% in Baden of areas are concerned
by changes in the baseline scenario). This is explained by the fact that Baden farms are
more diversified than Alsatian ones where corn is largely dominant in 2003, and corn is
the crop whose profitability is the most affected by the three scenarios. Diverse responses
are also observed for the different farm types which are more or less sensitive to the simu-
lated changes.
The large development of rape (for raw oil production) in the B2 scenario is a direct
response to increase of energy price that incites farmers to produce their own (bio)fuel.
Whatever the impact on nitrate contamination, it should be considered that this develop-
ment would generate a reduction in greenhouse gas emissions, i.e. an 8% reduction of the
agricultural emissions of CO
2
.
Evolution of farming practices

The simulated changes also impact corn fertilizing practices. Alsatian farmers reduce the
nitrogen application in response to fertilizer and energy price increase (this adaptation in
practices is simulated based on a production function incorporated in the model). In
Baden, fertilizer reduction is lower than in Alsace, partly because fertilizer use in the 2003
reference situation is already lower than in Alsace.
Groundwater abstraction for irrigation (for corn, vegetables and tobacco) is also reduced
by approximately 70% (50% in the A1 scenario and 80% in the B2 scenario). This is due
to the decrease of the area under irrigated corn, vegetables and tobacco. Irrigation is reduced
even more in Alsace where it was already more significant than in the Baden region in 2003.
The area under irrigated corn is reduced and even disappears in the B2 scenario, along with
tobacco (all scenarios). This induces a 6% reduction (in volume) in aquifer withdrawals for
the baseline scenario.
The 2015 scenarios suggest a development of Intermediate Nitrate Trap Crops (INTC)
due to the increase in area under winter cereals, which enables winter and autumn INTC
sowing. INTC area increases from 12% of the agricultural area (in 2003) to 40% (baseline
scenario) in Alsace and from 20% to 26% in Baden. These results must be interpreted with
caution, because economic data vary from one farmer to another, and, like most agro-
environmental measures, their implementation by farmers is often largely due to non-
economic factors (sensitivity to environmental aspects, acceptability).
Economical impact of scenarios on agriculture

These important production changes, however, suppose a restructuring of downstream
industries (agro-food and trade), which would have to be able to take the cereal production
(wheat and barley) and find markets for rape by-products (oil cakes for animal food or
heating). Indeed the models only simulate changes in agricultural supply, assuming that
the demand (at the local level) reflects global market changes hypothesised in the scenario
description.
Scenarios also have a significant effect on farm income and on the total value of agri-
cultural production. But the reduction of marketable agricultural production is partially
replaced by on-farm production of auto-consummed biofuel (no taxes to pay until today).
Changes in income level, including changes in CAP premium regimes, may result in certain
types of farms disappearing, whilst others may have to merge into larger units. The farm
typology described on the basis of the 2003 data is likely to evolve, and agri-environmental
measures designed today may not be well adapted to this new agriculture in the medium term.
Long-term impact on nitrate concentration

The consequences of these three global change scenarios were assessed using the chain of
integrated models described above. The results show that future nitrate average concentra-
tion in the aquifer will not significantly differ from one scenario to another in the short-
term (estimated at 19mg/l for the baseline and A1 scenarios, and 19.5 for the B2 scenario)
which can be explained by the long response time of the aquifer to surface stresses.
Differences are expected to increase in the long-term (average concentration of 16mg/l for
baseline and A1, 18.2mg/l for B2). The model also shows that the area where nitrate con-
centration exceeding the drinking water threshold value (50mg/l) will drastically fall from
17000 ha in 2005 to around 4000 ha for the baseline and A1 scenarios and 6000 ha for the
B2 scenario (Fig. 2). Surprisingly, scenario B2, which is assumed to depict a world with
more stringent environmental constraints, is also the worst scenario in terms of water pol-
lution due to the increase in industrial crops used for producing bio fuels.
6 CONCLUSION AND PERSPECTIVES
From a policy perspective, the simulations presented above clearly show that groundwater
improvement, which has been observed between 1997 and 2003, should continue at a rate
3000
5000
7000
9000
11000
13000
15000
17000
2005 2015 2025 2035 2045
A
r
e
a

w
h
e
r
e

[
N
O
3
]

>

5
0

m
g
/
l

(
h
a
)
Baseline
A1
B2
Figure 2. Simulated evolution of the area (in ha) where nitrate concentration exceeds 50mg/l
(adapted from LUBW 2006).
that will change to reflect global assumptions. Surprisingly, the B2 scenario, which re-
presents a future where CO
2
emission decreases, leads to the worst situation in terms of
nitrate concentration. This result highlights how one type of environmental policy (here green
house warming control) may negatively impact another policy (groundwater protection). This
statement clearly calls for more environmental integration when assessing the impact of
projects and policies. Assessment should include various criteria and look at the differ-
ent environmental effects of scenarios considered, i.e. on air, water, biodiversity, land-
scape, etc. This could help identifying possible oppositions or synergies between different
environmental policies.
From a methodological perspective, the integrated modelling approach presented in this
paper was first used to simulate the impact of contrasting global change scenarios. In its
philosophy, the approach aims more at exploring possible futures than forecasting what
will actually take place in the coming decades. The use of models is clearly perceived as a
way to integrate assumptions made by experts, to assess their consequences and construct
a shared vision of plausible futures. The development of several scenario variants is also
considered as one way to account for uncertainties related to global change evolutions,
such as unpredictable changes in world energy markets. The chain of models could also be
used to simulate the impact of alternative groundwater management options (actions) that
could be implemented to alter expected trends. This includes, for instance, simulating the
impact of various forms and levels of subsidies for specific practices, including organic
farming and possible subsidies for agro-environmental measures. This would, however,
require improving the economic model to account for possible changes in farm structures
(dynamic modelling).
REFERENCES
Alcamo J., Kreileman G.J.J., Bollen J.C., van den Born J., Gerlagh R., Krol M.S., Toet de Vries, H.
(1996) Baseline scenarios of global environmental change. Global Environmental Change, 6, 4,
261303.
Casper M., J., Grimm-Strele Th., Gudera S., Korte H., Lambrecht B., Schneider P., van Dijk J-D.,
Rinaudo & Finck M., (2005) EU-Project MoNit: Decision support system to assess the impact of
actions and changing frameworks on the nitrate load in the Upper Rhine Valley aquifer. Models
and scenarios. Paper presented at the EWRA conference, Menton: September 710, 2005.
Chambre dAgriculture Rgionale dAlsace (2003) Typologie des exploitations alsaciennes.
Schiltigheim: Chambre dAgriculture Rgionale dAlsace.
Deutscher Bauerverband (2003) Die Reform der Gemeinsamen AgrarPolitik.
European Commission. 2002. Implementation of Council Directive 91/676/EEC concerning the
protection of waters against pollution caused by nitrates from agricultural sources. Synthesis from
year 2000 Member States reports. COM (2002) 407 fin. Brussels: European Commission.
Graveline N., Rinaudo J-D. Segger V. (2005) Simulating the economic impact of groundwater pro-
tection scenarios on the farming sector of the upper Rhine valley EU-Project MoNit: Decision
support system to assess the impact of actions and changing frameworks on the nitrate load in the
Upper Rhine Valley aquifer. Models and scenarios. Paper presented at the EWRA conference,
Menton: September 710, 2005.
Hazell PBR, Norton RD (1986) Mathematical Programming for Economic Analysis in Agriculture.
Macmillan New York.
IPCC (2000) Summary for policy makers Emissions scenarios. IPCC special report. Intergovernmental
Panel on Climate Change.
LUBW (2006) Simulating future groundwater nitrate contamination final report of INTEREG III
A MoNit project (in French and German) Karlsruhe: Landesanstalt fr Umwelt, Messungen und
Naturschutz Baden-Wrttemberg. (Available at www.lubw.baden-wuerttemberg.de publikatio-
nen wasser)
MAAPAR (2004)-La nouvelle politique agricole commune (PAC)-Les modalits dapplication
nationale 200520062007. Paris: Ministre de lagriculture de lalimentation de la pche et des
affaires rurales.
Rgion Alsace (2000) Inventaire de la qualit des eaux souterraines dans la valle du Rhin Suprieur
1997. Strasbourg: Rgion Alsace.
Rinaudo J.-D., Blanchin R., Arnal C., Elsass P., Meilhac A., and Loubier S. (2004) Assessing the cost
of groundwater pollution: the case of diffuse agricultural pollution in the upper Rhine valley
aquifer. Water Science and Technology, 52, 9,153162
CHAPTER 37
Simplified vulnerability mapping of groundwater in the
Seine-Normandy Basin. Application to the BASOL
1
sites
Vincent Mardhel
1
, Annabel Gravier
1
, Jean-Yorick
Koch-Mathian
1
, Cline Nowak
1
, Jean-Luc Terreyre
2
,
Mireille Raguet
3
& Ccile Garnier
3
1
BRGM, Orlans cedex, France
2
MEDD, av. de Sgur, Paris, France
3
Agence de lEau Seine-Normandie, rue Salvador Allende, Nanterre Cedex, France
ABSTRACT: The Network Persistence and Development Index (IDPR) is a tool for national and
regional vulnerability mapping. Based on an analysis of the drainage patterns (calculated from a
Digital Elevation Model (DEM)) and the natural hydrological flow, it reflects the influence of the
underlying geological formations toward surface-water runoff or infiltration. The concept provides
an indirect approach to infiltration, knowledge of which is essential for drawing up any vulnerability
map, and replaces many of the criteria usually employed in the methodologies for producing vulner-
ability maps.
The vulnerability map described here relates to an area supervised by the Seine-Normandy Basin
Agency, and is the first that was compiled using the IDPR approach. The methodology was validated
by a working group made up of representatives from the French Ministry of Ecology and Sustainable
Development (MEDD), the Water Agency, the Regional Environment Division (DIREN), the
Regional Department for Industry, Research and the Environment (DRIRE) and the French
Geological Survey (BRGM). Since then, the same approach has been used for areas supervised by
the Loire-Bretagne Water Agency, for areas in the Nord-Pas-de-Calais and Aquitaine regions, and
also for the area supervised by the Rhone-Mediterranean-Corsica Agency where calculation of the
IDPR was begun in 2003.
The method is not designed for large-scale vulnerability mapping; the precision of the DEM and
river networks would not allow it. It is rather an indicator of a general state of vulnerability. The result-
ant map must be interpreted not as reflecting an absolute vulnerability, but as showing a hierarchy of
the systems in terms of their vulnerability. It enables the targeting of priority areas in terms of protect-
ing the resource. Its application to local pollution pressure points, such as those related to the BASOL
sites, undertaken in 2005 in the Seine-Normandy Basin in partnership with MEDD and the Water
Agency, has shown the value of such comprehensive maps in helping establish monitoring and inter-
vention priorities. Potential uses are many, in particular for monitoring groundwater quality in vul-
nerable areas downstream from classified installations.
1
BASOL BAse des Sites et SOLs pollus, ou susceptibles de ltre, sur lesquels ladministration a engag une
action (Database of polluted, or potentially polluted, ground where the authorities have intervened)
1 THE DRAINAGE NETWORK
The drainage network corresponds to all linear flow concentrated at the topographic sur-
face, which it gradually alters. It represents a state of the water cycle, i.e. the run-off or
stream flow that follows the easiest path, continuously minimizing the entropy (translated
from Deffontaines, 1990).
In 1972, Robert Prudhomme, one of the French forerunners in the analysis of drainage
networks, showed the value of this type of analysis for geological surveys because: The
features of the drainage network are highly dependent during their evolution on the state
of the drainage basin. He adds: the three parameters whose variation would appear to be
the most important in the study of drainage networks are topography, geological structure
and lithological composition of the subsurface. Prudhomme thus experimented with the
cartographic analysis of drainage networks and came up with a number of very pertinent
correlations. Kim Woo-Gwan continued this work and went deeper into the analysis of
drainage networks. In the 1990s, Deffontaines insisted on the value of the drainage net-
work for a morphostructural approach: The drainage network is modified mainly by
lithology and fracturing. At the same time, Riazanoff suggested automatic methods for
the extraction and analysis of drainage networks, revealing a few correlations that opened
the door to the automated computer analysis of the drainage network. The work of these
various authors shows that the pattern of the drainage network results from complex inter-
actions. Although water is by nature uncompressible and flows by the shortest route from
the high point (peak, ridge, etc.) to the low point (sea, lake, river, etc.), the conditions of
the stream course during its journey have an influence on the drainage pattern. The origin
of many of these factors is purely geological:
Lithology: its role is influenced by both its physical nature (thickness, permeability,
etc.) and chemical nature (sensitivity to dissolution, crystallization phenomena, etc.)
Structure: geological alternations, unconformities, lithospheric deformation, etc., mod-
ify the topography and shape the drainage network.
Regional slope: defined by a sense, a direction then an inclination, it is a determining
element in the orientation and structure of the watercourse.
Other factors resulting from a combination of varied types of constraint also come into
play. Their importance varies according to the scale at which one examines the drainage
network, one example being rejuvenation and eustasy. Thus, over and above its morph-
ology, the morphogenesis of a drainage network implies looking for possible correlations
between the network pattern and the nature of the subsurface. This has instigated a large
number of studies wherein the main underlying hypothesis is that the pattern of the
drainage networks retains a signature of the endogenous or exogenous phenomena.
2 DEFINITION OF THE IDPR (NETWORK PERSISTENCE AND
DEVELOPMENT INDEX), MORPHOLOGY VS. MORPHOGENESIS
The work cited above shows that over and above the geographic factor, the drainage net-
work can be considered as a morphostructural unit whose genesis reflects the correlations
between cause and shape. The IDPR looks at these correlations and highlights those that
are presumably geological. Arbitrarily, this approach minimizes the other factors and that
despite the fact that climate (paleoclimate), vegetation, human action play an important
role in the evolution of the drainage pattern. Thus, in the event of a perfectly homogeneous
environment with zero permeability, only slope and landscape morphology will influence
the emplacement of the watercourse. The search for thalwegs (i.e. lines of greatest slope),
which collect the runoff waters through gravity, should thus lead to a reproduction of the
drainage network. In reality, the natural drainage network differs from this theoretical net-
work, and it is this difference that reflects the environmental complexity. The idea under-
lying the IDPR, therefore, derives from the premise that as (a) the structure of the drainage
network, the thalwegs, depends on the natural ground relief, and (b) the structure of the
drainage pattern depends on the underlying geological context, (c) the difference between
the thalweg network and the drainage pattern reflects the influence of the subsurface envir-
onment. Thus, a basin underlain of very permeable material will generally have a low-den-
sity drainage pattern compared to the density of the calculated thalwegs. Conversely, a
basin underlain by impermeable, but soft and easily eroded rock, such as marl, will com-
monly show a higher density drainage pattern. So the IDPR becomes a means of indirectly
quantifying the transfer function of meteoric water from the surface to the subsurface. It
compares a theoretical network established according to the hypothesis of a perfectly
homogeneous environment (the Development Index ID) against the natural network
formed under the control of a heterogeneous geological context (the Network Persistence
PR). It presents an observed metrology of the difference between the two networks and
takes into account the role of the subsurface environment in the formation of the drainage
network and can thus be used for an indirect approach to the inherent nature of the ground
regarding surface-water infiltration or runoff.
3 NECESSARY INFORMATION LAYERS
Derived from the world of academic geographers, adopted by major projects that involved
the acquisition of large volumes of data that need to be managed in time and space,
Geographic Information Systems (GIS) have been developed since the 1980s for digitiz-
ing, conserving, analysing and synthetizing geographic information. Although many sci-
entists, geologists, geographers, hydrologists and geomorphologists have, since the 1950s,
been defining methods for analysing drainage networks, this type of analysis was never
truly applied, other than experimentally or sparingly, until the 1980s. The lack of applica-
tion can be explained by the length of the manual calculations and the unsuitability of the
methods for automating such analyses. The increasing power of the GIS has for some time
been placing the accent on automated extraction of the drainage network from the digital
elevation model (DEM). The IDPR combines the automation approach for the analysis of
drainage networks with the digitization of the natural networks.
4 THE NATURAL NETWORK
In the context of the Seine-Normandie Basin study, the natural drainage network, used for
the IDPR calculation, came from data in the IGN (French National Geographic Institute)
Carthage
database (Fig. 1). The attributes used for the analysis were the state and nature
of the watercourses.
Simplified vulnerability mapping of groundwater in the Seine-Normandy Basin 499
5 THE THEORETICAL NETWORK
The theoretical flow network is based on analysis of the thalwegs (Fig. 2) established from the
50m digital elevation model (DEM). Although this model is accurate, there are depressions
from which no outlet is possible and flat areas where it is impossible to determine the thalweg;
here it is necessary to process the areas in order to re-establish the potential flow. For the
depression areas, the corrections are guided by the hypothesis that water flows on the surface
and if there is a natural depression, the water fills this first before continuing its downstream
Figure 1. Drainage network extracted from the IGN Carthage
database.
Figure 2. Thalweg pattern.
path. A first processing technique assumes that depressions are artefacts that do not represent
a real form (Jenson and Domingue, 1998). Consequently one has only to fill them after hav-
ing determined the outlet: this is the so-called hole filling technique. A second technique
(Fairfield and Leymarie, 1991) aims at determining the outlet of the depression and then
reversing the flow path between the bottom of the depression and the outlet in re-establishing
the network continuity. Finally, a last method consists in identifying a very high area between
two depressions and then digging this area until the water in the depression flows down hill.
In practice, these techniques all give comparable results. The processing algorithm for
depressions has no significant influence on the network obtained through direct extraction
on the other hand, essential differences are found in changes to the DEM. In effect, although
the second indicated technique has no effect on the altitudes, the other two modify the DEM
and provide an indication as to the quality of the initial data and/or the nature of the observed
reliefs (dolines, endorheic areas, etc.). So as to simplify the tools that are used, the IDPR cal-
culation process uses the first method.
6 IDPR CALCULATION
The IDPR is the ratio between the distance of the closest actual watercourse and the dis-
tance of the closest calculated thalweg, and varies on scale of between 0 and n, arbitrarily
brought to a range of 0 to 2000.
IDPR 1000: Infiltration greater than surface runoff. The runoff water on natural
ground joins a drainage axis defined from a thalweg analysis without this being obvi-
ous through the appearance of a natural hydrological axis.
IDPR 1000: Infiltration and surface runoff are equal. There is conformity between
the availability of drainage axes associated with the thalweg and in situ flow.
IDPR 1000: Surface runoff greater than infiltration to the subsurface. The runoff
water on natural ground very rapidly joins a natural hydrological axis whose presence
is directly justified by a thalweg. An IDPR close to or equal to 2000 reflects a tempor-
ary or permanent stagnation of the water, leading to two different interpretations: if the
water table is close to the natural surface, (water course and wetlands), the ground is
saturated and the water no longer infiltrates; if the water table is deep, the runoff char-
acter may indicate an imperviousness of the natural ground. It is suggested that IDPR
values greater than 2000 are mainly assignable to wetland.
Figure 3 shows the result of the IDPR calculation for the whole of the Seine-Normandy
area, the grid being composed of 100m 100m cells. It is based simply on analysis of the
geomorphological and hydrological data and corresponds remarkably well with the inset
geological map. Two examples illustrate the correspondence between the IDPR results and
the geological formations.
6.1 Beauce Limestone
The Beauce water table represented by the complex Beauce Limestone aquifer is a hydro-
graphic unit that stretches from the Seine to the Loire. Represented by the unconfined
groundwater of the Beauce Limestone and its confined section beneath the Orleans Forest
and the Burdigalian overburden, the reservoir aquifer consists mainly of limestone within a
basin with an underlying impermeable clay. The confined groundwater in the Orleans Forest
is overlain by a clay bed a formational cover that is not shown by the 1: 1 million scale geo-
logical map (which is stratigraphic and not lithological), but which is clearly revealed by the
IDPR grid (Fig. 4).
Figure 3. IDPR grid applied to the Seine-Normandy Basin.
Figure 4. The Beauce as revealed by the IDPR and the 1:1 million scale geological map (BRGM).
6.2 Bray country
The northern part of the sedimentary Paris Basin consists mainly of exposures of the Upper
Cretaceous aureole beneath a chalk facies that is widely developed in the Somme area from
north of the Aisne to west of the Oise. Tertiary exposures (mainly Eocene sands and clays)
are found locally, deposited on the chalky substratum. Even more locally, older rocks of
Jurassic age crop out at the edge of the region, in the Bray Country inlier in the Beauvaisis.
These rocks are formed of relatively impermeable clay or calcareous marl. The block dia-
gram (Fig. 5) and associated cross section illustrate the morphology and geological structure.
The IDPR calculation grid for this region gave a very detailed response where the pres-
ence of the only slightly permeable Lower Cretaceous formations mark the structure of the
Bray Country anticline which forms this rather distinctive inlier.
7 MAPPING THE SIMPLIFIED VULNERABILITY INDEX
The notion of vulnerability is based on the idea that the physical environment in contact with
the groundwater provides a more or less high degree of protection where pollution is con-
cerned. The vulnerability is, therefore, represented by the given capacity of surface water to
reach the water saturated subsurface. Two types of vulnerability are distinguished: intrinsic
vulnerability and specific vulnerability (Schnebelen et al., 2002). The intrinsic vulnerability
is determined in order to represent the characteristics of the natural environment, which
determine the sensitivity of the groundwater to pollution from human activities. The specific
vulnerability is the term used to define the vulnerability of a groundwater body to a specific
pollutant or group of pollutants; it takes into account the properties of the pollutants and their
relationship with the various components of intrinsic vulnerability.
Figure 5. Bray country.
Figure 6. The IDPR response for Bray country.
Figure 7. Mapping intrinsic vulnerability in the Seine-Normandy Basin.
Many methods of determining groundwater vulnerability have been developed worldwide,
passing from the most complex with models that taking account of the physical, chemical and
biological processes in the saturated zone, to methods of weighting between different criteria
affecting the vulnerability (Gogu and Dassargues, 1998b). Among the latter, the so-called
index mapping with criteria weighting methods (Point Count Systems Models, PCSM)
appear to be the most relevant where field realities are concerned. They are also the best
known and currently most widely used (Gogu and Dassargues, 2000; Zaporozec and Vrba,
1994). These methods assume the identification of many indices that, at small scale, will be
substituted for the IDPR. Thus, the study that we present here endeavours to construct a sim-
plified intrinsic vulnerability of the Seine-Normandy Basin simplified because it uses only
two criteria:
The unsaturated zone (ZNS) that is crossed before the pollutant reaches the aquifer. Its
nature, permeability, thickness are determined from a compilation of piezometric data
from the uppermost groundwater and the DEM used for the IDPR;
The IDPR, which substitutes for data related to ground permeability and the aquifer
characteristics. It is an approximation of the characteristics of these environments for
which, as seen above, the IDPR qualitatively describes the permeability based on the
infiltration and runoff poles.
The formula for the calculation is: Vi (IDPR Weight * [IDPR Criterion]) (ZNS
Weight * [ZNS Criterion]) with: IDPR Weight ZNS Weight 1, and Vi ranging between
0 and 100. The adjustment of the weights retained by the different regional experts as
being the most representative from their acquired knowledge is fixed at 50%/50%.
8 COMPARISON OF THE IDPR/ZNS METHOD WITH EARLIER
REGIONAL VULNERABILITY MAPS
The mapping of the Seine-Normandy Basin was discussed and validated by a working
group made up of representatives of MEDD, the Water Agency, DIREN, DRIRE and
BRGM. Among the various control areas taken into account during the study, we illustrate
and compare the results for the regional maps in the Burgundy and Champagne areas.
In Champagne-Ardenne, the adopted reference document is the specific vulnerability
map for nitrates in the area of Upper Turonian and Senonian (Upper Cretaceous) outcrops.
The method elaborated by J. Ricour (1998) for an application to the chalk groundwater in
the Guines area (Pas-de-Calais) was used and adapted for compiling this map. The degree
of vulnerability is expressed as one of four categories (Negligible to zero vulnerability,
Low vulnerability, Moderate vulnerability and High vulnerability) represented on the map
by a colour gradient (cf. Legend). The result of the calibration test of the simplified aquifer
vulnerability analysis method is shown in Figure 8; comparison with the reference map
indicates a suitable correlation within this region.
In Burgundy, the reference map that was used is the Map of groundwater vulnerability
to phytosanitary products. The comparison is given in Figure 9 and shows a clear corres-
pondence with the degree of vulnerability assigned to the aquifers by the simplified
method using the IDPR.
9 EXAMPLE OF THE APPLICATION OF THE INTRINSIC VULNERABILITY
MAP TO (POTENTIALLY) POLLUTED SITES
The Seine-Normandy Water Agency and MEDD instructed BRGM in 2004 to develop
vulnerability assessment indicators for the groundwater below the (potentially) polluted
BASOL sites over the entire basin. A method of data crossing according to different pertin-
ent criteria was first developed with the study partners. Then a Geodatabase and a
Geographic Information System using Access 2000
and Arcmap
format were develop-

ed. The resultant tool, called BASOLHYD SN 2005, coupled with the GIS enabled pro-
cessing and monitoring to be prioritized at the BASOL sites, and also at any other local
pressure point. The criteria make it possible to classify and compare the sites between
themselves and combine the parameters relative to the pollution source (BASOL data), the
transfer to groundwater factors (obtained by the vulnerability GIS) and the potential tar-
gets (drinking water catchments, offtake volumes, water body protection issues). The GIS
linked to the digital BASOLHYD tool enables the analysis of different multicriteria scenarios
through modulating the weight assigned to each criterion (IDPR, ZNS, source, transfer,
Figure 8. Comparison of methods in the Champagne-Ardennes region.
Figure 9. Comparison of methods in the Burgundy region.
target), the display of the results in map form (Basin Atlas showing water bodies,
Dpartements and regions, and BASOL sites) and in table form sorted by site category.
Out of a total of 846 georeferenced BASOL sites in the basin suitable for multicriteria
processing, 35% are located immediately above high vulnerability aquifer formations and
47% above average vulnerability aquifer formations. Among the 281 sites considered as
priority, groundwater monitoring has been installed at 256, and site-processing actions
have been undertaken at 222. As an example, Figure 10 is a map of the BASOL sites in the
Picardie region.
10 CONCLUSION
The transition from the IDPR to the notion of vulnerability uses the criteria linked to the
thickness of the unsaturated zone (ZNS). This criterion allows one to rank the groundwater
target according to the distance crossed from the natural ground surface to the saturated envir-
onment. In the sedimentary domain, the weighting of the criteria linked to the thickness of
the unsaturated zone equals that of the IDPR in conditioning the groundwater vulnerability
index. The latter allows one to simplify the dataset required for constructing these maps.
The progress of the various surveys being carried out in mainland France allows an
anticipation of the full groundwater vulnerability coverage for the year 2007 (Fig. 11).
Such coverage has already been achieved for the whole of Slovenia (Mardhel et al., 2004).
The limit to the interpretation and exploitation of these vulnerability maps is deter-
mined by the process of drawing up the component data. A likely minimum scale of 1:
100,000 is envisaged with a vectorized hydrological base map at the same scale and a
Figure 10. Vulnerability analysis of BASOL sites in the Picardie region.
digital elevation model of 50m steps. The analysis of these networks from a DEM of 90m
steps (SRTM 90m
2
) gives a comparable result and allows one to envisage a European map
of this index (FOOTPRINT
3
project).
The application to the BASOL sites of the Seine-Normandy Basin has shown the value
of such maps for helping establish intervention and monitoring priorities. Other prospect-
ive uses are many, and include:
An extension of the method to the BASOL sites throughout France;
The use of the IDPR for surface-water vulnerability studies at watershed scale;
The mapping of vulnerability at a more local scale in the areas at stake, associated with
a sensitivity map;
The establishment of priorities for databanking the groundwater-quality monitoring
data downstream of classified installations;
A contribution to the selection of classified installations requiring localized groundwater
monitoring in the most vulnerable sectors (article 65b of the decree of 03/08/2001 rel-
ative to monitoring);
An aid to the adaptation of the monitoring networks linked to the European Water
framework Directive and to the installation of a common water monitoring for all
industrial sites located over a highly vulnerable zone.
Figure 11. Progress of the different surveys in France.
2
SRTM 90m: Void-filled seamless SRTM data V2, 2005, International Centre for Tropical Agriculture (CIAT),
available from the CGIAR-CSI SRTM 90m Database.
3
FOOTPRINT was published in the Official Journal of the European Communities on 30 October 2004. Its a
STREP (Specific Targeted Research Project). FOOTPRINT Creating tools for pesticide risk assessment and
management in Europe.
REFERENCES
Corbier P. and Mardhel V., 2003 Carte de vulnrabilit simplifie des bassins versants de la rgion
Pays de la Loire. Rapport BRGM/RP-53106-FR
Deffontaines B., (1990) Dveloppement dune mthodologie morphonotectonique Analyse des
surfaces enveloppes du rseau hydrographique et des MNT, Thesis, Universit Paris VI, 225
pages, 1990.
Fairfield J. and Leymarie P., 1991 Drainage networks from grid Digital Elevation Models. Water
Resources Research, pp. 709717.
Gogu R.C., Dassargues A., 2000 Current trends and future challenges in groundwater vulnerabil-
ity assessment using overlay and index methods, Environmental Geology, 39, 6, International
Journal of Geosciences, Springer Verlag, Germany, v.39 (6) April 2000, pp.549559.
Gravier A., Koch-Mathian J-Y., Mardhel V. and Nowak C., 2006 Croisement des donnes des sites
BASOL avec la vulnrabilit simplifie des nappes du bassin Seine-Normandie, rapport de syn-
thse, BRGM/RP-54257-FR.
Jenson, S.K. and Domingue, J.O. 1998 Extracting topographic structures from digital elevation
data for geographic information system analysis. Photogrammetric engineering and remote sens-
ing, pp. 15931600.
Kim W.G. 1978 Analyse cartographique du bassin de Nak-Dong (Core), Thesis, Universit
Bordeaux I, 95 pages, 1978.
Lacroix M., Martz L.W., Kite G.W. and Garbrecht J., 2002 Using digital terrain analysis modelling
techniques for the parameterization of a hydrologic model. Environmental Modelling and
Software.
Lawrence W. Martz and Jurgen Garbrecht, 1993 Automated Extraction of Drainage Network and
Watershed Data. Digital Elevation Models, 29, 6, 901908.
Machard de Gramont H., Mardhel V., Pointet T., Schomburgk S., en collaboration avec Caous J-Y.,
Caudron M., Corbier P., David-Rouxel E., Equilbey E., Ghyselinck M., Jauffret D., Maget Ph.,
Marchais E., 2003 Etude de sensibilit aux remontes de nappes Mthodologie et application
20 secteurs tests de France mtropolitaine. Rapport BRGM/RP-52340-FR.
Mardhel V. and Gravier A., 2005 Carte de vulnrabilit simplifie du bassin Seine-Normandie.
Rapport BRGM/RP- 54148-FR.
Mardhel V., Frantar P., Uhan J. and Miso A., 2004 Index of development and persistence of the river
networks as a component of regional groundwater vulnerability assessment in Slovenia.
International conference on groundwater vulnerability assessment and mapping. Ustro n, Poland,
1518 June 2004., Poland 15 to 18 June 2004.
Nowak C. and Mardhel V., 2005 Croisement des donnes des sites pollus, ou susceptibles de
ltre, de loutil BASOL et des donnes sur les nappes du bassin Seine-Normandie, rapport inter-
mdiaire BRGM/RP-53253-FR.
Prudhomme R., 1972 Analyse morphostructurale applique en Aquitaine occidentale et au golfe
de Gascogne Dfinition dune mthodologie cartographique interprtative, Thse, Universit
Bordeaux I, 364 pages, 1972.
Razianoff S. 1989 Extraction et analyse automatique des rseaux partir de MNT, Thse,
Universit Paris VII, 89 pages, 1989.
Schnebelen N., Platel J.P., Le Nindre Y.M. and Baudry D et al. 2002 Gestion des eaux souterraines
en Aquitaine Anne 5. Opration sectorielle. Protection de la nappe de lOligocne en rgion bor-
delaise Rapport BRGM/RP-51178-FR.
Zaporozec A. and Vrba J., 1994 Classification and review of groundwater and vulnerability maps.
In Vrba and Zaporozec (Editors), Guidebook on mapping Groundwater vulnerability. IAH, Verlag
Heinz Heise, Hannover, pp. 2119.
F. Aquifer monitoring systems
and management rules
CHAPTER 38
Characterization of pesticide concentrations in the
Craie du Nord aquifer system
Didier Renard & Chantal de Fouquet
Ecole des Mines de Paris, rue Saint-Honor, Fontainebleau
ABSTRACT: Although the number of measurements regularly increase, mapping the pesticide con-
centrations and their evolution over time remains difficult. The Craie du Nord aquifer system was
selected for a methodological study based on samples taken from 1997 to 2002 from three different net-
works: the information (good spatial coverage), the supply (preferential sampling) and the producer (few
stations) networks. The atrazine annual mean concentrations are calculated either from few stations with
long time series or on the whole data set; they are different but show similar evolutions with time. The
connection between concentrations and land use is minor. Data on atrazine concentrations from the
information network present a spatial structure, modelled through a space-time variogram. The sensitiv-
ity analysis shows that adding data from the supply network improves the accuracy of the estimation.
The maps of interannual variations are discussed in relation to the accuracy of the estimations. Finally,
recommendations are made on the systematic character and the regularity of the sampling.
1 INTRODUCTION
Even if the number of samples measured each year regularly increases, accurate mapping
of pesticide concentrations in aquifers remains difficult due to the following reasons:
Pesticides are a large variety of substances which have been used at different times and
for different durations.
The measurement accuracy (from the sampling procedure to the chemical analysis)
varies over time and depends upon the laboratory. The analytical uncertainty is
between 15% and 20% (Amalric and Baran, 2004).
The time series at the measurement stations are short. The sampling periodicity depends
on the function of the different networks (the information network is used to monitor
an area, while the supply network controls the quality of drinking water) and in some
instances it is even determined by the measured values themselves.
The wells used for monitoring purposes are usually drilled in order to provide drinking water,
not to characterize the quality of the whole aquifer system. A similar problem occurs at a
larger scale when characterizing the status of groundwater in France (Blum, 2004). Relevant
questions are: what is the relevant accuracy when mapping concentrations? Can temporal or
seasonal evolutions be detected? Are there any relations between the concentrations of the
different substances? The Craie du Nord aquifer system has been selected for quantifying pes-
ticide concentrations in relation to land use because of its large area, the large number of
measurements available and its vulnerability to surface pollutants (Amalric and Baran, 2004).
Located in formations dating from the Mesozoic era, the Craie du Nord system is mainly a
water table aquifer, except in the NW part where it is confined under Tertiary deposits (e.g.
Mardhel et al., 2005). Permeability is linked to fracture distribution and decreases with depth,
and is generally low below the first twenty or thirty meters. Besides, it is higher under the val-
leys than under the plateaux. The Craie du Nord aquifer system appears to form a large num-
ber of basins, the upper levels of which are exploited for drinking water.
2 SAMPLING CHARACTERISTICS
The sampling campaigns were carried out from 1997 to 2002. Two substances present rather
high concentrations: atrazine, a herbicide used during forty years or so on corn fields and
now prohibited, and one of its degradation products, desethyl-atrazine (DEA). More exten-
sive atrazine measurements were gradually taken in the NE area of the aquifer during this
period of time (Fig. 1). The overall statistics (mean, quantiles) cannot be compared from one
year to the other as they derive from different sets of sampling stations. A detailed study is
necessary in order to derive the importance of the spatial and the temporal variations.
Despite the large number of data (several thousands, see Table 1), few sampling stations
offer long time series. More than 30% of the stations have one measurement only, 28%
have two, with a maximum of 24 measurements for only two stations.
The area contains 40 zones which are mainly juxtaposed. 30% of the stations belong to the
same zone. All the other zones are each sampled by less than 5% of the stations. As the depth
of the measurement at each station is not given, the problem can be simplified by working in
two dimensions and considering that all samples belong to a same homogeneous aquifer.
3 QUANTIFICATION LIMITS
Each measurement is provided together with its quantification limit. In fact the presence
of a substance can be detected at a concentration greater than the detection limit, but it
2000 2002 1997
50km
Figure 1. Base map of station locations: () for a station and () for a measurement.
is not always possible to quantify it due to significant measurement uncertainties. This is
why the quantification limit is used, greater than the detection limit. Most measure-
ments lie below the quantification limit (Table 1), which varies for each substance, each
laboratory, and over time. For atrazine, the quantification limit is variously reported as
0.01, 0.02 and 0.05g/l. These values are not negligible with regard to the drinking water
standard, the current limit being 0.1 g/l per substance. In the rest of this study, the con-
centrations indicated as lower than a quantification limit are set to this limit. This trunca-
tion introduces a slight overestimation, which will be evaluated by a sensitivity analysis.
Conversely, setting the concentrations systematically to 0 produces an under-estimation,
which is problematic in view of the health risk involved. Elaborate methods dealing with
soft data can also be applied, which replace the measurements at intervals between 0 and
the quantification limit (Chils and Delfiner, 1999). However such sophisticated methods
require significant assumptions to be made on the statistical characteristics of the variable
(stationarity), which are not always fulfilled by the data.
In order to estimate the impact of the non-quantified concentrations, every measure-
ment is replaced by a random value drawn uniformly between 0 and the given quantifica-
tion limit. For atrazine, the overall mean decreases from 0.044 g/l to 0.036g/l, and the
variance slightly increases from 0.019(g/l)
2
to 0.021(g/l)
2
. There is no visible effect on
variograms. This variation of the mean highlights the necessity of a low quantification
limit, especially when compared to the drinking standards.
4 MEASUREMENTS FROM DIFFERENT NETWORKS
Three monitoring networks cover the area but they do not geographically overlap (Fig. 3).
The information network (269 stations) is designed to evaluate the quality of the aquifer
systems in time and to detect changes. The supply network (707 stations) controls the qual-
ity of the water to be put into supply. Only 68 stations belong to the producers network.
The statistics for atrazine concentration reflect this: the supply network has a lower mean
as it is focused on those stations where the atrazine concentration is lowest. This can also
explain the lower value for the maximum of the concentration, and thus the lower standard
deviation. But as the networks cover different geographical areas (Fig. 3), the explanation
is a little more complex. The experimental variance can be considered as a dispersion vari-
ance: even in the stationary case (when neither the mean nor the variance show any trend),
due to the spatial correlation, this variance increases with the area investigated. When the
variance increases, the histogram and therefore the maximum are modified.
Characterization of pesticide concentrations in the Craie du Nord aquifer system 515
Table 1. Number of measurements and percentage of quantified values per substance.
Substance Origin Number Quantified %
Atrazine Corn weed-killer 4374 44.7
Simazine Herbicide 4243 5.3
Terbuthylazine Corn weed-killer 3988 1.2
Isoproturon Corn weed-killer 3753 3.4
Diuron Herbicide 3751 6.0
Atrazine desethyl (DEA) Metabolite atrazine 3391 54.1
2-Hydroxy atrazine (HyA) Metabolite atrazine 3029 5.6
Similarly, the quantile-quantile plot (which compares the histograms, regardless of the data
location in time and space) shows the same difference according to the type of network: the val-
ues are systematically larger for the information network than for the supply network (Fig. 2).
5 SEMI-PREFERENTIAL SAMPLING OF THE SUPPLY NETWORK
Sampling is mostly carried out in April or October, i.e. with a six-month interval (Fig. 3). In
some stations, the sampling frequency is adapted to the concentration level: the frequency
0.0
0.1
0.2
0.3
0.4
0.5
0.0
0.1
0.2
0.3
0.4
0.5
0.0 0.1 0.2 0.3 0.4 0.5
0.0 0.1 0.2 0.3 0.4 0.5
Figure 2. Quantile-quantile plot for atrazine concentrations between supply (horizontal) and infor-
mation (vertical) networks. The line represents the first bisector.
(a)
25
0
1997 2003 2002 2001 2000 1999 1998
1997 2003 2002 2001 2000 1999 1998
50
75
100
125
150
175
200
225
250
275
300
25
0
50
75
100
125
150
175
200
225
250
275
300
(b)
Figure 3. Sampling according to the type of network. (a) Base map: information (), supply ()
and producers () (b) Number of measurements per month: information (red), supply (blue), and
producers (green).
decreases after a low value has been encountered and increases for a higher concentration,
in order to control the water quality. However, after a very large value is found, a station in
the supply network will be abandoned and there are no more measurements (Fig. 4). The
sampling is, therefore, semi-preferential, in space and time, in a complex manner. The dif-
ferences observed on the mean and the quantiles (Table 2 and Fig. 2) depend upon the net-
work, the location of the stations and the sampling frequency.
6 EVOLUTION OF CONCENTRATIONS IN TIME
Seven stations provide long time series with a maximum of 24 measurements sampled
over five or six years: one belongs to the producers, four to the supply network and two to
the information network (Fig. 4). These long time series are very useful to quantify the
temporal evolution of the concentration at fixed locations. The regression, which repre-
sents the mean concentration as a function of the date, clearly shows an overall decrease in
the atrazine concentrations, but with large temporal fluctuations. It is interesting to note
that the linear regression with time, based on the seven long time series, shows almost the
same negative slope as when calculated on the whole data set. But it cannot be deduced
that the decrease is the same everywhere, as shown by the time evolution at two different
stations (Fig. 5). Table 3 summarizes the statistics on the overall annual means, calculated
on all data set.
(a)
.100
.075
.050
.025
.000
1997
(b)
2003 2002 2001 2000 1999 1998
Figure 4. (a) Long time series: producers (), information () and supply () networks
(b) Regressions of the atrazine along time: 7 long time series (simple line), whole data set (thick line).
Table 2. Statistics for atrazine concentration as a function of the type of network (considered as
erroneous, the largest supply data have been removed).
Origin Number Minimum Maximum Mean Variation Coefficient
g/ g/ g/
Information network 1319 0.003 0.76 0.050 1.17
Supply network 2831 0.000 0.28 0.040 0.80
7 RELATIONS BETWEEN SUBSTANCES
For two stations with long time series selected from the supply network, the concentrations
of several substances are drawn as a function of time, showing different levels and evolutions
according to the station (Fig. 5). The temporal evolution is rather similar for the atrazine and
its metabolite (DEA), but the two curves are not parallel. The scatter plot between these two
substances, which is calculated using all the data, confirms the correlation (Fig. 6) with a
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
J an-97 J an-98 J an-99 J an-00 J an-01 J an-02 J an-03
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
J an-97 J an-98 J an-99 J an-00 J an-01 J an-02 J an-03
Figure 5. Atrazine (), DEA (), DIA () and HyA () measured in two stations of the supply
network (left: 00163X0081 right: 00167X0003/F1. Concentrations in g/l.
Table 3. Annual mean and standard deviation per station (all data set).
Number Mean Standard deviation
1997 139 0.058 0.060
1998 364 0.048 0.030
1999 472 0.047 0.041
2000 476 0.041 0.035
2001 495 0.039 0.032
2002 627 0.031 0.021
0.4
0.3
0.2
0.1
0.0
0.0 0.1 0.2 0.3 0.4 0.5
Figure 6. Correlation between atrazine (horizontal) and DEA (vertical). Measurements from the
seven long time series are highlighted.
rather moderate correlation coefficient (0.59), reflecting the dispersion of the diagram. This
diagram shows that the largest values for the two substances do not correspond to the same
measurements. This is also true for other substances.
Principal component analysis is a standard procedure, based on the covariance matrix, to
study the relationships between variables (Saporta, 1990). Several tests made with different
data sets, (varying the type of network, the substances and the time period) show that the
results of PCA are very unstable due to the large proportion of values measured below the
quantification limit, for substances other than atrazine and DEA. The variances and covari-
ances are linked to few high values which vary according to the selected samples (per year,
per network). When some suspect high values are discarded, the PCA results are greatly
modified (de Fouquet and Renard, 2004), and it is concluded that only the correlation
between atrazine and DEA can be used for mapping purposes. When additional information
is collected (Table 1), the atrazine data can be used to improve the estimation of DEA by a
cokriging procedure. The cokriging is also useful because it allows consistent estimations
of the concentration of each substance or of their sum (Chiles and Delfiner, 1999).
8 RELATIONSHIP BETWEEN CONCENTRATION AND LAND USE
The relationship between the pesticide concentrations and some environmental character-
istics such as land use can be used to improve the accuracy of the estimations, and several
variables are usually available on a regular grid covering the area of interest. The IDPR
(Index of Development and Soil Persistence) developed by BRGM (Mardhel et al., 2005;
Mardhel et al., 2007) regroups topography, geological structures and soil lithology infor-
mation in order to characterize the aquifer vulnerability (Fig. 7). IDPR is negatively corre-
lated to the thickness of the unsaturated zone (Fig. 8).
Figure 7. IDPR calculated for the Craie du Nord aquifer system, with location of stations.
Corine Land Cover data have been used in order to provide the land use, and in particu-
lar to derive the surface of the area dedicated to corn growing, where atrazine was mostly
used. The correlation is established per functional unit which is the area of the watershed
located above the station (Mardhel et al., 2005). There appears to be no significant corre-
lation between atrazine concentration and the IDPR, the thickness of the unsaturated zone
or the land use. Conversely, the correlation between atrazine concentration and the corn-
growing areas is slightly positive (Fig. 9) but the scatter diagram is dispersed. In particu-
lar, the large values experienced in the absence of corn growing can indicate some
non-agricultural use of atrazine. Indeed, corn is grown mainly on the plateaux, where the
runoff and the transmissivity are lower than in the valleys, and the corresponding atrazine
transport should be reduced. However, a possible contribution transported by groundwater
flowing from the corn-growing areas is not excluded.
80
70
60
50
40
30
20
10
0
0 1000 2000
Figure 8. Scatter plot between IDPR (horizontal) and thickness in meters of the unsaturated zone
(vertical). The experimental and linear regressions are reported. Correlation coefficient is 0.54.
0.15
0.10
0.05
0.00
0.00 0.05 0.10
0.8
0.7
0.6
0.5
0.4
0.3
0.2
A
t
r
a
z
i
n
e
0.1
0.0
0.00 0.05
Density Mais
0.10
Figure 9. Scatter plot between atrazine concentration and corn growing. The empirical regression
is represented. Magnification for small distances (right).
9 TEMPORAL VARIABILITY OF ATRAZINE
In what follows, only the atrazine concentration is examined.
The correlation between annual means per station is high in more recent years (close to
0.75) and almost systematically decreases when the time interval increases (Table 4). This
high correlation level implies that the accuracy of the estimation map of the mean annual
concentration can be improved by taking into account data from several neighbouring
years, although this requires the calculation of a time variogram.
The variogram represents (one half of ) the mean squared difference between pairs of
data as a function of the distance or the delay (Chiles and Delfiner, 1999). It measures the
spatial or temporal variability of the concentration. As the data from the information and
supply networks lead to different statistics, the spatial and temporal variability will be
studied for each network separately.
The half-yearly periodicity of the measurements affects the number of pairs for each lag
of the temporal variogram. The large variability at the origin relates to the measurement
error variance. The time variogram presents a structure for the information network, with
a larger variance consistent with the statistics in Table 2. In the case of the supply network,
there is no structure due to the presence of a few very high concentration values (Fig. 10).
Table 4. Number of pairs (upper triangle) and correlation coefficient (lower triangle) of the mean
annual concentrations per station.
1997 1998 1999 2000 2001 2002
1997 139 87 97 77 79 86
1998 0.59 364 201 201 217 202
1999 0.35 0.66 472 236 245 272
2000 0.35 0.62 0.66 476 376 394
2001 0.31 0.54 0.65 0.75 495 396
2002 0.40 0.58 0.47 0.69 0.74 627
0 100 200 300 400 500 600 700 800
0 100 200 300 400 500 600 700 800
41
147
730
195
670
177
522
180
476
175
0.000
0.001
0.002
0.003
0.004
0.005
0.006
0.000
0.001
0.002
0.003
0.004
0.005
0.006
62
786
1974
463
1415
394
1074 890
220
0 100 200 300 400 500 600 700 800
0 100 200 300 400 500 600 700 800
0.00000
0.00025
0.00050
0.00075
0.00100
0.00000
0.00025
0.00050
0.00075
0.00100
340
Figure 10. Temporal variogram for the supply [a] and the information [b] networks. Experimental
variogram calculated for 10 lags of 90 day, and fitted model.
10 SPATIAL VARIABILITY OF ATRAZINE
The space variograms of the concentration do not show any obvious anisotropy: therefore,
only the omni-directional variogram is represented.
The spatial structure presents a longer structure for the information network than for the
supply network where the range is smaller than 5km (Fig. 11). As for the time variogram, the
sill of the space variogram is much larger for the information network, because these data bet-
ter cover the investigated and better represent the high values in particular. Therefore, the var-
iograms of the information network will serve as a reference. The influence of preferential
sampling appears clearly, modifying the sill and the range on the experimental variogram.
11 SPACE TIME VARIOGRAM MODELLING
The space and time experimental variograms are jointly fitted by a model composed of a
nugget effect, a spherical component and a linear component. A classical geometrical
anisotropy between space and time is used in order to produce a tractable space-time model.
11.1 Cross-validation per network
Can the data from the supply network (more numerous but less representative) be of any help
to improve the estimation? The answer is provided by the cross-validation technique where a
known value is temporarily removed and estimated from the remaining information. The val-
idation error is the difference between the measured and the estimated value; the normalized
validation error corresponds to the validation error scaled by the kriging standard deviation
predicted by the model. The estimation, performed by kriging, partially removes the sam-
pling artefacts, as it attributes relatively smaller weights to data in dense areas (declusteriz-
ing). Conversely, kriging does not correct the preferential sampling of low or high values.
In the first case, the concentrations at the supply measurements are estimated using data
belonging to the information network. The validation errors lie within [0.008, 0.24]
with a slightly negative mean (0.0007g/): this estimation is unbiased, with a slight
0.000
0.001
0.002
0.003
0.004
0.005
0.000
0.001
0.002
0.003
0.004
0.005
0 10 20 30 40
(a) (b)
0 10 20 30 40
0 10 20 30 40 0 10 20 30 40
0.00000
0.00025
0.00050
0.00075
0.00100
0.00000
0.00025
0.00050
0.00075
0.00100
Figure 11. Spatial variograms for the supply [a] and the information [b] networks. Experimental
variogram calculated for 10 lags of 5km and fitted model.
dissymmetry towards high values corresponding to measured high concentrations. The scat-
ter plot of this normalized validation error with time shows no systematic trend (Fig. 12).
Conversely, what is the influence of the semi-preferential supply network? In the second
case, the information measurements are estimated using data from the supply network.
The validation error lies within [0.005, 0.72] with a mean of 0.0089g/, consistent
with the statistics in Table 2 and smaller than the lowest quantification limit (0.01 g/).
The histogram of the normalized validation errors shows that the estimation quality is
poorer here than in the first case, with some values between 4 and 8, and a maximum
around 30 instead of 4 (Fig. 13).
Due to the high variability of the time-space variogram at the origin, it is essential to use
the information optimally in order to improve the estimation accuracy. When a sufficient
number of data from the information network are available, data from the supply network
can also be used in the kriging procedure. But in the area where stations from the informa-
tion network are few, the estimation made mainly from the supply network may be locally
biased. The variance map of the kriging estimation error, calculated using the information
network only, can be used to delineate these areas.
(a)
1 0 1 2 3 4
0.00
0.05
0.10
0.15
0.20
0.25
(b)
1997 1998 1999 2000 2001 2002 2003
0
5
10
Figure 12. Estimation at the supply sample using data from the information network (a) Histogram
of normalized errors (b) Scatter plot of normalized error along time.
(b)
(a)
1997 1998 1999 2000 2001 2002 2003 0 10 20 30
0.0
0.1
0.2
0.3
0.4
0
10
20
30
Figure 13. Estimation at the information sample using data from the supply network (a) Histogram
of normalized errors (b) Scatter plot of normalized error along time.
11.2 Atrazine estimation maps
The mean annual concentration is estimated on 2km by 2km cells, from data of all net-
works, using the variogram model derived from the information network and a space-time
moving neighbourhood.
The estimation maps show the global decrease of the atrazine concentration from 1997
to 2002, with some high spots whose locations change over the years (Fig. 14). The stan-
dard deviation maps show that the overall accuracy improves with time, in particular
towards the north were measurements are added gradually, whereas the sampling becomes
sparser in the south (Fig. 15).
11.3 Inter-annual difference maps
In the same way, kriging can directly provide inter-annual difference maps (Chils and
Delfiner, 1999). These differences vary greatly between two consecutive pairs of years
(Fig. 16), even if these variations must be considered with respect to the precision given by
the kriging error standard deviation maps. Moreover, the difference maps show that the
evolution over time cannot be forecasted with a great degree of confidence.
The relative accuracy (ratio of the kriging error standard deviation over the estimated
annual mean) can locally reach 100% in 1997 and 1998, 130% from 1999 to 2001 and
190% in 2002. In the mapped area, the spatial mean of the relative accuracy is about 70%
for the first two years, and it increases then up to 84% in 1999, 92% in 2000 and 2001, and
113% in 2002. This is mainly due to the decrease of the annual mean.
At present, the estimation maps should be considered with caution, especially when
compared to drinking water standards. In particular, some high concentration areas may be
1997 2000 2002
Atrazine
0.150
0.125
0.100
0.075
0.050
0.025
0.000
Figure 14. Estimation maps of the mean annual concentration.
1997 2000 2002
St. dev
0.050
0.045
0.040
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
Figure 15. Kriging standard deviation maps of the mean annual concentrations.
underestimated locally if the sampling is stopped after a high measurement or if the local
accuracy is poor. More appropriate sampling is needed to improve the accuracy, or the rel-
ative accuracy of the estimated concentrations.
12 CONCLUSIONS
The supply network is designed to monitor the potability of the water; therefore, it induces
a preferential sampling process, which is not suited to providing an accurate image of the
whole aquifer system. Conversely, the information network is more regularly sampled in
time and space but the number of stations remains insufficient for accurate mapping of the
concentration of atrazine or other pesticides. For atrazine, the decrease in concentration
with time is confirmed globally and can be mapped. Nevertheless, large amplitude fluctu-
ations are observed both over time and space. No strong relationship has been established
between the concentration and the land use in this study: this may simply reflect some non-
agricultural use of atrazine and also that the use of atrazine has been reduced during the
period of the study. The other substances are even more difficult to evaluate due to the
small number of measurements, a large proportion of them being lower than the quantifi-
cation limit. Apart for the atrazine metabolite DEA, the lack of systematic joint measure-
ments makes it impossible to highlight possible links between the concentrations of these
substances, and to perform a joint estimation. Despite inadequate sampling, a space-time
structure has been observed and modelled for the atrazine concentrations. This model has
been used through a kriging procedure in order to map the mean annual concentrations, the
inter-annual difference and the corresponding accuracies. Although kriging accounts for
the data irregularities, it cannot always correct for preferential sampling, i.e. when the
sampling is linked to the measured values (supply network).
This study leads to several recommendations on the sampling strategy, in order to qual-
ify the status of the aquifers and to provide quantitative maps of concentrations. First, as the
pesticide concentrations show a decreasing trend with time, the proportion of measure-
ments below the quantification limit becomes larger; this makes the mapping of concentra-
tions more difficult. A solution would be to reduce the quantification limit, or at least to
keep a large gap between this limit and the drinking water standard. Optimally we would
2000-1999 2001-2000
0.020
0.015
0.010
0.005
0.000
-0.005
-0.010
-0.015
-0.020
Figure 16. Inter-annual difference maps where decrease is represented in blue.
need regular sampling in space and time, covering each area of interest. Practically, given
the locations of the stations, the sampling with time must be kept as regular and systematic
as possible: concentrations should be measured at a regular frequency, independently of the
measured values. For example, measurements should be continued even when the station
has been abandoned for water supply. To study the different pesticides simultaneously, a
regular and systematic multi-substance sampling procedure is recommended. However, if
some statistical links are established, the measurement frequency for these substances can
be reduced or adapted. As the concentration of each substance tends towards low values, it
would also be interesting to estimate the sum of concentrations for families of substances.
This study has demonstrated that geostatistical techniques can be used to calculate indi-
cators (such as the annual mean concentration per station). They also produce maps which
highlight the variability of the concentrations in time and space, together with the corre-
sponding uncertainties. Finally, the geostatistical framework can be enlarged in order to
address other problems, such as the probability of a concentration exceeding a given
threshold, either locally or over a given area or a period of time (Cinnirella et al., 2005).
This study is however a work in progress, since certain assumptions have had to be made,
i.e. the depth of the sampling was not known, and the behaviour of the watershed and the
meteorological conditions were not taken into account. Combining the process modelling
and the geostatistical modelling to map pesticide concentrations first requires an impor-
tant acquisition stage, then the validation of different data.
ACKNOWLEDGMENTS
This study was performed with the financial support of IFEN, Institut Franais de
lEnvironment. The authors thank A. Blum, A. Spiteri et C. Feuillet (from IFEN) and
N. Baran, V. Mardhel et A. Gravier (from BRGM) for their valuable advice and fruitful dis-
cussions. The authors also thank the two reviewers for their helpful remarks.
REFERENCES
Amalric L., Baran N. (2004) Caractrisation de la contamination des eaux souterraines par les
pesticides phase 1. Rapport final. BRGM/RP-53341-FR, 45pp, 13ill., 1 annexe.
Blum A. (2004) Ltat des eaux souterraines en France: aspects quantitatifs et qualitatifs. Etudes et
travaux n 43, IFEN (France).
Chils J.-P., Delfiner P. (1999) Geostatistics: modelling spatial uncertainty. Wiley.
Cinnirella S., Buttafuoco G., Pirrone N. (2005) Stochastic analysis to assess the spatial distribution
of groundwater nitrate concentrations in the Po catchment (Italy). Environmental Pollution, 133:
569580.
de Fouquet C., Renard R. (2004) Etude relative la caractrisation des eaux souterraines par les pes-
ticides. Rapport technique N-09/04/G, Centre de Gostatistique, Ecole des Mines de Paris.
Mardhel V., Pinson S., Gravier A. (2005) Cartographie de la vulnrabilit intrinsque des eaux
souterraines en rgion Nord-Pas-de-Calais. (BRGM/RP-54238-FR). 113pp.
Mardhel V., Gravier A., Koch-Mathian J.Y., Nowak C., Terreyre J.L., Raguet M., Garnier C. (2007)
Simplified vulnerability mapping of groundwater in the Seine-Normandy Basin. Application to
the BASOL sites. This volume.
Saporta G. (1990) Probabilits, analyse des donnes & statistiques. Editions Technip.
CHAPTER 39
Optimisation of groundwater monitoring networks:
application of geostatistics with case studies from a
granitic aquifer in a semi-arid region
Shakeel Ahmed, Aadil Nabi, Shazrah Owais & Dewashish Kumar
Indo-French Centre for Groundwater Research, National Geophysical Research Institute,
Hyderabad, India
ABSTRACT: Groundwater, an important constituent of the hydrological cycle and a major source
of water supply, is causing concern due to over-exploitation caused by population growth and exten-
sive agricultural use. Pollution is also increasing and is threatening our ecosystems and even the life
of future generations. Groundwater needs to be managed to maintain sustainability. Monitoring is
essential of management but needs to be cost-effective. The selection of optimal groundwater moni-
toring network to observe the quality and quantity of the groundwater depends on several criteria. In
the present study, two different geostatistical approaches were used on two monitoring variables to
develop a monitoring network, which provides sufficient and useful information of these variables to
delineate their spatial distribution. The study was carried out in a hard-rock aquifer. Optimisation of
the water-level monitoring network was carried out using a geostatistical estimation variance reduc-
tion method (EVRM) in an unbiased way with a preset degree of accuracy. The monitoring network
of 57 boreholes was reduced to 40 wells, in such a way that the variance of the estimation error does
not exceed the pre-set limit. A simple method was then developed with the cross-validation technique
of geostatistical estimation to analyse and optimise an existing network for monitoring fluoride. The
fluoride values from the existing monitoring network of 60 boreholes was geostatistically analysed
and the priority and/or redundancy of each measurement point was determined to provide an optimal
monitoring network of only 30 wells. It was observed that the optimised monitoring network provided
the same variability as that attained by the original monitoring network in both the cases.
1 INTRODUCTION
The success of any scientific study greatly depends on the quality and the quantity of the
available data. Scarcity of data and their collection at isolated locations makes it necessary
to adopt special procedures or an estimation technique to make sure that field measure-
ments are consistent with data requirements. Geostatistics based on the theory of regional-
ized variables has found many applications in groundwater hydrology, including parameter
estimation and predictive modelling. Geostatistical estimation variance reduction, and
cross-validation techniques are procedures that could be used to study the adequacy of a
monitoring network and to optimise it against a set of constraints. The advantage of the
geostatistical estimation technique is that the variance of the estimation error can be calcu-
lated at any point without the actual measurement at that point. Thus, the benefits accruing
from an additional measurement can be predicted.
The work by Hughes and Lettenmair, 1981; Carrera et al., 1984; Rouhani, 1985;
Loaiciga, 1989, Gao et al, 1996 etc. are examples of the application of geostatistical tech-
niques for optimising data collection networks. However, Agnihotri and Ahmed, 1997 have
pointed out some crucial ambiguities in these applications and thus a few modifications
have been carried out to make the procedure effective and useful. The present study
describes this modified procedure using case studies carried out for monitoring network
optimisation in a fractured granitic aquifer.
There are many approaches and constraints that can be applied. In the present work, the
monitoring network was optimised for water level and fluoride concentration using differ-
ent geostatistical approaches. A monitoring network consisting of 57 boreholes of which
25 were specially drilled to monitor the water level and 32 were selected from the existing
irrigation wells, was analysed geostatistically. The variance of the estimation error
obtained using the kriging technique on the grids in the entire area was the basis for dis-
carding some of the irrigation wells for a given accuracy in the maximum permitted vari-
ance of the estimation error. Finally, after a few iterations, a network of 40 wells was
adopted. In another case, fluoride values from 60 boreholes were analysed geostatistically
and the calculated variogram was validated by a cross-validation test. It was found that
monitoring fluoride from a network of only 30 wells selected according to a priority index
could provide almost the same variability as that obtained from monitoring all 60 wells.
2 GEOSTATISTICS: THEORETICAL BACKGROUND
Geostatistics, a special statistical technique, is widely used to analyse and map the distri-
bution of a given variable and its variations in space and time. The theoretical fundamen-
tals of the geostatistical methods (based on the theory of regionalized variables) are
described in various textbooks (Journel and Huijbregts, 1978; Marsily, 1986; Isaaks and
Srivastava, 1989, etc.). Geostatistics uses variograms to characterize and quantify space-
time variability, perform rational estimates and calculate the variance of the estimation
error. A generalized formula to calculate an experimental variogram from a set of scattered
data can be written as follows (Ahmed, 1995).
(1)
where
(2)
with
(3) d
N
d
N
d
i i l
N
d
i i l
N
d d

1 1

d d d d d

( ) )
d
N
z x d z x
d
i i i l
N
d
, [ (
( ,
]
1
2
2
Where d and are the initially chosen lag and direction of the variogram with d and
as tolerance on lag and direction, respectively. d and are actual lag and direction for the
corresponding calculated variogram. N
d
is the number of pairs for a particular lag and
direction. The additional equation (3) avoids the rounding-off error of preset lag times
(multiples of the initial lag are taken only in conventional cases) and the direction. It is
very important to account for every term carefully while calculating variograms. If the
data are collected on a regular grid, and d is taken as zero, equation (2) and equation
(3) will be simplified only for .
The second step involves the estimation of the parameter at unmeasured locations x
o
using a family of kriging techniques. The equations for estimation and the variance of the
estimation error are given below in brief.
(4)
i 1.. n
where
j
are the kriging weights, is the variogram, is the Lagrange multiplier and n is
the number of measurement points.
(5)
The kriged estimate at the estimation point x
0
as a function of the kriging weights
j
cal-
culated with the linear system (45) is given by :
(6)
where Z are the values of the parameter under study, the superscript e is estimated, and
f is observed, and the subscript i or 0 refers to the location of the point. The vari-
ance of the estimation error (
k
2
) becomes:
(7)
A cross-validation test (Ahmed and Gupta, 1989) to validate the variogram, a step in-
between the two above, is also performed, and the following norms are obtained while per-
forming the cross- validation test on the variogram:
(8)
and
(9)
Z Z
i
e
i
f
i
/ 2.0 i 1.n
Z Z
i
e
i
f
0.0 i 1.n

k i
i
n
2
1

( , ) x x
i o
Z Z
e
i
i
n
i
f
0
1
j
j
n
1
1
j
j
n
(
x x x x
i j i o
, ) , )
(
1
Optimisation of groundwater monitoring networks 529
is the standard deviation of the estimation error. The cross-validation test ensures that the
variogram represents the true variability of the parameter and that it is able to reproduce
the measured values. The cross-validation is performed by masking one known value from
the data set and estimating the same from the remaining values and the final variogram. If
equation (9) is not satisfied, either the variogram does not represent the true variability or
the data could be erroneous (Ravi Prakash et al, 1990). The values of equation (8) indicate the
difference of the estimated value from the measured values whereas a low value of equation
(8) suggests that the parameter could be estimated at this point and need not be measured.
Thus, based on the values of equation (8), a priority index of measurement could be
assigned to all the measurement points starting from the highest value of equation (8).
Among the most important uses of the regionalized variables theory is the local estima-
tion by the method of kriging as given by equation (6). Kriging is a general term that
embraces several estimation procedures (Krige et al., 1989). Kriging is unique and
accepted compared with other methods of estimation because its estimates are unbiased
and have minimum variances. In this sense it is optimal. Furthermore the estimation vari-
ances themselves can be estimated (equation 7), and so the technique can be used with
known confidence. Kriging is also an exact interpolation, i.e., the kriged value at a sam-
pling point is the measured value and the variance is zero. Laslett et al., (1987) compared
kriging with other techniques of interpolation and showed that kriging was the only one
that performed reliably in all circumstances.
The estimation of variance for simple kriging equations depends on knowing only the
variogram and the position of the observation points in relation to the point or block to be
estimated; it does not depend upon the observed values themselves. This can be used in
designing sampling schemes for mapping spatial variables. Burgess et al. (1981) and
McBratney et al. (1981) computed the estimation variances for estimates at points and
over blocks on regular grids for a range of sampling intensities. They plotted the variance
against the grid spacing and then determined the optimal spacing for a given precision
from the graph. Oliver and Webster (1987) used this technique and followed it by sampling
to map the particle size distribution of the soil. Webster and Burgess (1984) showed that
the approach can also be used to optimise the location of sites in order to select bulked
samples.
3 GEOSTATISTICAL OPTIMISATION OF THE MONITORING NETWORK
It is difficult to define or generalize the adequacy of data for a particular study, but avail-
ability of adequate measurements to capture the variability of the parameter is the key to a
successful study. Abundant measurements make the study easy, but the project becomes
extremely cumbersome and expensive, whereas a smaller number of data will make the
study superficial. It is difficult but important to determine the optimal requirement of data
for any study, and it often depends on the scientific objectives. The main two objectives on
which the optimisation was based in this study were that the optimised network should be
able to:
represent the true variability of the parameter under study, and
provide its estimates on a fairly fine grid with a desired accuracy in the form of the
variance of the estimation error.
The entire area is divided into a fine grid, the variance of the estimation error is calcu-
lated with a suitable kriging technique and it is compared with the preset or desired limit
of the variance for the estimation error. Thus, depending on the outcome of the compari-
son, a network is categorized as dense, sparse or near optimal. Then, iteratively, the net-
work is optimised either by discarding, adding or shifting the measurement points.
Another procedure is cross-validation tests, where the variogram is calculated from a
larger network and after finalizing the variogram, the cross-validation table is analysed.
The measurement points are then assigned a rank in ascending order based on the differ-
ence in the measured and estimated values. This priority index is the priority of their meas-
urements. Thus, the larger the difference, the lower is the index providing high priority for
further measurements (Ahmed et al, 2002). The borehole with the smallest difference is
given the highest index, simply because the observed value at that location is accurately
estimated from the remaining measurements and thus need not be measured. The highest
index will provide lower preference for further measurements.
In both cases, the top down approach was adopted where the spatial variability is deter-
mined from a larger data set and presumed that by removing a few values the variability
remains same.
4 DESCRIPTION OF THE AREA AND THE PARAMETERS UNDER STUDY
In a small watershed of 53km
2
(Maheshwaram watershed) near Hyderabad, India, ground-
water is present in a coupled system of weathered and fractured granitic rocks. The two zones
form a single often semi-confined aquifer. Large-scale fracturing and jointing has resulted in
the formation of huge boulders of granite. The joints in the area strike in a NE-SW direction.
There is a second set of horizontal joints with a frequency of about 5 to 10m. These are
infilled with weathered material and form good aquifers with high transmissivity.
The water levels are being monitored through a network of about 57 bore-wells of which
25 were specially drilled (IFP wells) to observe a comparatively undisturbed water table
and the other 32 were irrigation wells (IFW wells) (Fig. 1). The water level measurements
were made on a monthly basis for a period of almost one year. In addition, to investigate
the water quality of the aquifer, samples from about 60 wells were taken and water was
analysed mainly for the fluoride concentration. Only wells with pumps could be used for
sampling water for chemical analyses. In order to study the variation of fluoride with time,
it was decided to optimise the monitoring network and if possible to reduce its size to per-
mit more frequent monitoring.
5 OPTIMISATION OF WATER LEVEL MONITORING OF WELLS
IN THE MAHESHWARAM WATERSHED
It was decided to reduce the number of IFW wells from the total 57 observation wells (Fig. 1)
for several reasons:
all the wells can be monitored in the shortest possible time, preferably one day,
discard some of the irrigation wells fitted with pumps as it was difficult to monitor
static levels in these wells, and
reduce the cost of monitoring without loosing the monitoring benefits.

the purpose was to obtain an optimal monitoring network with 25 IFP wells and minimize the
IFW wells such that the kriging estimation of the water levels provides a standard deviation of
the estimation error of not more than 8m (against the average standard deviation of 12m of
the water level data) for the entire area. Through a special procedure, the IFW wells were
removed one by one and the resulting variances of the estimation error all over the area were
compared with the preset value. A number of combinations of boreholes were tried, particu-
larly where they were grouped. Finally, a network with 25 IFP wells and 15 IFW wells was
established for monitoring the water levels every month. It is clear that with the optimised
monitoring network, it is still possible to maintain the same magnitudes of
k
. Fig. 2 shows the
location of the wells in the optimal monitoring network. The area was further divided into
much finer grids and the entire procedure was repeated to verify that the same network holds
good for the finer grids as well. Such a comparison of resulting
k
is shown in Table 1.
6 RANKING OF MONITORING WELLS FOR FLUORIDE MEASUREMENTS
During July 2001, groundwater samples for fluoride analysis were collected from 60 wells
fairly evenly distributed in the area. The fluoride concentration in this area varies from 0.5 to
2.97mg/l with a mean value of 1.6mg/l. Although the area is not greatly affected by fluoride
0 1000 2000 3000 4000 5000 6000 8000 7000 9000 10000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
IFP-1
IFP-2
IFP-3
IFP-4
IFP-5
IFP-6
IFP-7
IFP-8
IFP-9
IFP-10
IFP-11
IFP-12
IFP-13
IFP-14
IFP-15
IFP-16
IFP-17
IFP-18
IFP-19
IFP-20
IFP-21
IFP-22
IFP-23
IFP-24
IFP-25
IFW-1
IFW-2
IFW-3
IFW-4
IFW-5
IFW-6
IFW-7
IFW-8
IFW-9
IFW-10
IFW-11
IFW-12
IFW-13
IFW-14
IFW-15
IFW-16
IFW-17
IFW-18
IFW-19
IFW-20
IFW-21
IFW-22
IFW-23
IFW-24
IFW-25
IFW-26
IFW-27
IFW-28
IFW-29
IFW-30
IFW-31
IFW-32
Figure 1. Location of observation/monitoring wells (57 wells).
problems, the situation is alarming as the mean exceeds 1.5mg/l, the WHO upper limit for
drinking water. The experimental variogram calculated with the fluoride values was fitted on
a theoretical (called spherical) model. The variogram parameters are reported below.
A cross-validation test (Ahmed and Gupta, 1989) was performed as described in section 2.
( ) 0.1 0.25
3
2 2000
1
2
(2000)
3
d
d d

3
1
)
d
d

2000
0.1 0.25 2000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
IFP-1
IFP-2
IFP-3
IFP-4
IFP-5
IFP-6
IFP-7
IFP-8
IFP-9
IFP-10
IFP-11
IFP-12
IFP-13
IFP-14
IFP-15
IFP-16
IFP-17
IFP-18
IFP-19
IFP-20
IFP-21
IFP-22
IFP-23
IFP-24
IFP-25
IFW-1
IFW-3
IFW-8
IFW-9
IFW-11
IFW-13
IFW-17
IFW-18
IFW-20
IFW-23
IFW-26
IFW-27
IFW-29
IFW-31
IFW-32
Figure 2. Location of optimised observation/monitoring wells (40 wells).
Table 1. Comparison of
k
for the existing and optimised network.
Grid Total measurement point MIN. MAX. MEAN VARIANCE
500m by 500m 57 1.56 10.30 5.03 3.78
500m by 500m 40 2.34 10.17 5.32 3.19
250m by 250m 57 1.53 11.82 5.49 3.84
250m by 250m 40 1.57 11.56 5.77 3.26
1000 2000 3000 4000 5000 6000 7000 8000 0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Counter Interval =0.5 m
2
3
5
6
8
9
11
Figure 3. Sigma (in meters) with 57 measurement wells (500m by 500m).
1000 2000 3000 4000 5000 6000 7000 8000 0
1000
2000
3000
4000
5000
6000
7000
8000
9000
Contour Interval =0.5 m
3
4
6
7
9
10
Figure 4. Sigma (in meters) with 40 measurement wells (500m by 500m).
Table 2. Priority index for selecting wells for the measurement of Fluoride.
Priority Well X in m Y in m F F Difference Diff/
index No. measured Estimated
51 1 5902.212 3344.458 1.97 1.897 0.0726 0.4507 0.1611
8 2 6219.405 2769.202 1.54 2.284 0.744 0.4627 1.6079
31 3 5094.45 3066.748 1.47 1.726 0.2561 0.4688 0.5463
10 4 5508.345 2391.547 1.81 2.372 0.5619 0.4083 1.376
43 5 5822.104 2147.406 1.73 1.854 0.1245 0.4536 0.2744
1 6 5636.137 2572.363 2.82 1.731 1.089 0.4052 2.6872
53 7 3222.197 4091.376 1.14 1.208 0.0678 0.415 0.1633
50 8 2961.081 4091.376 1.2 1.117 0.0833 0.4031 0.2067
26 9 1552.695 3613.776 1.89 1.527 0.3631 0.4376 0.8296
20 10 1477.736 3974.646 1.27 1.726 0.4561 0.4408 1.0347
58 11 3680.533 5991.858 1.51 1.531 0.0209 0.3994 0.0524
36 12 3794.402 6630.438 2.06 2.238 0.1775 0.368 0.4824
40 13 3830.45 6580.084 2.18 2.038 0.1419 0.3675 0.3862
9 14 3783.339 7058.447 2.46 1.878 0.5825 0.4571 1.2742
28 15 4516.714 7441.443 1.43 1.716 0.2856 0.3906 0.7312
16 16 5311.125 7380.408 1.76 2.248 0.4876 0.3474 1.4037
38 17 6480.521 7471.96 1.16 1.312 0.1524 0.3499 0.4356
47 18 6400.03 7816.809 1.26 1.148 0.1121 0.4474 0.2505
6 19 6530.493 5791.205 0.7 1.615 0.9152 0.5374 1.703
52 20 6366.652 3625.22 1.76 1.828 0.0681 0.3592 0.1896
33 21 6205.481 4360.693 1.54 1.774 0.234 0.4273 0.5476
44 22 5983.275 4710.883 1.02 1.138 0.1179 0.4537 0.2597
55 23 6394.499 3644.293 1.81 1.84 0.0295 0.359 0.0822
5 24 7313.841 3227.728 2.14 1.213 0.9271 0.4917 1.8854
30 25 7416.647 3830.45 1.6 1.329 0.2712 0.3624 0.7483
41 26 7383.269 3916.663 1.51 1.651 0.1407 0.3776 0.3726
24 27 7472.151 3835.791 1.21 1.627 0.4173 0.3602 1.1585
23 28 7533.377 3888.434 1.65 1.233 0.4175 0.3755 1.1118
57 29 4955.595 4810.828 1.15 1.129 0.0212 0.4182 0.0506
19 30 4858.321 5038.946 0.73 1.193 0.4625 0.4083 1.1328
14 31 4749.983 5569.189 1.21 0.706 0.5043 0.4514 1.1173
11 32 3663.939 6252.783 2.32 1.765 0.5554 0.3823 1.4527
60 33 3508.299 6283.301 1.97 1.975 0.0048 0.3762 0.0129
32 34 3469.389 6177.252 1.65 1.897 0.2472 0.3814 0.6482
3 35 5827.826 6402.319 2.69 1.657 1.0325 0.3624 2.8494
2 36 5786.055 6393.927 1.54 2.574 1.034 0.3608 2.8655
29 37 5722.159 6530.493 2.74 2.457 0.2828 0.3695 0.7654
13 38 5566.71 6499.976 1.81 2.335 0.5246 0.3894 24
22 39 5669.325 6618.994 2.97 2.543 0.4274 0.3776 1.1319
42 40 6488.913 7469.672 1.3 1.162 0.1375 0.35 0.393
56 41 7174.033 8463.782 1.26 1.289 0.0291 0.5541 0.0526
45 42 4127.806 3596.991 1.02 1.137 0.1167 0.3674 0.3177
25 43 2411.002 4243.964 0.9 1.301 0.401 0.4061 0.9875
35 44 2247.16 4458.35 1.16 1.35 0.1895 0.5065 20
27 45 2213.782 4546.851 1.49 1.159 0.331 0.382 0.8665
54 46 2680.51 3891.486 1.31 1.28 0.0301 0.4085 24
21 47 2230.566 3774.756 1.92 1.467 0.4532 0.4196 1.08
59 48 1858.252 3833.502 1.69 1.706 0.0156 0.4114 21
7 49 4449.957 4736.06 1.92 1.102 0.818 0.454 22
46 50 3833.312 4163.855 1.27 1.383 0.1131 0.4537 0.2494
(Continued )
Table 2. (Continued)
39 51 4642.981 7420.081 1.65 1.504 0.1458 0.3883 0.3755
15 52 5310.935 7378.119 2.25 1.762 0.4878 0.3474 1.4043
48 53 4190.939 3590.125 1.15 1.04 0.11 0.3693 0.2978
12 54 3888.052 6091.803 1.22 1.769 0.5488 0.4002 1.3712
34 55 4290.884 6413 1.64 1.437 0.2026 0.4283 20
18 56 4720.038 6204.718 0.96 1.425 0.4646 0.4439 1.0468
17 57 4354.971 2993.506 0.67 1.14 0.4696 0.462 1.0166
49 58 4087.942 3417.7 1.01 0.92 0.0898 0.3984 27
37 59 2814.978 3468.054 1.28 1.444 0.1641 0.4628 0.3546
4 60 6417.959 3960.913 2.72 1.728 0.9917 0.4167 2.3797
The number of wells to define the desired size of the monitoring network could be
decided on the basis of the available resources including man-power and analytical facili-
ties. Then a network could be prepared from Table 2 by picking the wells with decreasing
priority. Table 3 shows the statistics from the measured values for comparing the various
monitoring networks. It is clear that the measured minimum and maximum values are
present in all cases. The mean and the variance increase consistently as the number of
measurement points decreases. However, the change in the mean value is almost negligi-
ble. Figs. 5 and 6 show the contour lines for fluoride concentration based on the 60 and 30
measurements, respectively, from the network designed on the basis of the priority index.
The two contour lines show similar distributions and the regionalized picture is almost
identical.
7 CONCLUSION
The constrained optimisation of the monitoring network of the water levels with only 40
wells will ensure that all the wells are measured in the shortest possible time every month.
The revised network includes all the 25 wells without pumps and care has to be taken only
when monitoring the 15 private wells that are fitted with pumps for irrigation. This reduces
the effort of collecting the water level measurements. It was verified that the revised
network provides comparable accuracy in the form of the variance of the estimation error
as that obtained from the network of 57 measurements.
Table 3. Statistics of measured values of F for different networks.
Network size Statistics on the parameter values in mg/l
Minimum Maximum Mean Variance
60 (July 2001) 0.67 2.97 1.60 0.286
55 0.67 2.97 1.61 0.304
50 0.67 2.97 1.61 0.324
45 0.67 2.97 1.66 0.336
40 0.67 2.97 1.70 0.352
35 0.67 2.97 1.71 0.378
30 0.67 2.97 1.74 0.426
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
1
2
3
4
5
6
7 8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32 33
34
35 36
37 38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Iso-values of Fluoride content with location of 60 observation wells.
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
6
36 35
60
24
19
49
2
14
4
32
54
38
31
52 16
57
56
30
10
47
39
28
27
43
9
45
15
37
25
Figure 6. Iso-values of Fluoride content with location of 30 observation (Optimised) wells.
The other case study demonstrates a simple and new method developed using the geo-
statistics cross-validation technique to analyse and optimise an existing network for mon-
itoring fluoride concentrations. Using the result of the geostatistical cross-validation, it
was possible to assign a measuring priority index to all the measurement points and
depending on constraints such as finance and logistics, the network could be reduced in
size without incurring negative results. A number of monitoring networks with systemati-
cally reduced numbers of wells were prepared and compared.
In conclusion, it can be affirmed that the above techniques provide an unbiased and quan-
titative approach to arrive at an optimal network of measurement points. One can optimise
a monitoring network in a qualitative way but the optimisation of a measuring network
depending on many geological and hydrogeological parameters in the objective function as
numbers, was not previously possible. Thus, geostatistical techniques provide in a quantita-
tive framework an optimal or near-optimal network that can be used for practical purposes.
REFERENCES
Agnihotri, V., Ahmed, S. (1997) Analysing ambiguities in the data collection network design by
Geostatistical estimation variance reduction method, Jour. Environ. Hydrology, 5, paper 5,
(http://www.hydrweb.com).
Ahmed, S. (1995) An interactive software for computing and modeling a variograms. In Mousavi
and Karamooz (eds.) Proc. of a conference on Water Resources Management (WRM95),
August 2830. Isfahan University of Technology, Iran, 797808.
Ahmed, S., Gupta, C.P. (1989) Stochastic Spatial Prediction of Hydrogeologic Parameters: Use of
cross-validation in Krigings, In Proc. of Internat. Groundwater Workshop, Hyderabad, India,
Feb.March, 1989, (Gupta et al. eds.), Oxford and IBH Pub. Co., Vol III, p. 7790.
Ahmed, S., Betrand, F., Saxena, V.K., Subrahmanyam, K., Touchard, F. (2002) A Geostatistical
Method of determining priority of measurement wells in a Fluoride Monitoring Network in an
Aquifer. J. of Applied Geoschemistry, 4, 2, 576585.
Burgess, T.M., Webster, R., McBratney, A.B., (1981) Optimal interpolation and isarithmic mapping
of soil properties: IV Sampling strategy. J. Soil Sci. 32, p. 643659.
Carrera, J., Usunoff, E., Szidarovsky, F. (1984) A method for optimal observation network design for
groundwater management. J. Hydrol., 73, 147163.
Deutsch, C.V., Journel, A.G. (1992) GSLIB, Geostatistical software library and Users guide. Oxford
Univ. Press, New York, 340 p.
Gao, H., Wang, J., Zhao, P. (1996) The updated kriging variance and optimal sample design: Math.
Geology, v. 28, no. 3, p. 295313.
Hughes, J.P., Lattemair, L. (1981) Data requirements for Kriging: Estimation and Network design.
Water Resour. Res., 17,6, 16411650.
Isaaks, E.H., Srivastava, R.M. (1989) Applied Geostatistics, Oxford Univ. Press, 561 p.
Journel, A., Huijbregts, C. (1978) Mining geostatistics. Academic Press, New-York.
Krige, D. G., Guarascio, M., Camisanicalzolari, F. A. (1989) Early South African geostatistical techniques
in todays perspectives. Armstrong, M. (ed.) Geostatistics, volume I (Kluwer, Drodrecht) p.119.
Laslett, G. M., McBratney, A. B., Pahl, P.J., Hutchinson, M.F. (1987) Comparison of several spatial
prediction methods for soil pH. J. Soil Sci. 38, p. 32541.
Loaiciga, H.A. (1989) An optimisation approach for Groundwater quality monitoring network
design. Water Resour. Res. 25, 8, 17711782.
Marsily, G. de (1986) Quantitative Hydrogeology, Groundwater Hydrology for Engineers, Academic
Press, 440 p.
Matheron, G. (1971) Theory of Regionalized Variables and their Applications. Cahier du C.G.M.M.,
Fontainebleau, France, 211 p.
McBratney, A.B., Webster, R., (1981) The design of optimal sampling schemes for local estimation
and mapping of regionalized variables: II. Program and examples. Comput. Geosci. 7, p. 331334.
Oliver, M.A., Webster, R. (1987) The elucidation of soil pattern in the Wyre Forest in the West
Midlands, England. II. Spatial distribution. J. Soil Sci. 38, p. 293307.
Ravi Prakash, M, Ahmed, S., Gurunadha Rao, V.V.S. (1990) Stochastic Analysis of Transmissivity
and Specific Yield obtained from Dug Well Pumping Tests, In Proc. of Groundwater Investi-
gation, Management and Geophysical Technique, Lucknow, India, Dec, 1990.
Rouhani, S. (1985) Variance reduction analysis, Water Resour. Res., 21, 6, 837840.
Wackernagel, H. (1995) Multivariate Geostatistics: An introduction with applications. Springer,
256 p.
Webster, R., Burgess, T.M. (1984) Sampling and bulking strategies for estimating soil properties in
small regions. J. Soil Sci. 35, p. 12740.
CHAPTER 40
Use of piezometric networks for evaluating the quantitative
status of groundwater bodies: Example of the
Seine-Normandy Basin in France
Jean-Francois Vernoux
BRGM Service EAU, B.P. 36009, 45060 Orlans Cedex 2, France
ABSTRACT: Piezometric time series data are the basic element necessary, although not always
sufficient, for evaluating the quantitative status of groundwater bodies. This data must enable trends
in the evolution of the water status, and in particular degradation, to be highlighted so that the neces-
sary measures can be taken for sustainable management of the resource and to maintain the right bal-
ance between withdrawals of groundwater and preservation of the environment. The piezometric
data must, therefore, be compared to water withdrawal data. In addition, the choice of piezometers
is fundamental to the overall characterisation of groundwater bodies, given that most are dissimilar
both in terms of structure and hydrodynamics. This paper presents the example of the piezometric
network in the Seine-Normandy Basin, installed at the end of the 1960s by the BRGM. Different
steps have resulted in the constitution of a relatively dense network (more than 350 stations), but
with points unequally distributed throughout the basin and, in particular, of widely variable useful-
ness vis--vis the objective of evaluating the quantitative status of groundwater bodies. Work has
recently been carried out in collaboration with DIREN (French Regional Environment Department)
to analyse and rank the piezometers monitored by the BRGM. The ranking criteria were as follows:
time series greater than 30 years, comprehensible hydrogeological operation, influence of water-
courses, influence of pumping operations, uniqueness of the exploited aquifer and status with regard
to the risk of silting up or drying up. This work will make it possible to define the piezometric net-
work for monitoring the quantitative status of groundwater bodies as defined by the Water
Framework Directive. The last part of the paper presents an example of the use of piezometric data
for the management of groundwater in drought periods.
1 NEW PIEZOMETRIC NETWORK OPERATING REQUIREMENTS
The European Water Framework Directive (WFD), dated December, 22, 2000, sets out the
framework for the management and protection of water on a European scale along major
catchment basin lines. It requires Member States to attain good water status (surface water,
fresh water, coastal water and groundwater) before the end of 2015. But for groundwater
bodies, the WFD also requires, as a review of impacts of changes in groundwater levels,
the identification of those bodies of groundwater for which lower objectives are to be spec-
ified as a result of consideration of the effects of the status of the body on (1) surface water
and associated terrestrial ecosystems, (2) water regulation, flood protection and land
drainage, and (3) human development. The Directive implies the reinforcement of moni-
toring networks in order to:
Better understand resources in three respects, namely quantitative, qualitative and

anthropic effects, in order to be able to make informed decisions.
As a precautionary measure, monitor changes of all economically and ecologically rel-

evant resources, in order to ensure a sustainable, balanced and equitable supply of good
quality water.
Check the effectiveness of corrective actions to restore resources to their proper condi-
tion, wherever they have been implemented.
In France, in order to comply with this EU commitment, the MEDD (Ministre de
lEcologie et du Dveloppement Durable, French Ministry of Ecology and Sustainable
Development) initiated the set-up of a Water Information System (MEDD Ministerial
Circular dated March 26, 2002) that implements water status monitoring programmes in
order to build up a complete and coherent picture of each hydrographic district. This moni-
toring of groundwater bodies concerns not just the quantitative aspects but also the quali-
tative aspects, with the following requirements:
WFD compliant.
Provisions for monitoring the action plans provided for by the WFD.
Optimisation of policing water.
Alert measures vis--vis floods and droughts.
Informing the public.

These requirements have been formalised by a Cahier des charges pour lvolution des
rseaux de surveillance des eaux souterraines en France (Specification for updating
groundwater monitoring networks in France) issued by the MEDD through the WFD
Circular 2003/07 dated October 8, 2003 and completed by the WFD Circular 2005/14
dated October 26, 2005. This specification sets out the new monitoring network operating
requirements and provides for a major improvement to the existing networks, focusing on
three main aspects:
A considerable increase in the number of measuring points to assure that all water bod-
ies are covered on a national basis (Table 1) and, in particular, water resources subject
to crisis situations (high exploitation, floods, low water levels, etc.).
Information made available to the public more rapidly.
Cost control, through efficient technical and organisational choices.

These choices lead to new constraints regarding the design of the monitoring system. For
the quantitative monitoring of groundwater resources, this extends from piezometric level
measuring equipment to the dissemination of data on the Internet, with the use of process-
ing tools for analysing and managing alerts. This problem is common to other monitoring
networks, even though monitoring of the quantitative status (levels and flow rates) with
current techniques lends itself more easily to automation than qualitative network
measurements.
In the case of quantitative status monitoring networks, fluctuations in quantity are
directly linked either to fluctuations in the flow rate of a spring outflowing from the aquifer,
or to variations in the hydraulic head of the aquifer, which can be determined by measuring
water level variations in wells or boreholes. The measurement of the flow rate of a spring by
measuring the height or flow stage of water, or the measurement of the water level in a well
or borehole, are physical measurements that are already extensively automated. The data
stored on digital recorders is easily accessible for long distance data transmission.
2 IMPLEMENTATION OF PIEZOMETRIC MEASUREMENT
NETWORKS IN FRANCE
Piezometric networks began to be developed in the late 1960s as a result of local initiatives,
each with widely varying station densities and measurement frequencies. The necessity of
having at ones disposal a more accurate and reliable national network was formalised in
1999 with the creation of a French National Groundwater Network (RNES, Rseau
National des Eaux Souterraines), the outcome of a protocol between the French Ministry of
the Environment and the Water Authorities. The Circular dated October 26, 2005 stipulates
that the RNES will, in the longer term after restructuring, constitute the future WFD moni-
toring control network. This involves a meta-network corresponding to a series of points
belonging to elementary networks. These elementary networks meet a common objective,
but do not necessarily have the same contracting owner. The objective is to identify the sta-
tions in the elementary networks that will form the national network for monitoring the
quantitative status of groundwater bodies as defined by the WFD. This work has been car-
ried out in certain basins but has to be completed and harmonised at a national level before
the end of 2006. The creation of ADES (Accs aux Donnes des Eaux Souterraines, French
National Data Bank for Access to Groundwater Data) has made it possible to identify a
certain number of elementary piezometric networks, locate the measuring stations and
gain access to the results of quantitative measurements. The ADES now centralises the
data from networks installed by the Water Authorities, DIREN (Direction Rgionale de
lEnvironnement, French Regional Environment Departments), the BRGM (Bureau de
Recherches Gologiques et Minsires, French Geological Survey), local authorities and
public institutions. As of April 15, 2006, 67 quantitative monitoring networks were
Use of piezometric networks for evaluating the quantitative status of groundwater bodies 543
Table 1. Minimum density by type of water body as recommended by
the French ministry of ecology (MEDD, 2003)
Type of water body Minimum density (n/km
2
)
Sedimentary
Free and confined aquifer associated
Free acquifer dominant 1/500
Confined acquifer dominant 1/3000
Free and confined aquifer dissociated
Confined aquifer 1/3000
Free aquifer
Karst 1/500
Non Karst 1/500
Alluvium 1/500
Basement 1/7000
Volcanic edifice 1/7000
Intensely folded 1/7000
registered in ADES with 2833 measuring stations, 2592 of which are located in metropol-
itan France (Figure 1).
3 BACKGROUND AND OPERATION OF THE SEINE-NORMANDY
BASIN NETWORK
In 1996, following the recommendations of a report about sustainable management of
underground water (Martin, 1996), the Ministry of the Environment invited the Water
Authorities to take over as contracting owners and to boost their funding for networks pro-
viding information on groundwater on a national scale. In 1997, the AESN (Seine-Normandy
Water Authority) entrusted BRGM with the management of the piezometric network of
the Seine-Normandy Basin within the framework of an annual agreement, which was
renewed up to 2002. This period was used to assure the long-term viability of the network
and to improve it through the installation of measuring equipment (pressure sensors, data
acquisition units, modems) and local restructuring of the network (elimination of non-
relevant points, transfer from the basin network to local networks or vice versa). On the
other hand, since the network already comprised more than 200 measuring points, few new
points were created. Installing measuring equipment has made it possible to cut the data
Figure 1. Map of French quantitative monitoring stations (piezometers and springs) registered
in ADES.
availability time and it can now be accessed on-line on a dedicated and still fully opera-
tional website (http://seine-normandie.brgm.fr ).
The MEDD Circular dated March 26, 2002, complying with the WFD, inaugurated a
new organisation for managing and monitoring water data. This Circular appointed the
BRGM and DIREN contracting owners of the piezometric networks. The Seine-Normandy
basin network was expanded from 218 measuring stations in 2001 to 270 stations at the
start of 2006, to which must be added local authority networks, i.e. a total of 323 measur-
ing stations (Figure 2). Over the last four years, a considerable effort has been made to
modernise the equipment and equip stations with long distance data transmission facili-
ties, thus enabling them to be interrogated every two weeks, or even every week, and
thereby have access to data virtually in real time and thus anticipate crisis situations (cf.
example below of the use of piezometric data to manage drought problems in the Ile-
de-France, or Verjus, 2007).
As regards piezometers under BRGM contracting ownership, the process of recovering,
validating and loading the data in the data bank may be summarised as follows (Figure 3):
Interrogation of stations with long-distance data transmission facilities every two

weeks: operational check of the data acquisition unit; on-site intervention if necessary,
loading of data in the BRGM data bank.
During field inspections (every three months for stations equipped with data acquisi-
tion centres and every six months for stations equipped with long-distance transmis-
sion data acquisition units): probe measurement and sensor test; calibration of the
sensor, if necessary; recovery of data; loading in the data bank.
Figure 2. Status of piezometric networks in the Seine-Normandy Basin as of end 2005, with the
different contracting owners and delimitation of underground water bodies
While loading data in the data bank: correction of drift (if any drift in the sensor has been
observed); allocation of a data validity code, in accordance with the specifications defined
by SANDRE (French Data Reference Centre for Water, http://sandre.eaufrance.fr/).
In the case of manual stations: one probe measurement every month, with the measure-
ments fed into the data bank.
Data from data producers other than the BRGM are loaded into the data bank when
received.
4 DEFINITION OF THE WFD MONITORING NETWORK IN
THE SEINE-NORMANDY BASIN
The Seine-Normandy Basin piezometers registered in ADES are grouped together into
elementary networks, each managed by a unique contracting owner, including the BRGM
network, DIREN networks and local authorities networks. These elementary networks are
themselves grouped together for the purposes of management and network reporting into
meta-networks that may only include part of the previous network points, chosen to meet
the specific requirements of the meta-network grouping. The main ones are, by increasing
order of aggregation:
Departmental or regional networks stemming from the elementary networks of local
authorities, the BRGM or DIREN,
The basin network, under the responsibility of basin DIREN.
These meta-networks aggregate the data from stations managed by the different contract-
ing owners selected with a view to specific network reporting.
The national network for monitoring the quantitative status of groundwater corresponds
to the upper aggregate at the national level of points selected from basin networks. It is
intended for French State reporting to the European Commission under the responsibility
Archived in
BRGM
groundwater
data bank
Piezometer
monitored by the
BRGM
Data
recovery
every 3 or 6
months
Interrogation
every 2
weeks
Local Validation
Internet access
http://www.ades.eaufrance.fr
Other producer of data
Figure 3. Data production circuit of the Seine-Normandy Basin piezometric network
of the MEDD. This work of defining the basin meta-network is underway. All of the 244
piezometers managed by the BRGM have been ranked (Putot, 2006). Grouped together by
water bodies, the piezometers were analysed and compared by taking into account not just
the condition of the piezometer itself but also the quality of its time series, whether it is
influenced or not and its representativeness vis--vis the body of water with which it is
associated. In order to characterise the condition of the piezometers and assign them a
quality rating, several criteria were taken into consideration:
The length of the time series.
The reference levels (very low / very high water levels).
Good correlation of the time series with modelling (when it exists).
Correct understanding of the piezometer behaviour (based on modelling, correlation
with neighbouring points or under the same conditions, etc.).
The type of fluctuation (multiyear, mixed, seasonal, etc.).
Any external influences (pumping, surface network).
The major exploited aquifers (unique or not).
The status of the aquifer (unconfined or confined).
The depth of the piezometer.
The condition of the station (dry, silted up, blocked, etc.).
Out of the 244 studied piezometers, 212 have a sufficiently long time series for being
assigned a rating (Figure 4). The general trend is towards good monitoring point quality.
More than half of these are considered as good or very good. 5/6th of the piezometers are
suitable for monitoring water bodies. The remaining 15% include piezometers that are not
13%
30%
34%
20%
2%
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
1 2 3 4 5
Figure 4. Distribution of ratings allocated to piezometers.
representative of a water body (but which may be interesting on a local scale), those where
the station is in poor condition and requires overhaul work and those that will have to be
abandoned or replaced.
This ranking must firstly be validated at the local level, in particular for stations
assigned an intermediate rating, and secondly integrate a selection of stations under
DIREN and, if necessary, local authorities contracting ownership.
5 CHARACTERISATION OF THE QUANTITATIVE STATUS OF GROUNDWATER
BODIES IN THE SEINE-NORMANDY BASIN
The inventory carried out by the Water Authority (AESN, 2004) shows that no water body
in the basin is showing a sustained downward trend. However, those water bodies at greater
risk can be determined by comparing replenishment and withdrawal rates. Figure 5 shows
that a high proportion of the minimum replenishment of aquifers (observed over the last 30
years) is taken up by withdrawals (50 to 100%) on several water bodies. For four water bod-
ies, withdrawals even account for more than 100% of the minimum replenishment.
These trends at the water body scale nevertheless mask disparities that are observable
through the piezometric time series. Figure 6 illustrates these disparities in the case of
groundwater body 3103 (Brie Champigny Limestone) with 4 piezometers showing dif-
ferent trends. These differences are due to heterogeneous withdrawal distribution. These
phenomena must be taken into account when characterising the status of the water body
and its medium- and long-term evolution with, in parallel, an accurate evaluation of with-
drawals for each groundwater tapping point.
Figure 5. Impact of withdrawals on groundwater in the Seine-Normandy Basin (source:
Seine-Normandy Water Authority)
6 AN EXAMPLE OF THE USE OF PIEZOMETRIC DATA: MANAGEMENT OF
GROUNDWATER IN DROUGHT PERIODS IN ILE-DE-FRANCE
As part of a study funded by DIREN Ile-de-France, we used part of the piezometric series
supplied by the monitoring network to elaborate groundwater resource management aid
tools, particularly during drought periods (Normand et al., 2005, Verjus, 2007). These
management aid tools are of two types:
rainfall-piezometric level transfer models, which, for each piezometer, enable pre-
dictions to be made after a phase of calibration on the observations, by generating sev-
eral hundred climatic series (rainfall potential evapotranspiration) and by carrying
out a frequency analysis of the corresponding levels.
probabilistic models constructed from piezometers selected for predicting levels.

These models are built up from a frequency analysis at a monthly time step for the
piezometers by adjustment of a probability distribution function.
From all of the available piezometers managed by the BRGM Ile-de-France, those in
which the time series are too short, those that are too influenced by pumping operations,
those that exploit an alluvial aquifer (and thus influenced by a water course) and those that
exploit the deep and confined Albian Aquifer were ruled out. Certain piezometers were
also ruled out because their operation is poorly understood, which means that their
Rupreux
151
152
153
154
155
156
157
158
159
160
161
01/01/1968 01/01/1978 02/01/1988 02/01/1998
Signy-signets
91
92
93
94
95
96
97
98
99
100
01/01/1968 01/01/1978 02/01/1988 02/01/1998
Saint-Martin Chennetron
115
120
125
130
135
140
145
01/01/1968 01/01/1978 02/01/1988 02/01/1998
Montereau-sur-le-Jard
46
47
48
49
50
51
52
53
54
55
56
01/01/1968 01/01/1978 02/01/1988 02/01/1998
Figure 6. Evolution trends of the piezometric level for 4 piezometers belonging to the same water
body (3103)
behaviour cannot be explained on the basis of a rainfall level model due to significant dis-
ruptive elements, the analysis of which is beyond the scope of this study.
Twenty piezometers were selected and subjected to rainfall-level modelling followed
by projected simulations with the use of a generator enabling stochastic sequences of rainfall
and PE to be produced for the period of prediction (Figure 7). Regular updates of rainfall,
110
115
120
125
130
135
140
145
150
1990 1994 1998 2002 2006
Saint_Martin_C Modle 0.1
0.2 0.5 0.8
0.9 0.95 0.975
Figure 7. Saint Martin Chennetron piezometer: short-term prediction and adjustment of monthly
piezometric levels for December to a 3 parameter lognormal distribution
Figure 8. Example of the four thresholds taken into account in local authority orders in the Val
dOise, and actual piezometric fluctuation, Chalk aquifer, in Buhy, February to May 2006.
PE and piezometric level data enable short- and medium-term (one or two years) predic-
tions to be made, represented in the form of quantiles corresponding to the various return
periods. Probabilistic models constructed from piezometers selected for predicting levels
have been built from frequency analysis at a monthly time step of the piezometric data by
adjustment of a probability distribution function.
This statistical analysis is used by DIREN to define the thresholds likely to be taken into
account in local authority orders restricting water use. Four thresholds were defined by
DIREN. The vigilance threshold corresponds to a return period of about three years and
the serious crisis period corresponds to a return period between ten and twenty years
(Figure 8).
Piezometric levels:
Vigilance 44.40ma.s.l Crisis 43.50ma.s.l.
Alert 43.90ma.s.l. Serious crisis 43.00ma.s.l.
7 CONCLUSIONS
Measuring networks (management, alerts, etc.) are increasingly necessary for water man-
agement requirements, both at national and European levels (Le Nir, 2006). A new operat-
ing design is currently being finalised to meet the latest needs:
Additional measuring points for improved monitoring and sustainable management of
water resources with regard to human stresses.
Digital recording for an appropriate measuring frequency to better understand natural
phenomena.
Long-distance data transmission for the operational control and rapid dissemination of
information through on-line data banks.
Greater autonomy and reliability of the measuring systems in order to cut field work
costs.
Cost control by reducing the amount of field work required through the use of the most
modern technologies to deal with the optimum distribution of points and measurements
and the necessity of making better operational use of the data, including making it more
available to the public.
This finalisation of monitoring networks is only one step in a programme clearly marked
out by the timescale of the European Water Framework Directive, which requires a high
level of sustained reactivity over an extended period:
end of 2004: presentation of the status regarding the different uses of water and their
impact on water bodies; definition of water bodies where there is a risk that the environ-
mental objectives will not be reached by 2015; definition of protected areas subject to
special protective provisions.
end of 2006: finalisation of monitoring networks which will provide measurements
enabling aquatic environment quality to be compared between Member States.
end of 2009: presentation of a management plan setting out the objectives to be reached
in 2015, associated with a measurement programme (regulatory, financial, voluntary
agreements, etc.) able to meet these objectives.
end of 2015: review of reached objectives.
REFERENCES
AESN (2004) Etat des lieux du bassin Seine et cours deau ctiers normands, au titre de la directive
cadre europenne sur leau 2000/60/CE. AESN, Prfecture dIle de France, DIREN Ile de France,
120 pp. annexes (155 pp.) atlas cartographique (22 cartes).
Martin Y. (1996) Rapport sur la gestion durable des eaux souterraines. Conseil Gnral des Mines,
74 pp.
MEDD (2003) Cahier des charges pour lvolution des rseaux de surveillance des eaux souter-
raines en France, Circulaire DCE 2003/07 du 8 octobre 2003, 117 pp.
MEDD (2005) Complment au cahier des charges sur lvolution des rseaux de surveillance en
France, texte rsum, 17 Juin 2005, 21 pp.
Le Nir M. (2006) Rseaux pizomtriques en France: objectifs et utilisation. Gochronique, 97,
3032
Normand M., Pinault J.L., Seguin J.J., Vernoux J.F. (2005) Assistance au plan scheresse en Ile de
France. Analyse critique des donnes pizomtriques et prvision de niveaux non influencs.
Rapport BRGM/RP-54221-FR, 101 pp.
Putot E. (2006) Qualification du rseau pizomtrique du bassin Seine-Normandie. Rapport
BRGM/RP-54481-FR
Verjus, Ph. (2007) Taking groundwater into account in a drought plan, This Volume
Vernoux J.F., Girault F., Ceaux J.L., Gaudefroy M.J., Hugot V., Joublin F., Jegou J.P., Arnaud L.,
Wuilleumier A., Chabart M. (2005) Gestion du rseau pizomtrique du bassin Seine-Normandie,
bilan de lanne 2004. Rapport BRGM/RP-53856-FR, 182 pp.
CHAPTER 41
Taking groundwater into account in a drought plan
Philippe Verjus
Direction Rgionale de lEnvironnement Ile de France, 79, rue Benoit Malon, 94257 Gentilly
Cedex, France
ABSTRACT: The aim of a drought plan is to prepare contingencies for drought conditions.
Groundwater is part of that plan which needs to promote sufficient baseflow to small rivers and their
associated wetlands. A methodology was developed which comprises:
preliminary hydrogeological and ecological investigations to identify priority basins;
definition of correlations between minimum river flow and piezometric levels;
selection of reference piezometers;
determination of piezometric thresholds for imposing limitations or suspending the use of water.
The Beauce aquifer management strategy is used as an example.
1 INTRODUCTION
The head of any French Regional Authority can impose severe limitations on the use of
water or issue suspension orders pursuant to decree 92-1041 of September 24, 1992. The
latter measure was improved by a new plan, called the drought plan, which was implemented
after the long hot and very dry 2003 summer. The plan is being gradually improved in the
light of the results of the 2004 and 2005 low-water management experiments. The general
objective is to manage water shortages by ensuring priority supply. The measures taken
during drought periods must be progressive, focussed on the objectives and the severity of
the drought situation. They can be prescribed only for a limited period. It is thus a question of
defining, at catchment or basin scale, suitable activation thresholds and implementation
measures which are associated with the framework of administrative orders.
The roles of the Regional Authoritys Direction Rgionale de lEnvironnement (DIREN) are:
definition of thresholds and administrative orders in collaboration with the water police
authority and the users.
dissemination of information relating to river flows and groundwater levels on the
Internet: http://www.ile-de-france.ecologie.gouv.fr
technical support for the water bailiffs. In normal periods, information is given monthly
to provide an overall picture of the regional situation, but increased to weekly in periods
of low water levels.
2 IMPORTANCE OF TAKING GROUNDWATER INTO ACCOUNT
The national drought plan was drawn up in 2004 and was justified by the intense hot dry
period experienced in France in 2003. The thresholds defined in the plan concerned only
the rivers, but in the Ile de France region, they were extended in 2005 to include ground-
water. The key objective of taking groundwater into account is to safeguard the more vul-
nerable small rivers and wetlands located in upstream basins. Moreover, the Ile-de-France
is characterized by two very distinct river types. There are the large river basins: the
Marne, Seine, Yonne and Oise. Flow in these rivers depends primarily on the upstream
catchments, with only a minor contribution from tributaries in Ile-de-France and ground-
water local baseflow. The Marne and the Seine also receive water from upland storage
reservoirs, which can release water at times of low flow. There are also the smaller river
catchments which are fed mainly by groundwater during low-water periods. Groundwater
may also be the key to manage possible user conflicts. It can also allow piezometric thresh-
olds to be applied as prescribed tipping points, especially in catchments where there are no
flow measurement stations. The hydrometric and piezometric network density is quite
variable from one sector to another in Ile de France and it is essential to make the best pos-
sible use of the existing piezometric network.
3 SELECTION OF BASINS AND REFERENCE PIEZOMETERS
The reference piezometers should fulfil the following criteria:
be representative of the catchments with particular regard to the more fragile aquatic
ecosystems;
have meaningful and long-duration time series;
have a recognisable correlation between minimum river flows and low-water piezometric
levels;
be available for downloading on a weekly basis.
3.1 Basin choice
A good regional knowledge is a prerequisite. The DIREN has used a 3D database devel-
oped by the French Geological Survey (BRGM), which provides a basis for hydrogeo-
logical analysis. The first phase is to identify the priority areas, since available resources
are limited. Initially this will exclude:
baseflow to rivers that are subject to specific management (the Beauce aquifer for
example);
rivers with few fragile ecosystems;
small rivers whose flows are strongly influenced by urban discharges originating from a
different aquifer system (Seine, Marne or Oise alluvia).
The most interesting basins are those with fragile ecosystems that need to be protected.
The objective is to distinguish wetlands and small rivers that depend on a specific
groundwater level and where the risk of ecological damage is high, from those where this
risk is low.
An examination of the cross-section (Figure 1) demonstrates the vulnerability of the
Rmarde River to intense groundwater exploitation, in particular in the Beauce region.
In the upstream reaches of the Rmarde, there is a vulnerable wetland area that is located
on an anticline and at a high elevation. Conversely, the Mauldre River, which drains a syn-
clinal basin in an area with low topography, is less vulnerable.
3.2 Piezometer selection
The first selection relates to the intrinsic quality of the piezometers:
time series longer than thirty years;
understandable time series;
no river influence;
not screened in several aquifers;
good condition of the piezometers;
permanent water column;
reliable data transmission equipment.
Figure 2 shows the selected piezometers, groundwater catchments and hydrometric network.
3.3 Hydrogeologic relations between groundwater and topographic basins
Some groundwater catchments coincide with topographic basins. It is possible to detect
the similarities and differences between basins, with the aim of dividing them into groups,
the important one being a gaining river, or group of rivers, which receive groundwater
baseflow from the same aquifer. Figure 3 illustrates an example where the rivers Aubette
and Viosne, in the Vexin, occur in such a group, and drain a lower Eocene multilayered
aquifer. The Epte drains the Chalk and cannot be included in the same group. The rivers
Taking groundwater into account in a drought plan 555
Figure 1. Example of hydrogeologic analysis comparing the situation of Rmarde and Mauldre.
located beyond the Oise have impermeable beds, and they cannot be connected directly to
the regional groundwater system unless by leakage.
In some cases, rivers lose water into the underlying aquifers. It is then necessary to man-
age the rivers and groundwater independently during low flows. Figure 4 illustrates this
case for the Yerres River. Its upstream part receives some recharge from the Champigny
regional aquifer, its median section is perched and only the downstream section is fed by
significant groundwater baseflow.
Figure 2. Groundwater basins, representative piezometers and available hydrometric and ground-
water networks.
Figure 3. Example for grouping basins by hydrogeologic analysis.
Figure 4. Example of hydrogeologic analysis showing independence between rivers and regional
groundwater.
3.4 Correlation between piezometric levels and minimum river flows
The first step is to compare, on a monthly scale, minimum annual river flows and pieome-
tric levels. This semi-quantitative study makes it possible to detect if the variations are con-
sistent. Figure 5 shows an example where a weak correlation was obtained. However, the
correlation can be improved by taking minimum flows and piezometric levels over a three-
day period instead of average monthly values. This avoids the background effect caused
by variation in the minimum river flow, always likely to be influenced by runoff, and in
particular during summer storms. However, the most crucial problem is the difficulty of
finding comprehensible correlations. This difficulty arises from the low density of the
hydrologic and piezometric measurement network.
4 DEFINING THE DROUGHT THRESHOLDS
Four piezometric thresholds were defined by analogy to recommended river thresholds in the
drought plan. This required the development of a pragmatic methodology. The vigilance
threshold was defined statistically. It corresponds to the annual monthly minimum low-water
average with an approximately 3 to 5 year return period. It is useful to compare characteris-
tic return periods of piezometers to those of rivers, in cases where one observes a qualita-
tively satisfactory correlation. It seems that the vigilance threshold is consistent with that of
small rivers, where the characteristic return period is 5 years. On a monthly scale, the shorter
groundwater return period is due to the lower variability of aquifer levels compared to min-
imum flow in rivers, which are also influenced by runoff.
The definition of the reinforced crisis threshold is founded on analyses of long piezo-
metric time series in order to identify previous drought events and to examine the known
consequences. The objective of this method is consistent with that set for environmental
quality improvement, in particular when it is a question of mitigating the significant prej-
udicial effects of drought on aquatic environments. Under virtually identical climatic con-
ditions to those of the earlier drought, an attempt to reduce fish mortality could be tested
by maintaining a higher constant minimum flow, through progressive limitations of
non-priority uses. In more severe climatic conditions than the reference ones, ecological
damage or inevitable user conflicts could nevertheless occur. It is impossible to guard
against extreme climatic events beyond managing human activities. Finally, in the Ile-de-
France region, the 19921993 drought was regarded as a management reference drought
and the reinforced crisis threshold was fixed slightly above the low levels of that event
(Figure 8).
The alarm and crisis thresholds lie between the existing vigilance and reinforced
crisis thresholds.
Representative surface areas of the reference piezometers
It is advisable to specify the river basins represented by each reference piezometer
(Figure 7). For example, the Thmricourt piezometer hydrographic units are lAubette de
Correlation between average mounthly piezometric level at Allainville
and average mounthly flow of Essonne at Guigneville
0
1
2
3
4
5
6
128 129 130 131 132 133 134 135 136
Piezometric level (m)
F
l
o
w

(
m
3
/
s
)
Correlation between 3 days average piezometric levels
and flows
Figure 5. Examples of correlations between minimum annual flows and piezometric level, on a
monthly scale (top) or three-day scale (bottom).
PIEZOMETRIC EVOLUTION AND DROUGHT THRESHOLDS
MONTEREAU SUR LE JARD (77)
Figure 6. Montereau sur le Jard piezometric threshold.
Figure 7. Design of representative surface areas for reference piezometers.
Magny, la Valle du Roy, lAubette de Melan, le Ru de Montcient, la Viosnes and le
Sausseron.
The piezometric thresholds are linked to a reference piezometer. For various reasons,
this method was preferred to that of building an average indicator from several different
piezometers. Furthermore, calculating an average from piezometric data is only sensible
when the piezometers indicate the same relative variations. the choice depends on:
1. the most representative of the piezometric variations in a considered zone;
2. technically the best (remote transmission, no risk of breakdown or drying up);
3. the longest statistically representative time series.
In addition, some sectors have only a few piezometers, and it is not possible to develop an
average.
5 BEAUCE AQUIFER
The agricultural withdrawals in the Beauce aquifer are very high (75% of total with-
drawals). A brief summary of their management is given below:
5.1 Beauce aquifer management
The principle is to allot a guaranteed volume to each irrigating farmer at the beginning of
each year according to the prevailing groundwater level. This reflects the long reaction-time
of groundwater. The management is based on a piezometric indicator and three thresholds
that determine the total allotted volume. These thresholds, set up in 1999, have no relation-
ship with the drought thresholds. The principal rules are as follows. The total volume allot-
ted for irrigation is 450Mm
3
when the groundwater level is above the first threshold alarm;
a 0.9 or 0.8 reduction factor is applied if the groundwater level falls to the first or the second
threshold; nothing is envisaged beyond the third threshold, although this situation is critical.
5.2 Comparison between current management and drought-plan implementation
It is thus a question of comparing the current management system in the Beauce, with that
which would result from the drought-plan. From the Beauce indicator time series, virtual
Beauce and drought thresholds
Figure 8. Beauce and drought thresholds a comparison.
drought thresholds were established. Figure 8 presents the drought and the Beauce aquifer
thresholds.
The following relationships are apparent. The Beauce S1 threshold represents an average
groundwater situation and is consequently located well above the drought vigilance thresh-
old. The drought-vigilance threshold corresponds approximately to the Beauce S2 threshold.
The reinforced crisis threshold corresponds to the Beauce S3 threshold. Here, the drought
plan does not agree with the aquifer management plan. When the drought plan is implemented
during a reinforced crisis, no irrigation, except for market gardening, is authorized whereas
aquifer management still authorizes 360Mm
3
between thresholds S2 and S3 and does not
envisage any limitation beyond S3. This result tends to indicate that a revision of the manage-
ment plan is called for. The urgency of the revision depends on the groundwater level trend in
the Beauce aquifer. In 2005, it was the only aquifer in the Ile-de-France region showing an
average water level, whereas all other aquifers were at their minimum. In addition, the evolu-
tion of the situation could be predicted by means of a climate-piezometric-level model.
6 CLIMATE-PIEZOMETRIC-LEVEL MODELLING
The Tempo model was built by the Bureau de Recherches Gologiques et Minires, BRGM
(Pinault 2001). It generates piezometric variations using a climate-piezometric-level model
by means of transfer functions. Initially, the model of the Ile-de-France aimed to:
1. forecast short-term piezometric trends;
2. calculate, in the absence of measurements, piezometric levels for the piezometers that
had gone dry.
In fact, the short-term calculations still lack the required precision for crisis management,
but when used for medium-term forecasts, for the special Beauce aquifer case, the model
was very useful.
6.1 Beauce aquifer application
The Allainville piezometer is generally used to give the water bailiffs monthly information
on the Beauce aquifer. It is not used to calculate the Beauce indicator but it is correlated
with the indicator as well as with the Essonne river flow, which is the main Beauce aquifer
outlet in the Ile-de-France. It would be legitimate to analyse and calculate virtual drought
thresholds on the Allainville piezometer just as one would on the Beauce indicator. Once
the Allainville reference piezometer was calibrated, simulations of the Beauce aquifer
piezometric trends could be made for various climatic conditions. These estimated the
dates that the virtual drought thresholds would be reached. Figure 9 illustrates the results.
The simulation starts at the end of the last recorded piezometric time series in June 2005.
During 2005, the Beauce aquifer was not included in any drought-order implementations.
This decision was founded on the extrapolation of the piezometric trend. However, beyond
2005, it would have been difficult to make reliable forecasts without the model.
It appears that for the 2006 irrigation season, the situation is not likely to be serious
whatever the climatic conditions may be. For 2007, it appears that:
1. the vigilance threshold may be reached for a median climate scenario.
2. the alarm level would be exceeded for a climate scenario of drought with a 5-year
return period.
3. the reinforced crisis threshold would be exceeded for a climate scenario of a drought
with a 20-year return period.
If the above results are validated, they will have notable administrative consequences.
Indeed, for two years, it would not be necessary to manage any user conflicts. There would
thus be a two-year period in which to revisit the management of the aquifer to make it
compatible with the drought-plan implementation. This will allow time to deal with the
irrigation-water attribution system for the several thousand farmers, which will require
considerable rigour, and to develop aquifer management rules.
7 CONCLUSION
When groundwater flow toward rivers is taken into account in a drought management plan,
considerable progress is made in applying sensible water management rules. This is espe-
cially the case for prolonged drought events lasting several years. River flows are always
influenced by runoff, so decisions to impose limitations or suspend water use based only
on river flow thresholds do not take into account the complete picture of the drought
susceptibility.
REFERENCES
Pinault, J.L. (2001) Manuel utilisateur de TEMPO: logiciel de traitement et de modlisation des
sries temporelles en hydrogologie et hydrochimie. Report, BRGM/RP-51459-FR, 233 pp.
Normand, N., Pinault, J.L., Seguin, J.J., Verjus, Ph., Vernoux, J.F. (2005) Assistance au plan scher-
esse en Ile de France, analyse critique des donnes pizomtriques et prvision des niveaux non
influencs. Report, BRGM/RP-54221-FR, 128 pp.
Piezometric variations and predictions
with drought thresholds
Figure 9. Predictive simulation of the Beauce aquifer water level trends, compared to the drought
thresholds.
CHAPTER 42
Risk map of high natural background of trace elements in
the Water Authority Rhne-Mditerrane & Corse basin
Agns Brenot
1
, Ariane Blum
1
, Laurence Chery
1
, Romain Sonney
1
,
Laurent Cadilhac
2
& Thomas Pelte
2
1
BRGM Water Division, BP 36009, 45060 Orlans cedex 2, France
2
Agence de lEau Rhne Mditerrane Corse, Lyon cedex, France
ABSTRACT: This study identifies the areas where groundwater is likely to display high natural
background levels in trace elements in the Water Authority Rhne-Mditerrane & Corse Basin. The
natural occurrence of trace elements in the local geological formations drained by groundwater and
surface water is identified from published literature, existing trace element analysis in soils and the
identification of a mining index. These data, coupled with the geochemical status of the natural water
(pH, redox-conditions, temperature, concentration in dissolved oxygen ), allow the potential mobil-
isation of trace elements to be predicted. These and other data have been assembled within a GIS for-
mat to prepare an overall risk map for high natural background levels. For each defined area, depending
on the relevance of the information available to identify this risk, a degree of confidence or knowledge
(low, middle and high) is derived; interpretation of the map also promotes further study.
1 INTRODUCTION
As required by the Water Framework Directive in Article 17, a Daughter Directive on the
protection of groundwater against pollution was issued in December 2006. It specifies that the
good chemical status of groundwater shall be partly defined by Member States themselves.
But the evaluation of this good status requires knowledge of the natural geochemical back-
ground of water bodies in order to distinguish trace elements present naturally in the envir-
onment from those resulting from human inputs. Commonly baselines are also used
interchangeably with the term backgrounds. These terms are meant to imply something
about the natural tendency of systems in the absence of human influences such as land use.
To a large extend, it is important to decipher natural contributions for trace elements in nat-ural
water in order to establish relevant restoration strategies and to anticipate temporal evolu-
tion of the natural water quality. The project Fond Gochimique Rhne-Mditerrane &
Corse, conducted by the WATER division of BRGM, intends to identify the different areas
on the Water Authority Rhne-Mditerrane & Corse Basin where waters (groundwater and
surface water) are likely to display high natural background levels in trace elements. Elements
studied include all the elements considered as toxic or undesirable by the Drinking Water
Directive (98/60/EC) i.e. arsenic, barium, bore, fluorine, cadmium, chromium, mercury, cop-
per, nickel, lead, zinc, antimony, selenium, aluminium, silver, iron, manganese.
2 DATA AND METHODS
A literature search identified those areas where the geological context is likely to induce
trace element mobilisation from rocks to natural water. All the literature dealing with the nat-
ural occurrence of trace elements in soils, rocks and water, together with those dealing with
the geochemical status of natural water (pH, redox-conditions, T, and concentration in dis-
solved oxygen ) were collated. Existing trace element analysis in soils and the identifica-
tion of mining indices, essentially for crystalline areas, allowed geological units to be
identified in which trace elements are naturally present at high concentrations. Databases of
natural water considered in this study correspond to the ADES database (Accs aux Donnes
des Eaux Souterraines, French national groundwater database) for groundwater, to the RNB
(Rseau National de Bassin) and FOREGS (FORum of European Geological Survey) data-
bases for surface water. All these data were combined in Geographic Information Systems
(GIS) format with the Arcview software.
3 IDENTIFICATION OF AREAS PRESENTING HIGH TRACE
ELEMENTS BACKGROUND
A first step towards the interpretation of trace element groundwater data in the ADES, RNB
and FOREGS databases required an assessment of the sampling locations with regard to land
use distribution. Some of the sampling points are highly affected by human inputs that over-
ride the natural background trace element levels. In order to identify these sampling points,
a map presenting the main human pressures for trace elements has been created with land use
data from the Corine Land Cover 2000 database. The 44 classes of the initial Corine Land
Cover 2000 map have been grouped into seven simplified classes that relate to areas of
increasing impact or release of human-induced trace elements in natural water. The human
pressures established in this study include mining activities which also constitute human
amplification of natural weathering processes. The sampling points that were most affected by
human inputs in trace elements were eliminated. For groundwater and surface water sampling
points that are not affected by human inputs, further investigations were made in order to iden-
tify those areas presenting risks of high natural background in trace elements. Information
from the literature and from trace element analysis on soils was summarized in order to iden-
tify areas where trace elements are likely to be mobilized in natural water. Prediction of trace
element mobilisation from rocks to the dissolved phase has been well documented elsewhere
(Blum et al., 2002; BRGM, 2006). Thus the overall geological context associated with the
geochemical status of the natural water (pH, redox-conditions, temperature, concentration
in dissolved oxygen ) documented in the literature allow predictions of which elements
are more likely to be dissolved in natural water, depending on the considered area. This
information was combined within the GIS with concentrations of trace elements (arsenic,
barium, bore, fluorine, cadmium, chromium, mercury, copper, nickel, lead, zinc, antimony,
selenium, aluminium, silver, iron, manganese) recorded for groundwater and surface
water in all the databases (ADES, RNB, FOREGS). If high concentrations are compatible
with both the geological context and geochemical status, an area presenting a risk of high
natural background was then identified.
Interpretation of surface water data is more difficult because different media were
analysed: dissolved elements, suspended mater, sediments and bryophytes, and because the
locations of sampling points were not always relevant to identify natural background in trace
elements, as they are more affected by human inputs than groundwater. Concentrations meas-
ured for dissolved elements are highly variable, depending on the sampling period (high/low
river flow) and this makes it difficult to compare them. All these difficulties have been
considered in order to define the areas presenting risks of high natural background for sur-
face water.
Soil analysis of trace elements, relevant water analysis, articles explaining the origin of
trace elements in their geological context allow clear identification of the risk of high natural
background levels. However, some areas with few chemical analyses, especially for natural
water, allow only a prediction, based on knowledge of trace element mobilisation through nat-
ural water under similar geological conditions. Thus, in order to take into account these dis-
parities, a confidence level has been attributed to each area, depending on the abundance of
data. Three confidence levels have been defined, the criteria used are presented on Figure 1.
As an example, a low confidence level is assigned to areas where no data are available or
where the sampling points are affected by human inputs. Conversely, a high confidence level
is assigned to areas where all the data for water and trace element analysis in soils and water are
meaningful and allow identification of the risk of a high natural background in trace elements.
This leads to the location on a map of the areas where risks of high natural background in trace
element has been identified. For each area, the elements presenting the risk were defined, and
it was also specified if this risk affects groundwater, surface water or both (Figure 2).
4 ADVICE FOR DESIGNING MONITORING PROGRAMMES
Data available for the Rhne-Mditerrane & Corse Basin were not sufficient to apply
statistics in order to quantify the natural background for trace elements. Maps locating
high natural background in trace elements were used to design further surface and ground-
water data acquisition programmes. These programmes define a suitable sampling strat-
egy for guidance background values for trace elements. Depending on the confidence
level assigned to the areas presenting high trace element backgrounds, different objectives
Risk map of high natural background of trace elements 565
Confidence
level
assigned
High
Documented in the literature and trace element analysis in water
or
Trace element analysis in soils and water
or
Document in the literature and trace element analysis in soils and water
Medium
Documented in the literature
or
Documented in the literature and trace element analysis in soils
Data available
Low
No data available but knowledge on similar geological context
or
Trace element analysis in water
or
Trace element analysis in soils
Figure 1. Confidence levels assigned to the areas presenting high natural background in trace
elements.
Figure 2. Areas of high natural background in trace elements (in the background: simplified
geological units 1/1000000 scale).
for data acquisition programmes were allocated (Figure 3). In the areas defined with a low
confidence level, the objective was to identify the risk of high natural trace elements con-
centrations. For a medium confidence level, the objective was to specify the level and to
spatially locate the distribution of high trace element background. Finally for the areas
where the confidence level was high, further investigation should lead to the quantifica-
tion of the upper limits of the concentrations, which can be attributed to natural inputs
only, and would be used as guidance to establish the Environmental Quality Standards
(EQSs) and threshold values.
Risk map of high natural background of trace elements 567
Figure 3. Areas of high natural background in trace elements and location of sampling points for fur-
ther investigation programme (in background: simplified geological units 1/1000000 scale). Numbers
correspond to those assigned to areas where new investigations for trace element analysis in natural
water are recommended. Map corresponding to the French Alps domain (red rectangle on Figure 2).
The selection of points proposed for monitoring programmes took into account land use
distribution in order to minimize human inputs. Points corresponding to existing sampling
networks were preferentially used in order to reduce the costs. In areas where no existing
sampling points were relevant, new collection points were established. Depending on the
objectives for data acquisition programmes, different sampling media should be favoured.
Trace element concentration in suspended mater, sediments and bryophytes could only be
relevant to identify and specify levels of high natural background.
But in order to determine local background concentration, only analyses of dissolved
elements are suitable because EQSs for surface water are defined on these concentrations.
The choice of the sampling period is also of primary importance to minimize human inputs
and to represent the maximum concentrations reached, which can be attributed to back-
ground values, especially for surface water. As during low flow conditions, streams are fed
mostly by groundwater and soil leaching is limited, this period is most suitable for defining
natural background levels for trace elements. As hydrologic conditions during low flow
periods may vary from year to year, it is important to consider at least five sampling cam-
paigns for dissolved elements, distributed over a minimum of 3 years. For suspended mater,
sediments and bryophytes, one or two sampling campaigns are sufficient as the objective is
not to quantify local background concentrations. Finally, in order to harmonize compliance
checking throughout Europe, it is important to get comparable data sets. As measurement
uncertainty clearly has strong implications for the interpretation of analytical results in this
context, the EC will shortly define data-quality criteria on reference analytical methods that
would permit relevant inter comparisons (Coquery et al., 2005).
5 CONCLUSION
This work leads to define the location on a map of the areas where risks of high natural back-
ground in trace elements are likely to occur. For each delimitated area, a confidence level (low,
medium or high) has been attributed in order to take into account disparities depending on the
availability of data. It was not possible to quantify natural background levels for trace ele-
ments, because natural water analyses were insufficient to apply robust statistics. In order to
obtain guidance background values within future studies, new investigation programmes,
especially for surface water, will need to be designed. Depending on the level of confidence
assigned to the areas presenting high trace elements background, the risk map defines differ-
ent objectives for data acquisition programmes. These programmes could enable the concen-
trations due to natural background to be quantified by monitoring for those areas where a risk
has been identified. These studies are of interest to define the values of good chemical sta-
tus for all European surface and groundwater bodies depending on the local conditions. The
end of this process is to define EQSs and threshold values specifically for each water body.
REFERENCES
BRGM (2006) Guide technique Qualit des eaux souterraines Mthode de caractrisation des
tats de rfrence des aquifres franais. BRGM Editions, 248p.
Blum A., Chery L., Barbier J. et al. (2002) Contribution la caractrisation des tats de rfrence des
aquifres Franais. Outils et mthodologie. Rapport final. Rapport BRGM RP-51549-FR, 5 volumes.
Coquery M., Morin A., Becue A., Lepot B. (2005) Priority substances of the European Water
Framework Directive: analytical challenges in monitoring water quality. Trends in Analytical
Chemistry, 24, 117127.
CHAPTER 43
Groundwater and Soil Water System for Norway based
on daily simulations and real-time observations
Herv Colleuille
1
, Stein Beldring
1
, Zelalem Mengistu
1
,
Wai Kwok Wong
1
& Lars-Egil Haugen
2
1
Norwegian Water Resources and Energy Directorate (NVE), Hydrology department, Majorstua,
Oslo, Norway
2
The Norwegian University of Life Science (UMB), Department of Plant and Environmental
Sciences, Norway
ABSTRACT: NVE is developing a new method to produce soil-moisture and groundwater maps
based on daily model simulations and real-time observations. A distributed version of the HBV-
model is used to model hydrological processes for the entire land surface of Norway. The model
describes interfaces between the land surface and the atmosphere, soil moisture and groundwater
storage, groundwater flow and river flow. It provides a realistic framework for regional hydrological
modelling by integrating the contributions by several small-scale elements. The model operates on
a 1-km
2
grid and one-day resolution is used. The project is divided into three parts: (1) development
of a GIS-based system to visualize daily maps of groundwater and soil water situations based on
real-time and manual observations; (2) validation of the distributed model by comparing simulations
with observations, as well as with simulations of a physically-based 1-dimensional model (COUP)
at representative soil-monitoring stations; (3) based on (2), improving the distributed HBV-model by
refining some of the sub-routines used in this model. Development and implementation of methods
producing groundwater and soil-water maps for Norway provide national and local stakeholders
with tools for assessing groundwater-related risks in a new way.
1 INTRODUCTION
Norway was glaciated and deglaciated several times during the Quaternary period. As a
consequence bedrock is poorly weathered, and sedimentary deposits are young, sparse and
often thin. The landscape is characterized by fjords and mountains, glaciers, boreal and
alpine forests and a large number of lakes and bogs. Aquifers in Norway consist mainly of
small, highly permeable glaciofluvial deposits along streams and lakes located in typical
U-shaped valleys, and small precipitation-fed tills (low permeable moraines) in moun-
tains, both overlying fractured bedrocks without primary porosity (mainly crystalline and
metamorphosed hard rocks).
The groundwater level is controlled by the recharge-discharge mechanism, which is
determined by the physiographic and climatic conditions. The average annual precipitation
in Norway is about 1400mm, of which some 300mm evaporate (mostly between May and
October). The precipitation is not uniformly distributed. In the western part of the country
annual precipitation may exceed 5000mm, whereas some eastern valleys receive less than
300mm per year. On average, about half of the annual precipitation falls as snow. Snow
and soil frost affect recharge thereby distinctly influencing the recharge pattern. Soil frost
leads to a more rapid decrease of groundwater levels during the winter period and to
increased surface runoff when the snow melts. After the thaw of soil frost in the spring, the
groundwater levels show a sharp rise in response to the percolation of snowmelt.
Groundwater hydrographs for the interior of Norway have two falling and rising limbs
in keeping with the description of the groundwater fluctuation pattern for Norway, Sweden
and Finland by Kirkhusmo 1986, (Fig. 1). The first rise comes from the snowmelt and the
second one from rainfall at the beginning of the autumn. In the mountainous regions and
in northern Norway, precipitation does not appear as rainfall but as snow in the autumn
leading to one long recession stretching from summer to late winter until the snow melt
570 Aquifer Systems Management: Darcys legacy in a world of including water shortage
Figure 1. Top: Variation of groundwater levels in coastal, interior, and mountainous/artic regions
(examples of annual average water level from three stations). Bottom: Groundwater level recorded
in the largest unconfined and precipitation-fed aquifer in Norway (Gardermoen, South East
Norway). See Figure 4 for the location of the monitoring stations.
starts. In the coastal regions, recharge occurs generally in winter, and the groundwater level
drops in summer due to the high evapotranspiration. Annual fluctuations of the groundwater
table are large in tills and in bedrock (15m) and smaller in permeable glaciofluvial deposits
(01m).
Groundwater levels reflect the state of groundwater storage at a particular location and
time (van Lanen et al., 2004). Steep topography, thin deposits, cold and humid climate all
contribute to shallow groundwater levels in Norway. Norwegian aquifers generally have a
limited storage capacity and are thereby sensitive to extreme climatic change, controlling
hazardous events such as landslides, floods and droughts. Norwegian aquifers are fast
responding systems due to the rapid fluctuation of water levels in response to the annual
variability in recharge. In addition, a few aquifers in larger geological units due to the large
fluctuation of groundwater levels caused by the delayed and attenuated response to distinct
dry and wet periods (figure 1 bottom).
The soil water deficit is defined as the amount of depleted water between the field
capacity and the actual water content. The soil-water deficit estimates how much water the
soil profile can store before groundwater recharge can occur and is a good indicator of the
soil-water and groundwater conditions. The soil-water deficit may reach about 200mm in
Norway, depending on soil type, vegetation and actual evapotranspiration. Norway is very
rich in renewable fresh water resources, both as runoff per unit area and runoff per capita
(Pettersson, 2004). Water plays a significant role for important sectors of the Norwegian
society, e.g. hydropower industry, tourism, fish farming and water supply. Only about 15%
of the population in Norway is supplied with groundwater. However, groundwater supplies
1 in 3 of the large waterworks, and wells are commonly used to supply rural houses and
farms. The most obvious influence of groundwater on Norwegian rivers is the maintenance
of flow in dry periods. Baseflow studies show that 80 to 100% of the total river discharge
in the interior of Norway comes from groundwater in periods of low flow, typically in win-
ter and late summer (Colleuille et al., 2006). The percentage of groundwater is consider-
ably lower (3570%) in the south-western part of Norway due to the topography, the lack
of porous media within the mountains (fjord landscape) and the maritime climate.
The Norwegian Water Resources and Energy Directorate (NVE) is responsible for the
administration of Norways water and energy resources. The Hydrology department col-
lects water-related data covering the quantitative aspects of the land phase of the water
cycle, and is developing tools to manage national water resources (including flood and
drought forecasting, energy inflow for hydropower production, snow, glaciers, sedimenta-
tion, water temperature and groundwater).
The Hydrology department has launched a new project which aims to develop tools pro-
viding national daily information on groundwater and soil-water conditions. This project
may be divided into three parts: (1) development of a GIS-based system to visualize daily
maps on groundwater and soil-water situations based on real-time and manual observa-
tions. Each monitoring station is represented by a colour-coded symbol depending on the
current groundwater situation; (2) validation of the distributed model by comparing simu-
lations with observations, as well as with simulations performed by a one-dimensional
soil-water and energy-balance model (COUP (Jansson and Karlberg 2004)) at representa-
tive soil-monitoring stations; (3) based on (2), improving the distributed HBV-model by
refining some of the sub-routines used in this model (evaporation, soil frost, etc).
The purpose of this paper is to present the methods employed in this project and the first
results of the validation exercises.
Groundwater and Soil Water System for Norway 571
2 METHODS
2.1 Monitoring network
Currently NVE operates about 65 groundwater-monitoring areas using 80 observation
wells and has 15 stations for soil moisture and temperature measurement. These stations
are monitored in cooperation with the Norwegian Geological Survey (NGU), Norwegian
Institute for Agricultural and Environmental Research (Bioforsk), the University of Life
Science (UMB) and hydropower companies. This observation network is designed to
capture the effects of climate on soil-water and groundwater levels in key topographic set-
tings in all major physiographic and geological units in Norway. At soil-water stations,
measurements of soil moisture content are performed at 810 different depths, with pro-
file Probes Delta-T (TDR technology), along with soil-water potential by Watermark sen-
sors and tensiometers, soil temperature (thermistors) and groundwater levels (pressure
transducer). All the soil water stations and about 2/3 of the groundwater stations are auto-
mated with hourly measurement. By the summer of 2006, 35 wells will be equipped with
real-time capability. Automatic data transmission uses mobile phone technology and, in
remote locations, stations are powered by solar panels. All the data are automatically
stored in the national hydrological database operated by NVE. This database has a wide
range of functions for control, retrieval, presentation and analysis of data. Furthermore,
it has a flexible structure that enables data storage with various temporal and spatial
dimensions.
About 50% of the groundwater stations are located in the interior regions, 25% in
the coastal regions and 25% in the mountainous and arctic regions according to the classi-
fication presented in Fig. 1. The first groundwater-level observations were recorded
in 1949, but the majority of the groundwater stations were established in the 1970s
and 1980s, and the soil water stations in the beginning of the 1990s (Fig. 2). About 77%
of the 62 operative groundwater stations used in this work have over 20 years of
observations, and 26% over 30 years (16 stations). Ten stations have less than 10 years of
observations.
Figure 2. Number of years of observations of groundwater levels at each monitoring station.
Due to the sparse station network, which cannot cover all river basins in Norway, hydro-
logical models are used to describe the water and energy balances on a national scale.
2.2 Modelling tools
Two independent models simulating land surface hydrological and thermal conditions are
used. The physically-based COUP-model (Jansson and Karlberg, 2004) simulates one-
dimensional water and heat dynamics in a layered soil column covered by vegetation by
solving numerically the relevant differential equations. The main equations include the
laws of conservation of mass and energy together with flow equations for water (Darcys
law) and heat (Fouriers law). A detailed description of the model can be found at http://
www.lwr.kth.se/Vara%20Datorprogram/CoupModel. The model is run with a daily time
step, using precipitation, air temperature, wind speed, relative humidity and sun radiation
data as input. In addition, plant growth characteristics (e.g. Leaf Area Index, root depth)
and soil characteristics (e.g. soil-water retention curve, hydraulic conductivity) are neces-
sary inputs to the model. Simulations with the COUP-model are performed only for some
representative soil-water stations where observations are used for parameterisation of the
model. As the COUP-model uses detailed process descriptions at the point scale (evapo-
transpiration, snow and frost, percolation, storage, runoff), modelling of land surface
hydrology at larger spatial scales has to be performed by models that represent the system-
atic variations in the properties of the land surface.
The conceptual HBV-model is used for flood forecasting and water balance simulations
in the Nordic countries. The HBV model derives its name from the department at the
Swedish Meteorological and Hydrological Institute where it was developed: Hydrologiska
Byrns Vattenbalansavdelning. A good description of the model can be found in Bergstrm
(1995). A spatially distributed version of the HBV-model (Beldring et al., 2003) is used in
this project. The model performs water balance calculations for square grid-cell landscape
elements. The model is calibrated with a constraint such that model discretisation units
with an identical landscape classification are assigned the same parameter values. For
each grid cell the percentage of lake, glacier and land-use classes (sparse vegetation, sub-
alpine forest, forest, agricultural land) is determined based on nationwide GIS-informa-
tion. Every model element has unique characteristics, the input data are distributed, and
water balance computations are performed separately for every model element. The model
is run with a daily time step, using precipitation and air temperature data as input. It has
components for accumulation and ablation of snow, interception storage, soil-moisture
storage, evapotranspiration, groundwater storage and runoff response, lake evaporation
and glacier mass balance.
In HBV the rainfall reaching the ground and snowmelt leaving the snow pack infiltrates
into the soil. The soil profile is divided into 3 layers: the root zone layer (soil moisture
zone), the upper and the lower groundwater layer. The transport between the top layer and
upper groundwater layer depends on the soil-moisture content. From the soil-moisture
zone, the water can evaporate proportionally related to air temperature, the current season
and the water availability in the soil-moisture zone. From the upper groundwater zone the
water can percolate to the lower zone or flow laterally into discharge or storage. From the
lower zone the water can be drawn up to the soil-moisture zone or flow out as lateral dis-
charge. An overview of some important features of the COUP- and HBV-models is shown
in Table 1.
3 RESULTS AND DISCUSSION
3.1 Operational procedure
The Groundwater and Soil Water System is based on two totally independent sources of
information: observations and simulations. Observations are obtained from the monitoring
network, and simulations from the distributed HBV-model. The data processing for the
HBV-modelling involves automatic collection of meteorological observations and simula-
tions of daily snow, runoff, groundwater and soil-water conditions. The model is updated
every day with the last-day temperatures and precipitation observations. Forecasts are also
made for 6 days based on meteorological forecasts. The output data are retrieved in order
to get the desired information (e.g. groundwater conditions compared with long term aver-
age data). Hydrological maps are automatically produced based on the gridded output
Table 1. Comparison of the distributed HBV-model and COUP-model used in this study.
Distributed HBV-model COUP-model
Time resolution Daily Daily
Spatial resolution 1km
2
1-dimensional (1m)
Driving variables Air temperature, precipitation Air temperature, precipitation, relative
humidity, wind speed, global radiation
Snow/rain Threshold temperature Threshold temperature
Snow melt Degree-day-factor Simulated
Soil frost Not in the standard version Simulated
Interception Yes: Maximum interception Yes
storage
Infiltration No restriction Restricted by saturated hydraulic
conductivity (soil frost or low-
permeable soil)
Surface runoff No If rain intensity Ksat :surface
water is created. Runoff from the
surface water is described by 1st order
kinetics
Evapotranspiration
Potential Thornthwaite-type Penman-Monteith
Actual Soil moisture deficit Soil moisture deficit
Soil moisture storage Field capacity principle According to the soil hydraulic
properties (soil moisture
characteristics, hydraulic
conductivity)
Water flow in soil Upwards and downwards Upwards and downwards
(percolation and capillary (percolation and capillary rise)
rise)
Vertical discretisation 1 root zone layer Over 50 layers with their own
2 groundwater layers physical and soil characteristics
Output simulations
Each soil layer Storage of water, vertical flow Temperature, water potential, vertical
of water, discharge from flow of heat and water, storage of
groundwater layers water and heat, root water uptake
Others Snow water equivalent and Snow water equivalent and depth, soil
depth, groundwater level, frost depth, surface runoff, drainage
streamflow flow, groundwater level
data. Gridded maps are produced for the following variables: precipitation, air temperature,
evapotranspiration, snow-water equivalent, groundwater conditions, soil-water deficit and
streamflow. A Web- and GIS-based system already developed for producing snow maps
(Engeset et al., 2004) is used to distribute information on groundwater and soil-water situ-
ations. Currently, the gridded map archive is presented as a HTML-based intranet naviga-
tion service available only for NVE. An Internet service will be available after a one-year
trial. A historical archive of daily gridded data from 1st January 2005 is being established
using this system. The data processing for the COUP-modelling is not automated and is time
intensive. Data from COUP-simulations are, therefore, used first of all for comparison
purposes. However, a monthly presentation of the simulation-results for representative sta-
tions will be evaluated later. The data processing for the observations consists in collecting
all available groundwater-level data in the national hydrological database for the current
day. Observations are interpolated for a daily resolution, in order to fit the temporal reso-
lution of the models. A GIS-map with coloured dots representing different groundwater
conditions at each monitoring station is produced daily. The coloured dots represent ground-
water conditions as a percentile for the current day of the year. Statistics are compiled based
on available data in the reference period from 1975 to 2004. The groundwater condition is
classified by comparing the current observation with the reference period. For observation
values falling between the 25 and 75 percentile limits, the groundwater is considered to be
at normal conditions. For values greater than the 75 and 100 percentiles, the situation is
classified as high and very high respectively, illustrated by blue and dark blue dots. When
the values are smaller than the 25 and 0 percentiles, they are considered low and very low
and are represented by orange and red dots respectively. A black dot represents a well
which does not have sufficient data to produce statistics or where data are not available
(e.g. logger not transmitting). Additional information will be introduced later by showing
coloured triangles indicating increasing, stable or declining groundwater levels based on
the change over the last 3 days. Observation wells located in larger aquifer which show
attenuated change in groundwater levels will be represented by squares. The system, within
a GIS format enables (by clicking on a station) the presentation of the time-series plots
showing groundwater levels for the last 2 years together with corresponding 0-, 10-, 25-,
50-, 75-, 90- and 100 percentiles. Other information, e.g. on soil/vegetation characteris-
tics, automated/manual observation, logger type, measured parameters, are also available.
The maps presented in Fig. 3 depict groundwater and soil water conditions as simulated
by the distributed HBV-model on 15 September 2005. High precipitation combined with an
already high groundwater level and no soil-water storage capacity leads to an extremely high
groundwater level this day causing floods and landslides near Bergen on the south-west coast
(see also Fig. 5 bottom: Fana). At the same time the groundwater level in the south-eastern
part of Norway is very low in comparison with the normal period. Observations from soil-
water stations agree with this picture (Fig. 3) but show that the amount of soil-water deficit
seems to be overestimated.
3.2 Validation
The quality of the HBV model simulations of the groundwater and soil conditions are
evaluated by three independent procedures:
1. qualitative comparison between the gridded maps of groundwater conditions with the
coloured dots showing the corresponding observed condition. The gridded map of
Figure 3. Groundwater and soil-water conditions on 15-09-2005 simulated by the distributed
HBV-model. Groundwater volume (top left) and soil-water deficit (top right) are presented as
percentages of the long-term average (19611990). Map of soil-water deficit given in mm
(bottom right). Map showing both measured and simulated groundwater conditions on 15-09-2005
(bottom left).
soil-water deficit may also be compared with the available observations from soil-
water stations;
2. quantitative comparison between daily observed groundwater levels and simulated
groundwater volumes;
3. quantitative comparison between daily observations, simulated groundwater levels and
soil-water deficits obtained from the distributed HBV-model and the physically-based
COUP-model at some representative monitoring stations.
Figs. 3 (bottom left) and 4 show observed and simulated groundwater conditions for three
different hydrological situations in 2005 (spring, late summer and autumn). The agreement
between the observations and the simulations is mostly satisfactory: 75% of the simulated
groundwater conditions are identical with observations for these 3 days (140 observa-
tions). Note that the reference period for the HBV-simulations is for the time 19611990,
whereas the reference period used for the observation is 19752004 and often shorter. In
addition, the data from HBV-model are presented as a percentage of normal values while
the observations are presented as percentiles. These problems impair the validation pro-
cedure. When a new reference period for HBV-simulations is made available (expected in
summer 2006), systematic assessment of the results will be made with a focus on regional
and seasonal variations (all data will be presented with the same percentiles).
Fig. 5 shows the simulated groundwater volume and observed groundwater level at 2
monitoring stations. The first station (Abrahamsvoll) is located near Rros in a mountain-
ous area in Central Norway (750m.a.s.l.) and the second (Fana) near Bergen on the south-
western coast (50m.a.s.l.). The aquifers at these stations consist respectively of moraine
(silty sand) and glaciofluvial sediment (sand). The agreement between the simulations and
observations is relatively good for both stations with a correlation coefficient of about
Figure 4. Maps showing measured and simulated (distributed HBV-model) groundwater conditions
on 28-04-2005 (left) and on 20-10-2005 (right). The locations of monitoring stations are indicated.
0.75. Preliminary results from 10 monitoring stations show that the correlation coefficient
lies between 0.6 and 0.9. Two stations located in northern Norway behave differently (cor-
relation coefficient 0.3). At some locations and for some years the simulations show less
agreement in winter probably because the decrease of soil-water flow due to soil frost is
not considered in the HBV-model. Note that the correlation is better for logged data than
for manual observations. At Fana (figure 5), the correlation coefficient is 0.66 with man-
ual observations and 0.90 with logged measurements (after July 2003). Observed changes
in groundwater levels are often slower than simulated ones. This is especially evident in
summer at some locations. It is probably a consequence of an inappropriate representation
of the hydraulic conductivity for the soil type in the HBV-model.
Groundwater levels, soil-moisture storage and snow-water equivalent from simulations
with the COUP and HBV-models in the Groset catchment are presented in Fig. 6. Groset
is located at about 950m.a.s.l. in southern Norway (Figure 4). Note the dry period in 1996
characterized by a high deficit in soil moisture and a low groundwater level. The ground-
water levels are described quite well by both models, although there are discrepancies
between observations and model results. There are also differences between the two models,
which can be attributed to differences in simulated actual evapotranspiration. The water
storage of the two models agrees to some extent, but the total volumes differ due to different
Figure 5. Simulated groundwater volume (mm) and observed groundwater level (m) at two moni-
toring stations: Abrahamsvoll (top) and Fana (bottom). See maps in Fig. 4 for the locations of these
monitoring stations. The two stars in the bottom figure indicate two different events with extremely
high groundwater levels causing flooding and landslides near Bergen on the south-western coast
(September and October 2005).
representations of the hydrological processes (Beldring et al., 2005). The HBV-model
needs to consider only the volume of water active in the precipitation-runoff process, and
deep immobile groundwater is not necessarily included. It should be remembered that the
model simulations represent different areas.
Figure 6. COUP- and HBV-model results from the Groset catchment. Top: Observed and simu-
lated groundwater levels. Middle: Total water storage and soil water deficit. Bottom: Snow, soil
moisture and groundwater storage in the COUP-model.
Soil-water deficits simulated by HBV and COUP are quite different. According to obser-
vations, HBV may overestimate the soil-water deficit in summer and underestimate it in
winter. Comparisons between the HBV-model simulation and point measurements in Sweden
and Finland show that the HBV-model overestimates by 2050% the evaporation in spring,
after the snow cover has melted, and in late autumn (Engeland et al., 2004). The HBV-model
has an empirical parameterisation of the evaporation process which adjusts a long-term sea-
sonally dependent potential evaporation according to temperature. The COUP-model has a
better physical description of the evaporation process as it uses the Penman-Monteith
approach and all available meteorological information (Table 1). In addition, soil-water deficit
in HBV is only caused by evapotranspiration, whereas soil-water deficit in COUP includes
soil-water depletion following the fall in groundwater levels in winter which occurs especially
in headwater catchments with moraine sediment and large fluctuations in groundwater level.
It is caused by lack of recharge in winter and large groundwater discharge to streams and
lakes.
4 DISCUSSION AND CONCLUSION
As all hydrological processes have a variability that depends on local meteorological, geo-
logical, topographical and vegetation characteristics, area-averaged model simulations
cannot be expected to conform exactly to point measurements of the groundwater condi-
tions. The system presented here is considered an effective and satisfactory tool for pro-
viding a nationwide daily picture of the groundwater and soil-water conditions. However,
the HBV-model used in this system needs some improvements to give a better confidence
in the output results, especially if the simulations are to be used to analyse and forecast
extreme hydrological situations at the local scale. In particular, it should integrate:
1. better spatial resolution of the meteorological data and interpolation procedure;
2. better information of geological, soil and vegetation characteristics;
3. simplified multi-dimensional heat and water transport modules for better simulations
of soil-water deficit, evapotranspiration and soil frost;
4. a multi-criteria calibration procedure based on snow, soil moisture, groundwater and
streamflow.
Spatially distributed estimates of temperature and precipitation are needed as input to simu-
lations of the groundwater and soil-water conditions. The distributed HBV-model uses
available observations from the public meteorological network. Currently, temperature and
precipitation are observed at 84 stations. The data are corrected to account for temperature-
and precipitation dependence on altitude. The current model uses interpolated data from
the three nearby stations and gives good results at the regional scale. However, for reliable
results, especially at the local scale, it is necessary to work with good-quality meteorological
data to relevant spatial resolution. A new model, which will be operative in summer 2007,
will use precipitation and temperature grids at a spatial resolution of 1km
2
, generated auto-
matically by the Norwegian Meteorological Institute (Engeset et al., 2004).
The soil texture, structure, steepness, vegetation, and soil frost, as well as the character-
istics of the underlying geological structure control the amount of groundwater recharge
and surface runoff. For the time being, the poor availability of reliable information on
geology and vegetation at a small scale (1: 50,000), strongly limits the ability to improve
the quality of the HBV-simulations. As the model structure allows physically-based principles
to be used for describing the various processes in the subsurface and at the land-surface-
atmosphere interface, the distributed HBV-model may be improved by including better sub-
routines for calculation of evaporation and soil-water deficit. Calibration procedures, which
take into, account all available observations (snow, soil moisture, groundwater), and not only
streamflow, should decrease the number of model solutions and, therefore, give better results.
A historical archive of daily gridded data from about 1960 to the present will be estab-
lished and available on the Internet (www.seNorge.no) using the navigation and presentation
system developed previously at NVE for snow maps (Engeset et al., 2004). The purpose of
the Groundwater and Soil Water System is to provide national and local stakeholders with
tools for assessing the past, current and forecasted groundwater and soil-water conditions.
This system promotes the use of groundwater and soil-water information in analysing the
development of unusual hydrological events and may therefore be considered a good tool
for evaluating groundwater-related risks (floods, drought, landslides, hydro-electricity short-
ages, water restrictions).
ACKNOWLEDGEMENT
This work is funded by the Norwegian Water Resources and Energy Directorate (NVE). The
authors wish to thank engineers in NVE for contribution with data monitoring and database
development. We are especially grateful to Rune Engeset who initiated the Snow Map
System for Norway and Jess Andersen for developing necessary tools for the web-based
presentation. The authors also thank the Norwegian Meteorological Institute (met.no).
REFERENCES
Beldring S, Engeland K, Roald LA, Slthun NR, Voks A (2003) Estimation of parameters in a dis-
tributed precipitation-runoff model for Norway. Hydrology and Earth System Sciences, 7, 304316.
Beldring S, Colleuille H, Haugen LE, Roald LA, verlie T (2005) Climate change impacts on hydro-
logical processes in headwater catchments. Headwater Control IAHC konferanse. Bergen, Juni 2005.
Bergstrm S (1995) The HBV model. In: Singh, V.P. (Ed.), Computer Models of Watershed
Hydrology. Water Resources Publications, Highlands Ranch, 443476.
Colleuille H, Dimakis P, Wong WK, Tvedten S (2006) Interactions between rivers and aquifers and
their impact on management of water resources in Norway. Nordic Hydrological Conference
2006. Velje Danmark. NHP report (article accepted).
Engeset R, Tveito OE, Mengistu Z, Udns HC, Isaksen K, Frland EJ (2004) Snow Map System for
Norway. XXIII Nordic Hydrological Conference. Tallin. NHP Report No.48, Tartu.
Engeland K, Skaugen TE, Haugen JE, Beldring S, Frland E (2004) Comparison of evaporation esti-
mated by the HIRHAM and GWB models for present climate and climate change scenarios.
Norwegian Meteorological Institute report no. 17/2004.
Jansson PE, Karlberg L (2004) Coupled heat and mass transfer model for soil-plant- atmosphere sys-
tems. Royal Institute of Technology, Dept of Civil and Environmental Engineering, Stockholm.
Kirkhusmo LA (1986) The use of groundwater monitoring data from the Nordic countries.
Norwegian Hydrological Programme, NHP report no. 19, Norway.
Pettersson LE (2004) Water Resources in Norway. Nordic Hydrological Conference. Tallin, Estonia.
NHP report No. 48, Tartu.
Van Lanen HAJ, Frendekov M, Kupczyk E, Kasprzyk A, Pokojski W (2005) Flow Generating
Processes. Chapter 3 in Hydrological Drought. Processes and estimation methods for streamflow
and groundwater. Ed. Tallaksen LM and van Lanen HAJ, Elsevier.
Algeria 138, 191, 196, 270
Alluvial 117, 121, 128, 273, 275276, 278,
446, 449, 549
Anisotropy 67, 89, 522
Aquiclude 145, 263, 383, 462, 468
Aquitaine 253254, 266, 379380, 497
Aquitard 118, 177, 179, 184185, 195, 254,
263, 265
Argentina 239, 241, 245
Australia 106, 113, 226
Basalt 128, 130132, 239, 241, 247, 249,
251
Bedrock 138, 154, 283, 393, 569
Biodegradation 433
Borehole Cuttings, Drill Cuttings 4748, 52,
259
Boundary Conditions 35, 179180, 184,
265, 313314, 426, 461
Brazil 239, 241, 244, 247
Burkina Faso 137138, 141, 144, 146
Calibration 52, 74, 77, 83, 163164, 167,
182, 185, 453, 457, 475, 581
Carbonate 27, 57, 6466, 106, 138, 145,
233, 235, 259, 289
Cation Exchange 260
Chalk, Craie 42, 65, 79, 84, 381, 477,
513514
China 117118, 122, 125
Chloride 232, 244, 255, 283, 327, 373, 460,
464, 467
Clay 56, 77, 227, 230, 232, 236, 254, 298,
435436, 442
Climate, Climatic 127, 149, 192, 210, 451,
488, 561, 569, 572
Coastal Aquifer 130, 459, 462, 466
Conductivity, Resistivity (Electric) 3637,
65, 73, 215, 244, 309, 312313, 337, 339,
368, 464
Conjunctive Use 161163, 170172
Contaminants, Contamination, Contaminated
225, 236, 276, 389, 392, 399, 409410,
412, 419420, 430, 435, 467, 477,
485
Crystalline 92, 138, 163, 393, 564, 569
Darcy 3, 25, 35, 44, 55, 92
Darcys Law 34, 13, 15, 1820, 26, 92, 123,
232
Darcy-Weisbach Equation 1112
Detritic 293
Diffusion 19, 48, 63, 65, 90, 98, 254,
263266, 410, 435
Digital Elevation Model (DEM) 153, 321,
449, 499501, 505
Distributed Model 304, 472, 474, 571
Djibouti 127, 130134
Drinking Water 26, 243, 253, 269, 303, 384,
459460, 463, 466468, 477, 483, 485,
513, 515, 524
Droughts 169, 172, 542, 549, 553554, 557,
561, 571
Economic, Economy 9, 37, 122, 127, 138,
161, 207, 269, 379, 380, 391, 459, 466,
468, 472, 476477, 483, 485
Ecotoxicology 402
Endorheic 128, 150, 501
Environmental Tracer 226, 229
Epikarstic Aquifer 325
European Water Framework Directive (WFD)
471, 541, 551, 472, 474, 476, 543,
545546
Evaporation 118, 130, 150, 159, 226, 231,
277, 355, 580
Fault, Fractures, Lineaments 58, 60, 145,
152, 177, 185, 281, 319, 323, 343,
366, 380
Flow Lines 194, 244, 253, 286, 298
Fluoride 133, 247, 262, 528, 531532
Foggara, Galleries 197, 274, 396
Index
France 74, 82, 87, 253, 254, 291, 317, 333,
459, 541, 543, 549, 552, 554
Geochemical Background 563
Geophysics 78, 81, 88, 365
Geostatistics 178, 527528, 538
Geothermal Gradient 376, 382, 384
Geothermometry, Geothermometric,
Geothermometer 373, 375376
Germany 484, 487489
GIS 153, 178, 180, 392, 396, 475, 499, 506,
571, 573, 575, 564
Grain-size 3638, 4042, 44, 436
Granite, Granitic 151, 209210, 214, 318,
366, 368, 393, 531, 527
Groundwater Age 295, 372
Groundwater Bodies 471472, 476, 484,
541543, 548
Groundwater Chemistry 226227, 232233,
255, 400
Guarani 239240, 244, 249251
Hard-rock Aquifer 82
Hydraulic Conductivity 3537, 71, 76, 154,
166, 185, 210, 216, 254, 382, 392, 464,
573
Hydraulic Tests 93, 339
Hydrochemistry, Hydrogeochemistry 144,
147, 244, 282, 287
Hydrothermal 128, 239, 247, 365366, 369,
372373, 375, 380, 382, 384
IDPR 498, 501, 505, 508, 519
India 113, 531
Interference Test 335, 342, 344
Intrinsic Permeability 36, 38, 40, 56
Inverse Modelling 96, 304
Irrigation 131, 162, 170171, 180, 468,
487
Isotope, Isotopic 118, 229, 233, 235, 262,
283, 287, 355, 365, 369370
Japan 281, 283
Karst, Karstic 82, 289, 303, 317
Kriging 522524, 530
Leakage 227, 231, 263, 265, 464
Libya 113, 191, 195196, 202
Limestone 151, 333, 342344, 359,
381, 501
Logging, Wireline Logging 55, 57, 5962,
337
Mali 138139, 141, 143
Management 71, 127, 137, 162, 303304,
333, 379, 467, 471, 511, 549, 560561
Mauritania 138, 149
Mediterranean 210, 226, 291, 293, 300,
459460, 462
Metals, Heavy Metals 419
Mineral water 351, 354, 379
Model, Modelling 19, 56, 90, 123, 161, 175,
191, 253, 303, 351, 379, 407, 445, 483,
522, 561
Monitoring 313, 323, 340, 511, 527, 542,
546, 565, 572
Multi-layer 144, 195, 225226, 254, 273,
449, 461, 468
Natural Attenuation 407, 410411, 416
Network 25, 67, 74, 92, 122, 138, 294, 298,
300, 312, 343, 370, 393, 446, 498500,
515, 522, 528, 530, 541, 543544, 546,
572
Nitrogen, Nitrate 92, 328, 471472,
474475, 477, 484485, 487, 492493
Noble Gases 254
Norway 569573, 575, 577578
Nuclear Magnetic Resonance (NMR) 47, 52,
6061, 6365, 72, 77, 79, 93, 435438,
442
Nuclear Waste, Radioactive Waste 87, 333
Packer Test 9396, 336, 339340
Paleo-recharge 195
Permeability 33, 36, 4748, 50, 52, 55, 57,
59, 60, 63, 6667, 92, 96, 98, 339, 439
Permeability Measurements 87, 94, 9899
Pesticides 484, 513, 519, 525526
Piezometer, Pieozmetry, Piezometric Map
75, 141142, 156, 194, 273, 294, 311,
413, 451, 545546, 549, 554555, 558,
561
Piper Diagram 398
Poiseuille 3, 1213, 1819
Pore size 37, 61, 6364, 72, 76, 79, 93, 440
Porosity 40, 42, 44, 51, 53, 61, 6364, 72,
76, 79
Portugal 207208, 351, 353, 393
Prony Equation 11, 19
584 Index
Index 585
Protection, Protection Zone 278, 317318,
329, 379, 402, 435, 477, 563
Proton Magnetic Resonance Sounding (MRS)
7174, 7677, 7984
Pulse Test 9394, 96, 98, 340, 343
Pumping Test 303, 309, 311, 338339, 382,
384
Quality, Water Quality 27, 122, 243, 304,
365, 379, 384385, 462463, 466, 471,
484, 517, 531, 555, 567, 575
Radiocarbon 295, 297, 372
Rainfall 117119, 143, 226, 306, 549550
Recharge 118, 128, 155, 162, 186, 191, 209,
227, 229, 453
Redox 253, 255, 259260, 353
Relative Permeability 50, 56
Remote Sensing 188
Reserves 71, 82, 105, 113, 175, 191, 195,
303304, 325, 484
Reservoir Model 304, 315
Rriver-Aquifer Interaction 166, 169
Sahara 106, 149, 155, 191192, 194195,
200201, 203
Salt, Salinity, Salinisation 232, 276277,
278, 291, 299, 446, 452, 464
Sand 15, 17, 56, 270, 283, 254, 435436
Sandstone 56, 63, 106, 128, 138, 145, 151,
157, 187, 198, 226, 241, 245
Saudi Arabia 175176, 188
Seawater Intrusion 459461, 464 467
Sedimentary Rocks 105106, 128, 175
Simulation 169, 172, 196, 199, 259, 360,
383, 411412, 428, 451, 475, 489, 574
Snowmelt 570, 573
Socio-economic, Soci-politic, Policy 468,
472, 475476, 484, 486, 489
Soil Moisture, Soil-water, Saturation 233,
236, 360, 419421, 428, 440, 569,
571573, 575, 578, 580
Solute transport 36, 44, 464
Solvents, Chlorinated Solvents 407,
409410, 413414, 416
Sorption, Desorption 98, 419422, 431
Spain 370, 409, 459
Specific Storage 166, 169, 184185, 346
Spring 1415, 2627, 30, 304, 306, 308,
323, 373, 542
Subsidence 113, 122, 283, 465
Taoudenni (Taoudeni) Basin 137139,
144147, 149, 151155, 160
Tectonic 284287, 323, 345
Temperature 1213, 9495, 180, 287, 312,
353, 359, 361, 365, 367, 369, 373, 375,
580, 396
Thailand 161162, 174, 420, 421
Thermal Water 365366, 368369, 371373,
379, 381, 387
Total Dissolved Solids (TDS) 118, 129130,
132, 135, 231, 244245, 248, 276, 420
Trace Elements 262, 563565, 567568
Tracer Tests 318, 321, 324
Transmissivity 55, 76, 118, 128, 314, 340
Tritium 93, 98, 228, 231, 370
Tunisia 191, 195196, 201
Underflow 273, 275277
Urban 81, 83, 120122, 127, 391393
Uruguay 239, 241, 243244, 247
Vadose Zone 207208, 210, 214, 216, 219
Validation 260, 304, 522, 531, 571, 575
Variogram 521524, 530531
Vulnerability 81, 83, 328, 330, 497, 503,
505506, 555
Wadi 128, 152, 182, 185, 187, 270, 273, 275,
277
Water Budget 304, 320, 330
Water Content 73, 82, 427, 438
Wells 21, 55, 130, 132, 152, 162, 251, 381,
384, 450, 513, 531532
Afonso, M.J. 391
Aguiar, C. 207
Ahmed, S. 527
Aires-Barros, L. 351
Al Shoaibi, A. 175
Andrade, M. 351
Andr, L. 253
Atteia, O. 253
Aunay, B. 459
Baba Sy, M.O. 191
Baird, D. 225
Bakalowicz, M. 303
Baltassat, J.-M. 71
Barraqu, B. 471
Basto, M.J. 351
Bejranonda, W. 161
Beldring, S. 569
Bertone, F. 149
Besbes, M. 191
Blavoux, B. 291, 317
Blum, A. 563
Bobeck, P. 25
Borges, F.S. 207
Boucher, M. 71
Bour, O. 407
Brenot, A. 563
Brusset, S. 317
Bruxelles, L. 317
Cadilhac, L. 563
Carreira, P.M. 351, 365
Carvalho, J.M. 207
Carvalho, M.R. 351, 365
Casper, M. 483
Chamin, H.I. 207, 391
Chery, L. 563
Chotpantarat, S. 419
Colleuille, H. 569
Constantino, A.T. 207
Costa, I. 391
Cruchaudet, M. 333
DElia, M. 239
Dakour, D. 137
Danneville, L. 317
de Fouquet, C. 513
de La Vaissire, R. 291
Delay, J. 87, 333
Delhomme, J.P. 55
Derouane, J. 137
Distinguin, M. 333
Drfliger, N. 303, 459
Duarte, J.M. 207
Dupuy, A. 379
Duvail, C. 459
Egermann, P. 47
Elsass, P. 445
Franceschi, M. 253
Garnier, C. 497
Girard, J.-F. 71
Goblet, P. 407
Goff, F. 351
Graa, R.C. 351
Graine-Tazrout, K. 269
Graveline, N. 483
Gravier, A. 497
Grelot, F. 459
Grimm-Strele, J. 483
Guilhermino, L. 391
Guimares, L. 391
Guimer, J. 407
Gutierrez, A. 71, 175
Han, Z. 117
Handa, H. 281
Haugen, L.-E. 569
Hayashi, T. 281
Incio, M. 207
Inamura, A. 281
Jalludin, M. 127
Jordana, S. 407
Kerrou, J. 149
Koch, M. 161
Koch-Mathian, J.-Y. 497
Koller, R. 483
Koontanakulvong, S. 161
Kumar, D. 527
Lachassagne, P. 71, 303
Ladouche, B. 303
Lalbat, F. 291
Lambrecht, H. 483
Le Fanic, R. 379
Le Gal La Salle, C. 225
Le Strat, P. 459
Ledoux, E. 471
Legchenko, A. 71
Lenoir, P. 303
Lenormand, R. 47
Lepiller, M. 317
Love, A. 225
Luzio, R. 365
Macini, P. 435
Makino, M. 281
Mangin, A. 317
Mansour, H. 269
Marchet, P. 317
Mardhel, V. 497
Marchal, J.-C. 303
Margat, J. 105
Marques, J.E. 207
Marques, J.M. 207, 351,
365, 391
Martins, A. 207
Matias, H. 365
Matias, M.J. 351
Mekkaoui, A. 269
Mengistu, Z. 569
Merzougui, T. 269
Author index
Mesini, E. 435
Moix, P. 149
Monget, J.-M. 471
Montginoul, M. 459
Munaf, S. 175
Nabi, A. 527
Nakamura, T. 281
Nowak, C. 497
Noyer, M.-L. 71, 445
Nunes, D. 365
Owais, S. 527
Paris, M. 239
Pelte, T. 563
Perez, M. 239
Perrochet, P. 149
Poitrinal, D. 175
Pouchan, P. 379
Putthividhya, A. 419
Quiot, F. 407
Raguet, M. 497
Renard, D. 513
Renard, P. 149
Rinaudo, J.-D. 459, 483
Rocha, F.T. 207, 391
Rocha, L. 351
Rollin, C. 407
Ruiz, E. 407
Samper, J. 207
Santos, F.A.M. 365
Schafmeister, M. 35
Schwartz, J. 407
Seabra, C. 391
Simmons, C. 3, 225
Sonney, R. 563
Suthidhummajit, C. 161
Takahashi, H. 281
Takahashi, M. 281
Teixeira, J. 207, 391
Terreyre, J.-L. 497
Tujchneider, O. 239
Valarie, I. 303
Van Dijk, P. 483
Vandenbeusch, M. 175
Verjus, P. 553
Vernoux, J.-F. 541
Viavattene, C. 471
Viola, R. 435
Vouillamoz, J.-M. 71
Wijnen, M. 175
Wong, W.K. 569
Yasuhara, M. 281
588 Author index
SERIES IAH-Selected Papers
Volume 1-4 Out of Print
5. Nitrates in Groundwater
Edited by: Lidia Razowska-Jaworek & Andrzej Sadurski
ISBN Hb: 90-5809-664-5
6. Groundwater and Human Development
Edited by: Emilia Bocanegra, Mario Hrnandez & Eduardo Usunoff
ISBN Hb: 0-415-36443-4
7. Groundwater Intensive Use
Edited by: A. Sahuquillo, J. Capilla, L. Martnez-Cortina & X. Snchez-Vila
ISBN Hb: 0-415-36444-2
8. Urban Groundwater Meeting the Challenge
Edited by: Ken F.W. Howard
ISBN Hb: 978-0-415-40745-8
9. Groundwater in Fractured Rocks
Edited by: J. Krsny & John M. Sharp
ISBN Hb: 978-0-415-41442-5
10. Aquifer Systems Management: Darcys Legacy in a World of Impending Water
Shortage
Edited by: Laurence Chery & Ghislaine de Marsily
ISBN Hb: 978-0-415-44355-5
Forthcoming:
11. Groundwater Vulnerability Assessment and Mapping
Edited by: Andrzej J. Witkowski, Andrzej Kowalczyk & Jaroslav Vrba
ISBN Hb: 978-0-415-44561-0
12. Groundwater Flow Understanding
Edited by: J. Joel Carrillo R. & M Adrian Ortega G.
ISBN Hb: 978-0-415-43678-6

Aquifer Systems Management Darcy's Legacy in A World of Impending Water Shortage

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Aquifer Systems Management Darcy's Legacy in A World of Impending Water Shortage

Uploaded by

Copyright:

Available Formats

AQUIFER SYSTEMS MANAGEMENT

DARCYS LEGACY IN A WORLD OF IMPENDING WATER SHORTAGE

36 Aquifer Systems Management: Darcys legacy in a world of impending water shortage

40 Aquifer Systems Management: Darcys legacy in a world of impending water shortage

The volume of initial gas V

The gas saturation inside the cuttings S

The permeability K, adjusted using the curvature of the relaxation curve.

Measuring porosity and permeability on drill cuttings 51

size of cuttings down to 1mm diameter,

volume of cuttings around 1cm

permeability from the microDarcy to around 100mDarcy,

Containment capability of the host formation

Geomechanical properties Creation of damaged and disturbed zones associated with

Regional knowledge of geological and hydrogeological properties of the host rocks

A method consisting of interpreting the evolution of saturation profiles in an argilla-

Chemical effects due to interactions between drilling and formation fluids;

Physical-chemical and hydromechanical effects due to interactions between the test

Mechanical effects due to stress release (borehole-wall creep).

An initial passive pressure recovery and;

Steady-state tests (stationary);

Transient-state tests of the pulse-test type.

various hydrogeological ridges in Burkina Faso reveal different hydrogeological recharge

the piezometry indicates four main hydrogeological basins:

finally, on the subject of common cross-border water-resource management, note that

providing a complete 3-dimensional representation of the extent and boundaries of the

defining hydrogeological layers which could be used in a mathematical model to sim-

the SQG geological layer pinches out in a north-eastern direction;

EarthVision (Dynamic Graphics Inc.), a 3D geological modeller used for computing

A simplified, but representative, hydrogeological succession was defined. The object-

As far as possible, homogeneous hydraulic-parameter distributions were assigned to

The definition of a realistic conceptual groundwater model involving a stack of 13

An accurate reconstruction of the agricultural-abstraction time series in each cell of the

Group 1 Subgroup 1A involves water samples from: Federacin, Coln, Concordia

Group 1 Subgroup 1B is represented by Coln and Arapey groundwater samples with

Group 2 Subgroup 2A is defined by water samples of Villa Elisa, Gualeguaych,

Calcium carbonate equilibrium. A dominant Ca-HCO

(see, e.g., Back, 1966;

. In the Eocene sands, clay beds of variable

Dissolution of the salts in the Quaternary sand formations.

Intense evaporation (2153mm/y).

Reduction in input from the Erg caused by pumping.

Recycling of waste water and using it to irrigate the palm field.

) concentration indicates that groundwater with high Cl

-rich groundwater, spreading from the northwest to south-

-rich groundwater is also characterized by low D,

concentrations of groundwater in the tectonic zone.

-rich groundwater in more detail, and helps characterize the geolog-

) concentration (Fig. 2) clearly indicates that ground-

concentrations of more than 100mg/l occurs in Shobu, Ina, Shiraoka,

-rich groundwater is in good agreement with

concentration observed in this study is generally increased by

-rich groundwater zone, spreading from northwest to southeast,

contents of groundwater outside the Motoarakawa tectonic zone (such as in Washimiya,

concentration across the Ayasegawa and Kuki faults

) concentration. Unlike the Cl

concentration is observed between the groundwater

-rich deep-seated groundwater is

concentration of the groundwater in the tectonic zone.

still remains unflushed and, in

concentration, low D, low

before pumping due to initial

The measurement of the hydraulic heads before shaft sinking,

The characterisation of the overall transmissivity in the four boreholes,