You are on page 1of 233

Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.

org/
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

GEOPHYSICAL MONOGRAPH SERIES

NUMBER 17

FUNDAMENTALS OF
GRAVITY EXPLORATION

Thomas R. LaFehr
Misac N. Nabighian

Wei Liu, managing editor

Edward K. Biegert and Michal Ruder, volume editors

The international society of applied geophysics


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

ISBN 978-0-931830-56-3 (Series)


ISBN 978-1-56080-298-3 (Volume)

Society of Exploration Geophysicists


P. O. Box 702740
Tulsa, OK 74170-2740

© 2012 by Society of Exploration Geophysicists


All rights reserved. This book or parts hereof may not be reproduced in any
form without written permission from the publisher.

Published 2012
Printed in the United States of America

Cover figure courtesy of Guy Flanagan. Used by permission.

Library of Congress Cataloging-in-Publication Data

LaFehr, Thomas R., 1934- author.


Fundamentals of gravity exploration / Thomas R. LaFehr, Misac N. Nabighian ; Wei Liu,
managing editor ; Edward K. Biegert and Michal Ruder, volume editors.
pages cm. -- (Geophysical monograph series ; no. 17)
Includes bibliographical references and index.
ISBN 978-1-56080-298-3 (volume : alk. paper) -- ISBN 978-0-931830-56-3 (series : alk.
paper)
1. Gravity--Measurement. I. Nabighian, Misac N., author. II. Liu, Wei, 1969- editor. III.
Biegert, Edward K., editor. IV. Ruder, Michal, editor. V. Title.
QB334.L34 2012
526’.7--dc23
2012041177
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Dedicated to our wives, Arlys LaFehr and Aida Nabighian


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

This page has been intentionally left blank


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Contents

About the Authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii


Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

Chapter 1: Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Chapter 2: Principles of Attraction and Earth’s Gravity Field. . . . 5


Gravitational force. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Gravitational constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Gravitational potential. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
The earth’s gravity field. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
The geoid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
The standard International Gravity Formula. . . . . . . . . . . . . . . . . . 11
GPS and the geoid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Chapter 3: The Gravitational Potential and Attraction


of Mass Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Attraction of a spherical shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Components of attraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Analysis of potential fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Gravity calculations for simple geometries . . . . . . . . . . . . . . . . . . 25
Gravity calculations for 2D geometries. . . . . . . . . . . . . . . . . . . . . . 34
The logarithmic potential. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Green’s equivalent layer and the problem of ambiguity . . . . . . . . . 44

Chapter 4: Field Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49


Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Absolute-gravity measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Relative-gravity instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

v
Fundamentals of Gravity Exploration
vi  
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Gravity gradiometry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Field operations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Measurement uncertainty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Ambiguity related to survey design — Aliasing. . . . . . . . . . . . . . . 65

Chapter 5: Rock Density and Gravity Anomalies. . . . . . . . . . . . . . . 67


Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Typical near-surface rock densities. . . . . . . . . . . . . . . . . . . . . . . . . 67
Density and porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Constituent densities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Methods for deriving, measuring, and evaluating density. . . . . . . . 72
Definition of what causes a gravity anomaly. . . . . . . . . . . . . . . . . . 78

Chapter 6: Data Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81


Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Reduction of gravity survey data. . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Appendix A — Bullard correction. . . . . . . . . . . . . . . . . . . . . . . . . . 98

Chapter 7: Anomaly Interpretation Guidelines


and Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Purposes of gravity surveys. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Gravity calculations for an arbitrary model. . . . . . . . . . . . . . . . . . . 102
The fast-Fourier transform for calculating gravity effects. . . . . . . . 105
Anomaly shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Anomaly separation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
Spectral analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Depth determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Determination of anomalous mass. . . . . . . . . . . . . . . . . . . . . . . . . . 136
Interpretation of borehole gravity . . . . . . . . . . . . . . . . . . . . . . . . . . 138
Reservoir monitoring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Appendix A — The unit half-width circle (2D)
and ellipse (3D). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Appendix B — Application of Bott and Smith theorems . . . . . . . . 147
Appendix C — Corrections for incomplete integration
using Gauss’ theorem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Appendix D — Borehole-gravity distance/thickness
relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Contents  vii
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 8: Inversion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157


Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Density inversion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Geometric (boundary) inversion . . . . . . . . . . . . . . . . . . . . . . . . . . . 166

Chapter 9: Geologic Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . 169


Introduction to interpretation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Location of buried features by filtering and/or modeling . . . . . . . . 173
Example of salt with caprock. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Examples of seismic pitfalls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Example of borehole gravity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Borehole gravity in combination with surface gravity. . . . . . . . . . . 181
Integration of seismic and/or magnetic information
with gravity data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Mining applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Satellite gravity and satellite-derived gravity . . . . . . . . . . . . . . . . . 190

Appendix A: Fourier Transform. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199

Index. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

About the Authors

Thomas R. LaFehr received an A.B.,


College of Letters and Science, from the
University of California (Berkeley) in 1958;
an M.Sc. in geophysics from Colorado School
of Mines in 1962 (while working at the U. S.
Geological Survey); and a Ph.D. in geo-
physics from Stanford University in 1964.
He was employed as a geophysicist by Grav-
ity Meter Exploration Company in Hous-
ton from 1964 to 1969. At CSM from 1969
through 1992, LaFehr was an associate,
adjunct, and full professor and George Brown
Professor, taking leaves of absence during
which he was founder, consultant to, president, and chairman of EDCON.
He was also founder, president, chairman, and consultant for LCT. Since
1998, he has been a Distinguished Senior Scientist at Colorado School of
Mines.
LaFehr has published in Geophysics, the Journal of Geophysical Re-
search, the Bulletin of the Geological Society of America, and the Austra-
lian Oil and Gas Journal. He won three awards for best presentation at the
SEG annual meeting, was the SEG distinguished lecturer in 1971 and editor
in 1972–1973, was elected to honorary membership in 1979, was president
in 1983–1984, and received the Maurice Ewing Medal in 1997.

viii
About the Authors  ix
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Misac N. Nabighian received a B.Sc.


(honors) degree in geophysics in 1955 from
the Institute of Petrol and Gas in Bucharest,
Romania, and a Ph.D. in geophysics in 1967
from Columbia University in New York. He
began his career in Romania in 1955, first as
a party chief and then as an assistant profes-
sor at the Institute of Petrol and Gas. After
obtaining his Ph.D., he was employed by
Newmont Mining to carry out research and
develop new interpretation techniques in
various areas of mining geophysics. During
Nabighian’s tenure at Newmont, he devel-
oped, among others, the concept of the analytic signal for interpreting poten-
tial-field data and the “smoke-ring” concept to help interpret time-domain
electromagnetic methods. He retired from Newmont in 1997 and since then
has been a Distinguished Senior Scientist at Colorado School of Mines.
Nabighian is an honorary member of SEG and was the first recipient of
the Gerald Hohmann Award for excellence in electromagnetics. He was edi-
tor of the two-volume SEG publication Electromagnetic Methods in Applied
Geophysics and editor of the special issue of Geophysics devoted to time-
domain electromagnetic methods. He also served two terms as an associate
editor of Geophysics.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

This page has been intentionally left blank


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Preface

“Knowledge of the gravity field of the earth is important in the study


of our globe,” wrote W. Heiskanen and F. A. Vening Meinesz in 1958. We
would add that such knowledge is also very important in the study of local
earth features found in mining, petroleum, environmental, and other explo-
ration venues. This book is intended to aid all earth scientists engaged in
such studies.
Where we describe and emphasize analytical techniques, we do so in the
firm conviction that in their understanding lies the basis for future economic
discoveries and that an understanding of the limitations of techniques is as
important as their applicability. As discussed in this book, a rich mathemati-
cal substance is the basis for clearer understanding and innovative explora-
tion tools, but equations do not replace geologic breadth and scope.
No better example of the use of mathematics can be found than the de-
scription of the problem of ambiguity, eloquently described by Green’s iden-
tities. This eloquence, however, would be lost without a thorough grounding
in the acceptable solutions in practical geologic terms.
We have tried separately to acknowledge the uncountable sources from
which this book is derived. Some of our sources, such as our early teachers
and colleagues, living and gone, contributed to the framework with which
we view gravity methods and to our appreciation for their limitations.
Although many such sources will remain unknown to our readers, everyone
can appreciate the concept of community synergism and multiple collabora-
tions. We would therefore like to express our thanks to all those who have
been involved in our small but dynamic discipline.

xi
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

This page has been intentionally left blank


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Acknowledgments

Most of the material in this book is based on courses taught by both


authors at various times in the Geophysics Department, Colorado School
of Mines (CSM). We felt that a proper balance in presenting the gravity
method can be achieved by combining one author’s experience in the petro-
leum industry (TRL) with the other’s experience in the mining industry
(MNN). The final format of the book, however, was strongly influenced
by the excellent technical editing provided by Ed Biegert from Shell E&P
Technologies and Michal Ruder from Wintermoon Geotechnologies. We are
both extremely grateful for their patience and excellent suggestions and for
providing additional technical material when needed.
We are also indebted to Guy Flanagan of ConocoPhillips, who provided
the satellite gravity data and helped in writing the description of the tech-
nique. The chapter on inversion could not have been written without the
help of Yaoguo Li from CSM, who patiently guided us in streamlining this
important chapter. Mike Thomas from the Geological Survey of Canada,
Jeremy Cook from Newmont, and Jules Lajoie from Comtek Enterprises
Ltd. helped in obtaining some data related to mining applications. Camriel
Coleman, a student at CSM, helped in imaging the Heath Steele Stratmat
magnetic data, and Dionisio Uendro Carlos from Vale Mining Company
helped with sketching an important figure. Mark Ander from Ander Lab-
oratory LLC provided valuable information on various aspects of gravity
instrumentation.
Ed Biegert suggested the inclusion of various summarizing tables in
Chapter 3, and they were modeled after similar tables in the Russian gravity
handbook Gravirazvedka. Some material was extracted, sometimes verba-
tim, from the authors’ paper “Historical development of the gravity method
in exploration” (Nabighian et al., 2005), and as such, special thanks are due
to the other authors of that publication: Mark Ander, Tien Grauch, Richard
Hansen (deceased), Yaoguo Li, William Pearson, John Peirce, Jeff Phillips,
and Michal Ruder. The authors also acknowledge the contributions from
their cumulative experiences at GMX, EDCON, and LCT (TRL) and New-
mont Mining Co. (MNN).

xiii
Fundamentals of Gravity Exploration
xiv  
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

We also thank Ted Bakamjian, Jennifer Cobb, and the rest of the SEG
gang, our longtime friend, Jerry Henry, and our newfound buddy, Rowena
Mills, who ably performed as special editors.
Some of the material in this book is based on innumerable presentations
under the auspices of SEG and AAPG.
Finally, we would be remiss if we failed to acknowledge the enduring
support of our wives, Arlys and Aida, to whom this book is dedicated.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 1

Introduction

The goal of this book is to provide information about the principles,


understanding, and applicability of the gravity exploration method. This
book is intended to be suitable for classroom instruction and as a refer-
ence for anyone engaged in geophysical exploration, including those whose
specialties might be in another discipline but who would benefit from an
understanding of how gravity exploration can help them solve exploration
problems. For many decades, the 1971 SEG book by L. L. Nettleton (Geo-
physical Monograph Series No. 1, Elementary Gravity and Magnetics for
Geologists and Seismologists) has helped to fill this need, but it is limited
in scope (as its title implies) and is, of course, out of date, especially with
respect to modern exploration technology. This little book has been a best
seller, however, and it resides in the libraries of thousands of geologists and
geophysicists. It contains several classical and practical examples of how
the gravity method can be applied, and we have borrowed liberally from
these where they retain their long-held value. In 1995, Richard J. Blakely
published Potential Theory in Gravity and Magnetic Applications. This
book covers in depth much of which the Nettleton monograph lacks: the
principles of potential theory and the mathematical basis for the forward and
inverse techniques of interpretation.
Our book is intended to fill a need that is oriented more toward explo-
ration than the Nettleton monograph or the Blakely book, with more infor-
mation about the underlying principles and technology than the former
and clearer orientation toward the explorationist’s geologic goals than the
latter.
We expect that a relatively small minority of our readers will have an
interest in the mathematical basis that underlies the fundamentals of the
gravity method presented in this book. Readers who are in this minority,

1
2  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

however, require the mathematics. The majority of readers will have no


direct interest in the mathematics that underlie the gravity method. Nev-
ertheless, they can obtain the basic meaning of the subject matter. Some
explorationists will wish to “read through” the mathematics on first encoun-
ter and then later, at their own pace, study the mathematical basis in depth.
We have not, however, included laborious or copious derivations showing
how solutions to integrals can be derived, relying instead on the excellent
tables of integrals available at most good libraries (e.g., Gradshteyn and
Ryzhik, 1980). The same solutions can also be obtained using modern com-
puter software, which continues to evolve into ever more useful algorithms.
An example of the seeming contradiction of “requiring the mathemat-
ics” without needing it to understand the subject matter is given in the last
section of Chapter 3 (Green’s equivalent layer and the problem of ambigu-
ity). Most readers will feel well justified in skipping over Green’s identities,
but they would miss an important basic element in gravity exploration if
they also chose to ignore the entire section or even the chapter. The impor-
tance of mathematics can be obtained by reading the text. In combination
with the illustrations, the description of the problem of ambiguity gives the
reader an important understanding of how the gravity method should and
should not be applied.
investigation into the existence and behavior of gravity has a long and
distinguished history, starting at least as far back as the famous experiments
of Galileo (1564–1642) in the sixteenth century. Born in the year that Gali-
leo died (1642), Isaac Newton (1642–1727) was the creator of what is now
called classical mechanics or Newtonian physics, which forms the basis for
a rich wealth of mathematics relevant to gravity fields and their potential.
about the time of the Copernican Revolution, when the field of astron-
omy started its journey toward a better understanding of planetary motion,
observers such as Johannes Kepler (1571–1630) laid the groundwork for
Newton and those who followed him. Pierre-Simon, Marquis de Laplace
(1749–1827), wrote his monumental Celestial Mechanics (starting in 1799)
in five volumes, thus creating a very early underpinning to modern gravity
exploration, which began very late in the eighteenth century.
Baron Loránd Eötvös (1848–1919), in addition to his extensive interest
in law and politics, was a pioneer in the embryonic beginnings of mod-
ern gravity exploration, principally by inventing the torsion balance. Much
of the first geophysical exploration work was performed near the Gulf of
Mexico, and the first oil discovery based on geophysics was at the Nash
Dome in Texas following a torsion-balance survey there in 1924.
This book contains nine chapters, starting with this introduction and
ending with geologic applications. To interpret a gravity anomaly in geologic
Chapter 1:  Introduction  3
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

terms, we must first define it: the difference between the measured value and
the value predicted by a specific earth model. The nature of those measure-
ments, the details of the earth models, and the tools for examining the behav-
ior of gravity fields are the subject of this book. Generally, the specified earth
model (Chapter 6) is a simplification of the actual earth, and the resulting
anomaly field contains both regional and residual effects, the separation of
which remains as part of the interpretation (Chapters 7 through 9). This leads
to a second usage of the term anomaly: a series of values whose departure
from a normal or regional field is caused by the target(s) of interest.
For the student, the chapters are intended to be read in the sequence pre-
sented, and for the practitioner, in whatever order relevant to the exploration
problem at hand.
Chapter 2, “Principles of Attraction and Earth’s Gravity Field,” intro-
duces the reader to Newton’s gravitational force, the notion of potential,
the ellipsoid and the geoid, and the standard International Gravity Formula.
Chapter 3, “The Gravitational Potential and Attraction of Mass Distribu-
tions,” provides a mostly mathematical basis for a variety of mass sources,
along with an understanding of Laplace’s equation and its applicability both
inside and outside general distributions. It concludes, as discussed above,
with a discussion of Green’s equivalent layer and the problem of ambiguity.
Chapter 4, “Field Measurements,” discusses field operations, the acqui-
sition of absolute and relative gravity data, survey design, and the problem
of measurement uncertainty.
Chapter 5, “Rock Density and Gravity Anomalies,” describes typical
near-surface rock densities and how they are affected by porosity and its ten-
dency to decrease with depth owing to compaction. This chapter also treats
the variety of constituent mineralogy and its importance to rock density,
methods for determining density, and — in exploration — the all-important
differences in density.
Chapter 6, “Data Reduction,” recognizes that gravity observations are
strongly influenced by the nature of the field survey and the several environ-
ments that contribute to gravity variations unrelated to the geologic targets
we wish to study. Infield reductions, corrections for changes in elevation,
and the motion of the meter (if located in a moving vehicle, such as a marine
vessel or aircraft) are treated from the practical point of view of the explora-
tionist. The concept of isostasy is also included in Chapter 6.
Chapter 7, “Anomaly Interpretation Guidelines and Limitations,” starts
with the Bouguer anomaly, defined in Chapter 6, and proceeds to develop
the methods for analysis available to the explorationist, including anomaly
separation, depth-estimation rules, anomalous-mass determination, forward
calculations, the fast Fourier transform, and borehole gravity interpretation.
4  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 7 has four appendices dealing with the unit half-width circle, limit-
ing or maximum possible depth, corrections in the use of Gauss’ theorem,
and borehole gravity distance/thickness relationships. The purpose of the
appendices is to provide more detail without adding to the complexity of
the basic chapter.
Chapter 8, “Inversion,” treats a special subset of interpretation: the con-
struction of a geologic model based on a numerical procedure for which the
residual gravity anomaly is the input data set. Density inversion and geo-
metric boundary inversion are investigated in Chapter 8.
Chapter 9, “Geologic Applications,” covers a variety of important geo-
logic circumstances under which the gravity method has been and contin-
ues to be successful. These include salt structures, caprock, seismic pitfalls,
faults, borehole gravity, integration with seismic and magnetic data, and the
location of buried targets in mining applications.
Our wish is for the beginning student and the geophysicist early in his
or her career to develop an excitement for the technology as well as a firm
understanding of its applicability and limitation in the field of geologic
interpretation. For the seasoned interpreter, we hope the book will find a
place on the bookshelf to which one can turn in seeking answers to gravity
exploration questions.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 2

Principles of Attraction
and Earth’s Gravity Field

Gravitational force
Consider two small masses, M0 and M1, whose radii are very small in
comparison with the distance r between the masses (Figure 1). Newton’s
universal law of attraction states that each mass attracts the other by a force
F, whose amplitude F is in direct proportion to the product of their masses
and inversely proportional to the square of the distance between them:

M 0 M1
F=k , (1)
r2

where k is the universal gravitational constant equal to 6.67 × 10–11 m3/kg∙s2


in MKS units (i.e., distances measured in meters, mass in kilograms, and
time in seconds) or 6.67 × 10–8 cm3/g.s2 in cgs units (centimeters-grams-
seconds). The force caused by M1 acting at and on M0 is directed along r
toward M1; an equivalent force acting on M1 by M0 also is directed along r
but in the opposite direction, toward M0. The resulting force F is a vector
quantity given by

M 0 M1r
F = –k , (2)
r3

whose magnitude is stated in equation 1 and whose direction is in the direc-


tion toward the causative body. The negative sign is introduced to indicate
that r is measured from the source mass to the point of observation, whereas
the force F is directed in the opposite direction (i.e., toward the source).

5
6  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 1. Attraction between two small masses separated


M1
by distance r, with r̂ the unit vector directed from
F
r gravitational source to the observation point.
v

r
M0

We will consider M0 to be a mass residing at our point of observation


and M1 to be the source mass causing a field at the point of observation.
Newton’s second law states that a body’s acceleration is equal to the ratio of
the force acting on the body to its mass. Denoting the gravitational accelera-
tion with g, we have

F Mr
g= = – k 31 . (3)
M0 r

As Galileo1 observed in 1632 (without the benefit of the inverse square law),
the acceleration of a body is independent of its mass.
In the MKS system of units the gravitational acceleration is measured
in meters per second squared, whereas in the cgs system, it is measured
in centimeters per second squared. The cgs unit of gravitational accelera-
tion is known as a Gal (after Galileo), with 1 Gal = 1 cm/s2. Although only
one part in approximately 980 of the earth’s normal field, the Gal is much
too large for exploration work; hence, the milligal (1 mGal = 10−3 Gal) is
commonly used in surface exploration surveys and the microgal (1 µGal =
10−6 Gal) in borehole gravity work and in 4D gravity surveys (the fourth
dimension being time; see Chapter 9). In the geophysical literature, one also
encounters gravity data given in gravity units (1 g.u. = 0.1 mGal).

Gravitational constant
3
By replacing M1 in equation 1 with the mass of the earth, 4 3 π R ρm, where R
is the radius of the earth and rm is its mean density, we obtain that the product of
the gravitational constant k and the mean density of the earth is given by

3g
kρm = .
4π R
Hall (1963). Galileo’s experiments on acceleration were at least 10 years before
 1

Newton’s birth and more than 50 years before the publication of the Principia
Mathematica.
Chapter 2:  Earth’s Gravity Field  7
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The earth’s gravity g and its radius R can be measured so that the measure-
ment of either k or rm will lead to the determination of the other. Pierre
Bouguer in 1740 (see chapter 6) led an expedition to Peru (now Ecuador)
to determine arcs of the earth’s curvature at the equator which, although an
indirect requirement in their geodetic work, might have been the earliest
attempt to determine rm. His approach was to measure deviations in the
plumb line as affected by the high mountains in the Andes, but their mea-
surements were influenced strongly by isostatic effects (see chapter 6) that
were unknown at that time.
The earliest attempts to determine the constant k (which as we have seen
above leads to the determination of rm) were conducted in the laboratory by
Henry Cavendish. He used a torsion balance consisting of two small weights
that were deflected by interchanging the positions of two larger weights.
His result of 6.754 × 10−11 m3/kg∙s2 has been improved on in the succeeding
more than 200 years. However, this constant remains poorly determined in
comparison with any other basic physical constant because the gravity field
is much weaker than other fundamental forces.
The value accepted by the geophysical community is 6.67 × 10−11
m3/kg∙s2 or 6.67 × 10−8 cm3/g∙s2 in the cgs system2. For calculations desired
in milligals using length dimensions in kilometers and density in cgs units,
the factor is simply 6.67. For density expressed as kilograms per meter
cubed, the factor is 6.67 × 10−3. Although this value for k results in an aver-
age density for the earth of 5520 kg/m3, which is confirmed by independent3
means, its laboratory determination nonetheless contains a large uncertainty
in comparison with the other physical constants of the universe. However,
this issue is of very little concern in exploration work, as we will examine
in Chapters 6, 7, 8, and 9.

Gravitational potential
From equation 3, it follows that ∇ × g = 0. By Stokes’ theorem, this is
equivalent to g.dr = 0, which is a statement that the work done in moving
a unit mass is independent of the path taken. Such fields are called conser-
vative, and they can be represented as the gradient of a potential U. The
problem being spherically symmetrical for a point mass, we can express the
gradient of the potential in spherical coordinates, to obtain

 2Fixler (2007) reports a value of 6.693 × 10−11 with a standard error of the mean
of ± 0.027 × 10−11.
 3Verhoogen (1970, p. 617) reports an average earth density of 5517 kg/m3.
8  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

∂U 1 ∂U 1 ∂U
g = ∇U = lr + lθ + lϕ,
∂r r ∂θ r sin θ ∂ϕ

where lr, lq , and lj are unit vectors in the r, q, and j directions, respectively.
Because of angular symmetry, the derivatives with respect to q and j vanish,
and this equation reduces to

∂U M
l = – k 2 lr .
∂r r r

The solution to the above equation is simply

M
U=k + C.
r

We evaluate the constant C of integration by requiring U (the gravitational


potential) to vanish at infinity and obtain

M
U=k . (4)
r

To calculate the potential at an exterior point P for an arbitrary 3D body,


we integrate the potential of each elemental mass dm = r dV = r dw dh dV
over the entire volume V (Figure 2):

ρ ( ξ , η , ζ ) d ξ d η dζ
U ( x , y, z ) = k ∫ 1 . (5)
V (ξ – x )2 + ( η – y )2 + (ζ − z )2 2
 

The potential is a scalar quantity, thus simplifying some of our opera-


tions. Surfaces of equal potential are known as equipotential sur-
faces, and the gravitational attraction is always perpendicular to these
surfaces.

The earth’s gravity field


The earth is surrounded by its own gravitational field, which exerts
an attractive force on all objects. If the earth were a stationary nonrotat-
ing spherical body, then the strength of its gravitational attraction would
Chapter 2:  Earth’s Gravity Field  9
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

be constant over the surface of the x

earth, proportional to the earth’s


mass, and inversely proportional to
the square of the distance from the
center of the earth (see “Attraction y
of a spherical shell’’ in Chapter 3). dm ( , , ) = dv P (x,y,z )
The earth, however, is rotating
around its axis, which creates a cen-
trifugal force at every point on its
surface, being largest at the equa-
tor (about 3.4 Gal) and decreasing
toward zero as we approach the
poles (Figure 3). Thus, as a result
of rotation, the earth’s gravity field
will decrease from poles to the
equator.
The rotational potential caused z
by the centrifugal force is given by
Heiskanen and Moritz (1967) as Figure 2. Calculation of gravitational
potential at an exterior point P for
an arbitrary 3D body located in a
2 w2 r2, Cartesian-coordinate system.
where w is the angular velocity (7.292 × 10−5 rad/s) and r is the axial radius
shown in Figure 3. The maximum centrifugal force is less than 1/300 of earth’s
gravitational attraction.
The effect of the rotating earth, in addition to the centrifugal force,
results also in a flattening of the earth, with the final result being that the
earth now has a spheroidal shape (Figure 3). The equatorial radius of the
earth is Re = 6378.160 km, whereas the polar radius is Rp = 6356.775 km,
resulting in a flattening f given by

Re – Rp 1
f= = .
Re 298.257

Because of the difference between the polar and equatorial radii, it fol-
lows from equation 1 that the spheroidal earth will yield a larger gravi-
tational attraction at the poles compared with the equator. The combined
centrifugal force and flattening effects result in a difference of approxi-
mately 5.3 Gal between observation points at the equator and poles.
10  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The geoid
As mentioned above, the earth’s gravi-
tational field is normal to an equipotential
r
surface and defines the vertical at any loca-
P
F
tion. An equipotential surface of particular
g
interest is the one that coincides exactly
a
with the mean ocean surface of the earth
(assuming there are no tides or ocean cur-
rents) and extended through the continents
(such as with very narrow canals). The fact
that the mean ocean surface of the earth is an
equipotential surface for the earth’s gravita-
tional field can be explained by the fact that
Figure 3. The gravitational if this were not the case, then one would
force F from a nonrotating have a horizontal component of the earth’s
earth and the centrifugal force P gravitational field acting on the ocean water
combine to yield the observed and creating a gravitational current which
gravitational force g. is known not to exist (as opposed to known
ocean currents, e.g., the Gulf Stream).
This equipotential surface is known as the geoid, and it plays an important
role in gravity exploration (Figure 4). If we imagine a uniform rotating earth
(with oceans filled with rocks that have the same density as the continents and
the continents leveled to sea level), we would view an oblate ellipsoid, described
in the previous section and sometimes referred to as the spheroid. For such a uni-
form earth, the geoid and ellipsoid would be identical. The ellipsoid (or spheroid)
is an imaginary surface because the earth is irregular. The geoid departs from the
ellipsoid by dropping below it in oceans where seawater is less dense than the
rocks and by rising above it in continents where the mass per unit area increases.
The geoid is in essence an equipotential surface of the actual gravita-
tional field. The geoid is an irregular surface influenced by the underlying
masses. In the vicinity of a local excess mass which adds a potential ΔU to
the normal earth’s potential, the surface must warp outward to keep the total
potential constant (Figure 5).
Because of its complexity, the geoid is approximated by a rotating
oblate spheroidal surface of uniform density which, being very similar in
shape to an ellipsoid of revolution, is called the reference ellipsoid. The dif-
ference in height between the geoid and reference ellipsoid at most localities
is less than 50 m, with some exceptions. The gravitational field of the refer-
ence ellipsoid is known as normal or theoretical gravity and is used for the
computation of the gravity anomaly by removing the effect caused by the
earth’s oblateness and centrifugal acceleration.
Chapter 2:  Earth’s Gravity Field  11
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Deflection of the vertical

Plumb line

Topography

Ellipsoid

Geoid Geoid and sea level


Sea bottom

Figure 4. The geoid and the reference ellipsoid in relation to the earth’s
topography. The geoid coincides with the mean sea level and is an irregular
surface. The deflection of the vertical d is the local difference between the true
zenith (plumb line) and the theoretical vertical direction on a global ellipsoid.

The standard International Gravity Formula


During the nineteenth and twentieth centuries, increasing numbers of
measurements and international cooperation led to the so-called standard for-
mulas developed by and for geodesists but also used in exploration, relating
theoretical (or “normal”) gravity g to the latitude j of the station (in Gals):

1930: g = 978.0490 (1 + 0.0052884 sin 2 ϕ – 0.0000059 sin 2 2ϕ ),

1967: g = 978.031846(1 + 0.00523024 sin 2 ϕ – 0.0000058 sin 2 2ϕ ),


 1 + 0.00193185138639 sin 2 ϕ 
1980: g = 978.03267714  ,
 1 – 0.00669437999013 sin ϕ 
2

where the first two formulas are approximations and the last formula is
known as Somigliana’s equation, giving the theoretical gravity over the
reference ellipsoid. Note that the first term in the right-hand side of all the
equations is the value of the gravity field of the earth at the equator in Gals
(Geodetic Reference System, 1967).
12  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Plumb line perpendicular to the geoid

Geoid

If the disturbing mass is positive, Ellipsoid


the geoid is warped upward If the disturbing mass is negative,
(as shown). the geoid is warped downward
(as in the case of the oceans).

Figure 5. The effect on the geoid of a body with excess mass.

Most of the more than 10 million gravity stations acquired during this
period have been reduced using the 1930 formula (accepted by the Inter-
national Union of Geodesy and Geophysics meeting at Stockholm). The
formula was based on pendulum measurements taken in 1906 in Potsdam
which are believed to be in error by about 14 mGal. The differences between
the formulas are not important in exploration because the useful signal
we interpret is itself relative within a survey. It is important, however, that
we use the same formula for each survey when, as is often the case, multiple
surveys are integrated in a region. As we will see in chapter 6, the theoreti-
cal gravity is subtracted from the observed station gravity, a process known
as latitude correction. In the first several decades in the modern exploration
era, tables were created based on the 1930 formula, from which the latitude
correction could be determined for each field station. In modern explora-
tion, the value of theoretical (or normal) gravity is computed and removed,
usually by field computers.

GPS and the geoid


The heights obtained from GPS are typically heights above the refer-
ence ellipsoid. At least four GPS NAVSTAR satellites are used to deter-
mine three position coordinates and time. The position coordinates x, y, z
are geodetic coordinates (datum WGS-84) latitude, longitude and height h
above the reference ellipsoid (Figure 6). The height displayed on most con-
sumer handheld GPS receivers is, however, the orthometric height H, the
height above mean sea level (MSL) and the only one of interest in explora-
tion. Orthometric heights are thus the vertical distance from the geoid to
the surface of the earth and, by convention, the sign is considered positive
Chapter 2:  Earth’s Gravity Field  13
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

GPS

H
h Topography

Geoid N

Ellipsoid
Ellipsoid

Oceans Topography

Figure 6. The orthometric height H is obtained by adding (subtracting) to the


height h above the reference ellipsoid obtained by the GPS receiver the tabulated
ellipsoid-geoid separation height N.

as one moves radially outward. The conversion is done by interpolating


from a geoid-ellipsoid height separation model N (a lookup table in the
receiver’s firmware) and making the simple calculation of adding it to the
obtained height above the reference ellipsoid. Geoid-ellipsoid separation
heights in the conterminous United States range from about −8 m to −53 m,
and they display considerable variation in the mountains. By contrast, glob-
ally geoid-ellipsoid separation heights range from about +75 m to −100 m.
The geoid model for the earth is continuously refined. Different geoid
models will give different orthometric heights for a point, even though the
ellipsoid height (determined by GPS) might be very accurate. Therefore,
orthometric height should never be given without also stating the geoid
model used.
Although the orthometric height is the one commonly used in gravity
exploration, some geophysicists propose instead to use the height above the
ellipsoid. The main reason is that both latitude-correction estimates by the
International Gravity Formula and free-air correction estimates (see chapter
6) are designed to remove the gravity effects resulting from an ellipsoid of
revolution which can be calculated theoretically. In exploration for petro-
leum and for other minerals, the difference between the two approaches is
minimal because the geoid is a smooth surface. A global map of the geoid-
reference ellipsoid separation N is shown in Figure 7.
14  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 7. Undulations of the geoid. Values indicate the height in meters above
or below the surface that is very close to an ellipsoid of flattening 1/298.3. after
Guier and Newton (1965), Figure 1. used by permission.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 3

The Gravitational Potential and


Attraction of Mass Distributions

In the previous chapter, we investigated the gravitational acceleration g


and found that it can be derived from a scalar potential U. Based on these
definitions, we now investigate the gravitational attraction of a number of
mass distributions and derive some important consequences.

Attraction of a spherical shell


In Chapter 2, we found that to calculate the potential at an exterior point P
for an arbitrary 3D body, we have to integrate the potential of each elemen-
tal mass dm = r dV = r dw dh dζ over the entire volume V (equation 5 of
Chapter 2):
r (ξ , η,ζ ) dξ dη dζ
U ( x , y, z ) = k ∫ 1 .
V
(ξ − x ) + (η − y) + (ζ − z ) 
2 2 2 2

The potential is a scalar quantity, thus simplifying some of our oper-


ations. For example, suppose we would like to know the attraction at an
external point P of a hollow spherical shell that has a uniform surface den-
sity s   and radius a (Figure 1).
By using the potential from equation 4 of Chapter 2 and noting that
a very small element of mass dm located on the shell is equal to s ds, we
obtain
ds
U = ks ∫ ,
S
h
where the integration is carried out over surface S of the shell.

15
16  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 1. A uniform adθ


spherical shell.
dθ Uniform surface density
mass per unit area equals σ.
a
θ gr
r P

Spherical
shell

With h 2 = r 2 + a 2 − 2 a r cos q , ds = a 2 sin q dq dϕ and using spherical


coordinates, one obtains for points exterior to the shell (r > a)
p 2p p
a 2 sin q dq dϕ n q dq
sin
U = ks ∫ ∫ = 2 p ks a 2 ∫
0 0 r + a − 2 a r cos q
2 2
0 r + a 2 − 2 a r cos q
2

2p k s a p 2 p ks a 4 p k s a2
 = r 2 + a 2 − 2 a r cos q = [(r + a) − (r − a)] = .
r 0 r r
(1)

Letting M = 4 p a 2 s be the total mass of the spherical shell, we can


write the potential at an exterior point as
M
U ext = k . (2)
r
From equation 2, because of symmetry, the gravitational attraction of the
spherical shell is oriented along r and can be obtained by taking the deriva-
tive of its potential with respect to r:

∂U M
gr = = −k 2 .  (3)
∂r r

The obtained result demonstrates the nice property that the attraction of
a uniform spherical shell at an external point is the same as it would be if all
the mass of the shell were concentrated at the point in its center. Hence, for a
uniform solid sphere made up of concentric uniform shells, its attraction at
an external point would be the same as if all its mass were concentrated at its
center. To appreciate the elegance and simplicity of this approach, one can
attempt the determination of the gravitational attraction of the shell directly
without recourse to its potential.
Chapter 3:  The Gravitational Potential  17
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The same procedure can be used if the field point is inside the spherical
shell. In this case, in expression 1, we have (for a > r)

p
r 2 + a 2 − 2 a r cos q = (a + r ) − (a − r ) = 2 r
0 

and, as a result, the potential at an interior point becomes a constant


(independent of r) equal to

M
U int = k . (4)
a

Therefore, the gravitational attraction of the shell at an interior point (again,


found by taking the derivative of the potential) is everywhere equal to zero.
At an internal point of a uniform solid sphere, the attraction is a function
only of the part of the solid sphere between the respective point and the
center of the sphere. Similarly to the shell, the portion of the solid sphere
between the point of observation and the outer radius of the sphere would
exert no attraction at the interior field point.
Let us now calculate the potential at an interior point of a spherical shell
of variable finite thickness a – r, extending from an interior radius r to an
exterior radius a. From equation 4, the potential at the interior point of a
spherical shell of radius r is constant and is equal to

M 4p r2 s
u=k =k = 4 p k s r. 
r r

To obtain the potential of the thick spherical shell, we integrate this expres-
sion over r. By noting that s = r dr, where r is the volume density, we obtain

a
a2 r 2
U = 4 p k r ∫ r dr = 4 p k r ( − ) = 2 p k r (a 2 − r 2 ),  (5)
r
2 2

a result we will use later on.


A study of the gravitational potential is valuable as a general basis for
developing techniques for data reduction and interpretation, as we will
see in later sections of this chapter and in the following chapters of this
book.
18  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Components of attraction
Generally, we do not make measurements in the direction of the source
masses (this direction, for extended bodies, varies over a survey area), but rather,
we measure a component of the observed field. In a Cartesian-coordinate­
system (Figure 2), the potential at the point P resulting from mass m is given by
equation 4 of Chapter 2:
m
U=k ,
r

where r 2 = ( x − ξ )2 + ( y − η)2 + ( z − ζ )2 .
Because
∂U ∂U ∂U
g = ∇U = i +j +k ,
∂x ∂y ∂z
the components of attraction in this coordinate system are
∂U m
gx = = − k 3 ( x − ξ ),
∂x r
∂U m
gy = = − k 3 ( y − η),
∂y r
∂U m
gz = = − k 3 ( z − ζ ).  (6)
∂z r
The above expressions can be written in a compact form as

 ∂U   x − ξ
 ∂x   r 
 gx     
    ∂U  m  y − η
g =  gy  =  = − k , (6a)
∂y  r2  r 
 g     
 ∂U  z −ζ 
z

   r 
∂z 
with the terms in the last bracket representing the direction cosines between
the total gravitational attraction and the coordinate axes.
Let us examine these components of attraction at a point P, which is
above a horizontal uniform sheet or lamina of infinite extent in all directions
(Figure 3). By symmetry, the horizontal x- and y-components of g are zero,
and we have only the gz-component, which is measured predominantly in
modern exploration work.
Chapter 3:  The Gravitational Potential  19
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

γ P (x, y, z )

r
β

m(ξ, η, ζ )

Figure 2. A source point, m, and a field point, P, in a Cartesian-coordinate system. A


vector and its direction cosines; a, b, and g are the angles the vector makes with respect
to the x-, y-, and z-axes. The cosines of these angles are known as direction cosines.

For the lamina, we find the vertical component gz by directly integrating


over the surface S of the sheet:
cos a dm
gz = k ∫ .
S
r2

However, dm = s dS; therefore,


cos a dS
gz = ks ∫ .
S
r2
The last integrand above is the solid-angle1 element dw subtended by
dS at P (Figure 4). For an infinite plane, the solid angle varies from 0 to 2p,

A solid angle is the extension to three dimensions of the concept of radian in two
1

dimensions. An angle in radians is given by L/r, where L is the arc length subtend-
ing the angle and r is the radius of the circle. A solid angle is that fraction S of
the surface of a sphere that a particular object projects on, as seen by an observer
located at the center of the sphere. The numerical value of the solid angle is S/r2 and
is given in steradians, a dimensionless quantity. For a sphere whose total surface
area is 4p r2, the solid angle is 4p steradians.
20  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

and we then can write


2p
gz = k s ∫ dw = 2p ks . 
0
(7)

If the infinite sheet has thickness t, we note that s  = r t (where r is the vol-
ume density), and we obtain the well-known Bouguer plate or Bouguer slab
formula
gz = 2 p k r t ,  (8)

to which we will return many times.


P
x
gz

α
r
h

∞ ds

∞ z

Figure 3. An infinite thin rectangular plate or lamina at depth h and with constant
surface density s.

Figure 4. Definition of a solid n


α
angle. S
dS

P
Chapter 3:  The Gravitational Potential  21
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Analysis of potential fields


Gauss’ theorem

Let us consider a region R in which masses occur completely bounded


by the surface S, as shown in Figure 5. A source mass m, located at point P,
will produce a small flux dϕ  across an elemental surface ds:

m
dϕ = g ⋅ dS = k lr ⋅ n dS ,  (9)
r2
where lr is the unit vector along the direction of gravitational attraction g
produced by this mass and n is the normal to surface S. The amount of
flux across dS is proportional to the small solid angle dw and is either
+ m k dw or − m k dw , depending on whether the field exits or enters the
bounded space.
For a mass located inside the surface S (such as at P), the solid-angle
cone exits the region R one more time than it enters and the net flux is
− m k dw because the gravitational attraction produced by these masses is

n
Q dω g

R
S

g dω P
n

Figure 5. A closed irregular boundary enclosing all masses.


22  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

directed opposite to the normal n to the surface. By integrating over the


whole surface S, we obtain a complete solid angle of 4p, and the total flux
resulting from mass m becomes
ϕ = −4 p k m.

This expression can be generalized immediately to account for all masses M


located inside surface S:
ϕ = −4 p k M. (10)

Thus, we can write

ϕ= ∫ g ⋅ n dS = ∫ g n dS = −4 p k M ,  (11)
S S

which is Gauss’ theorem2, in which M is the total mass inside S and gn is


the normal component of gravity on the surface S. This is an extremely use-
ful theorem. It tells us that there is at least one unique piece of information
that a gravity anomaly can supply, namely, the amount of disturbing mass.
We make use of this theorem in the practical estimation of total mass in
Chapter 7, on interpretation. Gauss’ theorem is also useful for studying the
attraction of a few very simple models, such as the uniform sphere, cylinder,
and infinite slab or plate.
For masses located outside surface S, such as at a point Q in Figure 5,
the number of solid-angle entries is equal to the number of exits, and the net
flux through surface S is zero, or ϕ = 0 .

Laplace’s and Poisson’s equations

Using the divergence theorem, we can express Gauss’ theorem 11 as


ϕ= ∫ g ⋅ n dS = ∫ ∇⋅ g dV = −4 p k ∫ r dV .
S V V

Because g = ∇U and ∇ ⋅ ∇U = ∇ 2 U, we can write

∫ ∇ U dV = −4 p k ∫ r dV ,
2

V V

The divergence theorem often is called Gauss’ theorem, a term in exploration


2

usually reserved for the determination of total mass.


Chapter 3:  The Gravitational Potential  23
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

from which, by equating the integrands, it follows that

∇ 2U = −4 p k r. (12)
This is known as Poisson’s equation. At points of free space outside the
source region (r = 0), equation 12 reduces to

∇ 2 U = 0, (13)
which is known as Laplace’s equation.
It is easy to show that a vector field derived from a potential U that satis-
fies Laplace’s equation has both the curl and the divergence equal to zero.
Indeed, if we assume that A = ∇U, then it follows that ∇ × A = 0. Finally,
using ∇ ⋅ A = ∇ ⋅ ∇U = ∇ 2U = 0 completes the proof.
Let us examine Laplace’s equation in a Cartesian-coordinate system
when
∂ 2U ∂ 2U ∂ 2U
∇ 2U = + + .
∂ x 2 ∂ y2 ∂ z 2

Consider a point P, external to all sources. The potential at P is given by


1
expression 5 of Chapter 2. Letting r = (ξ − x )2 + (η − y)2 + (ζ − z )2  2 in
expression 5 of Chapter 2, we have

∂  1 x−ξ
  =− 3 ,
∂ x  r r
∂ 2  1  3 ( x − ξ )2 1
  = − 3
∂ x2  r  r5 r

and similar expression for the derivatives along the y- and z-directions. It
immediately follows, by adding corresponding terms, that
∇ 2 U = 0,
and Laplace’s equation holds for points located outside sources.
Now we consider the case for a point P located inside the mass source,
as shown in Figure 6. First we describe a very small sphere, radius e, inside
which P is located. The radius e is so small that the density of the small
sphere can be regarded as constant. Let U1 denote the potential resulting
from the mass outside the little sphere, and let U2 denote the potential result-
ing from the mass inside the little sphere. Now U = U1 + U2. By equation
13, ∇ 2 U1 = 0 .
24  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 6. Observation point P


located inside the mass.
ε
r
P (x, y, z)

To evaluate U2, we will derive the potential and the gravitational attrac-
tion at a point r inside the small sphere. The problem simplifies if we con-
sider the potential U2 to be composed of two parts. Let u1 denote the ­potential
resulting from the spherical shell contained between r and e, and let u2
denote the potential resulting from the remaining sphere of radius r. The
potential u2 for a sphere of mass m and radius r is given by
4 3
pr kr
m 3 4
u2 = k = = p r2 k r.
r r 3

From equation 5, we have for the potential of the spherical shell


u1 = 2 p k r (ε 2 − r 2 ).

The potential U2, resulting from the mass inside the little sphere, is now given
by
2p k r
U 2 = u1 + u2 = (3ε 2 − r 2 ),  (14)
3
from which

∂ U2 4p k r ∂ r 4p k r x
=− r =− ,
∂x 3 ∂x 3
∂ 2 U2 4p k r
=− ,
∂x 2
3
∂ 2 U2 4p k r
=− ,
∂y 2
3
∂ 2 U2 4p k r
=− ,
∂z 2
3

and ∇ 2 U = ∇ 2 U1 + ∇ 2 U 2 = −4 p k r , i.e., Poisson’s equation holds.


Chapter 3:  The Gravitational Potential  25
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Gravity calculations for simple geometries


The gravity meter (Chapter 4) is a very sophisticated weighing device.
It weighs in the direction of the plumb line that we call the vertical, which
is affected by the source masses under study. This effect is known as the
deflection of the vertical (defined in Chapter 2), which we will consider
again in Chapter 6. It is sufficient for present purposes to note that we are
interested primarily in the vertical component of gravity while considering
various geometric models.
The general form for calculating the vertical component of gravity
(1) decomposes the source distribution into small elements of mass dm
(Figure 7); (2) calculates, using Newton’s inverse square law, the gravi-
tational attraction caused by the mass dm in the direction, r, to the
­observation point; and then (3) multiplies the result by the cosine of the
angle between the direction r and the vertical axis z, which is given by
(z – z)/r. With dm = rdV and with volume dimensions of dx, dh′ , and dz,
the general form for the vertical component of attraction in the 3D case
becomes

d  1 r (ξ , η,ζ )(ζ − z ) dξ dη dζ
gz ( x , y, z ) = k ∫ r   dV = k ∫ ,  (15)
dz  r 
3

(ξ − x ) + (η − y) + (ζ − z ) 
V V 2 2 2 2

where r = (ξ − x )2 + (η − y)2 + (ζ − z )2 , and the integration is carried out


over the volume V of the body. Note that the above expression is nothing
more than the derivative in the z-direction of the expression for the potential
given in equation 5 of Chapter 2, i.e., gz = ∂U/∂z = Uz.
In the above integral (equation 15) and in equation 5 of Chapter 2, if
density r is constant, it can be taken outside the integral. In general, density
r is not constant (e.g., densities of sands and shales generally increase with
depth of burial because of compaction), and in such cases, r must remain
under the integral sign.
In gravity gradiometry (Chapter 4), one also measures the second-
∂ 2U
derivative terms of the gravitational potential. Letting U xy = and simi-
∂x ∂y
larly for other variables, the corresponding expressions are given in Table 1.3

3In this book, we will use both notations interchangeably, e.g., Uxy and gxy.
26  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

P (x, y, z)
dm (ξ, η, ζ ) = ρdv
r

Plumb-bob direction

z
Figure 7. Gravitational attraction of an arbitrary 3D body in a Cartesian system.

Equation 5 of Chapter 2 and equation 15 of this chapter state that


given the density and geometry of the causative source, we can calculate
uniquely its gravity potential and its gravity anomaly, both of which van-
ish at infinity. We call this the forward calculation, or the forward prob-
lem. However, equation 5 of Chapter 2 and equation 15 of this chapter
also show that in the absence of information about density and geometry,
an observed gravity anomaly cannot be satisfied uniquely. We call this
the inverse problem and will return to it at the end of this chapter and in
Chapter 7.
The vertical component of gravity can be calculated in two ways:

1) Calculate first the potential using equation 5 of Chapter 2 and then differ­
entiate with respect to z, the vertical axis.
2) Calculate the vertical component of attraction directly using expression
15 of this chapter or, in special cases, by using Gauss’ theorem. However,
such calculations can be carried out analytically for only a few simple
bodies.
Chapter 3:  The Gravitational Potential  27
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Table 1.  Gravitational potential and its derivatives.4


Function Formula
r (ξ , η, ζ ) dξ dη dζ
U k∫ 1

(ξ − x ) + (η − y) + (ζ − z ) 
V 2 2 2 2

(ζ − z ) r (ξ , η, ζ ) d ξ d η dζ
Uz k∫ 3

(ξ − x ) + (η − y) + (ζ − z ) 
V 2 2 2 2

( ξ − x ) r (ξ , η, ζ ) dξ dη dζ
Ux k∫ 3

(ξ − x ) + (η − y) + (ζ − z ) 
V 2 2 2 2

(ξ − x )(η − y) r (ξ , η, ζ ) dξ dη dζ
Uxy 3k ∫ 5

(ξ − x ) + (η − y) + (ζ − z ) 
V 2 2 2 2

(ξ − x )(ζ − z ) r (ξ , η, ζ ) dξ dη dζ
Uxz 3k ∫ 5

(ξ − x ) + (η − y) + (ζ − z ) 
V 2 2 2 2

(η − y)(ζ − z ) r (ξ , η, ζ ) dξ dη dζ
Uyz 3k ∫ 5
V
(ξ − x )2 + (η − y)2 + (ζ − z )2  2

 2(ξ − x )2 − (η − y)2 − (ζ − z )2  ( r (ξ , η, ζ ) dξ dη dζ
Uxx k∫  5
V
(ξ − x )2 + (η − y)2 + (ζ − z )2  2

 2(ζ − z )2 − (ξ − x )2 − (η − y)2  ( r (ξ , η, ζ ) dξ dη dζ
Uzz k∫  5
V
(ξ − x )2 + (η − y)2 + (ζ − z )2  2

4Other derivatives can be obtained by circular permutation.

Sphere

We saw previously that the potential at an external point P(x, y, z), result-
ing from a sphere with its center at a depth h, is the same as if all its mass
were concentrated at its center:
M
U=k ,
R
where R 2 = x 2 + y 2 + (h − z )2. To obtain the gravitational attraction, we dif-
ferentiate with respect to z to obtain (for z = 0)
28  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Mh
gz = k 3 . (16)
(x + y + h )
2 2 2 2

Some of the properties of the sphere anomaly and its higher-order deriva-
tives are given in Table 2 and are sketched in Figure 8.

Thin rectangular plate

For many simple geometries, equation 15 can be simplified. For exam-


ple, as we saw in Figure 3, if the depth extent (thickness) of the body is very
small with respect to its depth, we might consider its surface density (mass
per unit area) rather than a volume density. If its cross-sectional area is very
small, we might consider its lineal density (mass per unit length). Thus, the
number of required integrations is problem dependent.
We start with the horizontal thin rectangular plate shown in Figure 9,
for which we wish to know its vertical component of attraction gz. The
plate is at a constant depth h and extends along the x-axis from –a to +a and
along the y-axis from –b to +b, as shown in Figure 9. By dividing the plate

Table 2.  Gravitational anomaly of a sphere.


Location Maximum Depth to
of maxima (minimum) ­center of
Function Formula (minima)5 value sphere6

Uz kM
h x max = 0 max = kM h = ± 1.305 x1 / 2
3
h2
(x + y + h )
2 2 2 2

hx kM
Uxz −3 k M 5 x max = − h/2 max = +0.858 —
h3
(x + y + h )
2 2 2 2
x min = + h/2
kM
min = − 0.858 3
h

2 h2 − x 2 − y2 2 kM
Uzz kM 5
x max = 0 max = —
h3
( x 2 + y2 + h ) 2 2 x min = ± 2h
kM
min = − 0.036
h3

5 Because of symmetry, only the anomaly along the x-axis is considered.


6 Depth h is expressed as a function of abscissa x1 / 2, where the amplitude is half its maximum value.
Chapter 3:  The Gravitational Potential  29
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

1.2 Figure 8. Graphs


at arbitrary scales
1.0 of gz , gxz,, and gzz
gz
for a sphere.
0.8

0.6

0.4

0.2
gzz
gxz
0.0
–40 –30 –20 –10 0 10 20 30 40

–0.2 Sphere

into small mass elements dm = s ds = s dx dh, we can write the vertical


­component as

h dξ dη
a b

gz = k s ∫∫ 3 .
(ξ − x ) + (η − y) + h 
−a −b 2 2 2 2

Letting u = x − x and v = h – y, we can write

a− x b− y
h du dv
gz = k s ∫ ∫ 3 .
−a− x −b− y
(u 2
+v +h2
)
2 2

First we will integrate the integral with respect to u and then with respect
to v, and we will include the integration limits only at the end. Using Grad-
shteyn and Ryzhik (1980, p. 86, formula 2.271.5), we obtain, after integrat-
ing over u,

dv
gz = k s h u ∫ 1 .
( v + h )(u + v + h )
2 2 2 2 2 2
30  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 9. A thin rectangular x


plate or lamina with constant
surface density s. (x, y, 0)
h gz
–a +a
r
–b

dm = ds = d d

+b

This integral can now be solved using Gradshteyn and Ryzhik (1980,
p. 89, formula 2.284) to yield
a− x b− x
uv
gz = k s tan −1 . (17)
h u + v 2 + h2
2
−a− x −b− x

If we allow the sheet to become infinite in the y-direction (b → ∞), the


above expression reduces to
a−x a+x
gz = 2 k s (tan −1 + tan −1 ). (17a)
h h 
Note that if we allow the sheet to become infinite in all directions (a → ∞
and b → ∞), we obtain
gz = 2p k s ,

an expression we found before in equation 7. We will return to equation 17


in the chapter on interpretation (Chapter 7).

Vertical cylinder of finite depth extent

The calculation of the gravitational attraction of a vertical cylinder is


carried out more easily if we first derive the gravitational attraction of a
circular lamina of radius a located at a depth V   below the surface (Figure
10a). For an observation point P(0, 0, –z) situated along the lamina axis, we
obtain, after integrating first over j   and then over r,
Chapter 3:  The Gravitational Potential  31
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) P (0, 0, –z ) b) P (0, 0, –z )

Surface (z = 0) Surface (z = 0)

h1

ζ a

r
a

h2

Figure 10. Calculation of the gravitational attraction along the axis of a circular
lamina and a vertical cylinder. Both have a radius a, and the vertical cylinder has a
finite depth extent (h2 − h1).

a
2p a
( z + ζ ) r dr dϕ 1
gz = k r ∫∫ 3 = 2p k r ( z + ζ ) − 1 =
0 r 2 + ( z + ζ )2  2
0 r 2 + ( z + ζ )2 2
0

 1 1   z +ζ 
= 2p k r ( z + ζ )  z + ζ − 2 2 2
1 = 2p k r  1 − 2 1 .

  a + ( z + ζ )     a + ( z + ζ )  
2 2

To obtain the gravitational attraction of a vertical cylinder of a finite-


depth extent, we have to integrate the above expression with respect to V
from h1 to h2 (Figure 10b) to obtain

gz = 2 p k r  h2 − h1 + a 2 + (h1 + z )2 − a 2 + (h2 + z )2  .  (18)


 

A closed-form solution for the general case of observation points located


outside the z-axis also can be obtained (Nabighian, 1962). For an infinite
circular cylinder of radius a with top at a depth h, the gravitational attraction
at a point P(x, 0, 0) is given by
32  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

 1 − x2 p 
gz = 2 k r a  K ( k ) + (1 + x )2 + a 2 E ( k ) + a Λ0 (ϕ , k ) − pa  ,
 (1 + x ) + a
2 2
2 

(18a)
where a = h/a, K(k) and E(k) are complete elliptic integrals of the first and
second kind, Λ 0 (ϕ , k ) is the Heumann lambda function7,

4x a
k2 = , and sin ϕ = .
(1 + x )2 + a 2 (1 − x )2 + a 2
For x = 0, the above expression reduces to

gz = 2p k r ( a2 + h2 − h , )
which is the well-known expression of the gravitational attraction on
the axis of the cylinder. For a = 0 (outcropping cylinder), equation18a
reduces to
gz = 2 k ra [ (1 − x ) K ( k ) + (1 + x ) E ( k ) ] . (18b)

The gravitational attraction of a vertical cylinder of finite depth extent can


be obtained by subtracting two infinite cylinders.

Vertical prism of finite depth extent

To obtain the gravitational attraction for a vertical prismatic body


(Figure 11), one has to solve the following triple integral for the vertical
component of attraction,
a b h2
ζ dξ dη dζ
gz = k r ∫ ∫∫ 3 ,
( x − ξ ) + ( y − η) + ζ 
− a − b h1 2 2 2 2

for calculations carried out for z = 0.


Letting u = x − x and v = η – y, we can write
a− x b − x h2
ζ du dv dζ
gz = k r ∫ ∫ ∫ 3 .
− a − x − b − x h1
(u + v + ζ )
2 2 2 2

A computational method to evaluate equation 18a was published by Nagy (1965).


7

The Heumann lambda function is obtained by combining incomplete elliptical inte-


grals of the first and second kinds plus Jacobi’s zeta function.
Chapter 3:  The Gravitational Potential  33
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

P (x, y, 0)
gz

–a +a
–b
h1
y
h2

+b

z
Figure 11. Vertical prism with dimensions 2a and 2b and height (h2 – h1).

Again, we will first carry out the integration and apply the integration lim-
its only to the final result. The integration over V is straightforward to yield

du dv
gz = − k r ∫ ∫ .
u + v2 + ζ 2
2

We integrate over v using Gradshteyn and Ryzhik (1980, p. 86, formula


2.271.4) to obtain
gz = − k r ∫ ln ( v + u 2 + v 2 + ζ 2 ) du.

On integrating by parts,

 u u
gz = − k r u ln ( v + u + v + ζ ) −
2 2 2
∫v+ du  .
 u + v +ζ
2 2 2
u + v +ζ
2

2 2

To evaluate the integral in the above expression, we can write it in the form
34  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

u u u 2 ( u 2 + v 2 + ζ 2 − v)
I = ∫v+ (u 2 + v 2 + ζ 2 ) u2 + v 2 + ζ 2
du = ∫ (u 2
+ ζ 2 ) u2 + v 2 + ζ 2
du =

u 2 du (u 2 + ζ 2 − ζ 2 ) du
∫ u 2 + ζ 2 ∫ (u 2 + ζ 2 ) u2 + v 2 + ζ 2 =
− v

(u 2 + ζ 2 − ζ 2 ) du du du
∫ u2 + ζ 2
− v∫
u + v +ζ
2 2 2
+ vζ 2 ∫
(u + ζ ) u 2 + v 2 + ζ 2
2 2
.

The first two integrations are straightforward, and for the last integral, we
use Gradshteyn and Ryzhik (1980, p. 89, formula 2.284) to obtain
 u v u 
I = u − ζ tan −1   − v ln (u + u 2 + v 2 + ζ 2 ) + ζ tan −1  .
ζ  ζ u + v 2 + ζ 2 
2

When applying the limits from –b to +b for the variable v, the variable
u in the equation above will cancel. After collecting all terms and reverting
to the original variables, we finally obtain the gravitational attraction of a
truncated prism as

gz = − k r [ (ξ − x ) ln (η − y + R) + (η − y) ln (ξ − x + R)

ξ−x
−1 η − yξ − x a b h2
+ ζ tann − ζ tan −1  , (19)
ζ ζ R  −a −b
h1

where R 2 = ( x − ξ )2 + ( y − η)2 + ζ 2.
It is not difficult to see that although many simple geometries yield
closed-form or “exact” solutions to either equation 5 of Chapter 2 or equa-
tion 15 of this chapter, more complex simulations of the geology can be
solved only numerically. However, we should keep in mind that all such
models are fictitious in the sense that the actual geology does not behave
with densities and geometric boundaries everywhere constant. Regardless,
it is not uncommon to approximate a given geologic feature by the super-
position of multiple simple geometric bodies, a subject to be examined in
Chapter 7.

Gravity calculations for 2D geometries


If we can approximate the geology by assuming that its strike length
(distance along the y-axis in Figure 7) is extremely large in comparison with
Chapter 3:  The Gravitational Potential  35
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the other dimensions and that the density in that direction does not change,
the integral with respect to y in equation 15 could be taken first to large (but
finite) limits in both negative and positive directions and then examined as
those large limits approach infinity.

Horizontal cylinder of infinite length

The gravitational attraction caused by an infinite horizontal cylinder can


be calculated by using either equation 5 of Chapter 2 or equation 15 of this
chapter. However, the solution is easier to obtain using Gauss’ theorem, as
depicted in Figure 12.
By placing a Gaussian cylindrical surface of radius r, concentric with
the uniform cylinder of radius R, we note, by symmetry, that the gravity
attraction gr is constant on that surface and is directed toward the centerline
of the cylinder. By Gauss’ theorem,
1 1
M=
4p k ∫ g ⋅ n dS = 4 p k S ⋅ g .
S
r

For a given length L, the cylinder mass is M = p R2r L, and the Gaussian surface
M 1 R2 r
area is S = 2πrL. With = , one obtains from above
S 2 r

M 2 k p R2 r 2 k λ
gr = 4 p k = = ,
S r r

where λ = p R 2 r is the mass per unit length of the cylinder and r > R. We
normally measure the vertical component of gravity gz to yield

Gaussian surface 1n Figure 12. A uniform horizontal


S = 2πrL gr gz infinite cylinder surrounded by a
Gaussian surface.
r

Uniform cylinder
36  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

h 
gz = 2 k p R 2 r , (20)
x +h
2 2

where x and h are the horizontal and vertical distances, respectively, of the
observation point with respect to the center of the cylinder. Some of the
properties of the infinite horizontal cylinder anomaly and its higher-order
derivatives are given in Table 3 and are sketched in Figure 13.
For the case in which the cylinder degenerates to an infinite line
mass of mass per unit length equal to λ = π R2ρ, the above expression
reduces to

h
gz = 2 k λ . (21)
x + h2
2

We will have further occasion to make use of equation 20 of this chapter in


Chapter 7.

Horizontal prism of infinite length

To obtain the gravitational attraction for the case of the infinite horizon-
tal prism (Figure 14), one has to solve the following triple integral for the

Table 3.  Gravitational anomaly for an infinite horizontal cylinder of radius R.


Location of Depth to
maxima Maximum center of
Function Formula (minima) (minimum) value cylinder8
h 1
Uz 2 p kR 2 r 2 x max = 0 max = 2p k R 2 r h = ± x1 / 2
x + h2 h

xh 1
x max = − 0.577 h max = + 1.3 p kR r 2
2
Uxz − 4p k R 2 r
(x + h )
2 2 2 h
x min = + 0.577 h 1
min = − 1.3 p kR 2 r 2
h

h2 − x 2 2p kR 2 r
Uzz 2 p k R2 r x max = 0 max =
( x 2 + h 2 )2 h2
x min = ± 1.732 h
p kR 2 r
min = −
4h2
8Depth h is expressed as a function of abscissa x1 / 2 , where the amplitude is half its maximum value.
Chapter 3:  The Gravitational Potential  37
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

vertical component of attraction:


∞ h2
ζ dξ dη dζ
b

gz = k r ∫ ∫∫ 3 .
a −∞ h1 ( x − ξ ) + ( y − η) + ζ 
2 2 2 2

The integrations can be carried out using the expressions derived for the
vertical prismatic body (see the subsection titled “Vertical prism of the finite

1.2 Figure 13. Graphs


at arbitrary scales
1.0 of gz, gxz, and gzz
for an infinite
horizontal cylinder.
0.8 gz

0.6

0.4

gzz
0.2

0.0
–40 –30 –20 –10 0 10 20 30 40
gxz
–0.2

–0.4 Cylinder

a b Figure 14. Infinite-length horizontal


x prism of width 2b and height (h2 – h1).
gz

h1

h2

z
38  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

depth extent” in this chapter) to obtain


x−a b−x x−a
gz = 2 k s ς (tan
−1
+ tan −1 )+
ς ς 2
b−x
h2

ln ( x − a)2 + ς 2 + ln (b − x )2 + ς 2 . (22)
2 h1

Such a model is useful in representing the attraction caused by horsts or


grabens. From equation 22, we obtain

(a − x )2 + h22  (b − x )2 + h12 


gxz = k r ln  (22a)
(a − x )2 + h12  (b − x )2 + h22 

 h h h h 
gzz = 2 k r arctan 1 − arctan 2 − arctan 1 + arctan 2  . (22b)
 a−x a−x b−x b− x

2D thin sheet

In the third section of this chapter (“Analysis of potential fields”), we


found that the vertical component of attraction caused by a uniform thin
sheet (Figure 3) is proportional to the solid angle subtended by the sheet
at the point of observation, and we noted the general definition for the
solid-angle dw. For the case of the 2D thin sheet, shown in Figure 15,
the solid angle is extended to infinity in the directions perpendicular to the
plane of the figure and therefore can be represented by the plane angle q,
shown in the figure. By integrating in the y-direction from plus and minus
infinity, we can show that the solid-angle w is equal to 2q (note that for
an infinite plane, q = p, and we obtain the known result). For a thin sheet
whose surface density is s (mass per unit area), the vertical component of
attraction is

gz = 2 k s q . (23)

For a semi-infinite plate, expression 23 becomes

p x
gz = 2 k s [ + arctan ( )]. (24)
2 h

For an infinite plate, q is equal to p, s equals rt, and we obtain the famil-
iar Bouguer formula. Some of the properties of the 2D horizontal thin-
Chapter 3:  The Gravitational Potential  39
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

x Figure 15. 2D uniform


gz horizontal thin sheet.
h

sheet anomaly and its higher-order derivatives are given in Table 4 and are
sketched in Figure 16.

Semi-infinite finite step

Expression 24 can be used to determine the gravitational attraction of a


finite step (Figure 17). To achieve this, one has to replace x in equation 24
with x + z tan a and integrate over z from h1 to h2. We thus obtain

h2
p x + z tan a
gz = 2 k s ∫ [ + arctan ] dz. (25)
h1
2 z

The first integral above is straightforward, and in the second integral,


we make the change of variables:

x + z tan a x x dv
= v or z = and dz = − v
z v − tan a ( v − tan a )2

to obtain

x + z tan a 1
∫ arctan z
dz = − x ∫
( v − tan a )2
arctan v dv.

Using Gradshteyn and Ryzhik (1980, p. 210, formula 2.855), the integral on
the right-hand side above can be evaluated to yield

x + z tan a v − tan a 1 + v tan a


∫ arctan z
dz = − x cos 2 a ( ln
1 + v2

v − tan a
arctan v ).
40  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Table 4.  Gravitational anomaly of 2D horizontal thin sheet.


Maximum Depth to
Location of (minimum) top of thin
Function Formula maxima (minima) value sheet9
p x
Uz 2 k s [ + arctan ( )] x max = ∞ max = 2pks
2 h

h
Uxz 2ks x max = 0 max = 2 k s h = ± x1 / 2
x 2 + h2 h
x ks
Uzz 2ks x max = + h max = +
x + h2
2 h
x min = − h
ks
min = −
h
9 Depth h is expressed as function of abscissa x1 / 2 , where the amplitude is half its maximum value.

Figure 16. Graphs 35


at arbitrary scales
of gz, gxz, and gzz for 30
a 2D horizontal thin gz
sheet. 25

20

15

10 gxz

5
gzz

0
–30 –20 –10 0 h 10 20 30

–5
Semi-infinite thin sheet

–10
Chapter 3:  The Gravitational Potential  41
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

(0, 0) (x, 0) Figure 17. 2D finite


step.
2
r1 h1 z h2
r2 1

To infinity

Reintroducing the variable z, one obtains

x + z tan a
∫ arctan z
dz = ( z + x sin a cos a )

x + z tan a 
arctan + x cos 2 a ln z 2 + ( x + z tan a )2 . (26)
z

After applying the integration limits and using the notation shown
in Figure 17, we finally obtain to gravitational attraction from a finite
step as

 r 
gz ( x ) = p k s t + 2 k s (h2 q 2 − h1q1 ) + x (q 2 − q1 )sin a cos a + x cos 2 a ln 2  ,
 r1 
(27)

where t = h2 – h1 is the thickness of the finite step, r1 = h12 + ( x + h1 tan a )2


and r2 = h22 + ( x + h2 tan a )2 . Note that as x ® ∞, θ2 ® θ1, and r2 ® r1, we
obtain gz ( x ) = p k s t , i.e., half the value of a Bouguer slab, as expected.
For a = 0, expression 27 reduces to the expression of a semi-infinite
slab:

 r2 
gz ( x ) = p k s t + 2 k s (h2 q 2 − h1q1 ) + x ln  .(27a)
 r1 

Some of the properties of the semi-infinite slab anomaly and its higher-order
derivatives are given in Table 5 and are sketched in Figure 18.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

42  

Table 5.  Gravitational anomaly of a semi-infinite slab (vertical finite step).


Location of maxima
Function Formula (minima) Maximum (minimum) value x1 / 210

 r 
Uz p k r t + 2 k r (h2 q 2 − h1q1 ) + x ln 2  x max = ∞ max = 2πkp (h2 – h1) x1 / 2 = 0
 r1 

r2 h2
Uxz 2 k r ln x max = 0 max = 2 k r ln x1 / 2 = ± h1 h2
r1 h1
Fundamentals of Gravity Exploration

h2 h —
Uzz 2 k r (q1 − q 2 1 ) x max = + h1 h2 max = 2 k r (arctan − arctan 1 )
h1 h2
h1 h
x min = − h1 h2 min = 2 k r (arctan − arctan 2 )
h2 h1

Note: r1 = x 2 + h12 and r2 = x 2 + h22 .

10 x1 / 2 represents the coordinate where anomaly amplitude is half its maximum value.
Chapter 3:  The Gravitational Potential  43
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) 35 Figure 18.
Graphs at arbi-
30 trary scales of gz,
gz gxz, and gzz for a
25 semi-infinite slab.

20

15

10 gxz

5
gzz

0
–30 –20 –10 0 10 20 30

–5

–10

b) 0
x
h1
h2
To infinity

The logarithmic potential


In Chapter 2, we developed the Newtonian potential given by equation 5
of Chapter 2. For 2D targets elongated in the y-direction, we can integrate
the potential in this direction from minus infinity to plus infinity and can
ob­tain what is called the logarithmic potential:
1

U (x, z) = ∫ ∫ r (ξ ,ζ ) log r dξ dζ , 
ζ ξ
(28)

1
where r = (ξ − x )2 + (ζ − z )2  2 .
44  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

From equation 28, the gravitätional attraction for 2D targets can be writ-
ten as
ς−z
gz = ∫ ∫ r (ξ ,ζ ) 2 dξ dζ . (29)
ζ ξ
r

Treating geologic solutions in cross section only (as if the geology were
strictly two-dimensional) is often convenient for demonstrating concepts (as
we will see) and is useful in displaying the results of interpretations, but the
third dimension should not be ignored except in those cases in which the
strike length of the geology is sufficiently greater than the cross-sectional
dimensions. A rule of thumb sometimes used is that for bodies whose length
is four times the distance to the point of calculation, the error (overcalcula-
tion by assuming two-dimensionality) is a little less than 10%. In some cases,
it might be desirable to calculate corrections for the nonexistent “ends” of
the structure, but usually, if such errors are of concern, it is more appropriate
to use 3D algorithms.

Green’s equivalent layer and the problem of ambiguity


Green’s first identity11 can be derived from the divergence theorem
applied to the vector field F = V ∇ U , where U and V are scalar functions
of x, y, and z in the region R, with V once continuously differentiable and
U twice continuously differentiable. Green’s first identity relates a volume
integration to an integration over a surface S that completely bounds the
volume:
∫∫∫ R
(V ∇ 2U + ∇U ⋅ ∇V ) dv = ∫∫
S
V (∇U ⋅ n) ds,(30)

where ∇ 2 is the Laplace operator and n is a unit vector everywhere normal to


the surface S. If we let U be a harmonic function satisfying Laplace’s equa-
tion and we set V = 1, we obtain from equation 30

∂U
∫∫
 S ∂n
ds = 0. (31)

Thus the integral of the normal derivative of a harmonic function, over a


closed boundary surrounding the region in which the function is harmonic
and continuously differentiable, averages to zero.

Green’s three identities are derived from the divergence theorem and can be
11

studied in several mathematical texts, such as Kellogg (1929).


Chapter 3:  The Gravitational Potential  45
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Green’s second identity is written as

∫∫∫ R
(V ∇ 2U − U ∇ 2V ) dv = ∫∫ (V ∇U − U ∇V ) ⋅ n ds,(32)
 S

in which U and V are again continuous functions of x, y, and z in region R


and have continuous partial derivatives of the second order, and n is a unit
vector everywhere normal to surface S. Now, from an infinite number of
possibilities, we select any surface completely bounding the disturbing mass
that is also an equipotential surface, and we choose any point P which is
outside surface S. Let r be the distance measured from P. If we let U = 1/r in
Green’s second identity and let V be the potential resulting from the disturb-
ing mass, then
 1 1    1 1 
∫∫∫  V ∇ 2   − ∇ 2V  dv = ∫∫  V ∇  r  − ∇V  ⋅ n ds. (33)
R  r r  S r 

Because P is outside region R, therefore, by Laplace’s equation, the


first term on the left vanishes. By Poisson’s equation, the second term on
the left is just the potential at point P resulting from the material within S,
multiplied by 4p or 4pVP. Because V is constant on S, the first term on the
right vanishes by Gauss’ theorem. This leaves, after using the divergence
theorem,
1 2 1 ∂V
4p Vp = − ∫∫∫ ∇ V dv = − ∫∫ ds. (34)
Rr S r ∂n 

Thus, at any point outside S, the potential caused by a source inside S is the
same as it would be if all the material were spread over the equipotential
surface S with a surface density of
1 ∂V
− . (35)
4p ∂n 
The above relationship is called the Green’s equivalent layer. This is
a classic statement of nonuniqueness: A multiplicity of mass distributions
can cause an identical anomaly. Gravity measurements are usually but not
always limited to the earth’s surface. Therefore, the conditions stated above
can be met, except that it is usually unrealistic to concentrate 3D geologic
sources onto laminal surfaces without thickness.
Because surface distributions represent a class of bodies different from
those representing real geology, one might be inclined to dismiss Green’s
layer as not applicable to a large class of geologic problems. However,
46  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

mathematical fictions are often very helpful in visualizing real-world prob-


lems and sometimes are not bad geologic approximations.
For example, in the general case in which our observations are out-
side the causative sources, Laplace’s equation must hold. A large number
of solutions to Laplace’s equation leads to a variety of mathematical fic-
tions intended to represent rock boundaries occurring within the earth (Fou-
rier series and sinx/x methods are examples) and is used from time to time
in geophysical studies. Such methods can allow us to calculate the depths
to layers on which the mass is concentrated and, with the help of external
information, can allow us to identify the most likely depth in a given geo-
logic problem. It is important to understand the accuracy with which sur-
face distributions can be used to represent volume distributions, and we will
examine this question further in Chapter 7.
Before considering the 3D class of bodies in this chapter, we should
point out that the equivalency developed from Green’s theorem also applies
to other geophysical methods. However, we are concerned here only with
the gravity method, and we wish to make a clear statement about the prob-
lem of ambiguity. In exploration projects, we normally measure the grav-
ity field in an area of interest, and in modern times (almost always), we
also have available other geophysical data sets as well as some geologic
information.
In Chapters 6 and 7, we examine the various problems associated with
determining the observed gravity anomaly to be related to gz, defined in
equation 15, for which the density and geometry can be considered initially
to be unknown. The problem is clearly ambiguous if we lack information
about both the density and the geometry.
If, however, the observed anomaly caused by our target of interest is
unequivocally determined and either the density or the geometry is known
or can be assumed safely, then the remaining unknown quantity can be cal-
culated at least theoretically. With these constraints, we say that the solution
is unique. The questions of uncertainties that might affect our result with
respect to any of the steps required by data reduction and interpretation are
taken up in Chapters 6 and 7.
Perhaps the most referenced statement on the problem of ambiguity is
that of Skeels (1947), shown in Figure 19, in which each of seven geologic
boundaries (from an unlimited number) between the underlying basement
and overlying sedimentary section produces very nearly the same anomaly,
as shown. It has sometimes been assumed that these very different geologic
solutions produce exactly the same anomaly, but in fact, they were derived
empirically and, as with all such calculations, were performed to within
Chapter 3:  The Gravitational Potential  47
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Gravity anomaly Density Figure 19. Each interface has a


interfaces different density distribution, but
it causes nearly the same gravity
1
anomaly. After Skeels (1947),
2 Figure 1.
3
4
5

what was considered acceptable approximations. Nonetheless, Skeels’ cen-


tral point should not be missed: An excellent fit between the calculated and
observed anomalies does not by itself guarantee a correct interpretation.
In nearly every modern case, however, information in addition to equa-
tion 15 is available to the interpreter, and the extent of ambiguity can be
reduced significantly, as reported by Al-Chalabi (1971). By using parameter
hyperspace, he clearly shows many conditions under which ambiguity can
be limited. Nonetheless, Green’s equivalent layer, combined with the prob-
lems of ill conditioning12, requires that we consider alternatives to an initial
interpretation. It is relatively simple to replace any deep interpretive model
with one that is quite shallow, and the shallow model can be ruled out only
by geologic knowledge or understanding.
However, it is also possible to replace a model with a combination of
deeper structures within the precision of our measurements and our ability
to isolate the target anomaly. Consider, for example, in Figure 20, the shal-
low prism which causes the nearly 8-mGal anomaly depicted by the solid
line. We can find a deeper solution by making the central horst narrower
and thicker and adding two flanking drainage channels (or small grabens),
the calculated field for which is shown by the X symbols. This fit is found
through inversion after only two iterations and is very good. It can be made
better by more iterations, but the point has already been made that whether

12 Large changes in geometry, especially for deep sources, might cause only slight
changes in the calculated gravity field, rendering the accommodation of imperfectly
determined observed fields problematic.
48  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

caused by ill conditioning, anomaly definition and separation issues, errors


in data reduction, or the fidelity of the model, information in addition to
equation 15 is required in gravity interpretation. We will turn to these issues
in the remaining chapters.

mGal
8

–30 –20 –10 0 10 20 30


Distance (km)
Depth (km)
5

10

Interpretation A: Single prism, density contrast = +300 kg/m3. Anomaly is shown by a


solid black line.

Depth (km)

10

Interpretation B: Three prisms, density contrast = +300 kg/m3 (central), –300 kg/m3 (sides).
Anomaly is shown by X.

Figure 20. Two prism interpretations produce nearly the same anomaly. These
anomalies were calculated independently, but a similar exercise (including a
shallower nonprismatic-equivalent source) was published by Jung (1961).
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 4

Field Measurements

Introduction
The measurement of the earth’s gravity field, whether in absolute or
relative terms, is one of mankind’s greatest engineering achievements, the
accuracy of which can be on the order of one part in one billion of the earth’s
total field. During a period of more than one century, numerous instruments
have been invented, many of which have met with large commercial suc-
cess. It is not within the scope of this book to review the extensive history
of gravity instruments or to give details of instruments that are not now in
use. A complete description of all gravimeters mentioned in this chapter and
many others not mentioned here can be found in Nabighian et al. (2005),
along with their advantages and limitations.
In this chapter, we discuss absolute and relative instruments, gravity
gradiometry, field operations, measurement uncertainty, and ambiguity re-
lated to survey design.

Absolute-gravity measurements
The number of absolute-gravity measurements made at or near the
earth’s surface is still, in the early twenty-first century, only a very small
fraction of the total number of relative-gravity measurements made, which
is discussed below. Historically, absolute measurements were made using a
pendulum apparatus; modern devices use the free-fall technique.
The period of a simple pendulum is proportional to the square root of
its length and is inversely proportional to the square root of the local grav-
ity field. It is also a function of the amplitude of the pendulum’s swing,
which led to several refinements in the eighteenth and nineteenth centuries.

49
50  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Extensive geodetic surveys were conducted during that time. Schuler (1923)
noted that the period of a pendulum whose length is equal to the earth’s
radius would have a period equaling the orbital period of an earth satellite
at low altitude (about 84 minutes). Interestingly, this is the same time as the
round trip for an object dropped into a frictionless hole (any complete chord,
not just through the center) in a uniform nonrotating sphere whose density
is the same as the mean density of the earth. Pendulums were also used as
relative meters but have been superseded by modern gravity instruments
discussed below.
Zumberge et al. (1983) report the results of an absolute-gravity survey
in the United States in which 12 locations (requiring one day at each site)
resulted in a measurement accuracy of 10 μGal. These instruments are larger
and less portable than conventional relative meters, and both the purchase
price and the field-acquisition costs exceed those of conventional surveying
considerably. Absolute-gravity instruments now yield accuracies of about
1 μGal, are used routinely by academic and government institutions, and are
important in establishing and tying gravity networks and in studying earth
tides and crustal deformation. In recent years, in spite of the increased cost,
absolute-gravity surveys are also being used, usually in combination with
relative meters, in oil-field reservoir monitoring and in so-called 4D surveys
wherein the fourth dimension is time.
Figure 1 is a schematic diagram illustrating the free-fall method for
measuring absolute gravity. A laser interferometer is used to measure with
very high precision the distance an object is dropped. In 2004, the Bureau
International des Poids et Mesures accepted this method as an official pri-
mary method for the measurement of gravity. The time required for free fall
is measured with an atomic rubidium clock.

Relative-gravity instruments
Many gravity meters have been introduced since the 1930s. These, along
with their history and that of the torsion balance and pendulums which pre-
ceded them, are discussed in some detail by Nabighian et al. (2005), who
also include an extensive list of references on the subject.

Spring gravimeters
In the simplest of terms, the spring gravity meter works like a very
sophisticated device that can measure the elongation of a spring when sub-
jected to the weight of a test mass. Figure 2 shows that the device is brought
Chapter 4:  Field Measurements  51
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 1.  Principle of


Free-falling operation for absolute-
upper
mirror
gravity measurements.
Courtesy of Micro-g
LaCoste. Used by
permission.
Vacuum
chamber

to a “null” position at sta-


tion A, where gravity is al-
ready known or can be tied
to a known gravity sta-
tion. The device is moved
to station B and nulled
again. By careful calibra-
Interference
detector
tion of the instrument, the
change in gravity Δg can
be found by the change in
Interferometer the spring elongation Δd.
As an order of magnitude,
a change in gravity of 0.1
mGal would require the
measurement of a change
in the spring length Δd of
better than 10−5 cm.
To detect very small
changes in gravity of im-
portance to modern explo-
Stationary
lower mirror
ration, various techniques
are employed in building
instruments that can mag-
nify the very small cha-
nges in the spring. This is
usually accomplished by optical, mechanical, or electrical mechanisms.
Today, most land gravity surveys are carried out using one or more of a few
of the surviving types: the Worden and Scintrex (both using quartz springs)
and the LaCoste and Romberg (now part of the Micro-g company) and Bur-
ris (using metal zero-length springs) (LaCoste, 1934).
Figure 3 illustrates the concept of LaCoste’s zero-length spring, which
is now widely used in several gravity meters. The concept is for the system
52  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

to have an infinite period: By moving the test mass m located at the end of
a beam, the spring length S changes, but so does the angle between mg and
the moment arm b, and those effects cancel. In practice, to avoid the prob-
lem of not finding a null position or equilibrium point, the y-axis is tilted
a very small amount. The meter is quite sensitive and meets the practical
requirement for field usage.

Station A Station B

d d+ d

g+ g

Figure 2. Difference in gravity between stations related to difference in spring length.

Figure 3. Schematic diagram of


the LaCoste and Romberg zero-
length spring gravimeter. Courtesy
of Micro-g LaCoste. Used by
permission.

s
y

Pivot b

a
m

g + ∆g
Chapter 4:  Field Measurements  53
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) c)
Leveling
Level bubbles
screws
Eyepiece

Locking
knob

Adjustment dial Reading

b)

Figure 4. Spring gravimeters. (a) LaCoste and Romberg G meter. Courtesy of Micro-g
LaCoste. Used by permission. (b) Scintrex CG-5. Courtesy of Scintrex. Used by
per­mission. (c) Worden (SEG Virtual Geoscience Center, 2006).

In nearly all land, borehole, marine, and airborne surveys, a “still” read-
ing (in which the null position of the meter is determined) is taken at an
initial place, usually a base station (discussed below), at which the absolute
value of gravity is already known or can be determined by tying that station
to another station whose absolute gravity is known. Subsequent field sta-
tions or traverses are then obtained for which only differences in gravity are
determined, hence the term relative gravity.
Many of the gravity meters in use have a limited range of operation that
must be reset if the change in the earth’s gravity field within the survey area
exceeds the range over which the instrument can be nulled. The LaCoste
and Romberg instruments (Figure 4a) and a few others are designed for a
worldwide range and do not require resetting. Each gravity meter is cali-
brated before leaving the factory, leading to a calibration factor or table of
factors, enabling the user to convert gravity readings into the appropriate
units of gravity.
The operation is subject to two errors: (1) screw “backlash” and (2) meter
drift, or “spring hysteresis.” It is important for the operator to always ap­­
proach the null position by turning the screw in the same direction. If the
operator overshoots the null position, then the process should be repeated
54  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

by backing the screw position and reapproaching the null position. A good
operator can take a meter reading in just a few minutes.
All spring gravity meters capable of measuring small differences in
gravity suffer from a phenomenon known as instrument drift, which results
from the fatigue of internal components, whether made of metal or quartz,
and from small mechanical instabilities.
As discussed below and in Chapter 6, this is a time-varying function
requiring the reoccupation of stations with a frequency that depends in part
on the characteristics of the particular meter employed. A reoccupation rate
of one to two hours is commonly used in exploration work. Quartz springs
generally cause greater and more erratic drift than metal springs. Some
meters, particularly the LaCoste and Romberg, have drift characteristics
that can be treated as linear over a greater length of time and that improve
with instrument age.
Two other instruments that have found wide use in exploration are
the Scintrex CG5 (Figure 4b) and the Worden (Figure 4c). The LaCoste
and Romberg gravity meters have met with the most success and are
preferred by a majority of gravity surveyors, but of course, most meter
operators will be required to use the meter(s) their organization has in
its inventory.

Vibrating-string gravimeters
In the mid-twentieth century, several instruments were constructed
based on the principle of the vibrating string (Gilbert, 1949). An elastic
string vertically suspended with a mass at its end (under tension) vibrates
with a frequency directly proportional to the square root of the local grav-
ity field. The vibrating-string instrument was designed originally for work
in submarines (Wing, 1969). Later versions were developed for boreholes
(Howell et al., 1962; Goodell and Fay, 1964) because of the natural elonga-
tion of the instrument housing.
A more complicated version uses a vertically suspended double-
string and double-mass system. The second string and mass are mounted
below the first string and mass using a weak spring, and the entire system
is constrained at both ends. In this system, a difference in gravity results
in a difference in the tension of the two strings. Accordingly, the natural
frequencies of the strings are different, in proportion to the difference in
gravity.
In 1973, a double-string and double-mass vibrating-string sensor devel-
oped by Bosh-Arma was used to successfully obtain gravity measurements
on the moon during the Apollo 17 mission (Chapin, 2000; Talwani, 2003).
Chapter 4:  Field Measurements  55
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

That is the only time successful gravity measurements have been made by
humans on a celestial body other than earth.
Vibrating-string gravimeters have the advantage of generally being
physically smaller than spring gravimeters but with a larger dynamic range.
These instruments were subjected to considerable research and development
but were not competitive for land use and were superseded for borehole use
by the LaCoste and Romberg instrument discussed below.

Superconducting gravity meter


A superconducting gravimeter, the iGrav™ SG meter, is available but
has not yet found widespread applications in industry. These instruments
operate by suspending a diamagnetic superconducting niobium sphere
cooled by liquid helium in an extremely stable magnetic field. The current
required to generate the magnetic field that suspends the niobium sphere
is proportional to the strength of the earth’s gravitational field. Such gravi-
meters have extraordinary sensitivities of 1 nanogal (10−9 Gal) and a drift
of less than 0.5 µGal/month. Virtanen (2006) describes how an instrument
at Metsähovi, Finland, detected the gradual increase in surface gravity as
workmen cleared snow from its laboratory roof.
Land operations require that the meter is first leveled1 and then a null
reading is obtained by the operator or by automatic nulling electronics. If x
is the beam displacement, the appropriate differential equation of the motion
of the beam is (Nettleton et al., 1960)
x + δ x + cx = g + a + Ky,

where the first term is the beam acceleration; the second term is velocity mul-
tiplied by a damping coefficient δ; the third term is beam displacement mul-
tiplied by c, the net restoring force on the beam; g is the gravity acceleration
resulting from geology; a is the acceleration resulting from the motion of the
instrument; y is the spring tension adjustment; and K is an empirical coefficient.

 1 Some gravimeters made by Scintrex and by LaCoste and Romberg are self-lev-
eling for small, moderate ground inclinations. This is a very useful feature when
working in jungle areas where lines have to be cut and the Global Positioning Sys-
tem (GPS) is not operational everywhere because of tree canopy. The survey is done
with a helicopter that lowers the self-leveling gravimeter to the ground, takes the
reading, and retracts the gravimeter. The GPS-determined elevation in the helicop-
ter minus the cable length used to lower the gravimeter gives the station elevation.
56  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 5. Micro-g LaCoste air-


sea gravity system. Courtesy
of Micro-g LaCoste. Used by
permission.

In a stable land environment,


the first, second, and last terms
are zero. For marine and airborne
work, the damping coefficient δ is
made very large, the spring tension
is adjusted continuously, and a
feedback system keeps the instru-
ment close to the null position.
Inline and transverse gyroscopes
are used to maintain the platform
in a horizontal position with the
aid of horizontal accelerometers.
Details of this technique for
measuring gravity while the instru-
ment is in motion can be found in
LaCoste (1967). Techniques for
identifying the gravity signal g in
conditions where the motional ac-
celeration a might be more than 100,000 times greater are discussed in Chapter
6. The measurement of gravity on a stabilized platform is subject to cross-cou-
pling effects (an interaction between the horizontal and vertical accelerations)
for which an onboard computer has been designed (LaCoste et al., 1967). This
is discussed further in Chapter 6. An illustration of the Micro-g LaCoste air-sea
gravity system is shown in Figure 5.
Modifications to the standard land stationary instruments were introduced
in the last half of the twentieth century to enable gravity surveys in downhole,
offshore, and airborne environments. Currently, most of those instruments are
LaCoste and Romberg borehole or air/sea devices, but a few Bell Aerospace
and Bodenseewerk stabilized-platform instruments for continuous recording
are also in operation. At this writing, LaCoste and Romberg has the only com-
mercially available borehole instruments used primarily for petroleum explo-
ration, although Scintrex has recently developed a slim-hole borehole gravity
meter (Gravilog) which can also be used in mining applications. Both of these
instruments are land sensors made smaller, included in a sonde, and connected
to recording electronics by wires.
Chapter 4:  Field Measurements  57
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Satellite-derived gravity
The modern era of satellite radar altimetry, beginning with ­SEASAT in
1978, ushered in a golden age for imaging and mapping the global marine
geoid and its first vertical derivative, the marine free-air gravity field. The
SEASAT mission was equipped with oceanographic monitoring sensors and
a radar altimeter. The altimeter was designed to measure sea-surface topog-
raphy in an attempt to document the relief caused by water displacement
from large-scale ocean currents (e.g., the Gulf Stream) or water mound-
ing caused by local gravity anomalies within the earth’s crust and upper
mantle. Haxby et al. (1983) produce the first global marine gravity map
from SEASAT satellite altimeter data using interorbital track spacing of
about 180 km. The advent of a public-domain global marine gravity data-
base with uniform coverage and measurement quality provided a significant
improvement in our understanding of plate tectonics and had a significant
impact on regional exploration.
Figure 6 shows a map of a satellite-derived marine free-air gravity
field merged with a terrestrial gravity field. Understanding of the marine
free-air gravity field continues to improve as additional radar altimeter
data are acquired by new generations of satellites. The subject of gravity
measurements from satellites is treated in more detail in the section titled

–150° –120° –90° –60° –30° 0° 30° 60° 90° 120° 150°

60° 60°

30° 30°

0° 0°

–30° –30°

–60° –60°

–150° –120° –90° –60° –30° 0° 30° 60° 90° 120° 150°
mGal
–61.1 –31.6 –21.4 –14.9 –10.0 –5.6 –2.7 0.11.5 4.4 7.5 10.6 14.2 18.4 24.0 32.5 48.3

Figure 6.  Satellite-derived marine free-air gravity field merged with terrestrial
gravity field (Sandwell and Smith, 1997, 2001). Courtesy of D. T. Sandwell. Used
by permission.
58  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

“Satellite gravity and satellite-derived gravity” in Chapter 9 and will not be


discussed further here.

Gravity gradiometry
The conventional gravimeter measures a single component (the vertical
component) of the gravity-field vector. As we have seen in Chapter 3, the
gravity field can be represented as the gradient of a potential:

 ∂U 
 ∂x 
 gx   
∂U 
g = ∇ U =  gy  = .
 ∂y 
 g   
 ∂U 
z

 
∂z 

A gravity gradiometer measures the gravity gradient or how the gravita-


tional acceleration changes over distance in the horizontal and vertical di-
rections. A full-tensor gradiometer measures the changes in the x-, y-, and
z-directions of each of the three components of the gravity field gx, gy, and
gz, yielding the tensor

 ∂ 2U ∂ 2U ∂ 2U 
 dx 2 dx dy dx dz 
 U xx U xy U xz   
   ∂ 2U ∂ 2U ∂ 2U 
∇∇ U =  U yx U yy U yz  = ,
 dy dx dy 2 dy dz  (1)
 U U zy U zz 
zx  ∂ 2U ∂ 2U ∂ 2U 
 
 dz dx dz dy dz 2 

where Uxz is the gradient in the z-direction of the x-component of gravity and
similarly for the other components. The above tensor with nine elements
is symmetrical, e.g., Uxy = Uyx, and the diagonal elements are connected
through Laplace’s equation, thus leaving only five independent components
out of nine. Notice also that as a result of differentiation, the sources of grav-
ity gradients are not monopoles anymore.
The first gradiometer was the torsion balance, developed in 1886 by
Baron Loránd Eötvös. Two weights were suspended from a torsion fiber at
unequal heights. The weights were separated vertically and horizontally, so
they experienced different forces because of their spatial separations. From
Chapter 4:  Field Measurements  59
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

these, one can determine both the horizontal gradient of the vertical com-
ponent of gravity and the horizontal gradient of the horizontal component.
With careful measurement procedures, accuracies of a few Eötvös units
(1EU = 10−9/s2 = 0.1 mGal/km) could be obtained.
Such an instrument was used extensively in exploring for salt domes,
particularly on the U. S. Gulf Coast, culminating with the first geophysical
discovery of an oil field at the Nash salt dome in 1924. Although accu-
rate, the Eötvös torsion balance was slow and cumbersome, and it was sup-
planted by the now familiar gravity meter when it became available between
1935 and 1940.
The driving force behind the development of gravity-gradiometer sys-
tems in recent times has been their use on moving platforms. Usual airborne
gravimeters require significant corrections for the vertical acceleration of
the platform and velocity-dependent interactions with the rotation of the
earth (Eötvös effects; see Chapter 6). Although the use of the Global Posi-
tioning System (GPS) has greatly improved the situation, the above fac-
tors are still the main impediments to achieving high accuracies in airborne
gravity measurements. In principle, gradiometers are completely immune
to these effects, although in practice, the effects are always present at some
level. Regardless, gravity gradiometers have typically higher accuracy and
better spatial resolution than gravimeters do.
Today, there are two commercially available gravity gradiometers:
the full-tensor gradient (FTG) system used by Bell Geospace and ARKeX
(built by Lockheed Martin) and the Falcon system developed by BHP Bil-
liton, manufactured by Lockheed Martin and now operated by Fugro. The
FTG system measures the five independent elements of the full gravity
tensor shown in Figure 1, whereas the Falcon system measures the differ-
ential curvature gradients:

∂ 2U ∂ 2U ∂ 2U
− and ,
dx 2 dy 2 dx dy

which are then transformed into the more common vertical gravity gz and
vertical gravity gradient gzz during data processing to form maps.
Both systems are a direct result of gravity-gradiometry developments
by the U. S. Navy for use on its submarines. The FTG is used for land,
marine, submarine, and airborne surveys, whereas the Falcon is used for
airborne surveys only. In addition, Stanford University, the University of
Western Australia, Gedex, and ARKeX are all designing their own new air-
borne gravity-gradiometer systems.
60  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The FTG system uses three small-diameter gravity-gradient instruments


(GGIs) mounted on an inertial stabilized platform (Figure 7). Each GGI
contains four gravity accelerometers mounted on a rotating disk in a sym-
metrical arrangement such that each of the individual accelerometer input
axes is in the plane of the rotating disk, parallel to the circumference of the
disk and separated by 90º. The individual accelerometers consist of a proof
mass on a pendulumlike suspension that is sensed by two capacitive pick-off
rings located one on either side of the mass.
The signal generated by the pick-off system is amplified and converted
to a current that forces the proof mass into a null position. The current is
proportional to the acceleration. Vehicle accelerations are eliminated by
frequency separation in which the gradient measurement is modulated at
twice the disk-rotation frequency (which is 0.25 Hz), leading to a forced
harmonic oscillation. Any acceleration from a slight imbalance of oppos-
ing pairs of accelerometers is modulated by the rotation frequency. This
permits each opposing pair of accelerometers to be balanced precisely and
continuously.
Six gravity-gradient components are measured and referenced to three
coordinate frames. From these six components, five independent compo-
nents can be reconstructed in a standard geographic reference frame. The
remaining components of the gravity-gradient tensor are constructed from

12 accelerometers, three disks


Spin axis
xis

y-a x-a
12

36°

xis

Figure 7. The full-tensor gradient (FTG) system by Bell Geospace (made by


Lockheed Martin). Courtesy of Bell Geospace. Used by permission.
Chapter 4:  Field Measurements  61
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Laplace’s equation and the symmetry of the tensor. For a review of the FTG
sensor design, see Jekeli (1988) and Torge (1989).

Satellite-measured gravity gradiometry


The gravity field and steady-state ocean circulation explorer (GOCE)
and the gravity recovery and climate experiment (GRACE) satellites are
also capable of measuring the earth’s gravity gradients, which allows scien-
tists to have access to the most accurate model of the geoid ever produced to
further our understanding of the earth.
The GOCE satellite flies at a low altitude of 250 km and has six highly
sensitive accelerometers that measure gravity gradients. GRACE uses two
satellites flying about 220 km apart in a polar orbit at an altitude of 500 km.
It determines gravity components by making accurate measurements of the
distance between the satellites with an accuracy of a few microns, using
GPS and a microwave ranging system.

Field operations
Gravity data are acquired on the land surface, on the sea bottom, on
the sea surface, in the air, and in boreholes and mine shafts. In the early
twenty-first century, field operations are planned and executed in all these
environments except that underwater (sea-bottom) surveys have become
nearly extinct, having been replaced almost entirely by surface-ship opera-
tions2. The latter are compatible with modern 2D and 3D seismic opera-
tions being run simultaneously with gravity surveys. Gravity-only marine
surveys, wherein the cost of the ship and positioning system must be borne
solely as a gravity expense, have been superseded by gravity-gradiometry
surveys.
More than 10 million land gravity stations have been acquired through-
out all continents (Nabighian et al., 2005). Most of them are given an abso-
lute-gravity reference by tying local surveys to national and international
networks. Typically, in a new survey, base stations are established by carry-
ing absolute values from nearby networks. The new field stations are then
surveyed in a sequence (a “loop”), starting and ending with a measurement

 2Recently, underwater (sea-bottom) instruments have seen a resurgence, being


applied in time-lapse gravity surveys (or 4D, in which the fourth dimension is time)
for reservoir monitoring (see also the section titled “Time-lapse [4D-gravity] sur-
veys” in Chapter 9).
62  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

at the newly designated base3. Time-dependent corrections are applied as


described in Chapter 6, and the principal parameters (horizontal coordi-
nates, elevation, and absolute gravity) are recorded for each station.
The time duration of each loop should be short enough (one to two
hours) so that the drift of the instrument (discussed above) can be treated
as a linear function of time. In addition, one or more stations should be
repeated within the same loop (as a check on instrument drift) and one or
more at previously established stations in a different loop so that a statement
can be made as to the statistical repeatability of the survey. The station spac-
ing (meters or tens of meters for archaeological and engineering surveys
and tens, hundreds, or thousands of meters for exploration surveys) is deter-
mined based on the distributions and depths of the subsurface targets. One
rule of thumb is to separate stations not more than half the expected target
burial depths, although there are exceptions, as noted in the section below
titled “Measurement uncertainty.”
Marine and airborne gravity surveys entail the operation of moving-
platform gravity instruments along multiple traverses, called primary lines
and tie lines. Unlike land gravity measurements, which are made at rest,
marine and airborne gravity observations are subjected to both horizontal
and vertical motion, requiring filtering and corrections for the unwanted
motional effects. Modern GPS acquisition and processing technologies
have markedly improved the final gravity product in marine and airborne
environments.
Marine operations have the advantage that the vertical motion of the
ship oscillates about a constant elevation (near sea level), whereas airborne
operations typically fly at a constant barometric elevation for which the
average vertical coordinate can vary from traverse to traverse and from day
to day. In both cases, however, the intersection differences between the val-
ues obtained along the primary lines and those observed on the tie lines are
very important for adjusting the gravity data and for estimating the accuracy
of the survey. In offshore 3D seismic surveys, gravity data can be acquired
along lines that are very closely spaced, yielding a near redundancy and
enabling significant improvement in noise reduction.
More than 1000 wells have been logged in hydrocarbon explora-
tion using the LaCoste and Romberg borehole gravity meter (Figure 8).
However, the widespread use of this tool has been curtailed in the last
few decades because of instrument limitations: minimum hole diameter

 3Sometimes absolute-gravity meters can be used at the end of a line segment to


avoid returning to the designated base.
Chapter 4:  Field Measurements  63
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Downhole
electronics

Insulated
sensor

Shown with
restricted
movement
in sonde
Maximum
hole deviation
14.5°

Figure 8. LaCoste and Romberg borehole gravity meter. Courtesy of Micro-g


LaCoste. Used by permission.

of about 5 inches, hole deviation from vertical of not more than about
14°, and temperature limits that generally preclude depths much greater
than 12,000 to 15,000 ft (3700 to 4600 m). Poisson’s equation (equation
12 of Chapter 3) provides a basis for determining apparent bulk density
of a large volume of rock beyond the borehole and between stations. We
will examine this determination of density in Chapters 5 and 7.
Currently, instrument readings are static, similar to land measurements,
but are observed remotely using electronics connected to the meter by a
wire line extending to the depth of observation. As with land operations
discussed above, discrete stations are observed in loops to enable the iden-
tification and removal of instrument drift. In oil and gas work, the stations
are usually separated by 20 ft (7m), but smaller and larger intervals might be
incorporated, depending on the nature and expected distance from the well
to the geologic source.
The resolution of apparent density is affected by the relative accuracy of
the determination of station depths (generally, but not universally, 1 cm or
larger) as well as that of the gravity instrument, which is usually more than
5 μGal. The recently developed slim-hole Gravilog system by Scintrex does
not have some of the limitations of the BHGM system because it is operational
in boreholes inclined as much as 60º from the vertical and with an accuracy of
more than 5 μGal. The Gravilog system presently can survey in boreholes of a
64  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

minimum diameter of 57.2 mm (2¼ inches), but only to depths as much as


2000 m.
Research and development of new borehole tools to observe at higher
temperatures (greater depth) and to enable leveling at any hole deviation
continue to be very active. One of the principal goals in modern research is
to develop a gradiometer that can enable continuous logging.
Gravitec Instruments has developed a sensor technology conceived by
Veryaskin (1999) which employs a “string,” or “ribbon,” as the detection
device. The string device has the unique ability to detect gradient signals
with a single sensing element, as opposed to conventional gravitational
instruments which are based on multiple accelerometers. This instrument
is characterized by a lack of moving parts and very small size and weight.

Measurement uncertainty
Limitations on and resolution of our ability to interpret gravity anomalies
in geologic terms (Chapters 7, 8, and 9) start with the uncertainties inherent
in field measurements, and these uncertainties depend strongly on the nature
of the field operation. Figure 9 summarizes the achievable accuracies for
land, borehole, underwater, gradiometer, surface-ship, airborne, and satellite
operations. For land, borehole, and underwater surveys, discrete readings are
obtained, and observable wavelength is a function of station spacing (see the
section below titled “Ambiguity related to survey design — Aliasing”). The

Figure 9. Achievable 100


Smallest
accuracies for various amplitude
gravity surveys in terms measurable
10 Satellite
of shortest observable (mGal)
wavelengths (LaFehr and
1
Nettleton, 1967; Dransfield Surface-ship Airborne
et al., 2001; Fairhead and Underwater
Odegard, 2002; Nabighian 1/10
et al., 2005). Gradiometer

Borehole
1/100
Land

1/1000
Shortest wavelength observable (km)

1/100 1/10 1 10 100


Chapter 4:  Field Measurements  65
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

lower left end of the curve labeled “Airborne” can be achieved by slower-fly-
ing helicopters, whereas the upper right end represents fixed-wing operations.
Under acceptable surveying conditions, land and borehole gravity oper-
ations can achieve resolutions between 1 and 20 μGal; underwater (station-
ary) gravity between 0.1 and 0.2 mGal; surface-ship gravity about 0.1 mGal
over wavelengths of less than 500 m; and airborne gravity (fixed-wing)
about 1 mGal over wavelengths of less than 2-km half-wavelength from an
airplane and more than 0.5 mGal over wavelengths of less than 1-km half-
wavelength from a helicopter.
These performance figures are hotly debated, and it is often difficult
to find comparable data from different companies because there are many
ways to present resolution performance. Free-air anomalies over the oceans,
based on satellite measurements, are approaching resolutions of 2 to 5 mGal
over wavelengths of 5 to 10 km (Sandwell et al., 2003). Generally, reso-
lution limits resulting from instrument and operational considerations (in
well-run field surveys) are of less concern than those imposed by the con-
straints of data reduction and interpretation (Chapters 5 and 6).

Ambiguity related to survey design — Aliasing


For proper characterization of a given anomaly, the anomaly has to
be properly sampled. As can be seen from Figure 10, reducing the spa-
tial sampling rate also reduces the high-frequency content of the data,
resulting in a completely distorted picture. This process, known as alias-
ing, occurs when an anomaly is measured at an insufficient sampling rate.
Aliasing is an effect that causes different continuous signals to become

Figure 10. Example


of aliasing the data
(marked with ×) as the
sample interval is halved
successively.
66  Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

indistinguishable (or “aliases” of one another) when sampled. Aliasing


can be avoided by sampling at a frequency at least twice as high as that of
the waveform.
If the sampling rate is Δx, then the shortest wavelength that can be
defined accurately is

TN = 2 Δx,
and correspondingly, the highest frequency can be defined as
fN = 1/2 Δx,

which is known as the Nyquist frequency. If the data contain frequencies


higher than the Nyquist frequency, all the higher frequencies are folded
back as lower frequencies. In other words, a frequency fN + Δf is folded
back around fN and will look similar to a frequency fN – Δf, i.e., these two
frequencies are aliases of each other. Filtering with an antialiasing filter4
before sampling the frequencies above the Nyquist frequency is another
way to avoid aliasing.
As mentioned in the section above titled “Gravity gradiometry,” a good
rule of thumb is to separate stations by no more than half the expected target
burial depths to avoid aliasing. In the presence of both deep and shallow
sources, station spacing is chosen based on the depth to shallow targets,
even though the target of interest is at a greater depth. Not choosing such
spacing will alias the anomaly from the shallower targets, which will then
appear as deeper ones.

 4An antialiasing filter is a filter used before sampling the signal, to restrict the
bandwidth of a signal to approximately satisfy the sampling theorem.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 5

Rock Density and Gravity Anomalies

Introduction
Equation 15 of Chapter 3 is, in theory, a unique formula for perform-
ing the forward calculation that produces the gravity anomaly caused by a
subsurface density distribution. On the right side of that equation, under the
integral, are the density r and the geometric components of an element of
mass to be summed over the entire geologic body. In this chapter, we exam-
ine the nature of density in exploration, its determination, and finally, how a
lateral density contrast is required to cause an observable anomaly.

Typical near-surface rock densities


In Chapter 2, we found that the mean density of the earth is about
5500 kg/m3, but the densities of rocks of general interest in exploration (near
the earth’s surface) are much lower —1600 to 2600 kg/m3 for sedimentary
rocks, 2200 to 3300 kg/m3 for igneous rocks, and 2400 to 3500 kg/m3 for
metamorphic rocks.
Figure 1 gives an example of the density range of commonly encountered
rocks in the shallow earth’s crust compiled from various field studies. We see
that some correlation exists between geologic age and density — older rocks
tend to be denser, but this is not a safe guideline in interpretive work. In gen-
eral, mafic intrusive rocks (e.g., basalts, gabbros, and so forth) are denser than
felsic intrusive rocks (e.g., granites, rhyolites, and so forth). In sedimentary
basins, the porosity in sands and shales tends to decrease with depth because
of compaction, resulting in a general increase in density with depth for those
rocks, whereas salt, which is almost incompressible, is nearly constant in den-
sity with respect to depth. Fluids are low in density: 1000 kg/m3 for water,
about 1030 kg/m3 for seawater, and 600 to 900 kg/m3 for oil.

67
68  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Density (kg/m3)

2000 2400 2800 3200

Cenozoic

Mesozoic

Gabbro
Paleozoic Salt

Precambrian Mica schist


Gneiss
Shale
Quartz diorite Peridotite
Sandstone
Granite
Carbonate rocks

Figure 1. Densities of rocks.

Densities of a wide range of rocks and of the minerals from which they
are composed are tabulated in several references, such as Clark (1966).
Density variation in rocks of exploration interest is generally not large.
However, gravity interpretation is sensitive to the selected density contrast,
which can vary substantially. Data have been compiled to show this range
in Figure 2a.
We can easily see that the total range of absolute density for most explo-
ration projects is only a factor of two or less, but anomalies are caused by
lateral density contrast, not absolute density. A quick study of the density
ranges suggests that density contrasts can vary over a considerably larger
range, even reversing sign in some cases, as shown in Figure 2b. The increase
in density for sands and shales with respect to depth and the nearly constant
density for salt discussed above are demonstrated in Figure 2b.
Salt density of about 2200 kg/m3 (pure halite is nearer 2150 kg/m3, but
salt is often mixed with foreign constituents) is higher than that of the most
recently deposited sands and shales, which have very little overburden, and
is lower at greater depths where compaction is increasingly significant. The
depth at which salt density equals that of the surrounding sands and shales is
known as the crossover depth. The density-depth curve for sands and shales
varies from location to location. Onshore in Texas and Louisiana, crossover
depth can be as shallow as 700 m. However, crossover depth is progres-
Chapter 5:  Rock Density and Gravity Anomalies  69
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)
Igneous and metamorphic

Limestone

Shale

Sandstone

Soil and alluvium

Salt

1600 2000 2400 2800

Density (kg/m3)

b) 0

Salt
2
Depth (km)

4 Sand-
Crossover depth shale

Basement
8

10

1600 2000 2500 3000

Density (kg/m3)

Figure 2. (a) Density range for various rocks and for soil and alluvium.
(b) Density of sedimentary rocks as a function of depth over a salt dome.
70  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

sively greater in offshore regions, perhaps reaching depths of 1300 to 2600


m, and it is not a single point but can exist over a depth range of 300 m or
more (which can be defined as the nil zone).

Density and porosity


The bulk density of a rock rb is a function of the matrix density rm,
porosity f, and the density of the fluid rf occupying the rock’s cavities:

rb = rm (1 – f) + rf f.(1)

The fluid density can range from very low values, near zero if substantial
gas is present, to about 1030 kg/m3 or higher for seawater or brackish water.
We plot the bulk density for a range of matrix densities and porosities in
Figure 3, assuming that the cavities are filled with water that has a density
of 1000 kg/m3. A typical matrix for a wide range of rocks would be 2650 kg/m3,
i.e., that of quartz (SiO2). The matrix densities of anhydrite, dolomite, and
calcite are higher, as shown, and those of gypsum, halite, and sulfur are
lower, as also shown in Figure 3.

3000 Water-filled holes Anhydrite

2800 Dolomite
Calcite
Po Quartz
ro
2600 sit
Bulk density (kg/m3)

y(
%
)
0
2400
10
20
2200 Gypsum

Halite 30

Sulfur 40
2000

50

1800
2000 2200 2400 2600 2800 3000

Matrix density (kg/m3)

Figure 3. Bulk density as a function of matrix density for varying porosity.


Chapter 5:  Rock Density and Gravity Anomalies  71
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Note that the zero-porosity line shows that in the case in which the rock
cannot contain any fluid, the bulk density is everywhere equal to the matrix
density, and as porosity increases, bulk density decreases. This is consistent
with the information in Figure 2, which shows that sedimentary densities
generally increase with depth in sedimentary basins.
By rearranging equation 1, we obtain porosity in terms of bulk, matrix,
and fluid densities:
ρm − ρb
ϕ= .(2)
ρm − ρ f

This relationship is shown in Figure 4 for the common matrix density appro-
priate to that of quartz (SiO2). In this figure, the fluid density ranges from 0
to 1000 kg/m3. Of course, for zero porosity (indicated at the top of the graph),

0
Matrix density = 2650 kg/m3

10

20
Porosity (%)

30

Fluid densities (kg/m3)


40 0 200 400 600 800 1000

50
1600 2000 2400 2600

Bulk density (kg/m3)

Figure 4. Porosity as a function of bulk density and fluid density for a matrix of
quartz.
72  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

no fluids exist in the rock, so the bulk density of the rock in that case is equal
to its matrix density, the point where all lines in the graph converge.
In addition to indicating the dependence of bulk density on the porosity
and fluid contained in the rock, Figure 4 prepares us for an important applica-
tion of the borehole gravity meter (treated in Chapter 6): In a reservoir where
the matrix and fluid densities are known or can be assumed within a range
from core samples and/or gamma-gamma/density logs, the bulk density (de-
rived from borehole gravity) of a large volume of rocks can be used as an in-
dependent means for estimating porosity and therefore fluid volume. This
determination can have a distinct advantage because its estimate is based on
a much larger part of the reservoir than estimates by other logging tools are.
In Figure 4, for a bulk density (shown on the horizontal axis) of about
2300 kg/m3, the porosity would fall between 13% and 21% (shown between
the dashed lines on the vertical axis), depending on the fluid density. In the
second case, a bulk density (horizontal axis) of slightly less than 2000 kg/m3
yields a porosity of slightly greater than 30% if the gas-saturated fluid has an
average density of 400 kg/m3.

Constituent densities
Although matrix density varies substantially, it is often taken in oil and
gas exploration to be 2650 kg/m3, or that of silicon dioxide (SiO2). Common
exceptions are salt (2150 to 2160 kg/m3) and dolomite (2870 kg/m3). Clastic
sediments have bulk densities which are a function of grain size and compo-
sition (typically quartz and feldspars) and of porosity. Although porosity is
not a factor for igneous and metamorphic rocks, mineral assemblage is, and
the wide range of constituent densities gives rise to the range in rock densi-
ties shown in Figures 1 and 2. Table 1 lists constituent densities confined to
the range of 2000 to 3300 kg/m3.
The bulk density of any rock is equal to the sum of its constituent densi-
ties, each multiplied by the percent volume (fn) of the rock it occupies:

ρb = ρ1ϕ1 + ρ2ϕ 2 + ρ3ϕ 3 + ……

Methods for deriving, measuring, and evaluating density


Six methods are used widely for determining the densities of rocks in
exploration projects: (1) laboratory measurements of cores, cuttings, and
surface rock samples; (2) logging tools using the gamma-gamma instrument;
(3) the Nettleton profiling technique for surface or water-bottom topogra-
phy; (4) borehole gravity; (5) conversion of seismic interval velocities; and
Chapter 5:  Rock Density and Gravity Anomalies  73
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Table 1. Densities for constituents in the range of 2000


to 3300 kg/m3. Derived from Clark (1966).
Constituent Formula Density (kg/m3)
Salt NaCl 2160
Gypsum CaSO4-H2O 2320
Orthoclase KAlSi3O8 2550
Nepheline NaAlSiO4 2620
Quartz SiO2 2650
Anorthite CaAl2Si2O8 2760
Muscovite KAl2AlSi3O10(OH)2 2830
Dolomite CaMg(CO3)2 2870
Wollastonite CaSiO3 2910
Aragonite CaCO3 2930
Lime olivine Ca2SiO4 2940
Anhydrite CaSO4 2960
Bromellite BeO 3010
Andalusite Al2SiO5 3140
Enstatite MgSiO3 3200
Olivine Mg2SiO4 3210
Diopside CaMg(SiO3)2 3280

(6) inversion, either unconstrained or constrained by seismic structure, well


points, and/or other geologic information.
Every one of these methods has its limits as to effectiveness. Cores,
cuttings, and rock samples can be damaged or unrepresentative or can vary
because of the difference between atmospheric pressure in the laboratory
and in situ pressure in the earth. The gamma-gamma logging tool is subject
to errors in calibration, effects of hole rugosity, fluid invasion, or formation
damage, and it is not effective in cased holes.
The Nettleton profiling technique (Figure 5) is often used routinely as
part of data reduction. Several densities are selected in the calculation of the
Bouguer anomaly (Chapter 6). If we can assume that the topography does
not correlate with subsurface structure (this is a risky assumption because,
in fact, the topography might owe its existence to such correlation), then we
select the density that leads to the least correlation between the computed
anomaly and topography.
The example shown in Figure 5 indicates a density for the topographic
rocks of 2200 kg/m3, suggesting little or no correlation. If, however, a posi-
tive correlation is appropriate because of subsurface structure, then a lower
74  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Reduction
density
(kg/m3)
Reduced
anomaly
(mGal) 1800
20
2000

2200

2400
10

2600
Gravity profiles for
various densities
Elevation
(m)
4500
Topography
0
4400
4300

4200

4100
500 1000 1500
Traverse distance (m)

Figure 5. Nettleton profiling. After Nettleton (1971), Figure 5.

density for the topographic feature would be correct in this example. The
Bouguer correction requires the assumption of near-surface rock density
and is treated in Chapter 6.
If the topography (or bathymetry in marine work) does correlate with a
geologic feature below it, as would be the case of an erosion-resistant struc-
ture or low-density alluvium in a valley or drainage channel, it becomes a
matter for interpretation in selecting the most likely surface density. Even
so, Nettleton profiling can be a valuable source of surface-rock density
estimates.
Four of the six methods for determining density are indicated in Figure 6,
depicting a hill with a well drilled on the left side of the illustration. The
Nettleton profiling technique is applied to the gravity data taken on the
topography. Density for the hill rocks is based on the assumption that no
correlation exists in this example, and therefore, the appropriate Bouguer
reduction density is 2300 kg/m3. The subsurface is assumed to be a simple
Chapter 5:  Rock Density and Gravity Anomalies  75
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Gravity anomalies
(mGal)
Density
(kg/m3)
Density too low

2000

Correction density
for no correlation
2300

Density too high 2600

Hill

2300

2400

2500

2600

Core or Gamma- Borehole


cuttings gamma gravity
log meter

Figure 6. Four methods for determining rock density.


76  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

layered earth from which the laboratory-derived densities for the cores and
cuttings obtained by drilling yield reasonably average densities, perhaps
slightly low in comparison with those of buried rocks.
The gamma-gamma-based formation-density tool also shows a high-
frequency display, and for gravity-interpretation purposes, it should be cali-
brated for the rock types present and averaged over larger vertical distances.
The borehole gravity method is treated in Chapter 7.
Perhaps the most effective relationship between seismic velocity and
formation density in sedimentary rocks is that of Gardner et al. (1974):

r  =  a V1/4,
where r is density in kilograms per cubic meter, and the value of a is 310
if the P-wave seismic velocity (V) is given in meters per second and is 230
if the velocity is expressed in feet per second. Gardner’s relationship is
derived from empirical laboratory studies, characterized in Figure 7 and in
a log-log plot in Figure 8. As with any method based on seismic velocity,

24

20
Seismic velocity (kft/s)

16

12
Gardner

4
2000 2500 3000
Density (kg/m3)

Figure 7. The Gardner velocity-density relationship.


Chapter 5:  Rock Density and Gravity Anomalies  77
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Log of velocity
3.6 3.8 4.0 4.2 4.4
0.2

1.8
Gardner

2.0

Bulk density (g/cm3)


0.3
Log of bulk density

Sandstone
Rock salt
2.2
Shale
2.4
0.4
Limestone 2.6
Dolomite
2.8

Anhydrite 3.0
0.5
5 10 20
Velocity (kft/s)

Figure 8. Velocity-density relationships (Gardner et al., 1974).

the resulting density values are subject to errors in stacking velocity that
sometimes occur because of energy dispersion, but the data also indicate
that lithology is a major determinant.
Because the Gardner curve conforms well with and is central to the lines
plotted for shales and sandstones (Figure 8), it is used extensively in gravity
interpretation in the oil and gas industry, especially in the Gulf of Mexico. In
Figure 8, lines of equal acoustic impedance (not shown), along which seis-
mic reflections vanish, would be perpendicular to the Gardner curve, which
is a straight line in the log-log plot. Thus, for some geologic boundaries,
such as between some sands and shales, we expect weak seismic reflections
or even none at all.
It is also true that sufficient density contrast to produce gravity anom-
alies might be lacking, reinforcing the need for multiple tools. Figure 8
shows both the utility and the difficulty in using a single density-velocity
relationship for all rock types. Figure 8 also indicates that substantial den-
sity contrasts can occur where seismic reflections are weak or that strong
reflections can occur where density contrasts are not detectable, such as at
the crossover depth with salt, as shown in Figure 2. Density and velocity do
not always vary directly.
The sixth method for determining and using density contrasts in inter-
pretive work is depicted in Figure 9, along with the Nettleton profiling
78  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Observed gravity

Shallow
density
determination
Topography or
bathymetry
Depth or elevation

Deep density
determination
Time

Seismic reflections
Seismic section

Stacking velocity Interval velocity Interval density

Figure 9. Schematic diagram showing gravity, topography, and seismic section.

technique discussed above. The schematic diagram in Figure 9 suggests


the multifold nature of the problem — all horizontal density contrasts pro-
duce anomalies that are superimposed. Shallow features produce relatively
steep gravity gradients, but there can be much overlap in the character of the
anomalies. Seismic reflections are, of course, measured in time, and some
discrepancy can result when converting them to depth (required for inver-
sion of the gravity anomalies for density).
Both constrained and unconstrained inversion of gravity anomalies can
be influenced by errors in anomaly identification, which we will examine in
Chapter 6. The conversion of seismic interval velocities to density is one of
the most important techniques in oil and gas gravity exploration, but it suf-
fers from a lack of universal applicability.

Definition of what causes a gravity anomaly


The difference in density between that of the rocks of interest and that
of the adjacent rocks, or Dr, is what is needed for making geologic interpre-
tations. Two simple examples in which density contrasts arise from struc-
tural uplift are shown in Figures 10 and 11.
In Figure 10, the four horizontal beds, before being uplifted, produce
an unchanging Bouguer gravity anomaly (Chapter 3) equal to 2pkr1t1 +
Chapter 5:  Rock Density and Gravity Anomalies  79
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

2pkr2t2 + 2pkr3t3+ 2pkr4t4 (by the law of superposition), where r is bed


density and t is bed thickness. After uplift, a local anomaly is produced
depending on the geometry of the structure and the density contrasts, which
are the differences, r2 – r1,  r3 – r1,  r3 – r2,  r4 – r2, and r4 – r3.
Prior to uplift, the uniform horizontal beds provide an ideal geologic
environment for the seismic-reflection method, but with the absence of a

Gravity profile

Density contrasts for uplifted layered earth

1
ρ2 – ρ
1
ρ3 – ρ
1
2 ρ3 – ρ2

3 ρ4 – ρ2

4
ρ4 – ρ3

Figure 10. Density layers and contrasts in kilograms per cubic meter. After
Nettleton (1971), Figure 1.

ρ ∆ρ ρ

2200 kg/m3 200 2000

2200 0

2200
2400 200

2400 0
2400
2800 400

0 2800
2800

Figure 11. Horizontal density contrasts resulting from fault.


80  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

lateral density contrast, the gravity method is of no help. After uplift, the
geologic environment becomes more interesting and more problematic for
the seismic method (because of energy dispersion). It also becomes ame-
nable to the gravity method, as we will see in Chapter 6.
Without uplift or without a change in density within layers, no gravity
anomaly can occur. Both the magnitude of the uplift (geometry) and the
magnitude of the density differences are important factors in the production
of the anomaly, which is the sum of the superimposed effects with the con-
trasts indicated by the hachured zones in Figures 10 and 11.
Superposition of effects is also demonstrated in Figure 11, where a sim-
ple fault is depicted on the right side. The vertical extent to which a faulted
bed has the same density on both sides of the fault means that the absence
of density contrast produces no anomaly. However, where deeper beds that
have greater densities are upthrown against shallower beds that have lower
densities, an anomaly is produced from part of the subsurface structure, as
shown by the hachured areas in Figure 11.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 6

Data Reduction

Introduction
Lateral variations in the density of rocks cause variations in the gravity
field measured at the surface, and our central problem in gravity explo-
ration is to discover the nature of subsurface rocks, their constituents,
their structure, and their distribution. Toward this end, we use the theory
and tools developed and described in the first five chapters. In general,
the observed gravity value go is equal to the sum of the gravity anomaly
ga caused by the geologic masses we wish to study and the contribution
resulting from “noise,” gN . For present purposes, we will define the noise
contribution,

gN = g p + gg + gi + gd , (1)

as the sum of all unwanted effects, where gp represents all the effects caused
by variations in position, elevation, speed of the instrument, and so forth,
for which standard corrections apply; gg represents geologic noise effects
caused by unknown or uncertain geologic features other than our target(s) of
interest (discussed in Chapters 7 and 8); gi represents untreated instrumental
noise, such as nonlinear drift components in the instrument; and gd includes
survey design noise (aliasing), as shown in Figure 10 of Chapter 4.
In the data-reduction phase of gravity work, our goal is to identify and
remove the effects that make up the first term on the right side of equation 1,
gp, and to evaluate the potential magnitudes of the last two terms, gi and gd.
The intended result is an anomaly field in which all the unwanted contribu-
tions to measured gravity have been partly eliminated and partly minimized
and understood.

81
82  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

In establishing the rules, methods, and steps for data reduction in the
determination of gp, we refer to Chapter 2, where we defined three surfaces:
the topography and two equipotential surfaces defined as the ellipsoid and the
geoid (Figure 4 of Chapter 2). (A fourth surface, the imaginary ­projection
on which we establish latitude and longitude and perhaps other ­­­co­­­ordinates,
is generally understood as providing for a base map which gives ­geographic
orientation for the interpretation.)
The topography is the air-rock interface onshore and the water-rock
interface offshore, for which exist digital elevation models (DEMs) for
many surveyed areas. This surface, depicted in Figure 4 of Chapter 2, is
the actual surface on which we generally make measurements (onshore).
Offshore, we make measurements on the water bottom and on the sea sur-
face. (Measurements are also made in submarines by the military but not
generally in commercial exploration.) In addition, measurements are made
aboard aircraft and in boreholes.

Reduction of gravity survey data


We begin our discussion of infield reduction with typical land surveys.
Later, we will consider other types of data acquisition that require special-
ized treatment. All but a relatively small number of gravity stations continue
to be acquired by using relative-gravity meters, described in Chapter 4. This
means that prior to reducing the survey data, an already existing station for
which absolute gravity is known must be observed from a preexisting net-
work or carried to or near the new survey, thereby establishing the first base
station in the survey.1

Time variations
As we have seen, instruments have a characteristic drift resulting from
metal fatigue, generally but not always decreasing in value with time. Some
instruments drift more or less linearly with time; others can be erratic. Older
meters, like fine wines, often but not always improve with age and are

1 Although information on a large number of stations has been published (e.g.,


Woollard, 1958), with their principal facts and locations (many at airports), some
surveys are too remote and/or too urgently required to benefit from a tie with a
known or established network. In those cases, an arbitrary datum is used that is
independent of the world network. If available, absolute-gravity meters can be used
to establish the absolute-gravity value at one or more base stations.
Chapter 6:  Data Reduction  83
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

characterized by lower drift rates in their later years of operation. Quartz


springs (found in Worden and Scintrex instruments, for example) can be
notorious for producing very high drift rates.
Lacking clear reliability in drift behavior, instruments therefore must be
made to reoccupy the same station (typically a base station) and notice the
difference (the drift) in observed values. The shorter the time between base
readings, the more likely the drift can be removed accurately by assuming a
linear rate (or higher-order drift function using a larger number of repeated
stations) as a function of time. Thus, an observation is made at a base sta-
tion at the beginning and end of each loop of field stations to establish an
instrument-drift curve for that loop. In addition, it is also good field practice
to acquire at least one repeat station within the loop as a check on drift and
at least one repeat station from a different loop as an additional data point
for estimating survey accuracy.
As a good practice, surveying a loop should last no more than a maxi-
mum of one to two hours. Often, tears (abrupt step function changes) can
occur, caused by a bump to the meter, a rough transport between stations, or
extreme temperature changes. Although these changes are easily observed,
they are not so easily corrected, sometimes requiring the reoccupation of an
entire loop of stations.
In addition to instrument drift, gravity measurement is subjected to the
time variation of earth tides, the gravitational effects of which are a function
of both time and location. Most of this effect is caused by the well-known
positions of the moon and sun relative to the station location, but more com-
plicated secondary effects (much lower in amplitude) are also the result of
the diurnal deformation of the earth’s solid crust, ocean loading, and other
smaller contributions.
Although tidal gravity effects are calculated occasionally for gravity
survey work (Longman, 1959), this is not common practice. These effects
can be as much as 0.3 mGal over a six-hour period but are generally less,
and they fluctuate with an aberrant quasi-sinusoidal behavior as a function
of time, whose period is on the order of 12 hours (Figure 1). The tidal grav-
ity contribution to the measurement usually is removed effectively by the
same process that handles instrument drift — repeated observations at base
stations at reasonably close time intervals.
Instrument manufacturers calibrate every meter either over a known test
range or in the laboratory. This can result in a linear relationship to convert
the meter readings to milligals, or for meters with a worldwide range of
7000 mGal, it can result in a table of conversion factors. This conversion is
applied prior to obtaining the drift- and tidal-adjusted differences in gravity
between the base and field stations.
84  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 1. Typical earth tide


effects at midlatitudes. 0.4

0.3

0.2

mGal
0.1

0.0

–0.1

18 2 10 18 2 10
December 11 December 12
Time (h)

Latitude corrections
In Chapter 2, we studied the earth’s gravity field and its variations that
depend on shape and rotation. The expressions given in the section titled
“The standard International Gravity Formula” in Chapter 2 yield the value
of the earth’s gravity field at any point on the surface of the earth. This “nor-
mal” gravity field increases by about 5.3 Gal from equator to pole, and if not
corrected for, it will yield a north-south gradient in measured gravity data.
The latitude correction subtracts the normal field from measured gravity to
eliminate this gradient effect. In general, any of the expressions given in
“The standard International Gravity Formula” in Chapter 2 can be used, but
it is customary to use the expression

g = 978.031846 (1 + 0.00523024 sin 2 ϕ − 0.0000058sin 2 2 ϕ ). (2)

When working in a small project area, the above expression can be sim-
plified by differentiating it with respect to an element of arc, R dj, situated
on the surface of the earth (R being the radius of the earth) to yield a good
approximation of the normal gravity gradient,

g = 0.812 ⋅ sin 2φ mGal / km,

where j is the station latitude. To achieve an accuracy of 0.01 mGal, we


need to know the north-south location of our gravity stations to about
12.5 m, which is not difficult to obtain with modern Global Positioning
System (GPS) instruments (Geodetic Reference System, 1967).
Chapter 6:  Data Reduction  85
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Free-air correction
Let us examine the observed gravity across a topographic cliff with no
density contrasts in the subsurface, as shown in Figure 2. Because the eleva-
tion does not change from negative locations to the cliff’s edge, observed
gravity is almost zero on the left side of the graph. Observed gravity begins
to decrease near the cliff because of the upward attraction of the cliff (both
this decrease and the increase shown for observations at the top of the cliff
are discussed below).
As elevation increases (to the right of the cliff, as shown in Figure 2),
gravity measurements decrease abruptly because of the increased distance
from the center of the earth. This requires a correction known as the free-air
correction (FAC).
Although an arbitrary datum is sometimes used in gravity data reduc-
tion, we assume here the more common approach in which mean sea level
is accepted as the datum. We can calculate that theoretical gravity g(p) at a

Distance (m) Figure 2. Observed


0 gravity across a top-
–3000 –2000 –1000 0 1000 2000 3000
ographic cliff with
no density contrasts
–10
in the subsurface.
Vertical topography
Observed gravity
is exaggerated.
–20
anomaly (mGal)

–30

–40

–50

–60

Vertical cliff
x (distance) Relief = 200 m
86  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

point, p, which is h units above the datum by Newton’s law,

M M h h 2 ..... 
g(p) = k = k  1 − 2 + 3 − ,
( R + h )2 R2  R R2 
 h h2 
g(p) = g ( 0 )  1 − 2 + 3 2 − .....  , 
 R R 
h
g ( 0 ) − g ( p ) ≈ 2 g(0) , (3)
R
where g(0) is the gravity field on the datum and R is the earth radius.
The difference per unit change in elevation varies with latitude by about
0.02 mGal/m between the equator and the poles. Because both the earth’s
radius R and its theoretical gravity g(0) vary with latitude, we select a mean
for each, g(m) and R(m) in equation 3, and we obtain
h 
g ( 0 ) − g ( p ) ≈ 2 g(m) ,
R(m)
and the free-air correction term becomes

FAC = + 0.3086 mGal/m – (second-order term in equation 3,


(4)
      usually ignored).

With j as station latitude, the second-order term is usually written as

(0.00023 ⋅ cos2ϕ − 2 ⋅ 10 −7 ) mGal/m,

which is small at low elevations but can be large at high elevations. For exam-
ple, at 5000 m, the second-order term is 1.7 mGal. Although this term is not
usually incorporated into the free-air reduction, it is not difficult to do so.
The free-air correction can be very large. For example, at a 1000-m
­elevation, the correction is 308.6 mGal. If we require a precision of 0.01
mGal, then relative station elevations need to be known to about 3 cm,
which contributes substantially to the cost of a gravity station.
For completeness, we should also include in the free-air corrections the
atmospheric correction that accounts for the gravitational attraction of the
atmospheric masses above the gravity meter. This correction is necessary
because the value of normal gravity includes a component resulting from
the earth’s atmosphere, and without this correction, the gravity anomalies
will be underestimated.
Let us be clear that by making this reduction, we are not reducing the
data to a datum, i.e., obtaining at a fictitious station on the datum what we
Chapter 6:  Data Reduction  87
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

would have measured there if we had been able to do so. Instead, we are
simply accounting for the decrease in the measured value caused solely by
the station being farther from the earth’s center. In addition, we should be
clear that we are not accounting for local variations in the vertical gradi-
ent of gravity, which might be appreciable in the presence of large local
anomalous masses.

Bouguer correction
As the elevation increases, gravity measurements increase because of
the increased rock mass between the station and the survey datum; this
requires a correction known as the Bouguer correction. In Figure 2, at nega-
tive distance values to the left of the cliff, where the zero elevation is taken
to be the datum, both the free-air and the Bouguer effects are zero. How-
ever, at the higher elevation (positive horizontal distances), the Bouguer
effects cause the observed gravity to be greater than it would be if this effect
were not taken into account.
We have seen in Chapter 3 that an infinitely wide uniform section of
earth that has a thickness t, whose constant density is r, exerts a vertical grav­
itational attraction equal to

2p k ρ t,(5)

which is known as the simple Bouguer plate value. It is usually written as

0.04196 × ρ mGal/m.

This value is calculated for each station, and it is subtracted from the
measured value because the rocks between the station and the datum create
an increase in the measurement. The thickness t in equation 5 is the distance
between the station and the datum, usually taken to be the station elevation
where the datum is mean sea level. Often, in relatively flat terrain, this cor-
rection is the only accounting in data reduction for the rocks between the
station and the datum. In that case, the term simple Bouguer is applied to the
reduction process and to the resulting anomaly (see below).
If undulations in the earth’s surface are substantial, then an adjust-
ment called the terrain correction is also applied. High topography above
the station represents masses whose upward attraction at the station is not
included in the simple Bouguer term and causes a decrease in the measure-
ment; hence, its effect must be added to the measured data. Low topography
below the station represents mass deficiencies for which the simple Bouguer
88  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 3. A terrain-correction
Terrain compartment
compartment.

Station

term overcalculated; hence, the effect of such masses must also be added to
the measured value. Topographic elevations are usually estimated in com-
partments (Figure 3) to yield the topographic relief that exists between the
station and the average elevation within the compartment.
Where the terrain is severe, a “slope,” or “wedge,” model is used in the
compartment in some cases, rather than a simple prism. For many decades,
terrain corrections were made manually by estimating topographic eleva-
tions in compartments within concentric rings with increasing radii from
each station. The popular Hammer (1939) charts include terrain cover-
age to only 21 km from the station, whereas the comprehensive Hayford-
Bowie system (Swick, 1942) extends to 166.7 km. In most cases in modern
exploration, a digital elevation model can be acquired and used as a basis
for comprehensive corrections that use a variety of computer software
systems.
Most of these programs represent the earth as a collection of vertical
prisms with a flat top and with increasing dimensions farther from the sta-
tion (Plouff, 1977). Terrain corrections are obtained by summing up the
gravitational attraction of these prisms at the location of each station. Often,
the DEM provides for inadequate definition of the topography for the inner
zones near the station. If the terrain is sufficiently severe2, it is necessary
to supplement the DEM with field estimates of the topographic relief near
the station. In some modern surveys, in areas where extreme topography
is not defined by available digital terrain models, additional topographic
information can be obtained by surveying the key topographic landmarks
near the gravity station. These data are then used as supplementary digital
elevations.

2 Topographic irregularity is evaluated on a case-by-case basis, depending on the


terrain and on the desired survey accuracy. It is always good field practice to locate
stations away from abrupt changes in elevation if possible.
Chapter 6:  Data Reduction  89
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

We can estimate the severity of this potential problem by examining


the correction values tabulated for the inner zones of the Hayford-Bowie
system (Swick, 1942), the results for which are shown in Figure 4. Mul-
tiple errors for a given compartment-elevation error result from the variable
elevation relief (ranging, for these data, between 3 and 60 m) within each
zone. This chart is not intended as a means to determine actual errors but
rather is to help the survey planner determine the extent to which supple-
mentary elevations might be required. Larger errors can occur in very rug-
ged topography. For example, one of the four compartments in zone C, if
its elevation is in error by 62 m, can contribute an error of 4.7 mGal if the
total relief is 800 m.
Two exceptions to the rule to “always add the terrain correction” are:
(1) distant high topography that is actually below the station because of

Terrain-correction error by Hayford-Bowie zone

(Number of compartments/outer radius)


Zone B (4/68 m) Zone C (4/230 m) Zone D (6/590 m)
3

Terrain-
correction
error
(mGal)

0
0 6 12 18
Terrain-compartment elevation error (m)

Figure 4. Terrain-correction error resulting from an error in determination of the


compartment elevation for three Hayford-Bowie inner zones. The actual total
error for the station would be the accumulated sum for all compartments in error.
Topographic rock density in the Swick (1942) table is 2670 kg/m3.
90  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

earth curvature and (2) marine terrain corrections, which we will consider
below in the section on nonland reductions.
Earth curvature is taken into account in two ways: (1) in the direct
calculation of terrain effects, the computer algorithm should account (as
do the Hayford-Bowie tables, starting in zone J at about 9 km) for the
actual position of the topographic masses and deficiencies relative to the
station and (2) for the fact that the simple Bouguer plate does not curve
with the earth (this is the Bullard B correction treated in Appendix A of
this chapter).
In flat terrain, a constant error in Bouguer density will result in a con-
stant shift in the anomaly resulting from data reduction. However, it will not
affect the relative shape or gradient of the anomaly which, as we will see in
Chapter 7, are critical in interpretation. Variations in the density of the near-
surface rocks in flat terrain will produce anomalies that are not removed
in the Bouguer reduction process, but these can be incorporated into the
interpretation. If the topography is not flat, any error in Bouguer density will
result in unwanted artifacts of the data-reduction process, i.e., anomalies
that correlate with the terrain. Such correlations might help the interpreter,
as we have seen in Chapter 5, on rock densities.
As a general rule, one should always overlay a topographic map over a
Bouguer gravity-anomaly map and notice any correlations between them.
If the Bouguer density was chosen properly, there should be minimal corre-
lation between the two. In principle, one can use a variable Bouguer density
to overcome this problem (Vajk, 1956), but this is difficult to accomplish.
The simplest approach is to calculate separate Bouguer anomalies with a
few chosen densities and to use, in the various regions of the survey area,
the Bouguer map that correlates least with topography.

Gravity anomalies
Gravity measurements are very sensitive to changes in elevation, as can
be seen in Figure 2. The usual goal in exploration gravity work is to improve
our understanding of the subsurface. Thus, in the data-reduction process,
we would like to remove any effects that are not related to subsurface geo-
logic distributions. We would expect, therefore, because no subsurface den-
sity contrasts are present in Figure 2, that no anomaly would be present in
the results.
A gravity anomaly is defined as the difference between measured grav-
ity (i.e., station gravity after adjustment for time variations and network
ties) and theoretical gravity based on a defined earth model. The free-air
Chapter 6:  Data Reduction  91
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

anomaly is defined as

gfa = gs − λ + FAC.

The first term on the right side of the equation is station gravity, l is theo-
retical gravity, and FAC is the free-air correction, all defined in the previous
sections. Generally, this anomaly shows strong correlation with topography
even though the correction term, FAC, removes the direct effect of elevation
in terms of distance from the center of the earth.
For sea-surface surveys, the free-air correction is equal to zero (neglect-
ing the effect of tides), but the free-air anomaly nonetheless shows correla-
tion with bathymetry as a reflection of changes in the thickness of the water
column. We might think of the free-air anomaly as having been caused by
all the density contrasts within the earth, including the topographic rocks,
but not by the direct effect of changes in station elevation. The Bouguer
anomaly is defined as

gb = gs − λ + FAC − bc,

where bc is the Bouguer correction. If only the infinite plate (known as Bul-
lard A) is used in the reduction, this anomaly is called the simple Bouguer
anomaly. If terrain corrections (known as Bullard C) are added, it is called
the complete Bouguer anomaly.
Throughout most of the twentieth century, economics limited the ex-
tent to which terrain corrections were carried out. In the current era, with
­inexpensive computers and terrain models, every survey should be reduced
by using the complete Hayford-Bowie template, i.e., out to 167 km, and
curvature (Bullard B; see Appendix A of this chapter) should be applied
routinely. The Bouguer anomaly is usually the end product in data reduc-
tion, and it provides the starting point for most gravity interpretations.
The observed gravity depicted in Figure 2 is subjected to the correc-
tions defined here, and the resulting anomalies are shown in Figure 5. Both
the free-air and the simple Bouguer anomalies are near zero on the left
side, as we would expect for a region that has no subsurface density con-
trasts. However, at the cliff’s edge, the free-air anomaly takes a large step
up because of the abrupt change in elevation, and to the right, it increases
asymptotically to include the effect of the rocks between the stations and
the datum.
The simple Bouguer anomaly in this example, by contrast, is never
positive because (with the exceptions discussed above) the influence of ter-
rain features is negative. The simple Bouguer anomaly approaches zero to
92  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the right side because of the absence of subsurface density contrasts and
the increasing distance from the cliff edge. The flat zero-anomaly curve
is the complete Bouguer anomaly, which includes the terrain correction.
The effect of terrain reaches a maximum at the cliff’s edge and is indeed
very large, indicating the need for locating stations away from the abrupt
change in elevation and/or implementing additional surveying to define top-
ographic features with a precision consistent with the goals of the project.
The free-air anomaly does not overlay the simple Bouguer anomaly near
the base of the cliff because of the effect of interpolating between values at
stations (in this example) separated by 100 m.

Anomaly FAA SBA CBA


(mGal)
25

20

15

10

–5

–10
Ground-surface elevation = 200 m
Ground-surface elevation = 0

–2000 –1000 0 1000 2000


Distance from cliff (m)
Figure 5. Gravity anomalies across a topographic cliff with no density contrasts in
the subsurface: FAA = free-air anomaly, SBA = simple Bouguer anomaly, CBA =
complete Bouguer anomaly; all are defined in this chapter. Vertical separation in
ground surface is exaggerated. Topographic rock density is 2500 kg/m3.
Chapter 6:  Data Reduction  93
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Isostatic correction
On a global scale, Bouguer anomalies have a very strong inverse correla-
tion with station elevation: High/mountain stations yield Bouguer anomalies
superimposed over a long-wavelength negative background, whereas ocean
deeps produce Bouguer anomalies superimposed over a long-wavelength
positive background. From a geophysicist’s point of view3, this discovery
was made in 1749 by Pierre Bouguer during a French geodetic expedition to
measure the meridian arc. He found that at the base of the Andes, a plumb
bob was not deflected to the extent that calculations required. The idea fol-
lowed that at relatively shallow depths, isostatic equilibrium is attained such
that mountains are balanced by underlying mass deficiencies and ocean
depths by mass excesses.
Two major theories ensued, with many subsequent modifications to
each: (1) The Airy (1855) theory that mountains have roots and (2) the Pratt
(1855, 1859) theory that crustal densities vary horizontally. In both cases, a
depth of compensation occurs such that all columns from the earth’s surface
down to that depth will exhibit an equal amount of mass. In Figure 6, the
Airy theory is depicted by a variable depth to the base of the root, whereas
the Pratt system is depicted by a horizontally varying density function over-
lying a nearly planar depth of compensation.
Isostatic corrections were established by the geodetic community based
on modifications of these two basic models. Several models have been sug-
gested, the two most popular of which are the Airy-Heiskanen model and the
Pratt-Hayford model. The exact model used in the isostatic correction depends
on the relative amounts of compensation attributed to regional versus local
geology.
The resulting isostatic residual anomaly is generally much lower in
amplitude than the Bouguer anomalies. Isostatic anomalies have been
applied in the exploration industry occasionally, but not on a large scale.
Although it is often helpful in interpretive work to be aware of isostatic
effects, which can play an important role in identification of regional fields,
the corrections, which are based on uncertain models, are usually not applied
in exploration work over limited areas.
With a few exceptions (such as the Klamath Mountains in northern Cali-
fornia), the fact of isostatic equilibrium is well established. We can test this

3 Famous geodesist W. A. Heiskanen (Heiskanen and Vening Meinesz, 1958) sug-


gests that Leonardo da Vinci, among his many other accomplishments, suggested
that visible masses of the earth’s surface are in equilibrium. This is based on the
work of Delaney (1940).
94  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Densities in kilograms per cubic meter

2800 3000 2600 3000


2800 2900

3300 3300

Airy Pratt
Ocean
ρt

ρc ρ1
ρ2 ρ3 ρ4

ρm

Constant density crust Variable density crust

Figure 6. Examples of isostatic compensation.

theory for broad regions without depending on any basic model for compen-
sation by observing that the average free-air anomaly is zero if the region is
in compensation. That is because (unlike the Bouguer anomaly, which is cor-
rected for topography) the free-air anomaly has both topographic and com-
pensating mass components. We see below that Gauss’ theorem ­(Chapter
3) requires that the average free-air anomaly be zero over a broad region.
As the depth of compensation increases, the gravity components associated
with the compensating masses have greater horizontal extensions, requiring
that the free-air anomaly be averaged over regions as broad as 200 km.
Relating the Bouguer anomaly to the free-air anomaly and the Bouguer
correction,
Chapter 6:  Data Reduction  95
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

gb = gfa – bc.

It follows with the averages:

gb = g fa − bc(7)

From Gauss’ theorem, assuming that the region captures half the flux, the
mass deficiency, Md, where gb is a negative number, is given by
1 n
Md = ∑ gb,iδ Si .
2p k i=1
(8)

The mass excess of the topographic rocks MT is


n
M T = ∑ δ hi ∆Si . (9)
i =1

Isostatic compensation requires that equation 8 is equal to equation 9, result-
ing in
2p k n 1 n

n i=1
δ hi = − ∑ gb ,i . 
n i=1
(10)

The left side of the equation is the definition of the average Bouguer
­correction, whereas the right side is the average Bouguer anomaly. Sub-
stituting equation 10 into equation 7 shows that the average free-air
anomaly approaches zero when taken over a broad region. Studies have
shown that over large areas, the free-air anomaly does have an average
of near zero.

Eötvös corrections
For land and underwater measurements, the meter is leveled and at rest.
In all moving vehicles, a phenomenon known as the Eötvös effect is created
because the motion of the instrument modifies the effect of the earth’s rota-
tion, which is already treated as if the meter were at rest. This can be studied
in Figure 7, depicting an earth that has rotation w.
The maximum outward acceleration caused by the earth’s rotation
occurs at the equator and is equal to Rw 2. At latitude j, the component of
outward acceleration normal to the axis of rotation is Rw 2 cos2j. We mea-
sure gravity in the direction of the plumb line (essentially toward the earth’s
center, which further reduces the rotational effect by cosj), resulting in the
96  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

ω
Rω 2cos2Φ
R cos Φ
Rω 2cosΦ

Φ
Rω 2
R

Figure 7. Rotating earth on which a moving instrument is located at latitude j.

effect Rw 2 cos2j. Now we want to know the change in acceleration resulting
from the earth’s rotation,  w :

da = 2Rw cos2 j dw.

The change in w in terms of the east component of motion of the meter is

dw = V/R cos j.

The Eötvös effect is

E = 2V w cos j sin a,

where V is the vehicle’s velocity and a is its direction with respect to north.
To this, we add the actual outward centrifugal acceleration acting on the
meter, V2/R, which is, of course, independent of direction but is nearly con-
stant in a given survey.
The Eötvös correction is quite large. For a ship traveling easterly at
1 knot (kn)4 at 45° north latitude, the correction is 5.4 mGal, whereas for a
ship traveling at 10 kn at the equator, the correction is 75 mGal.

41 knot (kn)  = 1 nautical mile/hour  = 1.852 km/hour.


Chapter 6:  Data Reduction  97
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Marine reductions
Offshore gravity surveys require special attention so that corrections
unique to the marine environment account for the water layer and properly
tie with their land counterparts in the transition zone. Special corrections
might also be required for water-bottom surveys in deep lakes onshore or
in cases where the meter is on a tripod. In these cases, the datum is likely
to be below the water bottom, and the treatment of elevation must take this
fact into account. Figure 8 illustrates the onshore/offshore environment. The
onshore lake and tripod conditions are not shown but can be constructed by
placing the datum below h2.
The elevation of the land station is depicted at h1, the underwater station at
h2, and the surface-ship station at h3. The free-air and Bouguer corrections for
the land case are as described earlier in this chapter. However, note that for the
terrain-correction compartments that contain water, the appropriate density is
not that of the rocks but is the difference in density between the rock and water.
In some nearshore cases5, sea level fluctuates significantly, as depicted
by tidal displacement T. For the underwater location, h2 is negative, so that
the free-air and Bouguer corrections are reversed in sign with respect to land
locations (i.e., gravity stations are located below the datum). In addition,
a new term is introduced for the upward attraction of the overlying water
layer: 0.043 (h2 + T) mGal for seawater density. As in the case of the land
surveys, terrain corrections should take into account the proper density con-
trast of the rocks — water/rock contrast or air/rock contrast.
Surface-ship surveys have the distinct advantage of being located on
the datum. The free-air correction is zero, although this neglects the tidal
effects. Unfortunately, in the open ocean, accurate tidal behavior is usually
unknown, and the measurement errors that this creates must be treated in the
network adjustments discussed below.
It has been argued, largely by those in academic surroundings, that for
these offshore surveys, the free-air anomaly map is the preferable one for the
starting point in interpretation. In that case, we would treat the water layer
(which is generally quite well known) as part of the interpretation of the
subsurface. In the first major offshore survey jointly operated by more than
20 oil companies in 1965, industry voted (with minor argument) that the pri-
mary map would be of the Bouguer anomaly, and that has been the anomaly
of choice by industry ever since.

5In the Bay of Fundy and Cook Inlet, for example, some tides exceed 40 ft (12 m).
98  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Land
Gravity meter

h1
Surface ship
Actual sea level
T
Mean sea level

h3
h2
Underwater

Figure 8. Locations of onshore and offshore gravity stations.

In the surface-ship case, the Bouguer correction is

2pkh3(ρ − 1.03) = 0.0419 mGal (ρ − 1.03) h3,


where h3 is the water depth recorded by the fathometer. This correction
is added to simulate a land survey where water is converted to rock. The
same issue regarding appropriate density as in the case of land surveys also
occurs here. We recall that all but the distant zones in land terrain correc-
tions require a positive addition to the simple Bouguer plate.
That is not the case in surface-ship surveys precisely because the off-
shore stations are acquired on a relatively flat surface and do not view the
bathymetry as a land station views the topography — on the undulating
observational surface that contains the anomalous topographic masses. In
addition, marine terrain corrections might be larger for comparable relief
in the rocks because the solid angle subtended by rocks at the ship location
can be appreciable. Offshore southern California, for example, exhibits ter-
rain corrections ranging from −25 to +25 mGal.

Appendix A
Bullard correction
The Bullard B correction is an adjustment for the fact that the simple
Bouguer plate contains mass laterally beyond the earth and does not contain
existing mass where the earth’s surface dips below the plate, as shown in
Figure A-1.
This correction for curvature (Bullard B correction) modifies the sim-
ple Bouguer plate value (Bullard A) to that of a cap that has a ­surface
radius of nearly 167 km and a thickness the same as that of the infinite plate
(station elevation using a sea-level datum). This is equivalent to removing
Chapter 6:  Data Reduction  99
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Spherical cap:
166.735 km

Surface radius coincident with outer radius


of Hayford-Bowie terrain zones A through O

Simple Bouguer infinite plate


∞ ∞

Figure A-1. Geometry of spherical cap in relation to infinite Bouguer plate. After
LaFehr (1991), Figure 1.

all the plate above the earth’s surface and beyond 167 km, whether above
or below the earth’s surface (i.e., all of the slant-shaded zone in Figure
A-1) and adding the part of the cap below the plate (i.e., the solid black
zone). That part of the cap shown in stipple pattern is common to both the
cap and the plate and therefore does not enter into the curvature correction.
The sum of the stipple and black zones constitutes the entire spherical cap.
All dimensions are greatly exaggerated to clearly show the nature of the
correction. Following the methods described in Chapter 2, we can derive the
curvature correction B (for Bullard B),

B = 2p k ρ ( µ h − λ R ), (A-1)

where R is the earth’s radius to the station (Ro + h), and m and l are dimen-
sionless coefficients defined below. Equation A-1 can be used to calculate
the effect of curvature in the Bouguer correction.
The two dimensionless coefficients are μ and l:

 μ = 1/3h2  – h,
 l = 1/3{(d  +  fd  +  d   2)[(f − d  )2 + k]1/2 + p + m loge n/(f − d + [(f −d  )2 + k]1/2)},

where d = R0 /R, h = h/R, d = 3 cos2 a − 2, k = sin2 a, p = −6 cos2 a sin (a /2) +


4 sin3 (a /2), m = −3 sin2 a cos a, and n = 2[sin (a /2) – sin2 (a /2)]. R0 is the
normal earth’s radius to sea level, R is the earth’s radius to the station, and
h is the elevation of the station.
The angle alpha is the half angle subtended at the earth’s center by the
section of the earth’s surface at sea level for which the outer distance from
100  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the station is normally taken to be 166.7 km (or the outer radius of the Hay-
ford-Bowie zone O).
For more details about the exact solution, see LaFehr (1991). A simple
and generally quite adequate approximation to the Bullard B (BB) curva-
ture correction can be expressed as BB = Ah + Bh + Ch, where A × 10–3 =
1.46308, B × 10­7 = 3.52725, and C × 10–14 = 5.1.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 7

Anomaly Interpretation Guidelines


and Limitations

Purposes of gravity surveys


We have studied the mathematical basis for the generation of gravity
anomalies (Chapter 3), gravity instrumentation that enables gravity surveys
and generally available surveying methods for obtaining them (Chapter 4),
density variations and methods for determining rock density (Chapter 5),
and the reduction of gravity data in static and dynamic settings (Chapter 6),
which is intended to eliminate often very substantial measured effects that
are unrelated to the gravitational sources we wish to analyze. Both relative-
and absolute-gravity measurements are available in gravity exploration.
Six generalized purposes of gravity surveys can incorporate one or both
methods of measurement:

1) determination of the earth’s shape


2) determination of missile trajectories, a military application now seldom
used
3) tidal and earth elasticity studies
4) other time-dependent applications (such as the monitoring of reservoirs)
5) determination of physical constants
6) determination of the subsurface geology or other characteristics of the
earth’s structure

The last of these, the study of the subsurface, requires identification


of the anomalies associated with the geologic sources of interest (anomaly
separation) and an explanation of those anomalies in terms of the geology

101
102  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

that is the purpose of the investigation. We will now turn to those activi-
ties. In this chapter, we examine the guidelines and limitations of anomaly
interpretation. In Chapter 8, we examine inversion, a special case of inter-
pretation. In Chapter 9, we illustrate case histories to demonstrate practical
results of interpretation.
As we have seen in Chapter 3, three features of the causative bodies
must be present to produce observable anomalies: (1) sufficient density con-
trast with respect to the surrounding rocks, (2) sufficient geometric distribu-
tion (volume), and (3) sufficient proximity to the sensor. Each of these is
important in the evaluation of observed anomalies and in the simulation of
effects arising from proposed models of the subsurface geology.
Any quantitative interpretation of gravity anomalies in terms of sub-
surface mass distributions assumes some plausible geologic structure with
constant or variable density contrast. The parameters of the structure are
adjusted until its calculated anomaly agrees acceptably well with the ob­­
served anomaly. Inversion (see Chapter 8) can help to create an interpreta-
tion by using more complicated structures with variable density. In each
case, one needs first to calculate, at any observation point, the anomaly
caused by an arbitrary structure with any density distribution, a topic to
which we turn now.

Gravity calculations for an arbitrary model


Modern gravity exploration began in the 1930s, and for the first sev-
eral decades, it relied on a combination of characteristic curves for many
simple models and “dot charts” for manually adding the accumulative grav-
ity effects of irregular bodies. The charts typically were based on using
the method of solid angles defined in Chapter 3. For complicated geologic
structures, manual application could be very time-consuming. In the 1950s,
computers were beginning to change the methodology by direct calculation
of the gravity effects of polygonal and other simple bodies.
Then in the 1960s, grid-based systems were introduced for the forward
and inverse modes of computation. Computers were relatively slow then,
and computer memory was limited and expensive. It was important to take
these limitations into account when using early methods. Although com-
puter memory and speed are of less concern in the early twenty-first century,
they are not irrelevant as larger and more ambitious projects are undertaken.
A volume of mass can be approximated by a collection of rectangular
prisms, as shown in Figure 1. For points outside the source region, the ver-
tical component of the gravity contribution from each prism can be calcu-
lated using expression 19 of Chapter 3, and the total anomaly is obtained
Chapter 7:  Anomaly Interpretation  103
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

P (x, y, z)
x

Figure 1. Calculation of the gravity anomaly of a given target by superimposing


the response from rectangular prisms. The smaller and more numerous the
prisms, the more closely the sum of their effects will match the actual effect of the
causative body.

by adding (superimposing) the attraction from each individual prism. How-


ever, expression 19 of Chapter 3 requires the calculation of many logarith-
mic, arctangent, and square-root terms, which makes it cumbersome if not
impractical for day-to-day applications of gravity anomalies caused by sin-
gle bodies. Some papers in the geophysical literature have helped to address
this problem (e.g., Holstein, 2003; Nabighian et al., 2005).
A simpler approach is implemented by Talwani and Ewing (1960) by
assuming that the target can be approximated by a stack of infinitely thin
laminae (Figure 2). In this approach, one first obtains the response of each
lamina by integrating over its surface and then sums up (integrates) the
response from each lamina in the vertical direction. To simplify calculations,
each lamina is approximated by a polygonal shape, and the surface integral
over the lamina is reduced to an integral along the perimeter of the polygon.
Plouff (1975) expands the above technique by using laminae of finite
thickness with vertical sides and whose top and bottom surfaces also were
represented by a polygonal shape. This approach was used widely in calcu-
lating terrain effects for gravity data.
Barnett (1976) develops an analytical method for calculating the gravi-
tational attraction of a homogenous polyhedron-shaped 3D body. In this
method, the body is represented as being composed of triangular facets,
104  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) 3D view
P
x

y
dz

b) Plan view
x
rn + 1
n+1

n
rn

Figure 2. Approximation of attraction of (a) a body by a stack of laminae, each of


which is then approximated to have (b) a polygonal shape.

and the gravitational attraction of the body is obtained by integrating over


each facet and then by summing up the individual results. Because the inte-
gral over each facet can be calculated analytically and we need fewer facets
than laminae, this approach is much more cost-effective in calculating the
response from single bodies.
Okabe (1979) develops a similar approach for calculating gravity
anomalies. Pedersen (1978) develops frequency-domain expressions for
potential fields from arbitrary 2D, 2.5D, and 3D bodies, and those expres-
sions are simplified by Hansen and Wang (1988) for arbitrary 3D bodies. In
Hansen and Wang’s formulation, the gravity field of a polyhedron can be
expressed in the frequency domain as a summation over the N vertices of
the polyhedron. In the formulation of Hansen and Wang (1988), the contri-
bution of each vertex to the model can be computed separately without first
Chapter 7:  Anomaly Interpretation  105
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

decomposing the body into facets. The geometry of the body is described as
a collection of vertex coordinates and set of points which group the vertices
joined by edges.

The fast-Fourier transform for calculating gravity effects


With dm = r dV and with volume dimensions of dx, dh, and dz, the
general form for the vertical component of attraction in the 3D case is given
by equation 15 of Chapter 3:
d  1 ρ (ξ , η, ζ ) (ζ − z ) dξ dη dζ
gz ( x , y, z ) = k ∫ ρ   dV = k ∫ ,
dz  r  3

(ξ − x ) + (η − y) + (ζ − z ) 
V V 2 2 2 2

where the integration is carried out over the volume of the causative body
and r 2 = (ξ − x )2 + (η − y)2 + (ζ − z )2.
Expression 15 of Chapter 3 can be written in general as

gz ( x , y, z ) = ∫ ρ (ξ , η, ζ ) G
V
z ( x − ξ , y − η , z − ζ ) dξ dη dζ , (1)

where

d  1 ζ−z
Gz ( x − ξ , y − η , z − ζ ) = k   =k (2)
dz  r 
3
(ξ − x ) + (η − y) + (ζ − z ) 
2 2 2 2

is known as the Green’s function and represents the vertical gravitational


attraction at the observation point (x, y, z) resulting from a point mass of unit
density located at x, h, z. Similar Green’s functions also can be defined for
the gx and gy components of gravitational attraction.
Generally, the density function is nonzero only over a confined area of
the half-space, i.e., that represented by volume V. Because the density func-
tion is zero outside volume V, we can extend the limits of integration over
the entire half-space without any loss of generality and can write the above
expression as
∞ ∞ ∞

gz ( x , y, z ) = ∫ ∫ ∫ ρ (ξ , η, ζ ) G
−∞ −∞ z
z ( x − ξ , y − η , z − ζ ) dξ dη dζ . (3)

Expression 3 represents a convolution integral between the density func-


tion r and the Green’s function Gz. From Appendix A, “Fourier Trans-
form,” at the end of this book, a convolution in the time domain leads to
a multiplication in the frequency domain. Taking the Fourier transform of
106  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

equation 3, one obtains


F( gz ) = ∫ F[ ρ ( x , y, ζ )] F [Gz ( x , y , ζ − z )] dζ , (4)
z

where the Fourier integration is carried out over the x-, y-coordinates. To pro-
ceed further, we need the Fourier transform of the Green’s function. We have
∞ ∞
e − i ( u x + v y ) dx dy
F[ Gz ( x , y , ζ − z ] = k (ζ − z ) ∫ ∫ 3 =
−∞ −∞
[ x + y + (ζ − z) ]
2 2 2 2

∞∞
cos ux cos vy dx dy
= 4 k (ζ − z) ∫ ∫ 3.
0 0 [ x 2 + y 2 + (ζ − z ) 2 ] 2

The integration over x can be carried out using Gradshteyn and Ryzhik
(1980, p. 249, formula 3.773.6) to yield


K1 [u y 2 + (ζ − z )2 ]
F[ Gz ( x , y , ζ − z ] = 4 k (ζ − z ) u ∫ cos vy dy,
o y 2 + (ζ − z )2

where K1 (x) is the modified Bessel function of order 1.


We can now carry the integration over y using Gradshteyn and Ryzhik
(1980, p. 756, formula 6.726.4) to yield
u2 + v 2
F[ Gz ( x , y, ζ − z ] = 2 π k e − (ζ − z ) . (5)

With equation 5, expression 4 becomes


∫ F[ρ(ζ )]e
u2 + v 2 −ζ u 2 + v 2
F( gz ) = 2π ke z dζ . (6)
z

Expression 6 shows that for the most general case, we divide the body into
horizontal slices, take the Fourier transform of the density function for that
slice and, after weighting it by an exponential function dependent on the
depth of the slice, we sum up the results. To obtain the vertical component of
gravitational attraction, we then have to inverse-transform the above result.
Expression 6 yields a closed-form solution only for relatively few sim-
ple bodies, e.g., spheres (monopole), cylinders (horizontal line), vertical
lines and ribbons, and so forth (Blakely, 1995). The numerical calculations
required to carry out the integrations in expression 6 for the general case
are relatively cumbersome. At present, the preferred method is the direct
Chapter 7:  Anomaly Interpretation  107
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

evaluation of expression 15 of Chapter 3 using the methods developed in the


section above titled “Gravity calculations for an arbitrary model.” In some
special cases, e.g., reservoir simulation, the Fourier method does have some
advantages over the direct evaluation of expression 15 of Chapter 3 and is
used preferentially.
In one case, however, the Fourier-transform approach is definitely pref-
erable to the direct evaluation of expression 15 of Chapter 3. Parker (1973)
develops a technique for calculating the gravitational attraction of a source
layer of constant density with uneven top and bottom topography. Such a
model is useful in calculating isostatic residual gravity anomalies, in esti-
mating the effect of bathymetry or sedimentary basins, and so forth.
Following Parker (1973), we first develop the theory for a layer with a
flat bottom located at z = 0 and an arbitrary topography given by h(x, y, z) > 0
(Figure 3). For convergence purposes and because we can model only a
finite area of terrain, we will assume that the layer vanishes outside some
finite domain.
Letting a position in space be represented by the vector r = (x, y, z)
and with the z-axis positive upward, we can write the gravitational potential
resulting from this layer as

∞ ∞ h(r)
dV dξ dη dζ
U ( x , y, z ) = k ρ ∫ = kρ ∫ ∫ ∫ . (7)
V
r −∞ −∞ 0 ( x − ξ ) + ( y − η)2 + ( z − ζ )2
2

Taking the Fourier transform with respect to x, y, we obtain, after changing


the order of integration,
∞ ∞ h(r) ∞ ∞
e − i ( ux + vy ) dx dy
F[U (r )] = k ρ ∫ ∫ d ξ dη ∫ dζ ∫ ∫ . (8)
−∞ −∞ 0 −∞ −∞ ( x − ξ ) + ( y − η) + ( z − ζ )
2 2 2

z P (x, y, z)

r
z = h (r)

Q( , , )

z=0

Figure 3. Gravitational potential resulting from a source layer with uneven top
and with flat bottom.
108  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

After making the change of variables s = x – x and t = y – h, we can


use Gradshteyn and Ryzhik (1980, p. 419, formula 3.754.2; p. 736, formula
6.677.5) to obtain for the last integral
∞ ∞ u2 + v 2
e − ( ux + vy ) dx dy e − i ( z −ζ )
∫∫ ( x − ξ )2 + ( y − η)2 + ( z − ζ )2
= 2π e − ( uξ + vη )
u2 + v 2
. (9)
−∞ −∞

Introducing equation 9 into equation 8, we obtain


∞ ∞ h(r) u2 + v 2

∫ ∫e ∫
− z u2 + v 2 − i ( uξ + vη )
F[U (r )] = 2π k ρe d ξ dη dζ . (10)
−∞ −∞ 0 u2 + v 2

The integral over z can be performed analytically to yield
∞ ∞ 2 2
eh(r ) u + v − 1
∫ ∫e
− z u2 + v 2 − i ( uξ + vη ) (11)
F[U (r )] = 2π k ρe dξ dη.
u2 + v 2
−∞ −∞

Letting p2 = u2 + v2 and after expanding the last exponential function above


in a Taylor series, we obtain
∞ ∞ ∞
pn − 2 n
F[U (r )] = 2π k ρe − z p ∫ ∫e
− i ( uξ + vη )
∑ n!
h (r ) dξ dη. (12)
−∞ −∞ n =1

The above expression now contains the Fourier transform over various pow-
ers of h(r). We can then write

pn − 2
F[U (r )] = 2π k ρe − z p ∑ F [h n (r )]. (13)
n =1 n!

∂U
We have gz = and, after reverting to a z-axis directed downward in
∂z
the direction of gravitational attraction, we finally obtain the gravitational
attraction of a source layer with uneven top and flat bottom topography as

pn −1
F[ gz )] = 2π k ρe − z p ∑ F [h n (r )]. (14)
n =1 n!
It is easy to see that the above expression can be generalized imme-
diately for a constant-density source layer with uneven top and bottom to
obtain

pn −1
F[ gz )] = 2π k ρe − z p ∑ F [h n (r ) − d n (r )], (15)
n =1 n!
Chapter 7:  Anomaly Interpretation  109
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

where dn(r) represents the bottom topography. Parker (1973) shows that this
series converges fastest if the z = 0 plane is selected midway between the
minimum values of h(r) and d(r). If the density function is not constant
between the top and bottom surfaces, the above expression can be general-
ized as

pn −1
F[ gz )] = 2π ke − z p ∑ F [ρ (h n (r ) − d n (r ))]. (16)
n =1 n!
This formulation has proved to be extremely useful in calculating isostatic
corrections (see Chapter 6) and estimating the gravity anomaly of sedimen-
tary basins.
It is easy to see that for a Bouguer slab, i.e., h(r) = h = constant, expres-
sion 14, after taking the inverse Fourier transform, reduces to the Bouguer
slab formula gz = 2pkrh.

Anomaly shape
A comparison of anomaly shape for models that have different charac-
teristics can be examined in Figure 4. The bottom of the vertical 2D dike
is set arbitrarily at 50 units (to avoid the problem of an infinite maximum
amplitude). All horizontal distances are normalized to the depth of the center
of the sphere (which is also the depth to the center of the horizontal cylinder,
the top of the vertical bodies, and the termination point for the semi-infinite
horizontal slab). All maximum amplitudes are normalized to unity.
We note, as we also will observe in the section on depth estimation, that
the sphere anomaly decreases faster with distance whereas, as we have seen,
the infinite slab (not shown in Figure 4b) does not change at all.
Not surprisingly, the anomalies for 2D models produce less change over
the same horizontal distance than their 3D counterparts do. Of course, the
distance between our field measurements and the disturbing bodies plays a
major role in the appearance of the anomalies; everything looks like a point
source from far enough away. Both the geometry of the bodies and their
depth of burial are important in gravity interpretation.
As we now turn our attention to the geologic interpretation of anoma-
lies, we note that our models are almost always oversimplified versions of
the geology. Observed anomalies, however, do yield information about or
impose limitations on the nature and location of buried source rocks. The
interpreter’s role is twofold: (1) to identify that part of the observed anomaly
caused by the geologic feature of interest and (2) to accommodate the iden-
tified anomaly in terms of a reasonable geologic distribution. As we will see,
the shape of the anomaly can be diagnostic in this endeavor.
110  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Anomaly separation
Geologic modeling is the most effective anomaly-separation tech-
nique (Chapter 9), but other methods can be very useful. We might think
of the observed Bouguer1 anomaly map as the sum of the gravity effects of
all rock units in the project area and usually well beyond it for which a
horizontal Δ  r exists. Vertical changes in density do not cause relative (hori-
zontally changing) anomalies (Chapter 4). In simplest terms, the observed
anomaly,

O = R + r,

is equal to the sum of the regional (sometimes referred to as the “unwanted”


component) field R and the residual (sometimes referred to as the “wanted,”
or “target,” anomaly) field r.
It is obvious that without additional information such as a second equa-
tion identifying either the regional field, R, or the residual field, r, one can-
not be certain that the derived residual is in fact that which the geologic
target causes, no matter how sophisticated the attempt might be. Consider
the simple example shown in profile form in Figure 5.
The residual anomaly (shown in Figure 5b) is derived by subtracting the
assumed regional (shown as a dashed curve) from the observed anomaly. If
the regional is based on available geologic, seismic, magnetic, or well-log
data, the residual (in this example, a positive anomaly) might well be an
adequate approximation to that caused by the target geology.
However, we might have reason to place the regional field differently.
For example, we could place it above the observed anomaly, as shown in
Figure 6, thereby creating a double negative residual. It could be explained

  1Interpreters might wish to start their interpretations with the free-air anomaly
(see Chapter 6) and indeed might consider the Bouguer anomaly as “already an
interpretation.” Our view, however, is that the term observed Bouguer anomaly has
been standard terminology for several decades and that this anomaly is not gener-
ally viewed as an interpretation. We should be especially cautious in this regard in
the unusual cases in which a variable topographic density is used in data reduction,
and we should be mindful of the interpretational aspects of that approach. Should
we conclude that the topographic rock densities are a required element in the final
interpretation? Although they are often so accepted, that is by no means a require-
ment. In Chapter 6, we examine the relative meanings of the free-air and Bouguer
anomalies and can see why the latter is taken more often as a starting point in
interpretation, but we emphasize here that the former is also acceptable in that role.
Chapter 7:  Anomaly Interpretation  111
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)

1.0 Infinite slab (not shown below)


Amplitude (normalized to maximum)

0.8

Thick 2D dike
(bottom at great depth)
0.6
Vertical narrow pipe
(bottom at infinity)

t
ul
fa
0.4

ite
in
nf
i-i
m
Se
Infinite
0.2 horizontal cylinder
Sphere

0.0
–4 –3 –2 –1 0 1 2 3 4

b) 0
Distance (x)

1
Depth (z)

Figure 4. Vertical component of gravity anomalies for simple 2D and 3D models.

by two drainage channels, low in density, on the two flanks of what is shown
in Figure 5 as the positive residual — leading to an entirely different geo-
logic result.
For our present purpose, the scenario in Figure 6 can be ruled out
because we have ancillary information on the existence of a shallow salt
dome where the positive residual occurs. Even though external information
allows us to place the regional below the observed anomaly (in Figure 5),
we might still experience considerable ambiguity in the separation process.
Consider, for example, the case in which the regional caused by the
deeper part of the salt column (and influenced by surrounding salt struc-
tures) has in fact a much larger negative amplitude than the one assumed for
112  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 5. In this case, the resulting positive residual would be much larger
than that shown. Barring some constraint on the deeper salt column (which
might be supplied by drilling or deep seismic data), there is no a priori rea-
son to place the regional as close to the observed anomaly as shown, and

a)
Anomaly separation
by geologic constraint
or smoothing

O = R + r

Observed (O)

Regional (R)

b)
Residual (r)

Figure 5. Observed anomaly, assumed regional, and resulting residual.

Regional

Observed

Residual

Figure 6. Example of anomaly separation with regional above the observed


anomaly.
Chapter 7:  Anomaly Interpretation  113
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the larger the deep-salt negative, the larger the resulting positive residual.
Again, the problem is fundamentally ambiguous and can be solved reliably
only by applying geologic constraints, to which we will turn later in this
chapter and in the next chapters.
We can also characterize the observed gravity field in mathematical
terms. One once popular approach was to “fit,” in a least-squares sense, the
observed field directly to an nth-order polynomial surface by minimizing
the sum of the squared differences between the observed field and values
calculated from the polynomial. As we increase the order of the polyno-
mial, we can reduce the least-squares discrepancy between the mathemati-
cal definition and the observations2. Then we can “decompose” the field
into a regional and a residual by assigning only terms of the lowest degree
(perhaps as low as a simple linear gradient) to the regional field.
In an unpublished study by Tom LaFehr, this technique was applied to
approximately 50 salt domes in the Gulf of Mexico, with the added con-
straint that the regional polynomial was determined by not including the
local anomalous stations in the least-squares analysis, so that the resulting
regional would not be biased by the local sources which were the primary
reason for the study.
Interestingly, the resulting maximum amplitudes of the resulting resid-
ual anomalies by the polynomial method were invariably less than half the
maximum amplitude of those derived by the modeling technique (described
below). This illuminates the very problem with which we opened this dis-
cussion, i.e., a least-squares or other purely mathematical criterion might
not and generally does not quantitatively reflect the differences in gravita-
tional attraction amplitude between the target anomalies and those from all
other sources.
In the first several decades of the modern gravity-exploration era, grav-
ity maps were digitized routinely on a regular grid from which several sim-
plified operations on the observed field were carried out. One such approach
averaged the gravity-value points around the gravity station and considered
the averaged field to represent the regional field at that station. The regional
so determined (“grid residual”) was subtracted from the actual gravity mea-
sured at the station. The difference was considered to represent the residual
field at that station, i.e., the grid residual. Although this technique was quite
effective in outlining the position of many local geologic features, the final

  2Using orthogonal polynomials is another approach to solving this problem. One


advantage would be that if you change the order of the polynomial, you do not have
to recalculate the polynomial coefficients.
114  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

result was dependent on the number of points averaged and the distance be-
tween the farthest point and station location.
Another popular approach was the calculation of the second vertical
derivative, which is often used as an edge-detection tool. Laplace’s equation
(Chapter 3) relates this derivative to the sum of the horizontal second-order
derivatives, which can be obtained easily in space domain from the gridded
data by using finite differences.
Thus, from equation 13 of Chapter 3,

∂ 2 gz ∂ 2 gz ∂ 2 gz
∇ 2 gz = + + = 0,
∂ x2 ∂ y2 ∂z2
it follows that

∂ 2 gz  ∂ 2 gz ∂ 2 gz 
= − .
∂z2  ∂ x 2 + ∂ y 2 

Second vertical derivatives also can be calculated using Fourier-transform


techniques or spatial-convolution filters, but the above approach is the most
stable.
In a true second-derivative application, local contributions to the ob-
served field are enhanced, whereas the broader, regional components are
attenuated.
A final technique of anomaly separation is the graphical residual ap-
proach, in which the interpreter intuitively draws a regional field to yield a
residual compatible with the expected geology in the area of study.
In practical applications, the amplification of the shorter-wavelength
components tended to also amplify short-wavelength noise in the data sets,
and so the actual operators were modified (from the ideal) to give smoother
maps. In the next section, we develop the equivalent of these space-domain
operators in the frequency domain by using spectral analysis. Before the
introduction of digital computers, perhaps the most popular method used
in the separation of anomalies was the now seldom-used grid-residual ap-
proach.
Figure 7 shows a comparison between the grid and graphical residual
approaches. The observed anomaly consists of a large positive anomaly on
the right side, on which are superimposed two local positives. The graphical
residual reflects only the two local positives and results from the “smoother”
(the interpreter) deciding that the regional (unwanted) component consists
of the large positive on the right side. By contrast, the grid residual cannot
discriminate between the local positive curvature in the large positive and
Chapter 7:  Anomaly Interpretation  115
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Derivative anomaly

Observed anomaly

Graphical regional
Grid residual

Graphical residual

Figure 7. A comparison of the grid and graphical residual. After Nettleton (1971),
Figure 20.

that of the two local positives. Moreover, the grid residual produces both
positive and negative anomalies.
These two methods for extracting the residual anomaly from the ob-
served field dramatically demonstrate their very different purposes. The
graphical residual results from an interpreter applying geologic constraint —
in this case, the assumption that the large positive anomaly is caused by a
large intrabasement feature assigned to the regional geology. No negative
anomalies are included in this interpreter’s concept of the local geology.
The grid residual results from mathematical constraint that simply discrimi-
nates between the long-wavelength and short-wavelength components in the
observed field.
We started this section with two very different regional assumptions
(Figures 5 and 6) resulting in two geologic interpretations. In Figure 7, we
compare the grid-residual (anomaly-enhancement) result with the graphical
approach (anomaly separation).
The high-pass filter applied to the observed anomaly in the form of a
grid residual (or, similarly, using the second vertical derivative) is a purely
mathematical operation that does not involve any geologic constraints. Inter-
estingly, the vertical-derivative (high-pass) anomaly contains both positive
and negative anomalies, as if we have combined the two first-mentioned
cases (Figures 5 and 6) of placing the regional first below and then above
the observed anomaly.
Of course, the two anomalies shown in Figure 8 are not directly com-
parable, but they are shown together to indicate two approaches to the
116  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)
Anomaly separation

Residual

Geologic constraint

b)
Anomaly enhancement

Derivative

Mathematical constraint

Figure 8. (a) Anomaly separation versus (b) anomaly enhancement.

identification of residual anomalies. They have different units and serve dif-
ferent purposes.
In the next section, we develop the representation of any gravity field
in the Fourier domain. We can then design filters to enhance some anomaly
characteristics at the expense of others, in a manner similar to the grid resid-
ual demonstrated above. A brief list of enhancement techniques includes the
grid residual, second vertical derivative, downward continuation (in which
the field closer to the sources is computed), and band-pass filtering. These
residuals can be computed by convolution in the space domain or by mul-
tiplication in the frequency domain (see the section below titled “Spectral
analysis”).
Although the second vertical derivative is not an anomaly-separation
technique but simply a different view of the gravity data, we include it in
this discussion to illustrate its use in the location of possible geologic targets
of interest. One of several excellent examples showing the effectiveness of
enhancing local anomalies at the expense of regional ones is taken from the
Los Angeles Basin (Nettleton, 1971).
Figure 9a indicates the very high northeast-southwest gradient in the
observed field with little apparent visible correlation with the oil fields
(indicated in solid black). The change in contour spacing in this high-
gradient area produces in the second-vertical-derivative (Figure 9b) excel­
lent closures over the oil fields. However, because of noise in the data,
it is also possible to produce fictitious anomalies in the residual by this
technique.
Chapter 7:  Anomaly Interpretation  117
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) b)

1 mile

Figure 9. (a) Bouguer gravity, Los Angeles Basin, with (b) its second vertical
derivative. After Nettleton (1971), Figures 34 and 35.

Another useful technique to delineate target boundaries is the total hori-


zontal gradient, H, which is defined (Cordell, 1979) as
1
 dgz  2  dgz  2  2
H =   +  .
 dx   dy  

Using this approach at the approximate location of every boundary between


two media of different densities, one obtains a high value of the total gradi-
ent which, along with the adjacent points, defines a berm along the contact.
In the case of a localized target, the berm indicates approximately the lateral
extent of the target.
As we have seen, anomaly-separation techniques, unlike enhancement
approaches, are intended to preserve anomaly amplitude. The graphical
approach, discussed above, is qualitative and is subject to the bias of the
interpreter. The polynomial surface approach, also discussed above, is con-
strained by a least-squares differencing criterion that might not be relevant
geologically. By definition, it does not honor actual data points.
We can honor actual data points with a bicubic spline interpolation,
but we still need a method by which we can assign to it the potential-field
characteristics caused by the regional geology. The most effective anomaly-
separation approach (in which amplitude is preserved, unlike what happens
with enhancement techniques) is geologic modeling, an example of which
is given in Chapter 9.
118  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Spectral analysis
Interpretation of gravity data is facilitated by additional processing
to better emphasize various features of interest. Because there is a strong
relationship between the dominant wavelengths of a gravity anomaly and
the size, shape, and depth of the causative body, it becomes apparent that
Fourier-transform techniques will play an important role in providing fur-
ther insight in interpreting gravity data. Such an approach will help us to
better define and understand the nature of gravitational sources. No attempt
at interpreting these additionally processed data should be undertaken, how-
ever, without comparing them side by side with the data contained in the
Bouguer or free-air map.

Upward continuation

Shallower targets contain more energy in the shorter wavelengths than


1
deeper targets do and decay much faster with depth. Because of the 2
r
dependence, a target buried at a depth of 1 m will have an amplitude four
times higher than if the same target is buried at a depth of 2 m and 100 times
higher than if it is buried at a depth of 10 m. Thus, if we were to measure the
gravity field at a certain height above the original observational surface, the
resulting effect would be a stronger attenuation of shallower anomalies with
only minor attenuation of deeper ones. Fortunately, we do not have to liter-
ally carry out measurements at a higher elevation to achieve this purpose
because we can reliably calculate the field analytically to obtain what would
be measured at the new height3.
This upward-continuation operation has many useful applications in
practice. For instance, we can more easily interpret deep-seated intrusive
targets by greatly reducing the high-frequency noise created by shallower
sources. In addition, if we have two contingent airborne gravity surveys
flown at different elevations (or, say, a marine survey at sea level and an air-
borne survey), we can use upward continuation to bring the lower-elevation
survey to the level of the higher-elevation one, thus facilitating the merging
of the two data sets.
Upward continuation can be achieved easily by using Fourier-transform
methodology. We have seen in the section above titled “Anomaly separation”

  3Sometimes when the initial gravity survey covers only a small area, one cannot
reliably calculate the upward-continued field, and a higher-altitude survey becomes
necessary.
Chapter 7:  Anomaly Interpretation  119
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

that the Fourier transform of the gravitational acceleration for a completely


general target body is given by expression 6 as

∫ F[ρ(ζ )] e
u2 + v 2 −ζ u 2 + v 2
F[ gz ] = 2π ke z dζ , (17)
z

where the z-axis is positive downward. For what follows, it will be easier to
change equation 17 to a system of coordinates with z oriented upward, in
which case one obtains
0

∫ F[ρ(ζ )] e
u2 + v 2 ζ u2 + v 2
F( gz ) = 2π ke − z dζ for z ≥ 0, (18)
−∞

where we assumed all sources are situated below ground (z = 0). Writing the
above expression twice for two heights z1 and z2, with z2 > z1, we immediately
obtain
u2 + v 2
F ( gz 2 ) = F ( gz 1 ) e − h = F( gz 1 ) e − h p, with h = z2 − z1 and z2 > z1, (19)

where p2 = u2 + v2.
Thus, to continue the gravity data upward a distance h, one first would
take the Fourier transform of the data and, after filtering (weighting) with the
function e −h p, one uses the inverse Fourier transform to obtain the upward-
continued values. This procedure is common to all filtering operations
described in this chapter. The data are first Fourier-transformed and, after
applying the filtering operation, the data are inverse-Fourier-transformed to
obtain the desired result in space domain.
The upward-continuation filter is remarkably stable and well behaved,
is bell shaped, and has values ranging only from zero at infinite frequency to
one at zero frequency. From equation 19, it can be seen easily that upward
continuation is a low-pass operation designed to reduce the high-frequency
portion of the spectrum.

Downward continuation

In a manner similar to the previous development, one can also downward-


continue the gravity data to a lower elevation, as long as there are no sources
present along the way. Such an operation will be useful in better identifying
shallower sources because we are simulating taking measurements closer to
those sources. Using the same derivation as before, we thus obtain

u2 + v 2
F ( gz 2 ) = F ( gz 1 ) e h = F ( gz 1 ) e h p, with h = z1 − z2 and z2 < z1. (20)
120  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

In contrast with the upward-continuation filter, the downward-continuation


filter is not stable, having values ranging from one at zero frequency to infin-
ity at infinite frequency. To avoid the instability of this filter at higher fre-
quencies, it is customary to taper the filter gradually to zero at those higher
frequencies, with the taper cutoff frequency chosen based on the noise level
and downward-continuation distance. This approach will reduce the insta-
bility at high frequencies, but the results will be less accurate. Regardless,
downward continuation is used often in many practical applications when
shallower sources are the target.
A better behaved filter for the downward-continuation operator is given
by
ehp , (21)
F ( gz 2 ) = F ( gz 1)
1 + α (u 2 + v 2 ) e h p
where a is a regularizing parameter to be chosen according to the amount
of data noise (note that for a = 0, expression 21 reduces to expression 20).
The proof of expression 21 is beyond the scope of this book (see Tikhonov
and Arsenin, 1977).

First vertical derivative

We have seen that shallower sources attenuate faster with height in com-
parison with deeper sources. The rate of change of an anomaly with height
is given by its first derivative with respect to z. Thus the vertical derivative
d ( gz )
of gravitational acceleration will be larger over shallower targets than
dz
over deeper ones. Taking the z-derivative of expression 17, one obtains

 dg 
F  z  = 2π k u 2 + v 2 e z u 2 + v2
∫ F[ ρ (ζ )] e
−ζ u 2 + v 2
dζ (22)
 dz  0

or, more simply,

 dg 
F  z  = u 2 + v 2 F( gz ) = p F( gz ). (23)
 dz 

The first-vertical-derivative filter p = u 2 + v 2 is not well behaved, having


values ranging from zero at zero frequency to infinity at infinite frequency.
As before, a tapering of this filter is applied before doing the actual calcula-
tions. Regardless, the higher frequencies resulting from shallower targets
will be amplified, and thus, adjacent anomalies can be separated more easily.
Chapter 7:  Anomaly Interpretation  121
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Both a more stable and a more accurate first-vertical-derivative filter can


be calculated using Hilbert transforms (Nabighian, 1984). Starting from the
identity
− iu − iv
u 2 + v 2 = iu + iv ,
u2 + v 2 u2 + v 2
we can write (by recognizing the meaning of various terms)

 dg  − iu dg  − iv dg 
F z  = F z  + F  z  . (23a)
 dz  u +v
2  dx 
2
u +v
2  dy 
2

The x- and y-derivatives of the gravitational attraction can be computed reli-


ably in space domain, and their Fourier transforms are now multiplied by
filters that vary only with frequency from −1 to 1, thus avoiding any need for
tapering. The two filters above are the x- and y-components of the Hilbert
transform operator.

Second vertical derivative


A second vertical derivative is a measure of curvature. Because the
grav­itational anomaly of a shallow target decreases with height faster than
the anomaly of a deeper target, it follows that its curvature will be larger,
and this again can be used to emphasize shallower targets at the expense of
deeper ones. Taking once more the z-derivative of expression 22, one obtains


 dg 
F  z  = 2π k (u 2 + v 2 ) e z u 2 + v2
∫ F [ ρ (ζ )] e
−ζ u 2 + v 2
dζ (24)
 dz  0

or, more simply,

 dg 
F  z  = (u 2 + v 2 ) F( gz ) = p2 F ( gz ). (25)
 dz 

The second-vertical-derivative filter p2 = u 2 + v 2, like the first-vertical-


derivative filter, is not well behaved, having similar values ranging from
zero at zero frequency to infinity at infinite frequency. As before, a taper-
ing of this filter is required before doing the actual calculations. Although
the calculation of the second-derivative maps can be done in the frequency
domain using expression 25 and then transforming back to space domain,
it is easier and more stable to calculate such maps in space domain directly,
122  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

as shown below:
∂ 2 gz  ∂ 2 gz ∂ 2 gz 
= −  ∂ x 2 + ∂ y 2  , (26)
∂z2
where second derivatives with respect to x and y can be calculated easily in
the space domain using finite-difference expressions.
In early days, the second-derivative filter was used extensively when inter-
preting gravity data (Evjen, 1936), to emphasize shallow sources and because
the technique helps to better define the edges of those sources. The second-
derivative filter also helps in detecting and examining noise in data sets.

Directional and second-order derivatives

Using equation A-9 of Appendix A, “Fourier Transform,” at the end of


this book, one can calculate any other component of gravitational attrac-
tion, including the tensor components measured in gravity gradiometry, by
simple manipulations of expression 17 of this chapter. The derivations are
simplified greatly by first calculating the gravitational potential U and then
taking the appropriate derivatives. Because
∂U
gz = ,
∂z
by using the first-derivative filter defined above, it follows that

F( gz ) = u 2 + v 2 F(U ) = p F(U )
or
F ( gz )
F( gz )
F(U ) = . (27) =
u +v p
2 2

Using equation A-9 of Appendix A, “Fourier Transform,” at the end of this


book, it immediately follows that

F( gz ) F( gz ) ,
F( gx ) = iu = iu
u +v
2 2 p

F ( gz ) F( gz ) ,
F( g y ) = iv = iv
u +v
2 2 p
F ( gz ) F( gz ) ,
F( gxx ) = − u 2 = −u 2
u +v 2 2 p
Chapter 7:  Anomaly Interpretation  123
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

F ( gz ) F ( gz )
F( g yy ) = − v 2 = −v2 ,
u +v
2 2 p
F ( gz ) F ( gz )
F( gxy ) = − uv = − uv ,
u2 + v 2 p

F( gxz ) = iu F( gz ),

F( g yz ) = iv F( gz ),

F( gzz ) = (u 2 + v 2 F( gz ) = p F( gz ). (28)

The above expressions can be written in matrix form as

 iu 
 gx   p 
   
F  g y  =  iv  F ( gz )
 g   p 
z  
1 

and

 −u 2 − uv 
iu 
 gxx gxy gxz   p p
   
F  g yx g yy g yz  =  − uv − v 2  F ( gz ).
g g  iv 
 zx zy gzz   p p 
 iu iv p

The filters required to carry out the above calculations are again not well
behaved, and special care must be taken when applying them.

Analytic signal

The analytic signal, although used extensively in magnetics, is used


little in gravity techniques, primarily because of the sparser nature of grav-
ity data, which makes the calculation of derivatives less reliable (Nabighian
et al., 2005). For magnetic-profile data, the horizontal and vertical derivatives
fit naturally into the real and imaginary parts of a complex analytic signal
(Nabighian, 1972, 1974, 1984; Craig, 1996). In two dimensions (Nabighian,
124  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

1972), the amplitude of the analytic signal is the same as the total gradient.
For three dimensions, Roest et al. (1992) introduce the total gradient of
magnetic data as an extension to the 2D case. The results obtained for mag-
netic data can be extended to gravity data if one uses as input the horizontal
derivative of the gravity field. What is now commonly called the 3D analytic
signal should correctly be called the total gradient.

Matched filtering

If the spectrum of the expected signal is known, matched filtering can


help to locate the signal in a given data set. The matched filter has the same
spectrum as the desired signal. In potential-field methods, matched filtering
has been used primarily to separate data into anomaly components repre-
senting different source depths (Nabighian et al., 2005). The method first
was developed for use with magnetic data when Spector (1968) and Spec-
tor and Grant (1970) showed that logarithmic radial power spectra of aero-
magnetic map data contain straight slope segments that can be interpreted
as having arisen from statistical ensembles of sources or equivalent source
layers at different depths.
Spector (1968) applies matched filtering in both frequency and space
domains. Syberg (1972), who introduces the term matched filter, applies the
technique to modeling azimuthal variations within each band-pass. Rids-
dill-Smith (1998a, 1998b) develops wavelet-based matched filters, whereas
Phillips (2001) generalizes the Fourier approach of Syberg (1972) to sources
at more than two depths and explains how matched Wiener filters could be
used as an alternative to the more common amplitude filters.
An alternative to matched filters, based on differencing upward-con-
tinued fields, is developed by Jacobsen (1987). Cowan and Cowan (1993)
review separation filtering and compare results of the Spector and Grant
(1970) matched filter, the Cordell filter (Cordell and Grauch, 1982), the
Jacobsen (1987) filter, and a second-vertical-derivative filter on an aeromag-
netic data set from Western Australia.

Wavelets

The wavelet transform is emerging as an important processing tech-


nique in potential-field methods and has contributed significantly to the pro-
cessing and inversion of both gravity and magnetic data (Nabighian et al.,
2005). The concept of continuous-wavelet transform was introduced ini-
tially in seismic data processing (Goupillaud et al., 1984), whereas a form
of discrete-wavelet transform has long been used in communication theory.
Chapter 7:  Anomaly Interpretation  125
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

These were unified through an explosion of theoretical developments in


applied mathematics. Potential-field analysis — magnetic methods in par-
ticular — have benefited greatly from these developments.
Moreau et al. (1997) were the first to use continuous-wavelet transform
to analyze potential-field data for simple monopole sources. In a seminal
paper, Hornby et al. (1999) independently develop a similar theory and recast
commonly used processing methods in potential fields in terms of continu-
ous-wavelet transform. These wavelets are essentially different second-order
derivatives of the potential produced by a point monopole source. Methods
based on continuous-wavelet transform identify locations and boundaries
of causative bodies by tracking the extrema of wavelet transforms. Martelet
et al. (2001) apply a similar wavelet transform to gravity data to identify
geologic boundaries.
A second approach is applied primarily for data processing that uses
discrete-wavelet transforms based on compactly supported orthonormal
wavelets. Chapin (1997) applies wavelet transform to the interpretation of
gravity and magnetic profiles. Ridsdill-Smith and Dentith (1999), a paper
on processing aeromagnetic data, generally is applicable to gravity data as
well. Lyrio et al. (2004) improve on the concept of wavelet denoising in
signal processing and apply it to processing of gravity-gradiometry data by
first estimating the noise level in the wavelet domain and then removing the
noise accordingly.
Finally, discrete-wavelet transforms were used as a means to improve
numerical efficiency of conventional processing methods. Li and Oldenburg
(1997, 2003) compress the dense sensitivity matrix in 3D inversion to reduce
memory requirement and CPU time in large-scale 3D inverse problems. A
similar approach has been applied to the problem of upward continuation
from uneven surfaces by using equivalent sources (Li and Oldenburg, 1999).

Depth determination
The last section in Chapter 3 treats the problem of ambiguity and Green’s
equivalent layer. Simply stated, without external information, a large num-
ber of possible solutions exists, each of which, in a variety of configurations
and depths, will satisfy the observed anomaly. Generally, the actual depth to
the causative source lies somewhere between the earth’s surface (or deeper
in the presence of well data or other geologic or geophysical information)
and a limiting or maximum possible depth that we will discuss at the end of
this section.
Interpreters, however, often will calculate a depth to a structure by first
assuming that it can be approximated by a simple geometric model. Because
126  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

the deeper the target the wider its anomaly, some so-called half-width rules
are used frequently by assuming that the causative body is a sphere (3D,
in which the surface anomaly attenuates approximately the same in each
horizontal direction), a cylinder (2D, in which the surface anomaly is quite
elongated in one of its directions so that cross-sectional profiles are about
the same wherever they are taken near the maximum or minimum of the
anomaly), a vertical cylinder, and thin or thick dipping dikes. We briefly
develop the half-width rule for each below.

Sphere

As an example, we apply this technique to the radially symmetric gravi-


tational attraction of a sphere located at depth h below the origin of coordi-
nates. From equation 16 of Chapter 3, the sphere anomaly is given (for y =
0) by
Mh
gz = k 3 ,
( x 2 + h 2 )2
where x is the distance from the origin. The anomaly maximum is at x = 0
and is equal to
M
gz , max = k .
h2
To determine the distance from the anomaly center to the point x1/2 at which
its value is half the maximum value, we write

Mh 1 kM
k 3 = ,
(x2 + h )2 2 2 h2

which simplifies to
3
( x 2 + h 2 ) 2 = 2h 3 ,
2
x 2 + h2 = 2 3 h2 ,

and finally, h = 1.305 x1/2.


In other words, the depth to the center of the sphere is equal to 1.305
times the distance between the location of the maxima (or the minimal point
for a negative anomaly) of the gravity anomaly and point x1/2, where the
anomaly is one-half the maximum value.
Chapter 7:  Anomaly Interpretation  127
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Cylinder

For a horizontal cylinder using expression 20 of Chapter 3, one obtains


h = x1/2. These half-width rules, if the models approximate the actual geologic
sources, yield depths to the centers of the bodies. To obtain the depths to the
top surfaces of the features, one must assume the density contrast between
the source and the surrounding rocks and, based on the maximum (or mini-
mum) amplitude (see equation 16 of Chapter and equation 20 of Chapter 3
for the sphere and cylinder, respectively), calculate the radius of the feature.
Uncertainty in assigning the density contrast, therefore, leads directly to
uncertainty in the determination of the upper surface of the geologic feature.
In addition, the half-width rule for depth determination of both the cylinder
and the sphere applies equally well to an infinite family of thin horizontal
plates existing between the sphere (or cylinder) and the earth’s surface. This
observation leads to a rather simple and quick improvement in half-width
depth determination, and it is treated in Appendix A of this chapter.

Other simple geometries

For a vertical cylinder, one can use expression 18a of Chapter 3 to obtain
h = 1.732 x1/2. For other types of targets (thin dipping sheets, thick prisms,
and so forth), half-width rules are less reliable and depend strongly on their
depth extent, dip, width, and so forth. An alternate approach (Am, 1972) is
to use characteristic curves (nomograms) that relate various features of the
anomaly, e.g., location of points with the steepest gradient on both ascend-
ing and descending parts of the anomaly, half-steepest gradient points, max-
ima and minima locations, and so forth.
The half-width rule as described above applies only to localized targets
which have a bell-shaped anomaly associated with them. For targets that are
elongated horizontally, one can use some slightly modified rules.

Semi-infinite horizontal sheet

A semi-infinite horizontal sheet can represent a fault or other geologic


phenomena with an abrupt termination. As an example, the gravity effect of
a semi-infinite horizontal sheet is given by equation 24 of Chapter 3 as

π  x π 
gz = 2 kσ  + arctan    = 2 kσ  + θ  , (29)
2  h  2 
where s is the surface density and the angle q is defined in Figure 10. The
maximum gravity effect is at x → ∞ and is equal to gz, max = 2pks, where
128  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

35
g/ 2 k t
30

25

20 η= ¼ η= ¾
maximum maximum
amplitude amplitude
15

10

0
–20 –10 0 10 20 30
P Surface

To infinity

Figure 10. Gravity anomaly of a semi-infinite horizontal sheet.

s is the surface density, or mass per unit area, i.e., gz approaches the Bou-
guer formula asymptotically.
π kσ
At x = −h, expression 29 reduces to gz /gz , max = η = , which is
2
one-fourth of its maximum. By contrast, at x = h, expression 29 reduces to
3π kσ
gz /gz , max = η = , three-fourths of its maximum value. Thus we obtain
2
the depth to the thin sheet from

1
h=
2
( x 3 / 4 − x 1/ 4 ) ,
where x1/4 and x3/4 are the x-coordinates where the gravity anomaly is one-
fourth and three-fourths, respectively, of its maximum value.
Chapter 7:  Anomaly Interpretation  129
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

By taking the derivative of expression 29, we obtain

dgz 2 kσ h
= 2 ,
dz x + h2
whose half-maximum half-width is directly equal to h.
This treatment of the truncated bed assumes that the bed on the other
side of the fault trace is not present. The missing part of the faulted bed is
assumed to be either upthrown and eroded away or downthrown and so deep
as to be not relevant.
Figure 11 shows the more general case in which both members of the
anomalous bed are present along with their respective anomalies. This

40
Combined effect of two beds

Effect of infinite bed, not faulted

30

Anomaly from upper bed

20

10
Anomaly from lower bed

0
–20 –10 0 10 20 30

Figure 11. Two-sided fault. Shown are the gravity anomalies from each sheet
separately and the combined anomaly of both sheets. See also Figure 12.
130  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

model leads to the interesting gravity anomaly that has a relative high over
the upside of the fault and a relative low over the downside (assuming the
density of the bed is higher than that of the surrounding rocks). The anom-
aly amplitudes approach the infinite plate value at both infinities and pass
through that value directly over the fault trace. Recall that both sides of the
truncated bed have their one-half value at that crossing.
Assuming finite thickness for faulted beds (s = rt) and using the nota-
tion from Figure 12, we can write the equation of the combined effects
directly:

gz = 2 k ρt (π ± θ ), (30)

where q is the angle subtended at the field observation point by the verti-
cal line joining h1 and h2, the depths to the center of the two beds, respec-
tively. In this example, for positive values of x (which happens to be over the
upthrown side), q is added; for negative values of x, q is subtracted.
The distance between the highest anomaly value and the lowest (where
q is a maximum, qm) is equal to 2xc, and

Δgz = 4 k ρtθ m

or
Δgz
θm = . (31)
4 k ρt

mGal
4.8

4.4
∆gZ
2 k t

3.8
2xc
–10 km xc 10 km
m
h1 t Infinity

Infinity t h2

0 km

Figure 12. The two-sided fault anomaly.


Chapter 7:  Anomaly Interpretation  131
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

To obtain the distance xc, at which the anomaly achieves an extremum, we


set the derivative of equation 30 to zero and obtain

xc2 = h1h2 . (32)

We now have a measure of qm (equation 31) and the point at which it can
be constructed, xc, in terms of the two unknown depths h1 and h2. Two more
equations can be written by inspection of Figure 12:

h1
tan α =
xc
and
h2
tan (α + θ m ) = .
xc
Now we have three equations with three unknowns (h1, h2, and a), from which

1 − tan 2 α
θ m = arctan
2 tan α
or
θ
α = 45° − m . (33)
2
Now we can properly place qm and determine h1 and h2 by the intersection of
the rays of qm with the vertical fault trace. Alternatively, we can calculate h1
to determine the placement of qm:

θm
1 − tan
2 = h1
θm
1 + tan xc
2
or

sin θ m + cos θ m − 1
h1 = xc . (34)
sin θ m − cos θ m + 1

From a practical point of view, a considerable amount of anomalous


mass per unit area is required to cause a relative change in the anomaly
from its high to its low of more than 1 mGal. Nonetheless, this is an effect
which, when carefully examined, might be important in an exploration proj-
ect. Because the classical concept of a gravity anomaly over a fault typically
132  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

contains only part of the disturbed geology (the “up” or “down” anomalies
depicted in Figure 11), a misinterpretation might lead to the false conclu-
sion. For example, one might conclude that unconsolidated or lower-density
sedimentary rocks occur on the downthrown side, that the positive compo-
nent is mistaken for an anticlinal sedimentary structure on the upthrown
side, or that the two-sided fault anomaly analyzed here is transformed mis-
takenly in the anomaly-separation process into the classical fault anomaly.
In the latter case, the commonly constructed two parallel regional grav-
ity curves on the flanks of the anomaly (assuming the one-sided bed) can
result in an estimate of bed depth too shallow by a factor of as much as 2.5.
The example used for this section assumes a vertical fault trace. For a
thrust or reverse fault or for other more complex rock distributions, the sim-
ple formulas given here would not be appropriate. The intended purpose,
however, is not to provide the general solution to the two-sided fault prob-
lem but rather to alert the interpreter to the oversimplification of assuming
that only one side of the faulted bed is present. In most geologic environ-
ments, more complete 2D and 3D modeling is needed, as discussed below.

Euler deconvolution

Euler deconvolution has been used widely in automatic gravity and


aeromagnetic interpretations because it requires no prior knowledge of
density or the source magnetization or direction and assumes no particular
interpretation model. Strictly speaking, Euler deconvolution is valid only
for homogenous functions, and only very few functions as used in gravity
exploration are homogenous. Despite this, geologically plausible results can
be obtained in many cases with real data.
A function f(x) of a set of variables x = (x1, x2, x3,...) is homogenous if
it satisfies
f ( tx) = t n f ( x),
where n is the degree of homogeneity4. A function that is homogenous satis-
fies the equation (Hood, 1965)
x∇f ( x) = nf ( x).
The method was proposed by Hood (1965), who first wrote down Euler’s
homogeneity equation for the magnetic case and derived the structural index

  4A function of the type f = A/rn is homogeneous of degree −n. To avoid negative


numbers for the degree of homogeneity, we define a positive structural index (SI)
as N = −n.
Chapter 7:  Anomaly Interpretation  133
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

for a point pole and for a point dipole. The method was developed further
by Thompson (1982) and later was applied to 3D problems by Reid et al.
(1990).
The method uses three orthogonal gradients of a given potential quan-
tity as well as the potential quantity itself to determine the location and
dU
depth to a given target. As an example, for gz = = U z, the standard Euler
deconvolution equation can be written as dz

( x − x 0 ) U xz + ( y − y0 ) U yz + ( z − z0 ) U zz = − N ( Bz − U z ),

where x0, y0, z0 are the unknown coordinates of the source body center or
its top edge; x, y, z are the known coordinates of the observation point; Bz
represents an unknown constant background (regional) value for Uz; and N
is the structural index whose value depends on source type.
Mushayandebvu et al. (2001) and Nabighian and Hansen (2001) show
that the standard Euler equation is also satisfied if we replace the potential
function with the x- and y-components of its Hilbert transform (Nabighian,
1984). This added two more equations to the standard Euler equation, which
led to more stable solutions and allowed for the independent determination
of the structural index. The extension to gravity-gradient tensor data is pub-
lished by Zhang et al. (2000).
Using gridded data and applying the Euler equation to each point on an
n × n window, we obtain an overdetermined system of equations from which
the unknown quantities above can be determined. For standard Euler appli-
cation, we assume a structural index N  5 and then solve for the remaining
four unknown parameters x0, y0, z0, and Bz. The process is repeated for each
position of a moving window. The solutions are then clustered by both hori-
zontal location and depth for a cleaner representation. If we use the extended
Euler deconvolution, then structural index N also can be considered as an
unknown, and we solve now for five unknowns, x0, y0, z0, Bz, and N.
For some simple bodies, the SI values are shown in Table 1 and are one
less than the SI values for the equivalent magnetic values (Stavrev, 1997).
Other possible sources are not strictly homogeneous, and the application of
Euler deconvolution to such cases is an approximation at best.
A difficulty arises for the case of an infinite vertical contact for which
SI = 0 in the magnetic case (Reid et al., 1990), which leads to an SI = −1
in the gravity case. This is troublesome because it implies an increase of
gravity field strength with distance. However, the expression for an infinite

  5In standard Euler deconvolution, the vertical position and the structural index of
the source cannot be estimated simultaneously because they are linearly dependent.
134  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Table 1.  Structural index for selected simple bodies.


Body Gravity SI
Sphere 2
Horizontal cylinder 1
Fault (small step) 0

vertical contact does not give rise to a homogenous gravity field, and the
Euler method will yield only an approximate solution at best.
Tensor Euler deconvolution uses all gravity-gradient tensor components
and all components of the gravity-anomaly vector. In addition to the stan-
dard Euler deconvolution equation above, it also uses two similar equations
for the horizontal components:

( x − x 0 ) U xx + ( y − y0 ) U xy + ( z − z0 ) U xz = − N ( Bx − U x ),

( x − x 0 ) U yx + ( y − y0 ) U yy + ( z − z0 ) U yz = − N ( By − U y ).

The solution process is similar to the one mentioned above with the excep-
tion that we now have three times more equations than before, similar to the
case for extended Euler deconvolution. Assuming a value for the structural
index N, we now have to solve for the six unknown parameters x0, y0, z0, Bx,
By, and Bz.
Figure 13 shows some results from a gravity-gradiometry survey at the
Eugene Island area in the Gulf of Mexico between latitudes 27.9°N and
28.4°N and longitudes 91.3°W and 91.8°W, with a spacing of east-west lines
of 500 m and north-south ties each 2.4 km.
A structural index of 0.5 was found to give the best overall clustering
and linear grouping of solutions for this data set. Tensor Euler solutions gen-
erally are clustered more tightly and define linear features better than their
conventional counterparts do. These results suggest that many of the sources
are located close to or at the seabed, which is the strongest and shallowest
density boundary.

Limiting or maximum possible depth

If we assume that the source rocks have a density (which can vary) that
is either entirely greater than or entirely less than the host rocks, there is a
limiting or maximum possible depth below which the causative body might
not wholly be and still give rise to the observed gravity anomaly character-
istics. Following Bott and Smith (1958), a 2D approach toward determining
Chapter 7:  Anomaly Interpretation  135
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Uzy Tensor-grid Euler

28.3 40 28.3

20

0
28.0 28.0
–91.7 –91.4 –91.7 –91.4
26 +20
28.3 28.3

22 0

–20
18

14 –40
28.0 28.0
–91.7 –91.4 –91.7 –91.4
Uz Uzx

Color Depth (km) Color Depth (km)

Black 0.00 –0.50 Yellow 1.50 – 2.00

Blue 0.50 –1.00 Magenta 2.00 – 2.50

Green 1.00 –1.50 Red > 2.50

Figure 13. Gravity anomaly Uz and tensor gradients Uz x and Uz y in the Eugene
Island area with the determined depths and boundaries using tensor Euler
deconvolution. After Zhang et al. (2000), Figures 6 and 7.

the maximum depth is (Figure 14)


1
x1 − x 2 η 2
2D theorem: h ≤ , (35)
η −1
where η = gz ( x1 ) /gz ( x 2 ) > 1.
This concept has obvious utility in integrated geologic and geophysi-
cal problems in which something is known about the subsurface rocks. For
example, high gradients can preclude an anomaly being caused by a known
geologic feature, where the depth of that feature is below the maximum
or threshold depth. This and other theorems of Bott and Smith (1958) for
calculating the limiting depth, derived using the theory of inequalities, are
based on the definitions given in Figure 14, and they use, in some cases, the
maximum amplitude and the maximum horizontal gradient of the anomaly.
Unfortunately, however, although the limiting depth that results from the
application of equation 35 and the other theorems is a valid maximum depth
136  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

g (x)
z
gz (x1)

gz (max)
g (x )
z 2

x1 x2

Figure 14. Disturbing mass at limiting depth. After Bott and Smith (1958), Figure 1.

(i.e., the entire source cannot be deeper), it often is so large as to be mean-


ingless, depending on which points of the anomaly are selected. To solve this
problem and to give sound geologic meaning to the application of equation
35 and several other theorems, model studies have been applied over a signif-
icant range of applications. These results are tabulated in Appendix B of this
chapter, along with a selection of rules for the application of these theorems.

Determination of anomalous mass


We can define the most probable zone within which our geologic solution
lies as being approximately bounded horizontally by the map location of the
maximal anomaly gradients and vertically between the earth’s surface and/
or well depths (or maximal depths based on other geologic and geophysical
information) and the maximal or limiting depth based on the amplitudes and
gradients of the gravity anomaly, discussed in the previous section. The source
of the observed residual anomaly exists within this zone of possible solutions.
Chapter 7:  Anomaly Interpretation  137
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

What can we say about it? Of course, with well or seismic data or geo-
logic information, we might say quite a bit, and we will examine the results
of geologic modeling later. However, in addition to putting limits on the 3D
location of the causative bodies, we can consider one other item of information
uniquely determined by the gravity anomaly — the total amount of anomalous
mass. We developed Gauss’ theorem in Chapter 3 (equation 11 of Chapter 3):
1
4π k ∫s
M =–
(36)g ⋅ n dS,

where M is the anomalous mass, k is the gravitational constant, S is any sur-


face which completely bounds the mass, g is the attraction of gravity, and n
is the unit vector normal to the surface.
We now apply Gauss’ theorem (equation 36) to the surface shown in
Figure 15, which is composed of a semispherical surface and the corre-
sponding portion on the surface of the earth. Over the hemispherical sur-
face, the above integral reduces to
2π π
∂U
∫ ∫ ∂r r sin θ dθ dφ .
2

0 π
2

Because at large distances (radius of hemisphere ® ∞) the potential of


existing masses appears as having been caused by a point source, we have
∂U
r2 = kM, and the value of the above integral becomes –2pkM.
∂r

Observational surface
x

Anomalous
mass

y The enclosing Gaussian surface


includes the observational surface
and is external to all causative
masses.

Figure 15. Semispherical surface on which to apply Gauss’ law, which surface is
later extended to infinity in all directions. Gravity measurements are made only on
the surface of the earth.
138  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Thus we are left to integrate only over the horizontal surface of the earth,
and we obtain
1
2π k ∫s
M= gz dS , (37)

where gz is the vertical component of attraction. The integration in equation


37 assumes that the anomaly is measured and integrated out to infinity, which
of course we cannot do. However, we can apply corrections to account for the
incomplete integration, which are discussed in Appendix C of this chapter.

Interpretation of borehole gravity


In Chapter 5 (equation 19 of Chapter 5), we developed the working
equation for determining density for rocks between two gravity stations sep­
arated vertically by Dz. Ignoring the earth’s main (free-air) effect and assum-
ing that all the local rocks are uniform and extend laterally to infinity, we can
visualize the gravity attraction shown in Figure 16.
At all stations above the bed whose density is r, the attraction of the bed
is independent of distance from it and is shown as 2pkrt. Similarly, below
the bed, the attraction is upward (negative) and is equal to −2pkrt. Inside
the bed, the change in gravity resulting from the bed is a linear decrease,
passing through zero at the center of the bed because at that point, an equal
amount of mass exerts an upward attraction to that which exerts a downward
attraction. If we measure the gravity gradient in each layer (i.e., the differ-
ence in gravity values at two consecutive stations), we obtain null values
above and below the layer and a constant value inside the layer proportional
to the density of the layer from which density can be determined.
Attraction
–2 k t 0 2 k t
gz

Medium A (attraction caused


by layer is positive downward).
Uniform layer extends to infinity
in all horizontal directions.

Density = g t
z

z
Medium B (attraction caused
by layer is negative upward).

Figure 16. Density derived from borehole gravity for an infinite uniform layer.
Chapter 7:  Anomaly Interpretation  139
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) b) ρ ρ ρ ρ
1 2 3 4
ρ
ρ
∞ 1 ∞

∞ ρ ∞
2

∞ ρ ∞
3

∞ ρ ∞
4

Figure 17. Densities derived from borehole gravity for multiple uniform layers.

θ1
θ2
α a
∆Z
∞ ρ α max B ρ + ∆ρ

∆Z
D b
Figure 18. An infinite horizontal layer in which a change in density occurs at
distance D.

In Figure 17, four uniform horizontal layers that have infinite horizontal
extent are depicted with the borehole-gravity-derived density illustrated in
Figure 17b. Under those conditions, the density derived is the actual bulk
density of each layer. Now let us assume that the density of a layer changes
from the density of the layer r to r + Δr at distance D from the well. Figure
18 depicts this condition.
The vertical gravity attraction at any station in the well, gz(z), resulting
solely from the change in density Δr, can be defined as gB, meaning the
attraction of the mass B (having density r + Δr at distance D from the well).
At any two adjacent stations, 1 and 2, separated vertically by Δz, the vertical
attraction resulting from B is
g1 = gB
g2 = gB − gb + ga
Δ g = g2 − g1 = ga − gb ,
where ga is the attraction at station 2 resulting from the upper shaded zone
a, whose thickness is Δz, and gb is the attraction at station 2 resulting from
the lower shaded zone b, which also has thickness Dz.
If Δz is small relative to the other dimensions, the respective attractions
for the upper and lower shaded zones at station 2 are
140  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

ga = 2 k σ θ1 = 2 k Δρ Δz θ1
gb = 2 k σ θ 2 = 2 k Δρ Δz θ 2
gb − ga Δg
=− = −2 k Δρ (θ 2 − θ1 ) = −2 k Δρα ,
Δz Δz

where a is the angle subtended at the midpoint of the two stations in the well
by the change-in-density interface in the bed at a distance D from the well.
This angle reaches a maximum in the well in the center of the bed. Thus,
changes in density away from the well, Δ r, will cause a perturbation of the
otherwise bulk densities depicted in Figure 16 in all layers in which angle
a produces a measurable apparent density anomaly. The gradient, Δg/Δz,
in the working formula for borehole-gravity apparent density (equation 38)
is altered by the presence of the anomalous mass representing a change in
density of Δ r.
Thus, in the present example, the apparent-density anomaly measured
by the borehole gravity meter is
1 Δg α
Δ ρa = − = Δ ρ , (38)
4π k Δ z 2 π
and it is not surprising that bulk density is altered to apparent density at
measurements throughout the well. An example of this phenomenon, along
with a 3D case (horizontal circular disk) is given in Appendix D of this chap-
ter. The relationship between bed thickness and distance from the wellbore
for a variety of 2D and 3D models is also given in Appendix D.

Reservoir monitoring
Time-lapse gravity surveys (or 4D, in which the fourth dimension is
time) were developed to counter decreasing reservoir pressures in the Prud-
hoe Bay reservoir, where a water-injection program was initiated in Novem-
ber 20026. The major monitoring concern with the waterflood is to ensure
that water added in the gas cap does not prematurely flow downdip into the
oil-producing portions of the field where it could interfere with a highly effi-
cient gravity drainage mechanism. This topic is discussed in greater detail in
the section titled “Mining applications” in Chapter 9.

  6Szabó (2008) credits Loránd Eötvös as being able to detect in the early 1900s a
1-cm variation in the water level of the Danube River from a distance of 100 m by
increasing the sensitivity of his gravity compensator, one of the many instruments
he invented.
Chapter 7:  Anomaly Interpretation  141
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Appendix A
The unit half-width circle (2D) and ellipse (3D)
In the 2D case, we have seen that the depth to the center of the uniform
horizontal cylinder is exactly equal to the half-width of the anomaly — the
horizontal distance between the extremum amplitude (maximum if positive,
minimum if negative) and that point at which the amplitude is equal to half of
the extremum. This might not be definitive geologically, however, for two rea-
sons. The subsurface mass is highly unlikely to have this idealized form, and
even if it can be approximated by a cylinder, a large range of possible depths to
its upper surface results from the unknown density contrast. The second prob-
lem is discussed in this chapter in the section above titled “Depth determina-
tion.” The first problem regarding the geometry of the source is addressed here.
All models, of course, are fictitious in the sense that simple forms can
only approximate the actual geology, but only simple forms lend themselves
to closed-form or exact solution. Thus, although approximations are neces-
sary in the analysis of potential-field data, it is instructive to study the behav-
ior of anomalies that arise from simple forms. In selecting a model, it should
be convenient to use, flexible enough to meet varying geologic conditions,
accurate in its portrayal of geologic features, and economic, whether as a
tool for manual analysis or as a basis for a computer algorithm designed to
approximate an actual situation. The thin plate (or lamina) is such a model,
and although far from unknown to practicing geophysicists, its unique prop-
erties and high degree of adaptability are worth exploring.
In this appendix, we develop the valuable properties unique to the thin-
plate model and indicate a rule-of-thumb method for depth estimation. In
addition, this model can be the basis for very efficient iterative modeling of
any arbitrary 3D geologic feature.
For mathematical reasons, we require that the thin plate has no thick-
ness (i.e., the basic model is a lamina that has surface density s). We further
require that the thin plate be rectangular in shape and confined to a horizon-
tal plane. (We can apply this basic model to geologic bodies of any thickness
or irregularity.) We start with the 2D case in which the strike length of the
plate (perpendicular to the plane of the illustration) is infinite.
For a 3D plate, the vertical component of attraction, at a point of observa-
tion above the plate, is proportional to the solid angle w, subtended at the mea-
surement point by the plate. In the 2D case, the solid angle is equal to twice
the plane angle q. The vertical component of attraction, gz (Figure A-1), is

gz = 2 k σ θ (A-1)
142  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

or
 x+b x − b
gz = 2 kσ  tan −1 − tan −1  , (A-2)
 h h 
where b is the half-width of the plate, x the position of observation, and h the
depth to the plate. Note that if q = p  for a plate infinitely wide, then equation
A-1 becomes the Bouguer formula, where s = rt, for a plate of thickness t.
Equation A-2 permits us to find the loci of equal vertical attraction because
2bh
gz = 2 kσ tan −1 . (A-3)
h + x 2 − b2
2

Thus,

x 2 + (h − η)2 = b 2 + η 2 , (A-4)

where
b
η= .
g
tan z
2k σ
Equation A-4 is, of course, a circle whose center is located at (0, h) and
whose radius is b 2 + η 2 . In any cross section normal to the strike of the
infinite plate, the loci of points of equal vertical attraction are circles, as
shown in Figure A-2a. We might have started with a line mass that has a

gz

b b

Figure A-1. 2D thin-plate model of infinite strike length.


Chapter 7:  Anomaly Interpretation  143
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) h b)

Plate width = 2b

Figure A-2. (a) Loci of equal vertical attraction for a horizontal 2D lamina. As
distance h between the measurement plane and the plate is decreased, the step-
function nature of the anomaly (in the limit, from an angle subtended over the plate
of 2p to an offset angle subtended from the station to the plate of zero) becomes
obvious in the diagram. In addition, as plate width is decreased, approaching zero,
we obtain the family of circles for the horizontal line mass depicted in Figure A-2b.
(b) Loci of equal vertical attraction for a horizontal 2D cylinder.

lineal density of mass per unit length and obtained directly its loci of equal
vertical attraction and then obtained the result shown in Figure A-2b. It is
interesting to note that it can be thought of as a limiting case for the laminae,
whose width becomes vanishingly small.
Both models demonstrate graphically the continuation of gravity-anom-
aly data: By moving along the loci, we find where the same vertical attrac-
tion obtains at different elevations. Where the source mass is quite shallow
(i.e., close to the observational surface), the increased horizontal gradients
become quite obvious.
The lamina is also interesting, if generally impractical, as an equivalent
source. Any anomaly can be accommodated precisely by a surface distribu-
tion located on the observational surface, where the surface density at each
station is proportional to the gravity amplitude (gz) and the solid angle at
the station in question is equal to 2p. The effect of that particular thin plate
is zero everywhere off the plate but is equal to the anomaly at the station
directly on the plate.
The locus of equal attraction also leads to the requirement that very
high horizontal gradients (short wavelength) preclude deep source rocks.
Of course, in practice, we cannot observe step-function gradients (because
144  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

all rocks have thickness), but there are case histories remarkably consistent
with this idea (Ponce et al., 2009).
Now let us consider one particular circle from the family described by
equation A-4, the one for which the constant vertical attraction is equal to
one-half the maximum amplitude. Equation A-3 gives the vertical attraction
at any position x of the observational profile. If gz = 1/ 2 gz ( max ), then
2bh b
2 kσ tan −1 = 2 kσ tan −1 (A-5)
h +x −b
2 2 2
h
and

h 2 + b 2 = x 2, (A-6)

which describes a circle centered h units above the plate (i.e., the observational
surface), b units from the edge of the plate (i.e., over its center), and whose
radius is x (the distance obtained by using equation A-5 at which the anomaly
falls to half its maximum). We normalize horizontal and vertical distances to
the anomaly’s half-width and define this circle as the “unit half-width circle.”

Half-width depth rules: The unit half-width circle and ellipse

As noted in the section titled “Depth determination” in this chapter, the


horizontal distance between the maximum (or minimum in the case of nega-
tive anomalies) and half-maximum amplitudes is used to estimate the depths
to some geologic sources. In the 2D case, infinite horizontal cylinders are
sometimes invoked. If we equate the formula for the vertical attraction of
the cylinder whose depth is h, we obtain h = x1/2; the depth to the axis of the
cylinder is equal to the half-width distance.
In the 3D case, the vertical component of attraction of a uniform sphere
is set equal to its half-maximum amplitude, and we obtain h = 1.305 x1/2.
The depth to the center of the sphere is equal to 1.305 times the distance
between the peak amplitude and the place to which the amplitude falls to
half that value. In this appendix, we see that these rules of thumb are special
limiting cases for the more general family of thin plates.
A semicircle defined by equation A-6 is depicted in Figure A-3, the
lower limit of which is the location of an infinite horizontal uniform line
mass. By normalizing the x- (horizontal) and z- (vertical) axes to the half-
width (x1/2) of the anomaly, we define the entire universe of horizontal uni-
form lamina: Each thin horizontal plate (one of which is shown in Figure
A-3) subtended by this semicircle produces its half-maximum amplitude at
Chapter 7:  Anomaly Interpretation  145
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Maximum Half maximum


amplitude amplitude
at x = 0. at x = 1.
0 1.0
x

Thin horizontal plate

Unit
half-width
circle
Infinite
horizontal
line mass

Figure A-3. The unit half-width circle.

exactly the same position on the profile, namely, where the semicircle crops
out (or intersects the observational surface). The proof of this statement is
given in equation A-6.
The semicircle whose center falls under the maximum amplitude on the
observational surface and whose radius is equal to the anomaly half-width
(x1/2) is defined here as the unit half-width circle. We note that the infinite
line mass is the limiting case as the lamina width approaches zero, also yield-
ing its half-amplitude position exactly where the family of laminae yields its
positions. Figure A-3 suggests a rapid and significant improvement in esti-
mating the depth to a source that does not conform to the simple line mass.
For the more common anomalies that cannot be represented by cross
sections that have assumed infinite strike length, we replace the infinite hori-
zontal uniform lamina in Figure A-3 with a circular plate and the infinite line
mass with the uniform sphere. Although these laminae are two-dimensional,
with surface density (s) of mass per unit area, they can become reasonable
3D models by giving them small thicknesses, which we will do in the next
section. Again, by normalizing distances to the anomaly half-width (x1/2),
we obtain a near ellipse, as shown in Figure A-4. The minor axis is equal to
1.0, the half-width distance; the major axis is equal to 1.305, the depth to
the center of the sphere.
Because each member of the entire family of thin plates yields its half-
width at exactly the same place, namely where the half-width unit circle
(ellipse in the 3D case) crops out, we can see that the cylinder (2D) and
146  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Maximum Half maximum


amplitude amplitude
at x = 0. at x = 1.
0 1.0
x

Thin horizontal circular plate

Unit
half-width
ellipse

Uniform spherical mass

Figure A-4. The unit half-width ellipse.

sphere (3D) are special only if they are limiting cases of a range of oth-
erwise possible solutions. We make use of this concept in improving this
rule of thumb by considering at least one additional point of the anomaly.
Let h = the ratio of the value of the anomaly at any point of the anom-
aly to its maximum amplitude, gz(x)/gz(max); then h = ½, the half-width to
which the circle is normalized, h = ¼, and h = ¾ , as shown in Figure A-5,
can be used to refine the depth estimate by the following steps:

1) Find the maximum and half-maximum amplitudes on the profile or map.


2) Find the three-fourths amplitude position in terms of the normalized half-
width distance.
3) Obtain the plate at the intersection of (2) with the h = ¾ curve shown in
Figure A-5.
4) Check the result by repeating the process with the h = ¼ curve.

These depth-estimation rules are self-checking in two ways: (1) by


applying step 4 above for consistency of results and (2) by estimating plate
thickness based on the anomaly maximum amplitude and an assumed den-
sity contrast. If the plate thickness is large relative to its depth, further analy-
sis might be necessary.
Chapter 7:  Anomaly Interpretation  147
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Distance axis (x)



2D plate

Unit half-width
circle

Cylinder 2D depth-rule diagram


Depth axis (z)

Distance axis (x)


Circular plate

1.0

Sphere 3D depth-rule diagram


Depth axis (z)

Figure A-5. 2D and 3D depth-rule diagrams.

Appendix B
Application of Bott and Smith theorems
In the following theorems of Bott and Smith (1958), x1 and x2 are any hor-
izontal locations of the anomaly if the corresponding ratio of their amplitudes,
h = g1/g2, is greater than one. In addition, x and d are any numbers for which
2 gz ( x )
µ= > 1.
gz ( x + d ) + gz ( x − d )
The first three theorems below apply to 3D cases, whereas the last three and
corollary 4.1 apply to 2D cases.

x1 − x 2 η1 / 3
Theorem 1 (3D): h≤ .
η2 / 3 − 1
148  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Theorem 2 (3D): h ≤ d ( µ 2 / 3 − 1)−1 / 2.

48 5 gz , max
Theorem 3 (3D): h≤ .
dg
125
dx max
x1 − x 2 η1 / 2
Theorem 4 (2D): h≤ .
η −1
g (x)
Corollary 4.1 (2D): h≤ z .
dgz ( x )
dx
Theorem 5 (2D): h ≤ d ( µ − 1)−1 / 2 .

3 3 gz , max
Theorem 6 (2D): h≤ .
dgz
8
dx max
Theorem 4 (equation 23 of this chapter) serves as a good example of this
approach toward finding the limiting or maximum possible depth to a source,
by changing it to the following 2D function,

x1 − x η1 / 2
f = ,
η −1
where h = gz(x1)/gz(x) > 1, and f is the maximum possible depth.
For the case of the horizontal infinite uniform cylinder, we can substi-
tute the formula for the vertical component of attraction, take the derivative
of f with respect to x, set the derivative equal to zero, and find the “turning
point,” or the shallowest of all calculated maximum depths. In this case,
the minimum of maximum depths turns out to be the depth to the line mass
central to all cylinders (which have different densities) at depth h.
The 2D horizontal prism yields similar results, which are plotted in Fig-
ure B-1. If the parameter x1 is too large (15 and 30 in Figure B-1), no turning
point is achieved. If x1 is taken very near the maximum amplitude, a turning
point might not be achieved within the project area (i.e., it might occur too
far to the right in Figure B-1). The realization of the turning point is impor-
tant because it is far more diagnostic of the maximum depth of the causative
sources than are any of the other depths derived by using the Bott and Smith
(1958) theorems. All of the depths so derived are valid maximum depths,
but the turning points are clearly more useful in interpretive work.
The behavior of the 3D depth-estimation functions is similar to that of
their 2D counterparts. For thin bodies, the turning points occur very close
Chapter 7:  Anomaly Interpretation  149
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

g (x) Figure B-1. Limiting depth


curves for a thick 2D hori­
zontal prism. After LaFehr
(1964), Figure 6.

0 30 60

10 Turning point
x1 = 5
x1 = 15
Depth

20
x1 = 30

30 x1 = 0

to the actual depth of the source. For bodies with moderate thickness, the
turning points occur close to the central depth of the source. The turning-
point depth decreases with increasing body thickness but occurs closer to
the central depth than to the upper surface of the source. The turning points
themselves for sequential depth-estimation curves tend to decrease in depth
with decreasing horizontal separation between parameter x1 and the hori-
zontal coordinate where the turning point occurs.
For bodies that have great depth extent but limited horizontal dimen-
sions (“pipelike” bodies), the depths to their tops tend to be close to 0.27
times the depths of the turning points. It is instructive to compare the 2D and
3D results using the thin-plate model.
Three models are presented in Figure B-2: the square plate (3D), the
somewhat elongated plate (2.5D), and the infinitely extended plate (2D), each
at the same depth of 10 units and each shown in plan view at the bottom of the
figure. The 3D depth-estimation curves are always “correct” in the sense that
the bodies are never deeper than their respective turning points, but applying
the 3D function to the purely 2D model yields unsatisfactory results.
By contrast, the turning points for the 2D function are correct only for the
purely 2D model, in the sense that the bodies are shallower except in the 2D
case, in violation of the purpose of the theorem. Of course, one would not use
the 2D function for the square plate, but the average of the 2D and 3D results
for the middle (2.5D) case is an improvement over either function if used alone.
If properly applied, these functions can be very useful depth estimators for geo-
logic problems where we have a reasonable understanding of the geometry of
the sources. Again, if properly applied, they are always appropriate for deter-
mining the maximum possible depth to a source, regardless of its geometry.
150  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Gravity profiles

0
0 30 0 30 0 30
Depth (arbitrary units)

5 Depth-estimation curves

10 A A′ B B′ C C′

3D

15

Thin horizontal rectangular plates

To +infinity

A A′ B B′ C C′

Plan views To –infinity

Figure B-2. Comparison of 2D and 3D depth functions. After LaFehr (1964),


Figure 12.

Appendix C
Corrections for incomplete integration
using Gauss’ theorem
To obtain all the “upward” flux, the integration in equation 37 of this
chapter assumes that we integrate out to infinity. To study this effect of lim-
ited coverage, assume that the source is a point mass (or uniform spherical
Chapter 7:  Anomaly Interpretation  151
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

distribution) and that we integrate not to infinity but to a finite distance x, as


shown in Figure C-1.
If we express the distance to which we integrate in terms of the anomaly
amplitude as h, where
gz ( x )
η= ,
gz (max)
then for the spherical distribution (by substituting its expression in equation
16 of this chapter),

Mc = M (1 − η1 / 3 ),

and for the infinite horizontal line mass, where Mc is mass per unit length,
2 cos −1 η1 / 2
Mc = M .
π
We plot these functions in Figure C-2 and note that for these two simple
bodies, the fractional amount of mass (or mass per unit length in the 2D
case) obtained through limited integration of the anomalies is independent
of their depths. For these distributions, then, to obtain the total amount of
anomalous mass, we integrate only to a specified distance represented by its
value of h and then multiply the result by the appropriate factor representing

g (max)

g (– x )
g (x )

Region of integration

–x 0 x Distance

Figure C-1. Region of integration in application of Gauss’ theorem.


152  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Ratio of mass calculated to actual total mass


Mass calculated divided by actual total mass 1.0

.8

Infinite line mass (cylinder)


.6

.4

.2
Point source (sphere)

0.0
0 .2 .4 .6 .8 1
Anomaly value to which integration is carried out
as a fraction of the maximum amplitude

Figure C-2. Amount of mass (sphere) or mass/length (cylinder) calculated.

the missing part of the integration. The relationship between the fractional
amount of the total mass and the distance to which we integrate varies as a
function of the nature and depth of other distributions.
For extended bodies (LaFehr, 1965) of increasing ratios of width to thick­
ness, the fractional amount obtained in the integration is generally more
than that anticipated by the curves shown in Figure C-2. However, with
increasing body depth, the amount integrated converges on the appropriate
curve. For bodies with ratios of increasing thickness to width, the opposite
occurs — less mass is obtained for the same h.

Appendix D
Borehole-gravity distance/thickness relationships
In the section titled “Interpretation of borehole gravity” in this chapter,
we developed the concept of the apparent-density anomaly resulting from a
change in density away from the wellbore from r to Δ r and a given distance D.
Chapter 7:  Anomaly Interpretation  153
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Well

0.4
Depth km
= 1000 = 2000
0.0
0.6
0.2 0.7 1.0
Distance (km)

0.4 A B
Depth (km)

C
0.6

0.8

1.0

0 200 400 600 800 1000 1200


Density (kg/m3)

Figure D-1. Borehole-gravity density behavior for the model shown.

An example of this phenomenon is given in Figure D-1, where the


change in density at distance D from the well is 1000 kg/m3. Borehole-
gravity densities are depicted as step functions above and below the bed to
represent the discrete readings and intervals of borehole gravity surveys.
The steady increase in apparent density above the bed (and decrease below
the bed) is caused by the steady increase in angle a, culminating in A just
before the tool enters the bed. The major change on entering bed B is termed
the Poisson jump because it results from discontinuity in the second vertical
derivative of the potential (vertical gravity gradient approximated by bore-
hole gravity measurements) described in equation 17 of Chapter 5.
The horizontal derivatives in that equation, although not zero, are con-
tinuous across the geologic boundary to which the well is in normal inci-
dence. The Poisson jump can be diminished by deviated wells or by sloping
geology to which cosine corrections can apply. The apparent-density anom-
aly at its maximum inside bed is labeled C in Figure D-1. If a negative
change in density had been selected for this demonstration, the A and C
departures from bulk density would be negative.
Next we consider a change in density at distance D from the wellbore,
which is radially symmetrical about the well, as shown in Figure D-2, where
density is r and density beyond the disk is r  + Δr. Using the same method
154  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

ρ T ρ + ∆ρ

Figure D-2. A horizontal circular disk with a well along its axis.

as described above for the infinite layer, we obtain that the apparent-density
anomaly is proportional to the sine of the angle rather than to the angle
itself:
α
Δρa = sin Δρ . (D-1)
2
To appreciate how far from well D the change in density can occur and
still give rise to a measurable apparent-density anomaly and to compare
the infinite layer responses to those of the circular disk, we calculate the
straight-line relationships represented in equations 18 and 19 of this chapter.
These relationships are shown in Figure D-3.
The units can be in meters or feet or any other units if they are the same
for bed thickness T as for distance from wellbore D. Other configurations of
density and measured apparent-density anomalies can be constructed easily
from the equations, but in most cases, it would be more productive to model
the geology. Figure D-3 is intended only to show possible depths of penetra-
tion away from the well of the borehole gravity meter.
Usually, the borehole gravity meter is operated in wells that have an
independent density log, such as the gamma-gamma tool. In Figure D-4,
two porous zones are indicated in the central carbonate reef. The upper one
is offset (i.e., missed by the well), whereas the lower, thinner one is pen-
etrated by the drilling. The gravity measurements contain a response from
both zones, but the gamma-gamma tool “sees” only the lower, penetrated
zone because of the tool’s short radius of investigation. The density dif-
ference curve is borehole-gravity density minus gamma-gamma density.
Chapter 7:  Anomaly Interpretation  155
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)
120 ∆ ρ = 100 200
Infinite
horizontal
80 bed
400
T (bed thickness, arbitrary units)

40 800

20 40 60 80 100 120

b) 50 60 100 150

Circular
40 disk 200

300
20
500

20 40 60 80 100 120
D (distance from wellbore, arbitrary units)

Figure D-3. Relationship between distance from wellbore D and bed thickness
for different changes in density Δr (in kilograms per cubic meter), given an
apparent density anomaly signal of 30 kg/m3 for (a) an infinite horizontal bed and
(b) a circular disk.

For Figure D-4, the gamma-gamma-derived density is considered ideal


and does not reflect its usual high-frequency character. Normally, it is aver-
aged over the same interval as the separation of gravity stations and might
not yield the actual density of rocks immediately next to the well because of
formation damage and calibration errors. However, Figure D-4 is intended
only to help us understand how the greater penetration of the gravity meter
can be used to add geologic information. A case history of the density-
difference approach is given in Chapter 9.
156  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Density (g/cm3) Reference density


Origin ∆ ρ = 0.
0 2.4 Well 2.5 2.6
BHGM density

Ideal density
ρ = 2.4
100
Density difference
BHGM — ideal

200
Porous zones
Depth (ft)

T = 20 ft
ρ = 2.5 = 0.1
300 T = 5 ft
Reef
ρ = 2.6

400

ρ = 2.4

500
–0.10 –0.05 0 0.05 0.10
Calculated density difference (g/cm3)

Figure D-4. Density difference for reef model with porous zones. After LaFehr
(1983), Figure 4.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 8

Inversion

Introduction
We have seen previously that the calculation of the gravity anomaly
resulting from a given geologic structure, the so-called forward or direct
problem, is relatively straightforward and can be done with a high degree of
accuracy. In the most general case, the gravitational attraction is given by
expression 15 of Chapter 3:
ρ(ξ , η, ζ )(ζ − z )dξ dηdζ
gz ( x , y, z ) = k ∫ . (1)
3
V (ξ − x )2 + (η − y)2 + (ζ − z )2  2

Expression 1 is usually written in compact form as follows:

gz ( P ) = ∫ ρ(Q)G (P, Q)dV ,


V
z

where r(Q) is the density function at a point Q(x, h, z) inside volume V (usu-
ally the lower half-space), and
1ζ−z ζ−z
Gz ( P , Q ) = = 3
r2 r
(ξ − x ) + (η − y) + (ζ − z ) 
2 2 2 2

is referred to as a Green’s function and gives the vertical component of gravi-


tational attraction at observation point P(x, y, z) of an element of mass at dis-
tance r at point Q. In the above expression, r 2 = (ξ − x )2 + (η − y)2 + (ζ − z )2 ,
and (z − z)/r is the direction cosine between r and the z-axis. It is worth men-
tioning that the forward problem has a unique solution, i.e., gz is determined
completely from a knowledge of r(Q) and Gz (P, Q).

157
158  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The inverse problem is defined as an automated numerical procedure


that constructs a model of subsurface density distribution from measured
data using all prior information independent of data. In other words, given
gz and G(P, Q) in the above equation, we are asked to determine r(Q).
In contrast, however, to the forward problem, the typical inverse prob-
lem usually does not possess a unique solution. The interpreter must decide
among several solutions that satisfy the known or assumed geology for the
area under investigation while also fitting the observed data within certain
tolerances. That is because we have one set of measurements and two sets
of unknowns: density and causative body attributes. In addition, we have
only a finite number of data at our disposal, and they are inaccurate. This
leads to unstable solutions which fall into the category of so-called ill-posed
problems (Tikhonov and Arsenin, 1977), and only the introduction of some
a priori information about the model will alleviate the situation somewhat.
The problem can be simplified if one of the parameters is fixed. As an
example, Smith (1960) shows that a gravity anomaly has a unique solution
for a body of finite extent if its density is uniform and if the body is of con-
vex shape, i.e., any vertical line will not cross the body more than once. In
the most general case, it can be shown that to obtain a unique solution, it is
imperative that the dimensionality of data be the same as the dimensional-
ity of the underground density function. In other words, if the density var-
ies in three dimensions (x, y, z), we need accurate data in three dimensions
also.
In spite of the nonuniqueness of the problem, great strides have been
made in developing stable techniques for inverting gravity data in the gen-
eral case, and these techniques are now ubiquitous in the industry.
Data inversion can be carried out along two paths: density inversion and
geometric (boundary) inversion.

Density inversion
This approach to inversion tries to describe the underground geology
by finding a distribution of density contrasts within the earth that can repro-
duce the observed data within certain tolerances and with some conditions
imposed on the types of density structures possible. The earth normally is
divided into a large number of adjoining rectangular cells (Figure 1) situated
at fixed locations, and the inversion algorithm then attempts to determine
the density of each cell. Once that is done, areas of similar densities can be
combined to yield the various components of the underground structure. For
a truly 3D problem, the number of cells required to adequately represent the
subsurface can easily surpass tens of millions. Therefore, it is customary
Chapter 8:  Inversion  159
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

20 –20
–10
10 0
0 10
–10 20

–20

10

20

Figure 1. Underground density represented by a large number of rectangular cells,


each with a density to be determined by the inversion algorithm.

to increase the cell size as we go away from the area of interest so as to


decrease the size of the problem.
Most of the inversion algorithms to date are based on a technique intro-
duced by Tikhonov (Tikhonov and Arsenin, 1977) that penalizes model com-
plexity, i.e., it tries to find a solution that both fits the data and has the least
amount of model roughness. Tikhonov named this technique regularization.
The first step in applying Tikhonov’s method is to design the proper
model objective function that adequately describes the structural charac-
teristics of the model, e.g., smooth or sharp boundaries, expected dips in a
particular direction, preferential smoothness in a particular direction, and so
forth. The task then is to design a model norm such that the norm-minimizing
solution has those properties.
For gravity inversions, it is usually convenient to look at the departure of
actual underground densities from the densities of a reference model known
from some a priori information (e.g., a half-space, a layered earth, and so
forth). In addition, one would like to control the roughness of the model in
the direction of any of the three axis coordinates by minimizing the first-
order derivatives of the density function in that direction.
160  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Thus, in general, we can write a model objective function Fr(r) as follows:


2
 ∂( ρ − ρref ) 
Φ ρ ( ρ) = α ρ ∫ ( ρ − ρref )2 dv + α x ∫   dx
 ∂x
2 2
 ∂( ρ − ρref )   ∂( ρ − ρref ) 
+ αy ∫  dy + α z ∫   dz, (2)
 ∂y   ∂z

where the first term penalizes the departure between the reference model rref
and the constructed model r and the next three terms penalize the first-order
roughness of the determined model in the x-, y- and z-directions or, stated
otherwise, they try to keep the density variation as smooth as possible in
those three directions. The a coefficients in expression 2 can be chosen to
emphasize either the closeness to a reference model (ax, ay, and az small) or
the smoothness of the density distribution (ar small).
To obtain a numerical solution, the model objective function 2 is dis-
cretized over the rectangular cells in the model (Figure 1), and in each cell,
the density is assumed to be constant. Such discretization generally leads to
a large number of unknowns. For a 3D problem, it is not unusual to have
tens of millions of rectangular cells representing the underground geology
with a corresponding large number of unknowns.
To find the solution to the inverse problem, we have to find a model that
minimizes the model objective function and fits the data to a given toler-
ance. The observed data are subject to noise from various sources, e.g., inac-
curate locations, instrument noise, measurement errors, and so forth, and
are subject to errors caused by the discrepancy between our mathematical
density model and the actual density model within the earth.
In general, none of the above noises is well known in advance, and we
assume that the noise is Gaussian and write the data misfit in the form1
2
 d − di ,calc 
i=n

Φd = ∑  i  , (3)
i =1  σi

  1The expression for data misfit in equation 3 is known as an L2 norm, and it gener-
ally leads to smoother solutions for underground density distribution. For solutions
that require some abrupt changes in underground density distribution, an L1 norm is
i=n
di − di ,calc
often more appropriate and is defined as Φ d = ∑ .
i =1 σi
p
i=n
di − di ,calc
  Finally, an Lp norm defined by Φ d = ∑ σi
also can be used as a best
i =1
tradeoff between the L1 and L2 norms.
Chapter 8:  Inversion  161
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

where si are the estimated standard deviations assuming the errors are Gauss-
ian, di is the observed data at location i, and di, calc are the calculated values
from the model (the forward problem) at the same locations. With expres-
sions 2 and 3, the inverse problem, as formulated by Tikhonov, reduces to
minimizing the function

Φ ( ρ) = Φ d + λ Φ ρ ( ρ ), (4)

where 0 < l < • is a constant known as the regularization parameter, or


tradeoff parameter.
The minimization of equation 4 is usually unconstrained, and the ob-­
tained results will depend on the choice of parameter l. For very small val-
ues of l, the obtained model will fit the data as well as possible without
much input from model complexity, i.e., the obtained model will minimize
Fd without much input from Fr(r). Conversely, for very large values of l,
the model objective function Fr(r) will be minimized, and the data misfit
Fd will be large.
A solution that comes closer to an acceptable geologic model probably
will be obtained for a value of l between these two extreme values and will
lead to minimizing the model objective function and to fitting the data within
a certain tolerance. In other words, we are attempting to obtain a model that
is as simple as possible, devoid of many complications, while still fitting the
data to an acceptable level. All inverse problems are treated in this manner,
with minor differences related to the size of the model and/or whether the
problem is linear or nonlinear (Parker, 1994).
In general, the solution to equation 4 is underdetermined because we
have many more model parameters than data, especially for 3D problems. If
n is the number of data points and m is the number of model parameters and
because n < m, then we have m – n degrees of freedom and hence infinitely
many solutions.
The only remaining problem is the proper choice of the regularization
parameter l. The best solution to this problem is offered by the use of Tik-
honov’s curve (Tikhonov and Arsenin, 1977). To obtain the curve, we mini-
mize expression 4 for various values of regularization parameter l and then
plot on a log-log plot the continuous change of data misfit and model norm
as function of l (Figure 2).
Figure 2 shows that data misfit increases monotonically as parameter l
increases (left side of Figure 2), but it stays almost constant for smaller val-
ues of l even though the model norm changes rapidly in this area (right side
of Figure 2). In other words, for smaller values of l, even though the model
structure changes significantly, data misfit is affected little. By contrast, for
162  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Region with large values


( increases to the left)

Region with small values


∗ ( decreases to the right)

Figure 2. Tikhonov curve showing data misfit Fd versus model norm Fr function
of regularization parameter l. The transition point between the left and right
regions represents the best alternative among many possible solutions. The aster­
isk shows the location of the maximum curvature.

larger values of l, the misfit can be reduced greatly without a significant


change in model parameters. The transition point between the two regions
represents the best alternative among many possible solutions.
Hansen (1992) described this behavior as an L-curve and suggested that
the best value for l is the one obtained at the point of maximum curvature
shown in Figure 2. The left region, where data misfit can be reduced greatly
without large changes in the model norm, will better represent the longer
wavelengths of the model. By contrast, in the right region, where there is
little reduction in data misfit even though the model norm changes signifi-
cantly, the model is better represented by the higher frequencies, which gen-
erally are correlated to data noise.
The inversion algorithm consists of solving equation 4 for values of reg-
ularization parameter l and generating a Tikhonov curve. Using the value of
l at the point of maximum curvature in the L-curve, one generates a solution
that represents the best compromise between the two regions mentioned
above.
Another approach to obtaining the best possible value for regularization
parameter l is provided by the generalized cross-validation (GCV) tech-
nique (Golub et al., 1979; Wahba, 1990). In that approach, if one has n data
points, one can eliminate one data point at a given location and obtain a
solution to equation 4 with the remaining n - 1 data points for a given value
of regularization parameter l. Once a model is obtained though inversion,
one calculates the forward problem at location i of the missing data point
and notes the difference between the actual observed data and the predicted
one from the model, i.e., the difference d i = di, obs - di, pred.
Chapter 8:  Inversion  163
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

The above differences for the same value of l are then accumulated in
the cross-validation function GCV(l) as
n
GCV(λ ) = ∑ (d i , obs − di , pred )2. (5)
i =1

The procedure is repeated with a different value of l, and the value of


l that yields the minimum value for GCV(l) is chosen as the optimal one
because for that value, the data change is minimal when one discards arbi-
trary data points.
With some minor modifications of the above procedure, this is the
essence of the GCV technique. A typical graph outlining the GCV tech-
nique is shown in Figure 3, with the asterisk indicating the location of the
minimum value of l.
As shown above, the determination of regularization parameter l is
crucial because it controls the model complexity of the obtained solution
through inversion. Depending on data noise, some judicious tweaking in the
determination of the parameter l might be required. Regardless, that does
not guarantee that the obtained solution is necessarily the correct one. It
merely guarantees that the obtained solution yields an acceptable and inter-
pretable image of the underground geology. Much discretion and a great
deal of a priori information and geologic input are required to make geologic
sense of inversion results, and no blind faith in its outcome is warranted.
When inverting gravity data, we are faced with a few problems in addi-
tion to the nonuniqueness caused by the finite number of noisy data and the
other factors mentioned above.
As a first step, it is imperative to remove the regional gravity effect
before attempting any inversion. This regional removal is a crucial part of
GCV

Figure 3. Generalized cross-validation curve as function of regularization


parameter l. The location of the minimum is shown with an asterisk.
164  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

any interpretation, and much care has to be taken to obtain a residual map
devoid of effects not related to the underground target density distribution
(see the section titled “Anomaly separation” in Chapter 7). Because regional
removal is a subjective procedure, strongly dependent on interpreter bias, it
might be necessary to redo the inversion using different regionals. In addi-
tion, it is usually desirable to subtract the gravity effect of an assumed model
from data and then invert only on the residual anomaly.
As a second step, because gravity data obey Laplace’s equation, the solu­
tion suffers from an inherent ambiguity given by Green’s theorem, which
states that an infinite number of subsurface density distributions will satisfy
the data, even if data are known accurately. As a result, if the inversion pro-
gram is not constrained, it will tend to give solutions with densities concen-
trated toward the earth’s surface, thus giving no depth information about the
actual density distribution.
Li and Oldenburg (1996) proposed the use of a depth weighting func-
tion of the form
1
w( z ) = υ , (6)
(z + z0 ) 2

where z0 depends on observation height and cell size and u is chosen depend-
ing on the type of potential-field data we are trying to invert and is equal to
two for gravity data. This weighting function affects all terms in equation 2.
The model-objective function term then becomes

α ρ ∫ w 2 ( z )( ρ − ρref )2 dv ,

and the first-order roughness terms become, e.g.,

 ∂ [ w( z )( ρ − ρref ) ] 
2

αx ∫  dv , and so forth.
 ∂x 

Li and Oldenburg (1996) also found that imposing positivity and bound con-
straints on model densities helps to stabilize the obtained solutions.
One of the most successful approaches to the inversion problem of grav­
ity data is developed by Li and Oldenburg (1998), who formulate the gen-
eralized 3D inversion of gravity data by using the Tikhonov regularization
approach. A lower and upper bound are imposed on the recovered den-
sity contrast to further stabilize the solution. A similar approach has been
extended to the inversion of gravity gradient data (Li, 2001; Zhdanov et al.,
2004).
Chapter 8:  Inversion  165
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

More recently, there have been efforts to combine the strengths of these
two approaches. Krahenbuhl and Li (2002, 2004) formulate the base-of-
salt inversion as a binary problem, and Zhang et al. (2004) take a similar
approach for crustal studies. Interestingly, in the last two approaches men-
tioned above, the genetic algorithm (GA) has been used as the basic solver.
This is an area of growing interest, especially when refinement of inversion
is sought with constraints using increased prior information.
An example of density inversion is shown in Figures 4 and 5 for the
Heath Steele Stratmat copper-lead-zinc deposit in northern New Brunswick,
Canada. A more complete treatment of the inversion topic is beyond the
scope of this book. The interested reader is referred to Parker (1977, 1994),
Menke (1989), and Oldenburg and Li (2005).
Alternatively, one may invert for the density contrast as a function of
position in subsurface. Last and Kubik (1983) guide the inversion by mini-
mizing the total volume of the causative body. Guillen and Menichetti (1984)

Heath Steele Stratmat mGal


1000 2.0

1.8

1.6
750
1.4

1.2
Northing (m)

500 1.0

0.8

0.6
250
0.4

0.2

0 0.0
0 219 438 656 875
Easting (m)

Figure 4. Gravity-anomaly data consisting of 443 observations from the Heath


Steele Stratmat copper-lead-zinc deposit in northern New Brunswick, Canada.
The peak at (12800E, 10400N) is produced by a massive sulfide orebody and is
superimposed on linear anomalies produced by gabbroic intrusions. After Li and
Oldenburg (1998), Figure 11.
166  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Depth = 0 Depth = 40
0.75 0.75
0.62 0.62
0.50 0.50
1100 0.37 1100 0.37
0
Depth (m)

Depth (m)
130 0.25 130 0.25
260 0.12 260 0.12

m)

m)
0 0
550 550

g(

g(
0 0

in

in
Ea Ea
sti 500

rth

rth
ng sti 500
ng

No

No
(m (m
) )
1000 0 1000
0

Depth = 60 Depth = 80

0.75 0.75
0.62 0.62
0.50 0.50
1100 0.37 Depth (m)0 1100 0.37
0
Depth (m)

130 0.25 130 0.25


260 0.12 260 0.12
m)

m)
0 0
550 550
g(

g(
0 0
in

in
Ea Ea
sti 500 sti 500
rth

rth
ng ng
No

No
(m (m
) ) 0
1000 0 1000

Figure 5. Various depth slices of the density model obtained by inverting the field
gravity data shown in Figure 4. Positivity has been imposed during inversion.
After Li and Oldenburg (1998), Figure 12. Additional data courtesy of Y. Li and
D. W. Oldenburg. Used by permission.

minimize the inertia of the body with respect to the center of the body or an
axis passing through it. Although those approaches are effective, they usu-
ally are limited to recovering single bodies. For a more elaborate treatment
of this topic, the reader is referred to the above citations.

Geometric (boundary) inversion


The model-based geometric inversion assumes a geometric shape for
our target (e.g., spherical or cylindrical targets, polygonal or polyhedral
bodies, and so forth) and tries to determine its parameters (depth, dip, width,
and so forth) from observed data. In this approach, with some exceptions,
we normally have many more data than unknowns, and the overdetermined
problem reduces to solving equation 3 by using least-squares methods. Reg-
ularization generally is not required in this approach. In other words, the
objective function for this minimization problem reduces to only the data-
misfit term.
To make the problem tractable, however, because we are dealing with
variables that have different dimensionalities (depth, dip, width, and so forth),
Chapter 8:  Inversion  167
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

we have to normalize those variables to make them commensurate. Iterative


Newton-type techniques normally are used in solving such prob­lems.
The solution to this problem is generally simpler to obtain than the
solution to the density-inversion problem. However, if dependence between
model parameters and data is nonlinear or if model parameters are not inde-
pendent, there is some ambiguity because the objective function often exhib-
its local minima when plotted as a function of model parameters. Stated
otherwise, if we are trying to solve a problem involving n parameters (e.g.,
depth, dip, width, and so forth), then for each combination of values of these
parameters, we have a single value for the objective function. Thus the objec-
tive function can be represented by a surface in an n-dimensional space, and
this surface often has some local minima in addition to the sought-after glo­
bal minima.
As a result, the iterative procedures used in solving this problem might
end up in the wrong minima, depending on the starting point used for
iterations. In the absence of a reliable criterion to identify when we have
reached the global minima, we resort to accepting as a viable solution the
one obtained when we reach the same minima by starting from different
points.
In general, this approach is time-consuming, and often, Monte Carlo
techniques also are used in solving this problem. In the Monte Carlo tech-
nique, various models are selected at random from the many available. For
each model, one calculates the forward problem and evaluates the data mis­fit;
the models that best fit the data are then retained. To minimize the number of
sampled solutions, other techniques such as genetic algorithms, simulated
annealing, and other algorithms often are used also.
Bott (1960) first attempts to determine from gravity data the shape of
sedimentary basins by approximating them by 2D rectangular blocks and
adjusting their depths through a trial-and-error iterative process. Danes
(1960) uses a similar approach to determine the top of salt. This process
is extended to the 3D case by Cordell and Henderson (1968). Oldenburg
(1974) adopts the forward-modeling technique of Parker (1972) in the Fou-
rier domain (expression 14 of Chapter 7) to formulate an inversion algo-
rithm for basin depth by applying formal inverse theory.
Several papers follow on the same theme by extending the approach to
different density-depth functions or by imposing various constraints on the
basement relief (e.g., Pedersen, 1977; Chai and Hinze, 1988; Reamer and
Ferguson, 1989; Guspi, 1992; Barbosa et al., 1997). Recently, this general
methodology also has been used extensively in inversion for base of salt in
oil and gas exploration (e.g., Jorgensen and Kisabeth, 2000; Nagihara and
Hall, 2001; Cheng et al., 2003).
168  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

A similar approach has been used to invert for the geometry of isolated
causative bodies by representing them as polygonal bodies in two dimen-
sions or as polyhedral bodies in three dimensions (Nabighian, 1972; Peder-
sen, 1979; Roest et al., 1992; Moraes and Hansen, 2001), in which the ver­tices
of the objects are recovered as the unknowns.
The inversion procedures mentioned above are to be used with caution
because they are not black-box operations. Often, it is recommended to redo
the inversion with a different model objective function to obtain a sense of
the resolving capabilities of the method.
The geophysicist who uses inversion must have a good understanding
of local geology and of the limitations of the mathematical and geophysi-
cal methods involved. When applied judiciously, inversion methods have
proved to have a very beneficial effect in geologic interpretation of geo-
physical data. As such, geophysical inversion is still an ongoing topic for
researchers worldwide.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chapter 9

Geologic Applications

Introduction to interpretation
In this book, we have studied the mathematical basis for understand-
ing gravity anomalies, the gravitational nature of the earth on and in which
we make our measurements, gravity instrumentation which enables grav-
ity surveys, gravity inversion which provides for a tool by which we can
determine possible (in some cases probable) sources of gravity anomalies,
and the reduction of gravity data in static and dynamic settings which is
intended to eliminate substantial measured effects that are unrelated to the
geologic sources we wish to analyze. We have listed briefly the reasons and
generally available methods for performing gravity surveys. Now we turn to
gravity interpretation, the purpose of which is to improve our understanding
of the subsurface in geologic terms.
We enjoy a robust literature on the successes of the gravity method (failures
tend to go unreported) embracing a very wide range of geologic and engineer-
ing targets, and one approach is to list those to show the sometimes remarkable
effectiveness of the gravity method. Our purpose in this book, however, is to
provide the reader of the gravity-exploration method with a basic understand-
ing from which one can proceed to (1) determine if the method is applicable
toward an improvement in one’s understanding of the geologic problem under
investigation, (2) optimally plan a survey by which the appropriate data can be
obtained, (3) properly reduce the data to the desired anomaly field, (4) separate
the observed anomaly field into components in an effort to isolate the target(s)
of interest and, (5) if possible, determine the nature and distribution of the
geologic source that causes the resolved anomaly.
As a starting point, we will assume that the gravity method is an appro-
priate geophysical tool and that the survey has been optimally planned, exe­
cuted, and reduced to an observed anomaly field. This chapter has the dual

169
170  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

purpose of illustrating several geologic applications (steps 4 and 5 above) and


­assisting in the evaluation and planning of new projects (steps 1 and 2 above).
We learned in Chapter 6 that the anomalous field (the result of data reduction
and the starting point for data interpretation) could be a free-air, Bouguer, or
isostatic anomaly map. Although any of those can be used as a starting point
and occasionally all are used, we will assume for purposes of this chapter
that the data have been reduced to the Bouguer anomaly by using a constant
Bouguer density (as opposed to a variable density, which sometimes is used).
To explain the observed anomalies, the interpreter must account for
them in terms of any departures in density from the Bouguer reduction den-
sity between the datum used in data reduction and the observational surface
as well as in terms of density contrasts below the datum. Hence, our starting
reference is the actual surface on which our measurements are made.
We depart from the tradition that gravity data are used primarily for recon­
naissance purposes. Although this was true in the first decades of exploration,
it is misleading now for two reasons. The locations of basins are now gen-
erally known, and gravity data play an important role in detailed prospect
evaluation for which case histories are abundant. Add to this the role of bore-
hole gravity, gravity gradiometry, gravity acquired along 3D seismic lines,
reservoir monitoring, and detailed engineering applications, and it is clear
that modern gravity exploration is not primarily a reconnaissance method.
Gravity measurements continue to be used in our exploration of the
earth at all scales — studies of the shape of the whole earth, regional inves-
tigations on a continental scale involving the concept of isostasy discussed
in Chapter 6, prospect evaluations for oil and gas and mineral exploration,
and very local engineering and archaeological investigations.
In this chapter, we start with the Bouguer anomaly (developed in Chap­
ter 6) and examine it to understand the nature of the source rocks that cause
it. This anomaly has amplitude, shape, and gradient characteristics that
yield valuable information, if not (as in some cases) a d­ efinitive solution
about the geologic features under study. As we have seen, two elements
within the source rocks are necessary to produce observable anomalies:
(1) ­sufficient lateral density contrast with respect to the surrounding rocks
and (2) ­sufficient geometric distribution (volume).
Figure 1 is a general schematic for hydrocarbon exploration. If structural
and/or stratigraphic hydrocarbon traps contain or are associated with (as
in the case of basement-controlled structures) sufficient density anomalies,
an observable surface gravity anomaly can be an important aid in explora-
tion of the region. Generally, in sedimentary basins, sand and shale densi-
ties increase with depth because of compaction, as depicted in Figure 2.
Not shown in this illustration are other geologic phenomena of importance
Chapter 9:  Geologic Applications  171
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Depth to basement

Structural trap

k Stratigraphic trap
Seal roc
rvoir
Rese
Oil and gas

Source rock containing


organic material

Basement

Figure 1. Schematic diagram depicting the origin and accumulation of oil and gas
in a sedimentary basin which overlies basement rocks.

0
Salt

10 Sand/shale
Depth (kft)

20

Basement

30
1.5 2.0 2.5 3.0
Density (g/cm3)

Figure 2. Typical rock densities in a sedimentary basin.


172  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

in exploration, such as faults and carbonate reefs. The regional geologic


­framework, including local behavior of the Moho discontinuity, also might
be included in gravity interpretations.
Salt, which is very nearly incompressible, exhibits a constant density of
about 2200 kg/m3, somewhat less for pure halite and somewhat more depend-
ing on impurities in the salt. At most localities, basement rocks are igne-
ous or metamorphic and have high densities, in the range of 2700 to 3100
kg/m3. Gravity anomalies result from lateral density contrasts; very shallow
salt against sand/shale causes positive anomalies, whereas deep salt causes
negative anomalies. Positive basement structure usually causes positive
­
anomalies, and they are generally much broader than the superimposed salt
dome or shallow intrasedimentary anomalies.
Figure 3 shows a geologic example that illustrates both anomaly separa-
tion and the accommodation of the residual anomaly.

9.0
Regional
7.0
Observed

5.0
mGal
Total calculated
Residual
–2.0

Interpolated negative
–4.0
Positive residual 1.0
g/cm3 Calculated
mGal
2.2 2.6

Sediments
Caprock
–5
Dome
10 kft section
–10
Calculated
salt
Salt
–15

–20
kft

Figure 3. Salt-dome interpretation, Gulf of Mexico. After Nettleton (1971), Figure 49.
Chapter 9:  Geologic Applications  173
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Note that in Figure 3, anomalies resulting from both regional and deep
(negative) salt effects (and any other structures, including lateral sediment
changes) are shown and removed, leaving the residual caused by both shal-
low salt and caprock. The depth at which sedimentary rocks and salt have
the same density is known as the crossover depth and is shown in Figure 3
in the density-depth diagram left of the depicted salt dome. If, as shown
here, the dome is shallower than this depth, it produces a positive compo-
nent that will augment the positive contribution of the caprock, if any exists
at the top of the dome.
The problem of identifying the positive anomaly (shown here as the dif-
ference between the interpolated negative and the residual after removal of
the regional field) is discussed below in the section titled “Example of salt
with caprock.”

Location of buried features by filtering and/or modeling


It might be the case that the simple location of oil-bearing rocks (or
mineral deposits) on a map is quite sufficient toward the successful explora-
tion of an area. In section titled “Gravity calculations for simple geometries”
in Chapter 7, we described the second-derivative technique and used as an
example an oil discovery in the Los Angeles Basin. This approach is par-
ticularly well illustrated by the Bouguer anomaly at the Cement oil field in
Oklahoma, shown in Figure 4.
Elkins (1951, p. 48) writes, “The strong regional completely masks the
gravity picture of the field, which shows plainly on the second derivative
picture of Figure 18 [Figure 4b in this book], made with a grid of mile
spacing. In fact, the zero second derivative contour matches quite well the
outline shown for the [oil] field.”
A second approach toward finding the location of buried structures and
other geologic features of interest involves the integration of gravity data
with other geophysical and geologic information, very often including well
data, as depicted in Figure 5a. Here, the Bouguer anomaly map shows the
locations of Precambrian basement contacts previously determined from
wells. The anomaly strongly correlates with increasing basement depths,
which can be modeled based on an inversion of the Bouguer anomaly con-
strained by the Precambrian, as determined by well data.
Scales, culture, and other locations are not available for this example,
but the east-west distance across the prospect is approximately 15 miles (24
km). This example is intended only to help the reader visualize the tech-
nique and effectiveness of modeling and is not presented as a geologic
interpretation.
174  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)

b)

Figure 4. (a) The Bouguer anomaly at Cement oil field (shown in outline) in
Oklahoma. After Elkins (1951), Figure 17. (b) The second vertical derivative of
the Bouguer anomaly. After Elkins (1951), Figure 18.

Figure 5a depicts two geologic features trending northwest — one


approximately centered over the northernmost well, the other near the
southeastern boundary of the map — which produce gravity anomalies
(components in the observed Bouguer field) whose steep gradients cannot
be accommodated by density contrasts occuring at the depth of the base-
ment. The relatively large changes in gravity-anomaly amplitude over short
horizontal distances can be produced only by density contrasts at relatively
shallow depths, as we have seen in studies of maximum possible depths in
Appendix B of Chapter 7.
These features show up nicely in Figure 5c, which illustrates the intra-
sedimentary anomalies. Unlike the second-derivative approach shown in
Figure 4b, the anomalies shown in Figure 5c have amplitudes in milligals
and can be modeled more easily to determine the locations and sizes of the
Chapter 9:  Geologic Applications  175
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) b) c)

Figure 5. (a) Map showing Bouguer anomaly modeled by using a deep basement-
sedimentary interface. (b) Depth-to-basement map derived from the Bouguer
anomaly as constrained by well data. (c) Intrasedimentary residual anomaly map
derived by subtracting the basement-caused gravity from the observed Bouguer
anomaly map. Courtesy of EDCON. Used by permission.

buried structures. This approach can be enhanced by using seismic data,


well-log information, and electromagnetic and/or magnetic data for possible
basement or volcanic delineation if magnetic rocks are present.

Example of salt with caprock


As shown in Figures 2 and 3, salt existing at depths shallower than the
crossover depth produces a positive anomaly, as shown by the darker shaded
region in Figure 6a. The density contrast producing the shallow-salt compo-
nent of the positive anomaly is the difference between the salt density and
the nominal sedimentary densities (without uplift).
In some cases, as shown by seismic data, the sedimentary layers into
which the salt has been intruded show clear signs of having been uplifted.
The uplift brings denser sedimentary formations, originally deposited at
greater depth, into the shallower section, giving rise to local positive anom-
alies that correlate with the shallow part of the salt dome. In other cases,
the seismic data indicate undisturbed or only slightly disturbed sedimentary
horizons in the shallow section.
In any case, the uplift anomaly can be modeled by using the seismic
data to determine altered or disturbed density-depth function (Figure 2).
An example is the lower shaded area, approximately the lower third of
the positive anomaly, shown in Figure 6b. If the shallow part of the salt
176  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)
Caprock Total positive residual anomaly
residual
Anomaly amplitude

Shallow salt efffect

Sedimentary uplift effect

b)

Cap

Sedimentary uplift

Sedimentary uplift
Salt
Depth

Figure 6. Positive anomalies associated with a shallow salt dome.

dome contains caprock (examples are anhydrite, gypsum, and limestone),


it also might contribute to the observed positive anomaly (depending on
the density of the caprock material), as shown by the upper part of the
positive anomaly in Figure 6. (This figure is a schematic diagram w­ithout
scale to help the reader visualize the component effects of the positive
anomaly.)
As shown in Figure 3, the anomaly associated with a salt body can con-
tain at least three components: (1) the regional field, which can be influ-
enced strongly by surrounding salt domes; (2) the negative effects of the
deep salt occurring below the crossover depth; and (3) the positive anomaly
which, as shown in Figure 6, can have at least three causes.
In caprock exploration for sulfur deposits, the best method for finding
and rating potential targets is to integrate gravity surveys with seismic-
reflection data (for determining sedimentary uplift and, rarely, the top of
salt) and with seismic-refraction data (for mapping the top of the dome). In
addition, seismic velocities can be useful for estimating rock densities. The
reverse is also true; gravity data are used in prestack depth migration.
Chapter 9:  Geologic Applications  177
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Examples of seismic pitfalls


In the previous example (Figure 6), we see the importance of integrating
as much geophysical and geologic data as are available in a given project.
In part, this requirement derives from the problem of ambiguity (Chapter
3) inherent in gravity analysis. However, this need also results from other
ambiguities that are inherent in all geophysical methods. Tucker and Yor-
ston (1973) describe and illustrate a large number of geologic examples that
yield multiple interpretations if based on the seismic method alone.
The cross section shown in Figure 7a indicates the reflections in two-
way traveltime and the derived P-wave interval velocities of each horizon. If
the apparent rollover of the seismic reflections is caused geologically, then
the zone under and left of the fault could represent an attractive hydrocar-
bon play. However, it is also possible that this phenomenon is caused not by
geology but is an artifact of the shadow zone under the fault resulting from
changes in velocity (Tucker and Yorston, 1973).
The velocities shown in Figure 7a are converted into densities by using
Gardner’s equation (Chapter 5) and are reduced to the contrasts shown in
Figure 7b by removing a constant 2000 kg/m3. This is a sizable fault anom-
aly that will not be missed in either the seismic or the gravity data.
If the geometric rollover shown in Figure 7a does not exist in the geol-
ogy (i.e., an artifact of seismic processing), then the anomaly shown in
Figure 7b is the anomaly that would be observed in the gravity data. If,
however, the rollover does exist in the geology, then it would produce the
anomaly shown in Figure 7c after removing the large fault-caused anom-
aly shown in Figure 7b. This asymmetrical negative anomaly is caused by
the smaller density contrasts of the beds under the fault, which have been
caused to roll over.
The anomaly in Figure 7c exhibits a minimal amplitude of about –0.3 mGal,
which is definitely observable in most land surveys and is measurable in high-
quality marine work.
By contrast, if we apply the same modeling procedure to a reverse
fault (Tucker and Yorston, 1973) by using the data shown in Figure 8a, we
obtain the densities and density contrasts shown in Figure 8b and 8c, along
with the produced gravity anomalies. Although the fault anomaly shown in­
Figure 8b would not be difficult to measure, the fault rollover-induced resid-
ual shown in Figure 8c of only –0.06 mGal would be very difficult for all but
the h­ ighest-quality high-precision land surveys. The problem is made even
more difficult by the presence of other geologic sources.
These examples provide for supplementing seismic data analysis with
independent information about the prospect and can be used to determine
whether the gravity method is applicable to the project in the first place.
178  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 7. (a) Normal a)


0
fault as derived
from compressional- Interval velocity (ft/s)
6000
wave seismic data.
Horizontal dimension

Time (s)
of the seismic section 8500
0.5 6300
is approximately 2
miles. After Tucker 10,000
and Yorston (1973), 8900
Example 2-A, p. 10;
Example 2-B, p. 11. Seismic section 10,500
(b) Density contrasts 1.0
and gravity anomaly b)
derived from seismic
Gravity anomaly (mGal)

–20
velocities. Gravity
calculations are made
by extending the
horizons horizontally.
(c) Density contrasts –25
caused by shallower Distance (kft)
beds forced deeper 5 15
0
under the fault. 630
Density 650
Depth (kft)

kg/m3 460 440


3
370 340

c)
Gravity anomaly (mGal)

–0.1

–0.2

–0.3

5 Distance (kft) 15
0
Depth (kft)

3
Density contrasts
–190
(kg/m3) –90 –37
6
Chapter 9:  Geologic Applications  179
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a) Figure 8. (a) Reverse


Interval velocity (ft/s) fault as derived
0.8
9500 from seismic data.
Horizontal dimension
8400 of the seismic section
Time (s)

1.0
9900 is approximately 7
miles. After Tucker
8800 and Yorston (1973),
10,200
Example 3-A, p.
1.2 12; Example 3-B,
p. 13. (b) Densities
Seismic section
9200 and gravity anomaly
derived from seismic
b)
velocities shown in
Gravity anomaly (mGal)

Figure 8a. Gravity


calculations are made
121
by extending the
horizons horizontally.
120 (c) Density contrasts
and gravity anomaly
Distance (kft)
5 15
caused by shallower
Density Velocity Density Velocity beds forced deeper
(kg/m3) (ft/s) (kg/m3) (ft/s) under the fault.
2270 9500 2200 8400
Depth (kft)

4
2290 9900 2230 8800
2310 10,200 2250 9200
2670 2620

c)
Gravity anomaly (mGal)

–.03

–.06

Distance (kft)
5 15

Density contrasts (kg/m3)


–70
Depth (kft)

4
–60
–60
–50
8
180  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Example of borehole gravity


In Chapter 7, we derived the working equations for borehole gravity
surveys and discussed the apparent density and apparent density anomaly
based on borehole gravity measurements. We studied the density function
for a well that penetrates a reef with porous zones, both intersected by a well
and offset from it.
Figure 9 shows a case history of the density-difference approach (Ras-
mussen, 1975). This carbonate reef had been drilled and deemed to be non­-
BHGM

FDC (gamma-gamma)
BHGM — FDC (gamma-gamma)

Difference (g/cm3) Measured density (g/cm3)


2.5 2.55 2.60 2.65
–.15 –.10 –.05 0

6600

6620

6640

6650

6665
Depth (ft)

Zone of
Interest
6685

6700

6710

6730

6755

Figure 9. Borehole gravity-derived and gamma-gamma densities in a Michigan well.


Chapter 9:  Geologic Applications  181
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

productive, based on all other information. Before the well was to be plugged
and abandoned, the oil-company researchers persuaded the field engineers to
allow the logging of one of the wells with the borehole g­ ravity meter.
Observing the difference between the gamma-gamma-derived densities
and the gravity-derived densities shown in Figure 9, the researchers decided
that a zone just below 6665 ft would be promising. They set isolation pack-
ers above and below the zone, perforated the well, and injected acid. The
well began to produce at the rate of 500 barrels of condensate and 52 million
ft3 of gas per day. This was one of the major successes of the borehole grav-
ity meter (Bradley, J. W., personal communication, 1975).

Borehole gravity in combination with surface gravity


The apparent-density anomaly as measured in a well by the borehole grav-
ity meter can identify a zone of interest in terms of its depth, but this anomaly
does not yield any information about the horizontal direction from the well of
the causative body. In Chapter 7, we studied methods for determining how far
from the well a source can exist and still obtain a measurable response. Again,
this gives no information about the horizontal location of the geologic feature.
In Figure 10, we depict a well that shows two locations of anomalous
density behavior derived from borehole gravity data. The shallower of the
two correlates with an anomaly in the surface gravity field left of the well,
whereas the deeper of the two correlates with an anomaly in the surface
gravity data right of the well. (The 2D format is used to describe how the

Total mass from surface anomaly Figure 10. Schematic only,


showing surface gravity in
combination with borehole
gravity.

Volume
from
BHGM
anomaly
Density = mass/volume and porosity from integration.
182  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

two gravity data sets are used in combination. In an actual field example,
mapped data would indicate locations in any horizontal direction.)
The volume of the source can be estimated by vertical and horizontal
limits derived from the borehole data and surface gradients, respectively,
whereas the total amount of anomalous mass can be estimated by applica-
tion of Gauss’ theorem (Chapter 7) to the surface anomalies, together yield-
ing an estimate of the anomalous density of the source rock.

Integration of seismic and/or magnetic


information with gravity data
Seismic, magnetic, electrical, and gravity surveys often complement
one another, each yielding independent information about the subsurface.
In rugged terrain, seismic surveys are expensive and/or very difficult and
time-consuming to acquire.
A good example is the Wyoming overthrust belt (Figure 11), where
extensive gravity surveys are used to extend seismically derived geologic
structure into regions where seismic data have not been obtained. This
might help to justify an increase of the seismic budget to enable additional
data acquisition or to help position well sites that are not confirmed directly
by the existing seismic data.
Seismic reflections and interval velocities calculated for relatively hori­
zontal formations often deteriorate with increasing geologic structure because
of geometric dispersion of energy. Seismic data are often but not always
excellent if the beds are flat. In such beds, no change in gravity occurs

Figure 11. Schematic Wyoming overthrust belt


only, projecting Seismic-constrained gravity
seismically derived interpretation extended into
regions between and beyond
structure using gravity
seismic control
data.

ly
anoma
uguer gravity
observed Bo
ted and
Calcula

izon
ic hor
Seism
Chapter 9:  Geologic Applications  183
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

laterally if and only if the densities are uniform. In these cases of flat-lying
geology, the opportunity to determine formation density by using interval
velocities can be excellent. If for various reasons the seismic data are not
of high quality, it might not be possible to determine formation densities by
integrating seismic with gravity data.
With the occurrence of folding and faulting, flat beds are ­transformed
into far more interesting and productive structural and stratigraphic
traps, with an attendant degradation of the interval-velocity calculation
and therefore the determination of density. However, it is often safe to
project the densities derived where the beds are flat into the adjacent
areas where they are not and to use those densities in the modeling of
gravity anomalies in the more interesting (and more complicated) geo-
logic regions.
In some regions that contain massive salt formations, problems can
occur in the interpretation of seismic events. For example, strong seismic
reflections issuing from the top of a salt horizon might not be supported
by any comparable reflections from the base of salt or the base-of-the-salt
reflections might grade from poor to nonexistent. However, salt usually
exhibits strong density contrasts with respect to the surrounding rocks
(unless in the crossover zone discussed above). In such cases, forward
gravity modeling often can lead to the conclusion that a seismically de-
rived uplift in the salt is a bed with relatively uniform thickness. Alterna-
tively, gravity modeling might suggest a zone where the thickness of the
salt increases dramatically (as in the case of a salt dome, stock, or keel).
Similarly, gravity modeling, constrained by seismic information where
seismic data are of high quality, can be continued through zones of poor seis-
mic record quality or at least can add confirmation and source-identification
information to the seismic interpreter’s conclusion. Hence, gravity interpre-
tation can help to reduce exploration risk.
Gravity and magnetic data often complement one another, especially in
regions where the economic basement is also of high density and/or mag-
netic susceptibility and where volcanic rocks comprise part of the geologic
sequence.

Mining applications
Neves-Corvo massive sulfide deposit, Portugal
The densities of massive sulfide deposits contrast sharply with those
of host rocks, which makes gravity prospecting an attractive exploration
tool for such targets. However, because of cost and the relatively slow rate
184  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

of coverage of ground surveys over large areas, gravity surveys in mining


applications generally are used as a follow-up to anomalies detected by
other methods. However, there have been exceptions when gravity was used
to obtain systematic coverage over large areas with ground or airborne sur-
veys, leading to the discovery of significant ore deposits.
The potential of a regional gravity survey is illustrated best by the dis-
covery of the Neves-Corvo Group of massive sulfide deposits in the Por-
tuguese pyrite belt after the completion of a regional survey on 100-m and
200-m grids (Leca, 1990). The large sulfide deposit at Neves-Corvo (> 300
million tonnes) yielded a small gravity anomaly of 0.4 to 0.6 mGal near the
village of Neves (Figure 12).
The deposit occurs within an exposed part of the Iberian pyrite belt in
a volcanic sedimentary section and is buried at depths ranging from 300

N
0 500 1000

(m)

Neves
Corvo

Graca

Zambujal

Figure 12. Regional Bouguer gravity map around the village of Neves, Portugal,
extracted from the regional gravity survey of the larger Algaré region. The contour
interval is 0.1 mGal, and the Bouguer reduction density was 2.5 g/cm3. The left side
of the map (solid lines) was surveyed using a grid 100 m × 100 m, whereas the right
side (dashed lines) was surveyed on a grid 250 × 125 m. After Leca (1990), Figure 3.
Chapter 9:  Geologic Applications  185
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

to 700 m (Figure 13). The deposit density varied between 2900 and 3100
kg/m3, whereas the density of surrounding rocks was 2500 kg/m3. Gravity
played a key role by indicating the presence of excess mass, despite the fact
that the deposit was discovered as a result of a multidisciplinary exploration
program, with geochemical anomalies indicating possible sulfide sources
downdip. Leca (1990) states that without geophysics, particularly without
the distinct Bouguer anomaly, Neves-Corvo would not have been discovered.

The Abra base-metal deposit, Western Australia


The Abra base-metal deposit is a deeply buried, large, low-grade body
that has no surface geologic or geochemical expression. An aeromagnetic
survey flown along lines spaced 400 m apart detected a bull’s-eye 400-nT
anomaly. Based on the data provided by a later airborne survey along lines

mGal
0.3 Gravity

0
Ground surface

100

200

300
Graywacke
Various
schists
Tuffite
Purple/black
400 schists
Massive
sulfide
Acidic tuff
100 m
500
m

Figure 13. Geologic cross section over the Neves-Corvo massive sulfide deposit
and the associated gravity anomaly. After Leca (1990), Figure 10.
186  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

spaced 200 m apart and a follow-up with a ground magnetic survey, a 200-m
percussion hole was drilled in 1977 with inconclusive results.
In 1981, a more detailed ground magnetic survey was undertaken, along
with a gravity survey. The gravity survey was conducted on a grid of 100 m ×
50 m and showed a gravity high superimposed on a sloping regional field.
After subtracting from observed gravity data a regional component, deter-
mined from the widely spaced data available at the Australian Geological
Survey Organization, the residual magnetic and gravity data were modeled
using a proprietary ellipsoid modeling program.
The modeling result showed the target to be ellipsoidal with a thickness
of 300 m, lateral dimensions of 600 × 1000 m, and a depth to the top of 270
m (Figure 14). A subsequent hole based on this interpretation intersected
200 Mt of subeconomic mineralization of iron, barium, lead, silver, copper,
and gold at a depth of 260 m.
This survey shows the utility of gravity surveys in screening and better
delineating a target defined by other geophysical methods.

Kimberlite exploration
Kimberlite is a type of potassic volcanic rock best known for sometimes
containing diamonds. Airborne magnetic and electromagnetic surveys, along
with mineral geochemistry, have been standard tools for exploration for kim-
berlite. Airborne gravity gradiometry was added recently to the existing arse-
nal and has helped to better identify kimberlite targets. The targets generally
are characterized by higher electrical conductivity and lower density and less
often by higher magnetic susceptibility and/or remanent magnetization.
The example shown in Figure 15 was flown over the central part of the
Ekati tenement in the Lac de Gras kimberlite province in Northwest Ter-
ritories, Canada (Rajagopalan et al., 2007). Kimberlite intrusions in that
area are often but not always associated with a crater lake. The area was
flown with a fixed-wing and a helicopter Falcon gravity-gradient system.
Data were acquired on east-west lines spaced 100 m apart and at a nominal
survey clearance of 80 m for fixed-wing surveys and 50 m for heli­copter
surveys.
In addition to gravity-gradiometry data, the survey also included mea-
surements of electromagnetic and magnetic data. The fixed-wing airborne
gravity-gradiometry data showed that more than half of the known kimber-
lites have associated gravity anomalies.
Figure 16 shows a comparison of fixed-wing and helicopter airborne
gravity gradiometry over the Ekati kimberlite pipe. This kimberlite would
not have been selected as a target in the original data, whereas in the
Chapter 9:  Geologic Applications  187
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

South Total magnetic intensity North nT


a)
55,600

55,200

54,800

mGal
b) Bouguer gravity anomaly 6

l
iona
Reg
2

0
c) nT
Total magnetic intensity 55,600

55,200
Model
Observed
(filtered) 54,800
d) mGal
Residual gravity 1.2

Observed 0.0

Model
–1.2
AB3 m
e)
0
Mineralization

Ellipsoid
model 500

1000 m
1000

Figure 14. Profiles of magnetic and gravity data modeled using a single ellipsoid.
After McInerney et al. (1994), Figures 4 and 6.
188  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 15. Vertical gravity gradient. All pipes (diamond symbols) shown here, with the
exception of Kaspa, are associated with gravity-gradient anomalies. The pipe anomaly
is accentuated by the presence of water. Target A has a weak gzz anomaly, whereas
target B appears to be more typical of a deep-lake response. After Rajagopalan et al.
(2007), Figure 2.

G00: EW grid profile


Fixed wing Helicopter Eötvös

–50
500 m

Fixed wing
Helicopter
Figure 16. Comparison of fixed-wing and helicopter airborne gravity gradiometry
over the Ekati kimberlite pipe, Northwest Territories, Canada. Both methods
detect the pipe, but the helicopter system shows higher resolution. After Reed and
Witherly (2007), Figure 17. Used by permission.
Chapter 9:  Geologic Applications  189
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

helicopter data, it shows as a standout anomaly with an amplitude of 100


Eötvös.
The geophysical data for the selected area show that no single method
would have identified all known pipes, but all known pipes would have been
discovered by integrating all three data sets: electromagnetic, gravity gradi-
ometry, and magnetics. The combined use led directly to generation of new
targets and discovery of new pipes.

Time-lapse (4D-gravity) surveys


Time-lapse gravity surveys (or 4D, in which the fourth dimension is time)
were used to counter decreasing reservoir pressures in the Prudhoe Bay reser-
voir in Alaska, where a water-injection program initiated in November 2002
was designed to operate for approximately 20 years. The major monitoring
concern with the waterflood is to ensure that water added in the gas cap does
not prematurely flow downdip into the oil-producing portions of the field,
where it could interfere with a highly efficient gravity drainage mechanism.
Using gravity modeling of reservoir simulations, Hare et al. (1999) cal-
culate that an overall measurement precision of about 10 mGal is required
in the time-differenced gravity to effectively monitor the waterflood. From
1994 through 2002, a series of field experiments was conducted to develop
a suitable technique for acquisition of 4D microgravity data in the Arctic
(Ferguson et al., 2007; Ferguson et al., 2008; Hare et al., 2008). The reader
is referred to those papers for a thorough understanding of field procedures
and problems encountered in obtaining ­reliable data.
Since 2003, the Prudhoe Bay 4D microgravity surveys have been con-
ducted by exclusively using absolute-gravity meters because they do not
drift like relative-gravity meters and can be calibrated to precise time and
distance standards compatible with international standards. Approximately
300 stations in a 150-km2 area were reoccupied in each survey year, with
absolute-gravity measurements of better than 5-mGal precision and GPS
geodetic measurements that have centimeter precision. Because the res-
ervoir is at a depth of 2.5 km, the expected anomalies will be broad and
smooth. With more than half the stations in the bay and the others on land, it
was necessary to conduct the survey during winter when the bay and tundra
are frozen and safe for vehicle transportation.
A complete description of field techniques, applied corrections, and
modeling results can be found in Ferguson et al. (2008). See also the spe-
cial section on 4D gravity monitoring in Geophysics (Biegert et al., 2008).
Figure 17 shows the 4D gravity maps for all possible time intervals using
the 2003, 2005, 2006, and 2007 surveys at Prudhoe Bay. The 4D gravity
190  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

measured over various epochs (2005–2003, 2006–2003, and 2007–2003)


successfully modeled and tracked the mass of water injected since late 2002.

Satellite gravity and satellite-derived gravity


Measuring gravity from satellites has the advantage of covering large
areas that could be difficult to access otherwise. However, because the mini-
a) d)

b) e)

70

c) f) 56

42
Gravity (µGal)

28

14

–14

–28

Figure 17. 4D gravity maps for epochs (a) 2005–2003, (b) 2006–2003, (c)
2007–2003, (d) 2006–2005, (e) 2007–2005, and (f) 2007–2006 at Prudhoe Bay,
Alaska. After Ferguson et al. (2008), Figure 5.
Chapter 9:  Geologic Applications  191
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

mum altitude at which a satellite can orbit is 160 km, the resulting gravity
anomalies will be insensitive to shorter-wavelength variations.
Kaula (1987) divides the satellite systems for gravimetry into four cate-
gories:

1) displacement of a satellite in orbit relative to a tracking station on the


ground — CHAMP
2) displacement of a satellite and the sea surface with respect to each other —
SEASAT, TOPEX, GEOSAT, ERS-1
3) displacement of two satellites with respect to each other — GRACE
4) relative displacement of two masses within a single satellite — GOCE

The first category is the principal one used for much of the land area of
the earth, although the nonuniform distribution of tracking stations presents
problems sometimes.
Converting sea-surface height variations, derived from satellite altimetry,
to free-air gravity is not new. GEOSAT and ERS-1 satellites have provided use-
ful data. However, the advent of publicly available, worldwide satellite gravity
and topographic data sets for offshore areas has made a substantial impact on
worldwide exploration. The explorationist is no longer limited by widely spaced
academic or expensive speculative gravity surveys with limited coverage.
Instead, satellite-derived gravity is now used routinely to get a first-pass
look at the gravity field around the world and to define basin structures that
exhibit anomalies with wavelengths as short as 15 to 20 km (Sandwell and
Smith 1997, 2001). Plate boundaries, transforms, sediment loads, regional
basin structure, and crustal boundaries can be evaluated easily with a data
set of relatively consistent quality and spacing. Free-air gravity data can
be quickly downloaded, gridded, mapped, and used in conjunction with
publicly available topographic data to create Bouguer anomaly grids and
maps. Free-air and Bouguer grids then can be filtered and modeled to further
define tectonic and structural features.
The example displayed here is from offshore Brazil (Figures 18, 19, and
20). Seafloor transforms, the margin free-air anomaly, Florianopolis Ridge,
the Abrolhos volcanic complex, and so forth can be identified readily.
In combination with regional sediment-thickness grids and by either
assuming an average density or using a simple density-versus-depth func-
tion (such as Sclater and Christie [1980] or Chappell and Kusznir [2008a]),
relatively simple 2D or 3D gravity models can be constructed and inverted
to yield estimated depth to Moho and crustal-thickness maps. More complex
approaches that incorporate thermal corrections resulting from lithospheric
thinning (Chappell and Kusznir, 2008b) also can be used. The results of
192  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

–45 –40 –35

mGal
–20 –20 36.5 25.9
19.5
9.9 14.4
5.6
1.9 –1.6
–5.0
Degrees latitude

–8.8
–12.1 –15.6
–19.2
–25 –27.6 –23.4
–25 –32.3
–38.0 –46.3
–60.2

–N–

–30 –30
0 200 400
km

–45 –40 –35


Degrees longitude

Figure 18. Free-air anomaly. The black line represents the coastline of Brazil.
Courtesy of Guy Flanagan. Used by permission.

–45 –40 –35

mGal
–20 –20 278.2
210.1 235.7
189.5
171.8 154.3
139.5 125.4
111.8 96.8
Degrees latitude

83.2
55.1 69.5
–25 21.4 38.2
–25 2.3
–20.4 –53.8
–109.6

–N–

0 200 400
–30 –30 km

–45 –40 –35


Degrees longitude

Figure 19. Bouguer anomaly. Courtesy of Guy Flanagan. Used by permission.


Chapter 9:  Geologic Applications  193
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

–45 –40 –35


400-km high-pass-filtered
Bouguer gravity

Abrolhos volcanics

Vitória-Trinidade chain
–20 –20 mGal
Free-air effect 26.1

sin
19.8

Ba
16.0
13.0

os
10.4
mp
7.8
5.6
Ca
3.6
Degrees latitude

Transforms 1.5
sin

–0.7
Ba

gh

–2.7
hi

–4.7
os

–25 –25 –6.8


er
nt

–9.3
ut
Sa

–11.8
–14.6
–17.9
–22.9
–31.1
Possible
failed rift
–N–
0 400
–30 –30
km
Florianopolis Ridge

–45 –40 –35


Degrees longitude

Figure 20. Bouguer anomaly after applying a 400-km high-pass filter. Courtesy of
Guy Flanagan. Used by permission.

such analysis are essential inputs to basin modeling and thermal calcula-
tions, which are used to determine hydrocarbon thermal maturity, a key risk
component in exploration. 
Figure 21 shows an example of a simple 2D data inversion. Inputs were
regional bathymetry, regional basement, and an initial assumed depth to
Moho near the coast of 35 km with single densities of 1030, 2400, 2850,
and 3300 kg/m3, respectively, for water, sediments, crust, and mantle. As
can be seen from the inversion, one can easily obtain an initial idea of the
variable thickness and nature of crustal thinning along the profile by using
freely available public data.
194  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

a)
Free-air gravity (mGal)

0
Observed
–85
Calculated
–170

–255

Distance (km)
0 98 196 294 392
0
Water ρ = 1030 kg/m3
Sediments ρ = 2400
Depth (km)

10

20
Crust ρ = 2850
30

Mantle ρ = 3330
b)
Free-air gravity (mGal)

0 Calculated
Observed
–85

–170

–255

0
Water ρ = 1030 kg/m3
Sediments ρ = 2400
10
Depth (km)

Crust ρ = 2850
20
Mantle ρ = 3330
30

40

Figure 21. An example of data inversion offshore West Africa, (a) preinversion
and (b) postinversion. Density is expressed in kilograms per cubic meter. Courtesy
of Guy Flanagan. Used by permission.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Appendix A

Fourier Transform

The Fourier transform is a mathematical operation that decomposes a


function into its constituent frequencies. This appendix will give a brief
review of Fourier-transform theory as a tool for mapping functions of time
or distance (space) into functions of frequency or wavenumber.
In gravity exploration, we are concerned mostly with functions that
depend on distance (e.g., gravity profiles or maps), in contrast with seis-
mic exploration, in which the main output (seismic-record trace) is rep-
resented as a function of time. For a complete coverage of this topic, the
reader is referred to the excellent textbooks by Papoulis (1962) and by
Bracewell (1965).
The Fourier transform F(w) of a function f(x) is defined as

F (ω ) = ∫
−∞
f ( x ) e − i ω x dx, (A-1)

with the requirement that the absolute value of the function f(x) satisfies the
condition

∫ f ( x ) dx < ∞.
−∞
For functions that depend on distance (space), the variable w above is
called a wavenumber, and it has units of inverse distance. It is related to
wavelength l by the relation

2p
ω= .
λ

195
196  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

For functions of time instead of space, the variable w is the angular fre-
quency and is related to frequency f by the relation

w  =  2pf.

The Fourier transform F(w) is generally a complex function which has


the property that F(0) represents the total area under the curve f(x) over the
interval of integration.
The Fourier transform given by equation A-1 has an associated inverse
operation (the inverse Fourier transform). We have


1

∫ F (ω ) e dω .
iω x
f (x) = (A-2)
2p −∞

The Fourier transform can be generalized easily for multidimensional


functions. For a 2D function f (x, y), the corresponding Fourier transform
expressions are

∞ ∞

F (u, v ) = ∫ ∫
−∞ −∞
f ( x , y) e − i ( u x + v y ) dx dy,

∞ ∞
1
∫ ∫ F (u, v) e
i (u x + v y )
f ( x , y) = du dv , (A-3)
4p 2 −∞ −∞

where u and v are the wavenumbers in the x- and y-directions, respec­tively.


The Fourier transform has some interesting properties that are used for
various derivations in the main text of this book. The proofs are straightfor-
ward and usually involve integration by parts.
Letting the symbol ↔ represent Fourier transform, we can prove the
following properties:
Symmetry — If f (x) ↔ F(w) and if f (x) is a real function, then the real
part of F(w) is a symmetrical function with respect to w = 0, whereas the
imaginary part is antisymmetrical.
Linearity — If f (x) ↔ F(w) and g(x) ↔ G(w), then

a f (x) + b g(x) ↔ a F(w) + b G(w),

where a and b are arbitrary constants.


Appendix A:  Fourier Transform  197
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Scaling — If f (x) ↔ F(w), then

1 ω
f (ax ) ↔ F ( ), (A-4)
a a
where a is an arbitrary constant.
Shifting — If f (x) ↔ F(w), then

f ( x − x 0 ) ↔ F (ω ) e − i ω x0 . (A-5)

Differentiation — If f (x) ↔ F(w), then

dn
n
f ( x ) ↔ (iω )n F (ω ). (A-6)
dx

Convolution theorem — The convolution of two functions f (x) and


g(x) leads to a function h(x), defined as

h( x ) = ∫
−∞
f (t ) g( x − t ) dt.

Taking the Fourier transform of h(x), one immediately obtains

H(w)  =  F(w) G(w). (A-7)

Parseval’s formula — Parseval’s theorem states that the sum (or inte-
gral) of the square of a function is equal to the sum (or integral) of the square
of its transform:

∞ 2 ∞ 2

2p ∫
−∞
f ( x ) dx = ∫
−∞
F (ω ) dω . (A-8)

The above expressions can be generalized easily for 2D functions of the


form f (x, y). As an example, expression A-6 becomes

dn
f ( x , y) ↔ (iu)n F (u, v ),
d xn

dn
n
f ( x , y) ↔ (iv )n F (u, v ), (A-9)
dy
198  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

where u and v are again the wavenumbers in the x- and y-directions, respec-
tively.
Similarly, by defining the 2D convolution as
∞ ∞
h ( x , y) = ∫∫
−∞ −∞
f ( p, q) g (x − p, y − q) dp dq ,

one obtains

H(u, v) = F(u, v) G(u, v). (A-10)


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

References

Airy, G. B., 1855, On the computations of the effect of the attraction of the mountain
masses as disturbing the apparent astronomical latitude of stations in geodetic
surveys: Transactions of the Royal Society [London], ser. B, 145, 101–104.
Al-Chalabi, M., 1971, Some studies relating to nonuniqueness in gravity and mag-
netic inverse problems: Geophysics, 36, 835–855, http://dx.doi.org/10.1190/1
.1440219.
Åm, K., 1972, The arbitrary magnetized dyke: Interpretation by characteristics:
Geoexploration, 10, no. 2, 63–90, http://dx.doi.org/10.1016/0016-7142(72)
90014-2.
Barbosa, V. C. F., J. B. C. Silva, and W. E. Medeiros, 1997, Gravity inversion of
basement relief using approximate equality constraints on depths: Geophysics,
62, 1745–1757, http://dx.doi.org/10.1190/1.1444275.
Barnett, C. T., 1976, Theoretical modeling of the magnetic and gravitational fields
of an arbitrarily shaped three-dimensional body: Geophysics, 41, 1353–1364,
http://dx.doi.org/10.1190/1.1440685.
Biegert, E. K., J. Ferguson, and X. Li, eds., 2008, Special section on 4D gravity
monitoring: Geophysics, 73, no. 6, WA1–WA180.
Blakely, R. J., 1995, Potential theory in gravity and magnetic applications: Cam­
bridge University Press.
Bott, M. H. P., 1960, The use of rapid digital computing methods for direct gravity
interpretation of sedimentary basins: Geophysical Journal of the Royal Astro­
nomical Society, 3, no. 1, 63–67, http://dx.doi.org/10.1111/j.1365-246X.1960
.tb00065.x.
Bott, M. H. P., and R. A. Smith, 1958, The estimation of the limiting depth of
gravitating bodies: Geophysical Prospecting, 6, no. 1, 1–10, http://dx.doi.org
/10.1111/j.1365-2478.1958.tb01639.x.
Bracewell, R., 1965, The Fourier transform and its applications: McGraw-Hill.
Bradley, J. W., 1974, The commercial application and interpretation of the bore-
hole gravimeter, in Contemporary geophysical interpretation: A symposium:
Geophysical Society of Houston.

199
200  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Chai, Y., and W. J. Hinze, 1988, Gravity inversion of an interface above which
the density contrast varies exponentially with depth: Geophysics, 53, 837–845,
http://dx.doi.org/10.1190/1.1442518.
Chapin, D. A., 1997, Wavelet transforms: A new paradigm for interpreting grav-
ity and magnetics data?: 67th Annual International Meeting, SEG, Expanded
Abstracts, 486–489, http://dx.doi.org/10.1190/1.1885940.
Chapin, D. A., 2000, Gravity measurements on the moon: The Leading Edge, 19,
no. 1, 88–90, http://dx.doi.org/10.1190/1.1438472.
Chappell, A., and N. Kusznir, 2008a, An algorithm to calculate the gravity anom-
aly of sedimentary basins with exponential density-depth relationships: Geo-
physical Prospecting, 56, no. 2, 249–258, http://dx.doi.org/10.1111/j.1365-
2478.2007.00674.x.
Chappell, A. R., and N. J. Kusznir, 2008b, Three-dimensional gravity inversion for
Moho depth at rifted continental margins incorporating a lithosphere thermal
gravity anomaly correction: Geophysical Journal International, 174, no. 1, 1–
13, http://dx.doi.org/10.1111/j.1365-246X.2008.03803.x.
Cheng, D., Y. Li, and K. Larner, 2003, Inversion of gravity data for base salt: 73rd
Annual International Meeting, SEG, Expanded Abstracts, 588–591, http://dx
.doi.org/10.1190/1.1817995.
Clark, S. P. Jr., ed., 1966, Handbook of physical constants, rev. ed.: Geological
Society of America Memoir 97.
Cordell, L., 1979, Gravimetric expression of graben faulting in Santa Fe coun-
try and the Española Basin, New Mexico: New Mexico Geological Society
Guidebook, 30th Field Conference, 59–64.
Cordell, L., and R. G. Henderson, 1968, Iterative three-dimensional solution of
gravity anomaly data using a digital computer: Geophysics, 33, 596–601,
http://dx.doi.org/10.1190/1.1439955.
Cordell, L., and V. J. S. Grauch, 1982, Mapping basement magnetization zones
from aeromagnetic data in the San Juan Basin, New Mexico: 52nd Annual
International Meeting, SEG, Expanded Abstracts, 246–247.
Cowan, D. R., and S. Cowan, 1993, Separation filtering applied to aeromagnetic
data: Exploration Geophysics, 24, no. 4, 429–436, http://dx.doi.org/10.1071
/EG993429.
Craig, M., 1996, Analytic signals for multivariate data: Mathematical Geology, 28,
no. 3, 315–329, http://dx.doi.org/10.1007/BF02083203.
Danes, Z. F., 1960, On a successive approximation method for interpreting gravity
anomalies: Geophysics, 25, 1215–1228, http://dx.doi.org/10.1190/1.1438809.
Delaney, J. P., 1940, Leonardo da Vinci on isostasy: Science, 91, no. 2371, 546–
547, http://dx.doi.org/10.1126/science.91.2371.546.
Dransfield, M., A. Christensen, M. Rose, P. Stone, and P. Diorio, 2001, FALCON
test results from the Bathurst mining camp: Exploration Geophysics, 32, no. 4,
243–246, http://dx.doi.org/10.1071/EG01243.
References  201
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Elkins, T. A., 1951, The second derivative method of gravity interpretation: Geo­
physics, 16, 29–50, http://dx.doi.org/10.1190/1.1437648.
Evjen, H. M., 1936, The place of the vertical gradient in gravitational interpreta-
tions: Geophysics, 1, 127–136, http://dx.doi.org/10.1190/1.1437067.
Fairhead, J. D., and M. E. Odegard, 2002, Advances in gravity survey resolution:
The Leading Edge, 21, no. 1, 36–37, http://dx.doi.org/10.1190/1.1445845.
Ferguson, J. F., T. Chen, J. Brady, C. L. V. Aiken, and J. Seibert, 2007, The 4D
microgravity method for waterflood surveillance II — Gravity measurements
for the Prudhoe Bay reservoir, Alaska: Geophysics, 72, no. 2, I33–I43, http://
dx.doi.org/10.1190/1.2435473.
Ferguson, J. F., F. J. Klopping, T. Chen, J. E. Seibert, J. L. Hare, and J. L. Brady,
2008, The 4D microgravity method for waterflood surveillance: Part 3 — 4D
absolute microgravity surveys at Prudhoe Bay, Alaska: Geophysics, 73, no. 6,
WA163–WA171, http://dx.doi.org/10.1190/1.2992510.
Fixler, J. B., G. T. Foster, J. M. McGuirk, and M. A. Kasevich, 2007, Atom interferom­
eter measurement of the Newtonian constant of gravity: Science, 315, no. 5808,
74–77, http://dx.doi.org/10.1126/science.1135459.
Gardner, G. H. F., L. W. Gardner, and A. R. Gregory, 1974, Formation velocity
and density — The diagnostic basics for stratigraphic traps: Geophysics, 39,
770–780, http://dx.doi.org/10.1190/1.1440465.
Geodetic Reference System, 1967: International Association of Geodesy Special
Publication No. 3.
Gilbert, R. L. G., 1949, A dynamic gravimeter of novel design: Proceedings of the
Physical Society of London, 62B, 445–454.
Golub, G. H., M. Heath, and G. Wahba, 1979, Generalized cross-validation as a
method for choosing a good ridge parameter: Technometrics, 21, no. 2, 215–
223, http://dx.doi.org/10.1080/00401706.1979.10489751.
Goodell, R. R., and C. H. Fay, 1964, Borehole gravity meter and its application:
Geophysics, 29, 774–782, http://dx.doi.org/10.1190/1.1439417.
Goupillaud, P., A. Grossmann, and J. Morlet, 1984, Cycle-octave and related trans-
forms in seismic signal analysis: Geoexploration, 23, no. 1, 85–102, http://
dx.doi.org/10.1016/0016-7142(84)90025-5.
Gradshteyn, I. S., and I. M. Ryzhik, 1980, Tables of integrals, series, and products:
Academic Press.
Guier, W. H., and R. R. Newton, 1965, The earth’s gravity field as deduced from the
Doppler tracking of five satellites: Journal of Geophysical Research, 70, no. 18,
4613–4626, http://dx.doi.org/10.1029/JZ070i018p04613.
Guillen, A., and V. Menichetti, 1984, Gravity and magnetic inversion with mini-
mization of a specific functional: Geophysics, 49, 1354–1360, http://dx.doi
.org/10.1190/1.1441761.
Guspí, F., 1992, Three-dimensional Fourier gravity inversion with arbitrary density
contrast: Geophysics, 57, 131–135, http://dx.doi.org/10.1190/1.1443176.
202  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Hall, A. R., 1963, From Galileo to Newton: Dover Publications.


Hammer, S., 1939, Terrain corrections for gravimeter stations: Geophysics, 4, 184–
194, http://dx.doi.org/10.1190/1.1440495.
Hansen, P. C., 1992, Analysis of discrete ill-posed problems by means of the L-curve:
SIAM Review, 34, no. 4, 561–580, http://dx.doi.org/10.1137/1034115.
Hansen, R. O., and X. Wang, 1988, Simplified frequency-domain expressions for
potential fields of arbitrary three-dimensional bodies: Geophysics, 53, 365–
374, http://dx.doi.org/10.1190/1.1442470.
Hare, J. L., J. F. Ferguson, C. L. V. Aiken, and J. L. Brady, 1999, The 4-D micro­
gravity method for waterflood surveillance: A model study for the Prudhoe Bay
reservoir, Alaska: Geophysics, 64, 78–87, http://dx.doi.org/10.1190/1.1444533.
Hare, J. L., J. F. Ferguson, and J. L. Brady, 2008, The 4D microgravity method for
waterflood surveillance: Part IV — Modeling and interpretation of early epoch
4D gravity surveys at Prudhoe Bay, Alaska: Geophysics, 73, no. 6, WA173–
WA180, http://dx.doi.org/10.1190/1.2991120.
Haxby, W. F., G. D. Karner, J. L. LaBrecque, and J. K. Weissel, 1983, Digital imag-
es of combined oceanic and continental data sets and their use in tectonic stud-
ies: Eos, Transactions, American Geophysical Union, 64, no. 52, 995–1004,
http://dx.doi.org/10.1029/EO064i052p00995.
Heiskanen, W. A., and H. Moritz, 1967, Physical geodesy: W. H. Freeman & Co.
Heiskanen, W. A., and F. A. Vening Meinesz, 1958, The earth and its gravity field:
McGraw-Hill Book Company, Inc.
Holstein, H., 2003, Gravimagnetic anomaly formulas for polyhedra of s­patially
linear media: Geophysics, 68, 157–167, http://dx.doi.org/10.1190/1.1543203.
Hood, P., 1965, Gradient measurements in aeromagnetic surveying: Geophysics,
30, 891–902, http://dx.doi.org/10.1190/1.1439666.
Hornby, P., F. Boschetti, and F. G. Horowitz, 1999, Analysis of potential field data
in the wavelet domain: Geophysical Journal International, 137, no. 1, 175–196,
http://dx.doi.org/10.1046/j.1365-246x.1999.00788.x.
Howell, L. G., K. O. Heintz, and H. Barry, 1962, The development of a high-
precision downhole gravity meter: Geophysics, 31, 764–772, http://dx.doi.
org/10.1190/1.1439808.
Jacobsen, B. H., 1987, A case for upward continuation as a standard separation
filter for potential field maps: Geophysics, 52, 1138–1148, http://dx.doi.org
/10.1190/1.1442378.
Jekeli, C., 1988, The gravity gradiometer survey system (GGSS): Eos, Transactions,
American Geophysical Union, 69, no. 8, 116–117.
Jorgensen, G., and J. Kisabeth, 2000, Joint 3-D inversion of gravity, magnetic,
and tensor gravity fields for imaging salt formations in the deepwater Gulf of
Mexico: 70th Annual International Meeting, SEG, Expanded Abstracts, 424–
426, http://dx.doi.org/10.1190/1.1816085.
References  203
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Jung, K., 1961, Schwerkraftverfahren in der angewandten geophysik: Akademische


Verlagsgesellschaft, Geest & Portig K. G.
Kaula, W. M., 1987, Satellite measurement of the earth’s gravity field, in C. G.
Sammis and T. L. Henyey, eds., Methods of experimental physics, 24, Part B,
Geophysics — Field measurements: Academic Press, 163–188.
Krahenbuhl, R., and Y. Li, 2002, Gravity inversion using a binary formulation: 72nd
Annual International Meeting, SEG, Expanded Abstracts, 755–758, http://dx
.doi.org/10.1190/1.1817367.
Krahenbuhl, R., and Y. Li, 2004, Hybrid optimization for a binary inverse problem:
74th Annual International Meeting, SEG, Expanded Abstracts, 782–785, http://dx
.doi.org/10.1190/1.1851320.
LaCoste, L., N. Clarkson, and G. Hamilton, 1967, LaCoste and Romberg stabil­
ized platform shipboard gravity meter: Geophysics, 32, 99–109, http://dx.doi.
org/10.1190/1.1439860.
LaCoste, L. J. B., 1934, A new type of long period vertical seismograph: Physics, 5,
no. 7, 178–180, http://dx.doi.org/10.1063/1.1745248.
LaCoste, L. J. B., 1967, Measurements of gravity at sea and in the air: Reviews of
Geophysics, 5, no. 4, 477–526, http://dx.doi.org/10.1029/RG005i004p00477.
LaFehr, T. R., 1964, Gravity analysis of anomalous density distributions with ap-
plications in the southern Cascade Range: Ph.D. thesis, Stanford University.
LaFehr, T. R., 1965, The estimation of the total amount of anomalous mass by
Gauss’s theorem: Journal of Geophysical Research, 70, no. 8, 1911–1919,
http://dx.doi.org/10.1029/JZ070i008p1991.
LaFehr, T. R., 1983, Rock density from borehole gravity surveys: Geophysics, 48,
341–356, http://dx.doi.org/10.1190/1.1441472.
LaFehr, T. R., 1991, An exact solution for the gravity curvature (Bullard B) correc-
tion: Geophysics, 56, 1179–1184, http://dx.doi.org/10.1190/1.1443138.
LaFehr, T. R., and L. L. Nettleton, 1967, Quantitative evaluation of a stabilized
platform shipboard gravity meter: Geophysics, 32, 110–118, http://dx.doi.org
/10.1190/1.1439849.
Last, B. J., and K. Kubik, 1983, Compact gravity inversion: Geophysics, 48, 713–
721, http://dx.doi.org/10.1190/1.1441501.
Leca, X., 1990, Discovery of concealed massive-sulphide bodies at Neves-Corvo,
southern Portugal — A case history: Transactions of the Institution of Mining
and Metallurgy, Section B, Applied Earth Science, 99, B139–B152.
Li, X., 2001, Vertical resolution: Gravity versus vertical gravity gradient: The
Leading Edge, 20, no. 8, 901–904, http://dx.doi.org/10.1190/1.1487304.
Li, Y., and D. W. Oldenburg, 1996, 3-D inversion of magnetic data: Geophysics, 61,
394–408, http://dx.doi.org/10.1190/1.1443968.
Li, Y., and D. W. Oldenburg, 1997, Fast inversion of large scale magnetic data us-
ing wavelets: 67th International Annual Meeting, SEG, Expanded Abstracts,
490–493, http://dx.doi.org/10.1190/1.1885941.
204  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Li, Y., and D. W. Oldenburg, 1998, 3-D inversion of gravity data: Geophysics, 63,
109–119, http://dx.doi.org/10.1190/1.1444302.
Li, Y., and D. W. Oldenburg, 1999, Rapid construction of equivalent sources us-
ing wavelets: 69th Annual International Meeting, SEG, Expanded Abstracts,
374–377, http://dx.doi.org/10.1190/1.1821026.
Li, Y., and D. W. Oldenburg, 2003, Fast inversion of large-scale magnetic data us-
ing wavelet transforms and a logarithmic barrier method: Geophysical Jour­
nal International, 152, no. 2, 251–265, http://dx.doi.org/10.1046/j.1365-246X
.2003.01766.x.
Longman, I. M., 1959, Formulas for computing the tidal acceleration due to the
moon and the sun: Journal of Geophysical Research, 64, no. 12, 2351–2355,
http://dx. doi:10.1029/JZ064i012p02351.
Lyrio, J. C. S. de O., L. Tenorio, and Y. Li, 2004, Efficient automatic denois-
ing of gravity gradiometry data: Geophysics, 69, 772–782, http://dx.doi.org
/10.1190/1.1759463.
Martelet, G., P. Sailhac, F. Moreau, and M. Diament, 2001, Characterization of
geological boundaries using 1-D wavelet transform on gravity data: Theory
and application to the Himalayas: Geophysics, 66, 1116–1129, http://dx.doi
.org/10.1190/1.1487060.
McInerney, P. M., A. J. Mutton, and W. S. Peters, 1994, Abra lead-zinc-copper-gold
deposit, Western Australia: A geophysical case history, in M. C. Dentith, K. E.
Frankcombe, J. M. Shepard, D. I. Groves, and A. Trench, eds., Geophysical
signatures of Western Australian mineral deposits: Geology and Geophysics
Department (Key Center) and UWA Extension, University of Western Australia,
Publication 26, and Australian Society of Exploration Geophysicists, Special
Publication 7, 119–132.
Menke, W., 1989, Geophysical data analysis: Discrete inverse theory: Academic
Press.
Moraes, R. A. V., and R. O. Hansen, 2001, Constrained inversion of gravity fields
for complex 3-D structures: Geophysics, 66, 501–510, http://dx.doi.org/10
.1190/1.1444940.
Moreau, F., D. Gibert, M. Holschneider, and G. Saracco, 1997, Wavelet analysis
of potential fields: Inverse Problems, 13, no. 1, 165–178, http://dx.doi.org
/10.1088/0266-5611/13/1/013.
Mushayandebvu, M. F., P. van Driel, A. B. Reid, and J. D. Fairhead, 2001, Magnetic
source parameters of two-dimensional structures using extended Euler decon-
volution: Geophysics, 66, 814–823, http://dx.doi.org/10.1190/1.1444971.
Nabighian, M. N., 1962, The gravitational attraction of a right vertical circular cyl-
inder at points external to it: Pure and Applied Geophysics, 53, no. 1, 45–51,
http://dx.doi.org/10.1007/BF02007108.
References  205
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Nabighian, M. N., 1972, The analytic signal of two-dimensional magnetic bodies


with polygonal cross-section — Its properties and use for automated anom-
aly interpretation: Geophysics, 37, 507–517, http://dx.doi.org/10.1190/1
.1440276.
Nabighian, M. N., 1974, Additional comments on the analytic signal of two-dimen-
sional magnetic bodies with polygonal cross-section: Geophysics, 39, 85–92,
http://dx.doi.org/10.1190/ 1.1440416.
Nabighian, M. N., 1984, Toward a three-dimensional automatic interpretation of
potential field data via generalized Hilbert transforms: Fundamental relations:
Geophysics, 49, 780–786, http://dx.doi.org/10.1190/1.1441706.
Nabighian, M. N., V. J. S. Grauch, R. O. Hansen, T. R. LaFehr, Y. Li, J. W. Peirce,
J. D. Phillips, and M. E. Ruder, 2005, The historical development of the mag-
netic method in exploration: Geophysics, 70, no. 6, 33ND–61ND, http://dx.doi
.org/10.1190/1.2133784.
Nabighian, M. N., and R. O. Hansen, 2001, Unification of Euler and Werner decon-
volution in three dimensions via the generalized Hilbert transform: Geophysics,
66, 1805–1810, http://dx.doi.org/10.1190/1.1487122.
Nagihara, S., and S. A. Hall, 2001, Three-dimensional gravity inversion using simu-
lated annealing: Constraints on the diapiric roots of allochthonous salt struc-
tures: Geophysics, 66, 1438–1449, http://dx.doi.org/10.1190/1.1487089.
Nagy, D., 1965, The evaluation of Heuman’s lambda function and its application to
calculate the gravitational effect of a right circular cylinder: Pure and Applied
Geophysics, 62, 1, 5–12, http://dx.doi.org/10.1007/BF00875282.
Nettleton, L. L., 1962, Gravity and magnetics for geologists and seismologists:
AAPG Bulletin, 46, no. 10, 1815–1838.
Nettleton, L. L., 1971, Elementary gravity and magnetics for geologists and seis-
mologists: SEG Geophysical Monograph Series No. 1, http://dx.doi.org/10​
.1190/1.9781560802433.
Nettleton, L. L., L. LaCoste, and J. C. Harrison, 1960, Tests of an airborne gravity
meter: Geophysics, 25, 181–202, http://dx.doi.org/10.1190/1.1438685.
Okabe, M., 1979, Analytical expressions for gravity anomalies due to homogeneous
polyhedral bodies and translations into magnetic anomalies: Geophysics, 44,
730–741, http://dx.doi.org/10.1190/1.1440973.
Oldenburg, D. W., 1974, The inversion and interpretation of gravity anomalies: Geo-
physics, 39, 526–536, http://dx.doi.org/10.1190/1.1440444.
Oldenburg, D. W., and Y. Li, 2005, Inversion for applied geophysics: A tutorial,
in D. K. Butler, ed., Near-surface geophysics, Chapter 5: SEG Investigations
in Geophysics Series No. 13, 89–150, http://dx.doi.org/10.1190/1
.9781560801719.ch5.
Papoulis, A., 1962, The Fourier integral and its applications: McGraw-Hill.
206  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Parker, R. L., 1973, The rapid calculation of potential anomalies: Geophysical


Journal of the Royal Astronomical Society, 31, no. 4, 447–455, http://dx.doi
.org/10.1111/j.1365-246X.1973.tb06513.x.
Parker, R. L., 1977, Understanding inverse theory: Annual Review of Earth and
Planetary Sciences, 5, no. 1, 35–64, http://dx.doi.org/10.1146/annurev.ea.05
.050177.000343.
Parker, R. L., 1994, Geophysical inverse theory: Princeton University Press.
Pedersen, L. B., 1977, Interpretation of potential field data — A generalized in-
verse approach: Geophysical Prospecting, 25, no. 2, 199–230, http://dx.doi
.org/10.1111/j.1365-2478.1977.tb01164.x.
Pedersen, L. B., 1978, Wavenumber domain expressions for potential fields from
arbitrary 2-, 2½-, and 3-dimensional bodies: Geophysics, 43, 626–630, http://
dx.doi.org/10.1190/1.1440841.
Pedersen, L. B., 1979, Constrained inversion of potential field data: Geophysical
Prospecting, 27, no. 4, 726–748, http://dx.doi.org/10.1111/j.1365-2478.1979
.tb00993.x.
Phillips, J. D., 2001, Designing matched bandpass and azimuthal filters for the sep-
aration of potential-field anomalies by source region and source type: 15th
Geophysical Conference and Exhibition, Australian Society of Exploration
Geophysicists, Expanded Abstracts, CD-ROM.
Plouff, D., 1975, Derivation of formulas and FORTRAN programs to compute grav-
ity anomalies of prisms: National Technical Information Service Publication,
U. S. Department of Commerce, 243–526.
Plouff, D., 1977, Preliminary documentation for a FORTRAN program to compute
gravity terrain corrections based on topography digitized on a geographic grid:
U. S. Geological Survey Open-file Report 77-535.
Ponce, D. A., J. M. G. Glen, A. E. Egger, C. Bouligand, J. T. Watt, and R. L.
Morin, 2009, Geophysical studies in the vicinity of the Warner Mountains and
Surprise Valley, northeast California, northwest Nevada, and southern Oregon:
U. S. Geological Survey Open File Report 2009-1157.
Pratt, J. H., 1855, On the attraction of the Himalaya Mountains and of the elevat-
ed regions beyond the plumb-line in India: Transactions of the Royal Society
[London], ser. B, 145.
Rajagopalan, S., J. Carlson, and D. Wituik, 2007, Kimberlite exploration using in-
tegrated airborne geophysics: Australian Society of Exploration Geophysics,
Extended Abstracts, 1–5, http://dx.doi.org/10.1071/ASEG2007ab115.
Rasmussen, N. F., 1975, The successful use of the borehole gravity meter in north-
ern Michigan: The Log Analyst, SPWLA, 41, no. 5, 3–10.
Reamer, S. K., and J. F. Ferguson, 1989, Regularized two-dimensional Fourier grav-
ity inversion method with application to the Silent Canyon caldera, Nevada:
Geophysics, 54, 486–496, http://dx.doi.org/10.1190/1.1442675.
References  207
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Reed, L. E., and K. E. Witherly, 2007, 50 years of kimberlite geophysics: A review:


Fifth Decennial International Conference on Mineral Exploration, Paper 47.
Reid, A. B., J. M. Allsop, H. Granser, A. J. Millett, and I. W. Somerton, 1990,
Magnetic interpretation in three dimensions using Euler deconvolution: Geo­
physics, 55, 80–91, http://dx.doi.org/10.1190/1.1442774.
Ridsdill-Smith, T. A., 1998a, Separating aeromagnetic anomalies using wavelet
matched filters: 68th Annual International Meeting, SEG, Expanded Abstracts,
550–553, http://dx.doi.org/10.1190/1.1820491.
Ridsdill-Smith, T. A., 1998b, Separation filtering of aeromagnetic data using filter-
banks: Exploration Geophysics, 29, no. 4, 577–583, http://dx.doi.org/10.1071
/EG998577.
Ridsdill-Smith, T. A., and M. C. Dentith, 1999, The wavelet transform in aeromag­
netic processing: Geophysics, 64, 1003–1013, http://dx.doi.org/10.1190/1​
.1444609.
Roest, W. R., J. Verhoef, and M. Pilkington, 1992, Magnetic interpretation using
the 3-D analytic signal: Geophysics, 57, 116–125, http://dx.doi.org/10.1190
/1.1443174.
Sandwell, D. T., and W. H. F. Smith, 1997, Marine gravity anomaly from Geosat
and ERS 1 satellite altimetry: Journal of Geophysical Research, Solid Earth,
102, no. B5, 10039–10054, http://dx.doi.org/10.1029/96JB03223.
Sandwell, D. T., and W. H. F. Smith, 2001, Bathymetric estimation, in L. L. Fu
and A. Cazenave, eds., Satellite altimetry and earth sciences: Academic Press,
441–457.
Sandwell, D., S. Gille, J. Orcutt, and W. Smith, 2003, Bathymetry from space is now
possible: Eos, Transactions, American Geophysical Union, 84, no. 5, 37, 44.
Schuler, M., 1923, Die Störung von Pendel und Kreiselapparaten durch die
Beschleunigung des Fahrzeuges: Physikalische Zeitschrift, 24, no. 16, 334–350.
Sclater, J. G., and P. A. F. Christie, 1980, Continental stretching: An exploration of
the post-mid-Cretaceous subsidence of the central North Sea Basin: Journal
of Geophysical Research, Solid Earth, 85, no. B7, 3711–3739, http://dx.doi
.org/10.1029/JB085iB07p03711.
SEG Virtual Geoscience Center, 2006, Worden gravity meter, http://virtualmuseum.
seg.org/pict0665.html, accessed 29 August 2012.
Skeels, D. C., 1947, Ambiguity in gravity interpretation: Geophysics, 12, 43–56,
http://dx.doi.org/10.1190/1.1437295.
Smith, R. A., 1960, Some formulae for interpreting local gravity anomalies: Geo­
physical Prospecting, 8, no. 4, 607–613, http://dx.doi.org/10.1111/j.1365-2478
.1960.tb01736.x.
Spector, A., 1968, Spectral analysis of aeromagnetic maps: Ph.D. thesis, University
of Toronto.
208  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Spector, A., and F. S. Grant, 1970, Statistical models for interpreting aeromagnetic
data: Geophysics, 35, 293–302, http://dx.doi.org/10.1190/1.1440092.
Stavrev, P. Y., 1997, Euler deconvolution using differential similarity transforma-
tions of gravity or magnetic anomalies: Geophysical Prospecting, 45, no. 2,
207–246, http://dx.doi.org/10.1046/j.1365-2478.1997.00331.x.
Swick, C. A., 1942, Pendulum gravity measurements and isostatic reductions: U. S.
Coast and Geodesic Survey Special Publications 232.
Syberg, F. J. R., 1972, A Fourier method for the regional-residual problem of poten­
tial fields: Geophysical Prospecting, 20, no. 1, 47–75, http://dx.doi.org/10
.1111/j.1365-2478.1972.tb00619.x.
Szabó, Z., 2008, Biography of Loránd Eötvös (1848–1919): The Abraham Zelmanov
Journal: Journal for General Relativity: Gravitation and Cosmology, 1, vii.
Talwani, M., 2003, The Apollo 17 gravity measurements on the moon: The Leading
Edge, 22, no. 8, 786–789, http://dx.doi.org/10.1190/1.1605083.
Talwani, M., and M. Ewing, 1960, Rapid computation of gravitational attraction of
three-dimensional bodies of arbitrary shape: Geophysics, 25, 203–225, http://
dx.doi.org/10.1190/1.1438687.
Thomas, M. D., 1999, Application of the gravity method in mineral exploration:
Case histories of exploration and discovery, in C. Lowe, M. D. Thomas, and
W. A. Morris, eds., Geophysics in mineral exploration: Fundamentals and case
histories: Geological Association of Canada Short Course Notes 14, 101–113.
Thompson, D. T., 1982, EULDPH: A new technique for making computer-assist-
ed depth estimates from magnetic data: Geophysics, 47, 31–37, http://dx.doi
.org/10.1190/1.1441278.
Tikhonov, A. V., and V. Y. Arsenin, 1977, Solution of ill-posed problems: John
Wiley & Sons.
Torge, W., 1989, Gravimetry: Walter de Gruyter.
Tucker, P. M., and H. J. Yorston, 1973, Pitfalls in seismic interpretation: SEG Geo­
physical Monograph Series No. 2, http://dx.doi.org/10.1190/1.9781560802365.
Vajk, R., 1956, Bouguer corrections with varying surface density: Geophysics, 21,
1004–1020, http://dx.doi.org/10.1190/1.1438292, http://dx.doi.org/10.1190/1
.9781560802365.
Verhoogen, J., F. J. Turner, L. E. Weiss, C. Wahrhaftig, and W. S. Fyfe, 1970, The
earth, an introduction to physical geology: Holt, Rinehart and Winston, Inc.
Veryaskin, V. A., 1999, Apparatus for the measurement of gravitational fields: U. S.
Patent 5,962,781.
Virtanen, H., 2006, Studies of earth dynamics with superconducting gravimeter:
Academic dissertation, University of Helsinki.
Wahba, G., 1990, Spline models for observational data: Society for Industrial and
Applied Mathematics.
References  209
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Wing, C. G., 1969, MIT vibrating string surface-ship gravimeter: Journal of


Geophysical Research, Solid Earth, 74, no. 25, 5882–5894, http://dx.doi.org
/10.1029/JB074i025p05882.
Woollard, G. P., 1958, Results for a gravity control network at airports in the United
States: Geophysics, 23, 520–535, http://dx.doi.org/10.1190/1.1438499.
Zhang, C., M. F. Mushayandebvu, A. B. Reid, D. Fairhead, and M. E. Odegard,
2000, Euler deconvolution of gravity tensor gradient data: Geophysics, 65,
512–520, http://dx.doi.org/10.1190/1.1444745.
Zhang, J., C.-Y. Wang, Y. Shi, Y. Cai, W.-C. Chi, D. Dreger, W.-B. Cheng, and Y.-H.
Yuan, 2004, Three-dimensional crustal structure in central Taiwan from gravity
inversion with a parallel genetic algorithm: Geophysics, 69, 917–924, http://
dx.doi.org/10.1190/1.1778235.
Zhdanov, M. S., R. Ellis, and S. Mukherjee, 2004, Three-dimensional regularized
focusing inversion of gravity gradient tensor component data: Geophysics, 69,
925–937, http://dx.doi.org/10.1190/1.1778236.
Zumberge, M. A., J. E. Faller, and J. Gschwind, 1983, Results from an absolute
gravity survey in the United States: Journal of Geophysical Research, Solid
Earth, 88, no. B9, 7495–7502, http://dx.doi.org/10.1029/JB088iB09p07495.
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

This page has been intentionally left blank


Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Index

Abra base-metal deposit, equivalent layer, 125 theorem, 150–152


Western Australia, 185–186 ambiguity, 43, 46, 47, 53 reservoir monitoring, 140
ellipsoid modeling program, and survey design, 53 spectral analysis, 118–125
186 amplitude, mGal, 177 unit half-width circle (2D)
ground magnetic survey, 186 analytic signal, and magnetics, and ellipse, 141–147
Abrolhos volcanic complex, 123 anomaly separation, 3, 101,
offshore Brazil, 191, 193 andalusite, density, 79 110–117
absolute and relative Andes, 7, 99 by geologic constraint or
instruments, 53 anhydrite, density, 76, 79, 83 smoothing, 112
absolute-gravity measurements, velocity-density relationship, 83 anomaly-separation techniques,
53–54, 55 anomalies, observable, 102 and preservation of
accuracy, 54 anomalies, relative to density anomaly amplitude, 117
and earth tides and crustal of rock, 74 anomaly shape, 109
deformation, 54 anomalous mass, determination anorthite, density, 79
and oil-field reservoir of, 3, 136–138, 182 antialiasing filter, 70
monitoring, 54 and Gauss’ theorem, 137 anticline, 132
absolute-gravity meters, 88, 189 estimation by Gauss’ Apollo 17 mission, 58
and relative-gravity meters, theorem, 182 aragonite, density, 79
compared, 189 anomaly, gravity, as related to atmospheric component of
absolute-gravity survey, United density contrast, 86 normal gravity, 92
States, 54 anomaly, observed, relative atomic rubidium clock, 54
acceleration, and earth’s to regional field and attraction, gravitational,
rotation, 101–102 residual field, 110 components, 18–20
accuracy, achievable by various anomaly-enhancement attraction, principles, and
gravity surveys, 68 techniques, 115, 116 earth’s gravity field, 5–14
aeromagnetic map data, power anomaly interpretation geoid, 10–11, 12–14
spectra, 124 guide­lines and limitations, GPS and the geoid, 12–14
aeromagnetic survey, 185 101–156 gravitational force, 5–6, 7
airborne gravity gradiometry, anomalous mass, gravitational potential, 7–8
fixed-wing and helicopter, determination of, 136–138 standard International Gravity
188 anomaly separation, 110–117 Formula, 11–12
airborne gravity gradiometry, in anomaly shape, 109 universal gravitational
exploration for kimberlite, borehole gravity, constant, 6–7
186 interpretation of, 138–140 attraction, two masses, 6
Airy-Heiskanen model, 99 borehole-gravity distance/ Australia, 124
Airy theory, 99 thickness relationships, average Bouguer anomaly, 101
Algaré region, Portugal, gravity 152–156
survey, 184 depth determination, 125–136 band-pass filtering, 116
aliasing, and its effects, 69–70, fast-Fourier transform for basalt, 73
87 calculating gravity base-of-salt reflections, 183
and frequency of samples, 70 effects, 105–109 bathymetry, 80, 97, 107
and Nyquist frequency, 70 gravity calculations, arbitrary Bay of Fundy, 103
avoidance of, 70 model, 102–105 beam displacement, 57
relative to waveform, 70 gravity surveys, purposes, Bell Aerospace stabilized-
alluvium 75, 80 101–102 platform instruments, 60
and soil, density, range, 75 integration, incomplete, berm, 117
ambiguity, and Green’s corrections, using Gauss’ Bessel function, modified, of

211
212  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

order 1, 105 Bouguer density, variable, 96 continuous-wavelet transform,


BHGM system, 67 Bouguer effects, 93 124–125
bicubic spline interpolation, 117 Bouguer formula, 38, 128, 141 Cook Inlet, 103
Bodenseewerk stabilized- Bouguer gravity anomaly, 79, Copernican revolution, 2
platform instruments, 60 84–85, 182, 187 Cordell filter, 124
borehole gravity, 3, 4, 78 Lac De Gras kimberlite Corvo, Portugal, sulfide
and bulk density, 78 province, Northwest deposit, 184
borehole gravity, density, Territories, Canada, 187 crater lake, and kimberlite
derivation, 138–140 Bouguer gravity-anomaly maps, intrusions, 186
apparent density, and anomaly, and topographic maps, 96 cross-coupling effects, and
140 Bouguer gravity map, 170 measurement of gravity
infinite horizontal layer, 139 Bouguer gravity map, Neves, on stabilized platform, 60
multiple uniform layers, 139 Portugal, 184 crossover depth, 74, 75, 76, 173
uniform infinite layer, 138 Bouguer map, 118 crossover zone, salt, 183
borehole gravity, 180–181 Bouguer plate, 20, 96 cross-validation function, 163
borehole gravity, with surface Bouguer plate value, 104–105 crustal-thickness maps, and
gravity, 181–182 Bouguer reduction density, 80, gravity models, 191
and apparent-density anomaly, 170 cylinder, horizontal, infinite
lateral distance to, 181 Bouguer slab, 20, 41, 109 length, surrounded by
borehole gravity, interpretation bound constraints on model Gaussian surface, 35–36, 37
of, 138–140 densities, 164 cylinder, vertical, finite depth,
borehole-gravity distance/thick- Brazil, offshore, geology, gravity calculations for,
ness relationships, 152–156 191–193 30–32
borehole gravity meter, 78, 154 bromellite, density, 79
borehole gravity meter, and bulk density, 76, 77, 78 damping coefficient, δ, 58, 60
detection of reservoir Bullard A, 104 data inversion, West Africa,
rock, 180–181 Bullard A, infinite plate, 92 offshore, 193
borehole gravity work, 6 Bullard B correction, 96, data reduction, 88–106
Bott and Smith theorems, 104–106 gravity survey data, 88–106
application, 147–150 Bullard C, and terrain data reduction and interpretation,
Bouguer, Pierre, 7, 99 corrections, 97 and gravitational potential, 17
Bouguer anomalies, compared Bullard correction, 104–106 data-reduction phase of work,
with isostatic residual Burris spring gravimeter, 55 goal, 87
anomaly, 99 da Vinci, Leonardo, 93
Bouguer anomaly, 3, 97–98, calcite, matrix density, 76 deep interpretive models,
100, 101, 103, 110, 170, calibration factor of replacement by shallow
173, 175, 183–185, 192 instruments, 57 interpretive models, 47
and Bullard A, 97 California, 99, 104 deflection of the vertical, 25
and Bullard C, 97 Campos Basin, offshore Brazil, densities, 73–76, 78–84, 193
and correction for topography, 193 determining, methods, in
100 Canada, 165, 186 exploration, 78–84
and free-air anomaly, 98 caprock, 4, 175–176 igneous rocks, 73
average, 101 carbonate rocks, density, 74 metamorphic rocks, 73
complete Bouguer anomaly, Cartesian-coordinate system, rock, near-surface, 73–76
97, 98 with source point and rock, relative to geologic
depth-to-basement map, field point, 19 age, 74
derived, 175 Cavendish, Henry, 7 sedimentary rocks, 73
intrasedimentary residual Cement oil field, Oklahoma, water, sediments, crust,
anomaly map, derived, 175 Bouguer anomaly, 173–174 mantle, 193
map, 103 second vertical derivative, 174 density, absolute, compared
Neves-Corvo sulfide deposit, CHAMP, gravimetry satellite with density contrast, 74
Portugal, 183–185 system, 191 density, apparent, resolution
observed, 110 channels, 111 of, 67
offshore Brazil, 192 circular permutations, density, earth, average, 7
simple Bouguer anomaly, derivatives, and density, inconstant, 25
97, 98 gravitational potential, 27 density, measuring and
Bouguer anomaly map, 110 closed-form solution, spheres, evaluating, 78–84
Bouguer correction, 80, 93–96, cylinders, vertical lines, density, rock, 3, 76– 78, 81
97, 100, 101, 103, 104 ribbons, 106 density, variation, Δρ, distance
Bouguer density, 170 complete Bouguer anomaly from wellbore and bed
Bouguer density, error in, (CBA), 97, 98 thickness, 155
effects, 96 conservative fields, 7 density, vertical variation, and
Index  213
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

horizontally varied earth tides, time variation, 89 fluid density, range of, 76
anomalies, 110 earth tides and crustal forward calculation, 26
density contrasts, 74, 85, 86 deformation, 54 forward gravity modeling, 183
horizontal, related to fault, Ecuador, 7 forward modeling, 167
85, 86 Ekati kimberlite pipe, Lac forward problem, 26, 157
density difference curve, 154 de Gras kimberlite 4D gravity maps, six periods,
density distributions relative to province, Northwest Prudhoe Bay, Alaska, 190
gravity anomalies, 47 Territories, Canada, 186 4D gravity monitoring, 189
density inversion, 4, 158–166 Ekati tenement, Lac De Gras 4D gravity surveys, 6, 54, 65
density layers and contrasts, kimberlite province, 4D microgravity data, Arctic, 189
illustrated, 85 Northwest Territories, Fourier domain, gravity field
density log, 154 Canada, 186 represented in, 116
density of rock, interpreted ellipsoid, 10, 88 Fourier method, and reservoir
from interval velocity, elliptic integrals, 32 simulation, 107
“horizontal” strata, 183 enstatite, density, 79 Fourier series, 46
density-versus-depth function, Eötvös, Baron Loránd, 2, 62, 140 Fourier transform, 106, 107,
gravity models, and Eötvös corrections, 101–102 108, 109, 114, 118, 119,
crustal-thickness maps, 191 Eötvös effect, and motion of 121, 195–198
depth determination, 125–136 meter, 101–102 and convolution theorem, 197
depth-estimation rules, 3 Eötvös effects, 63 and multidimensional
depth functions, 2D and 3D, Eötvös torsion balance, 63 functions, 196
compared, 150 Eötvös units, defined, 63 and Parseval’s formula,
depth of compensation, 99 equator, and outward 197–198
depth-rule diagrams, 2D and acceleration caused by and wavelength, 195
3D, 147 earth’s rotation, 101–102 and wavenumber, 195
depth weighting function, 164 equipotential surfaces, 8 functions of time instead of
differential curvature gradients, ERS-1, gravimetry satellite space, 196
63 system, 191 gravitational acceleration,
digital elevation model (DEM), Eugene Island, Gulf of Mexico, target body, 119
88, 94 134, 135 Green’s function, 106
dike, 109, 126 Euler deconvolution, 132–134 inverse, 109, 119, 196
dimensionality of data, and excess mass, and gravity, properties, 196–197
dimensionality of Neves-Corvo sulfide techniques, 118
underground density deposit, Portugal, 185 theory, 195
function, 158 Fourier-transform theory, 195
diopside, density, 79 failed rift, offshore Brazil, free-air anomaly, 100, 101, 110
direction cosines, 19 Bouguer anomaly, 193 and isostatic compensation, 100
direct problem, 157 Falcon gravity-gradient system, average, 101
discrete-wavelet transforms, and 186 free-air anomaly, offshore
orthonormal wavelets, 125 Falcon gravity gradiometers, 63 Brazil, 192, 193
discretization, 160 fast Fourier transform, 3, free-air anomaly map, 103, 118
divergence theorem, 22, 43 105–109 free-air correction (FAC),
dolomite, 76, 78, 79, 83 fathometer, 104 91–93, 97, 103
double negative residual fault rollover-induced residual, free-air correction estimates, 13
downward-continuation 177 free-air gravity anomaly, 96–97,
filter, 120 faults, 172 98
downward-continuation fault trace, gravity anomaly, and bathymetry, 97
technique, 116 130, 132 free-air gravity data, and
drift, 88 field measurements, 53–70 Bouguer anomaly grids, 191
field operations, 53 free-air gravity data, and
earth, average density, 7 field point, 17 satellites, 191
earth, geoid model, 13 finite step, semi-infinite, free-air gravity map, 170
earth, mean density, 73 gravitational attraction, free-fall method for measuring
earth curvature, 96 39–42 absolute gravity, 54
earth satellite, 54 Finland, 59 free-fall technique, 53
earth’s crust, diurnal first-vertical-derivative filter frequency-domain expressions,
deformation, 89 p, 120 potential fields, 2D, 2.5D,
earth’s gravity field, 8–9 flattening of earth, and 3D bodies, 104
earth’s shape, determination gravitation attraction at frictionless hole, 54
of, 101 poles, 9 FTG system, 63, 64, 65
earth tide effects, midlatitudes, Florianopolis Ridge, offshore sensor design, 65
90 Brazil, 191, 193 full-tensor gradient system
214  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

(FTG), 63, 64 gneiss, density, 74 gravity-gradient tensor, 64


GOCE, gravimetry satellite gravity gradiometer, 62
gabbro, density, 73, 74 system, 65, 191 gravity gradiometry, 25, 62–65,
Galileo, 2, 6 graben, 47 122, 186
gamma-gamma/density log, 78, GRACE, gravity recovery and data, fixed-wing, compared
82, 154 climate experiment, 65, 184, with helicopter, 186
Gardner curve, 83 191 satellite-measured gravity
Gardner’s equation, 177 graphical residual approach for gradiometry, 65
Gardner velocity-density anomaly separation, 114 surveys, 65
relationship, 82, 83 granite, density, 73, 74 tensor components, 122
Gaussian noise, 160 Gravilog system, 60, 67–68 gravity meter, 25, 155
Gaussian surface, 35 gravimeters, 53 penetration, and geologic
Gauss’ law, 137 gravitational constant, 6–7 information, 155
Gauss’ theorem, 4, 21–22, gravitational force, 5–6, 7 gravity recovery and climate
26, 35, 45, 46, 100, 101, Stokes’ theorem, 7 experiment, 65
137, 150–152, 182 gravitational potential, 7, 8, gravity survey data, reduction
and determination of total 25, 27 of, 88–106
mass, 22 gravity, topography, and Bouguer correction, 93–96
and divergence theorem, 22 seismic section, 84 Bullard correction, 104–106
and estimation of anomalous gravity, 2D data inversion, Eötvös corrections, 101–102
mass, 182 offshore West Africa, 193– free-air correction, 91–93
corrections, 4 194 gravity anomalies, 96–98
masses inside, 21–22 gravity, vertical component, isostatic correction, 99–101
masses outside, 22 calculation of, 26 latitude corrections, 90
region of integration, 151 gravity analysis, inherent marine reductions, 103–104
utility of, 22 ambiguity, 177 time variations, 88–90
generalized cross-validation gravity and magnetics, gravity surveys, purposes,
(GCV) technique, 162 complementary data, 183 101–102
genetic algorithm (GA), 165, gravity anomalies, quantitative, gravity units, 6
167 assumptions for, 102 Green’s equivalent layer, and
geologic applications, 169–194 gravity anomalies, vertical ambiguity, 2, 3, 43–48, 125
borehole gravity with surface component, simple 2D Green’s first identity, 43
gravity, 181–182 and 3D models, 111 Green’s function, 157
buried features, location gravity anomalies and lateral Green’s functions, gravitation
by filtering and modeling, density contrast, 172 attraction, 105
173–181 gravity anomaly, 3, 22, 84–86, Green’s layer, 45
mining applications, 183–189 96–98, 158, 178 Green’s second identity, 45
satellite gravity and satellite- gravity calculations, arbitrary Green’s theorem, 164
derived gravity, 190–194 model, 102–105 Green’s three identities, 43
seismic and magnetic gravity calculations, simple grid-based systems, 102
information integrated geometries, 24–34 grid residual approach, 113,
with gravity data, 182–183 gravity calculations, 2D 114–115, 116
time-lapse (4D-gravity) geometries, 34–42 compared with graphical
surveys, 189–190 gravity compensator, 140 approaches, 115, 116
geometric rollover, 177 gravity data, inversion, and compared with grid approach,
geoid, 10–11, 12–14, 88 removal of regional 114–115
geoid, and GPS, 12–14 effect, 163–164 ground magnetic survey, Abra
geoid, described relative to gravity data integrated with base-metal deposit, Australia,
ellipsoid and spheroid, 10–12 seismic and magnetic 186
geologic modeling, and data, 182–183 Gulf of Mexico, 2, 83, 113, 134,
anomaly separation, 117 Wyoming overthrust belt, 135, 172
geologic modeling and 182–183 Gulf Stream, 10, 61
anomaly-separation gravity field, components, 62 gypsum, 76, 79
technique, 110 gravity field and steady-state gyroscopes, 60
geometric (boundary) ocean circulation explorer,
inversion, 4, 166–168 65 half-space, 159
GEOSAT, gravimetry satellite gravity-field vector, vertical half-width depth determination,
system, 191 component, 62 127
global marine gravity map, 61 gravity-gradient data, inversion half-width rules, 126–132
Global Positioning System of, 164 halite, density, 74, 76
(GPS), 12–14, 59, 63, 65, 66, gravity-gradient instruments Hammer charts, 94
90 (GGI), 64 harmonic function, 44
Index  215
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Hayford-Bowie inner zones, Klamath Mountains, California, and second vertical derivative,
compartment elevation, 95 99 117
Hayford-Bowie system, 94, 95 Louisiana, 74
Hayford-Bowie template, 97 Lac de Gras kimberlite L2 norm, 160
Hayford-Bowie zone, 106 province, Northwest
Heath Steele Stratmat copper- Territories, Canada, 156 magnetic and gravity data,
lead-zinc deposit, Canada, LaCoste and Romberg air/sea modeled, single ellipsoid,
165–166 devices, 60 188
Heiskanen, W. A., 93 LaCoste and Romberg borehole magnetics, and analytic signal,
Heumann lambda function, 32 devices, 60 123
high-pass filter, 115 LaCoste and Romberg borehole marine free-air gravity field,
Hilbert transform, 121, 133 gravity meter, 66–67 and terrestrial gravity
homogenous polyhedron-shaped and limitations of borehole field, 61
3D body, gravitational diameter, 66–67 marine reductions, 103–104
attraction 103 LaCoste and Romberg zero- marine terrain corrections, 96,
horst, 47 length spring gravimeter, 104
56, 58, 59 mass, excess, effect on geoid,
Iberian pyrite belt, 184–185 drift characteristics, 58 10–12
iGrav SG meter, 59 lamina, at depth, with constant mass distributions, gravitational
ill-posed problems, and density, 20 potential and attraction,
unstable solutions, 158 lamina, circular, gravitational 15–48
infinite slab, 109 attraction along axis, 31 attraction, components, 18–20
instrument drift, 58, 66, 67 lamina, constant surface gravity calculations, simple
integration, incomplete, density, gravitational geometries, 24–34
corrections, using Gauss’ anomaly, 30 gravity calculations, 2D
theorem, 150–152 laminae, finite thickness, for geometries, 34–42
International Gravity Formula, detection of gravity Green’s equivalent layer, and
3, 11–12, 13, 90 anomaly, 103 ambiguity, 43–48
interval velocity, degradation laminae, infinitely thin, for logarithmic potential, 2D
of calculated quantities, and detection of gravity targets, 43
of estimated density, 183 anomaly, 103, 104 potential fields, analyses of,
interval velocity, seismic, Laplace, Marquis de, 2, 3 21–24
converted to density, 84 Laplace operator, 44 spherical shell, attraction of,
intrusive targets, deep-seated, 118 Laplace’s and Poisson’s 15–17
inverse problem, 26, 158, 160 equations, 22–24, 114 masses completely bounded
inversion, 79, 157–168 and divergence theorem, 22 by surface, 21
density inversion, 158–166 and points located inside and matched filtering, and magnetic
geometric (boundary) outside, 23–24 data, 124
inversion, 166–168 mass with observation point mathematical fictions and real-
inversion, and local geology, inside, 24 world problems, 46
and limitations of methods, Laplace’s equation, 23, 44, 45, mean sea level (MSL), 11,
168 46, 62, 65, 164 12, 91
inversion, base-of-salt, 165 vector field derived from, 23 measurement, earth’s gravity
inversion, results of, and laser interferometer, 54 field, 50, 53
geologic interpretation, 163 lateral density contrast, 73 measurement uncertainty, 53
inversion algorithms, 159 latitude corrections, 12, 90 meridian arc, measurement
isostasy, 3, 170 layered earth, 159 of, 99
isostatic anomaly map, 170 L-curve, 162 metal fatigue, and drift, 88
isostatic correction, 99–101, 109 least-squares criterion, 113 meter drift, 57
isostatic residual gravity least-squares fit, 113 mica schist, density, 74
anomalies, 107 least-squares methods, 166 Michigan, 180–181
iterative Newton-type limestone, 75, 83 Micro-g LaCoste air-sea gravity
techniques, 167 limiting depth, 134–136 system, 60
lineal density, 28 microwave ranging system, 65
Jacobi’s zeta function, 32 lithospheric thinning, and mining, and gravitimetric data,
Jacobsen filter, 124 gravity models, 191 183–189
logarithmic potential, 2D Abra base-metal deposit,
k, the universal gravitational targets, 43 western Australia, 185–186
constant, 5, 6–7 loop, 65, 67 kimberlite, exploration for,
Kepler, Johannes, 2 Los Angeles Basin, 173 186–189
kimberlite, exploration for, Los Angeles Basin, Bouguer Neves-Corvo massive sulfide
186–189 gravity, 116, 117 deposit, Portugal, 183–185
216  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

missile trajectories, 101 oil, density, 73 140, 189–190


model norm, 159 oil-field reservoir monitoring, 54 water injection program,
models, 34, 141 olivine, density, 79 189–190
criteria for selection of, 141 orthoclase, density, 79
fictitious, 34 orthogonal polynomials, 113 quartz, density, 76, 78, 79
simple, and closed-form or orthometric height, 12–13 constituent density, 78
exact solution, 141 and gravity exploration, 13 quartz diorite, density, 74
Moho discontinuity, 172, 191, quartz springs, 55, 58, 89
193 parameter hyperspace, 47 compared with metal springs,
monopoles, 62 Parseval’s theorem, 197–198 58
Monte Carlo techniques, 167 pendulum, simple, period, 53
moon, 58, 89 pendulum apparatus, 53 rectangular plate, infinite, thin,
gravity measurements on, 58 pendulums, 54 at depth, with constant
mountains, and balance by peridotite, density, 74 density, 20
mass deficiencies, 99 Peru, 7 rectangular plate, thin,
muscovite, density, 79 physical constants, gravitational attraction,
determination of, 101 calculations for, 28–30
Nash salt dome, 2, 63 pickoff rings, 64 rectangular prisms, 102, 103
NAVSTAR satellites, 12 pickoff system, 64 and approximation of volume
nepheline, density, 79 plate, thin, rectangular, constant of mass, 102–103
net restoring force, 58 surface density, and calculation of gravity
Nettleton profiling, 83 gravitational anomaly, 30 anomaly, 103
Nettleton profiling technique, plate tectonics, 61 reef, model of, density,
78, 79–80 point dipole, 132 difference related to
Neves, Portugal, sulfide deposit, point pole, 132 porosity, 156
Bouguer gravity map, 184 Poisson jump, 153 reefs, 172, 180–181
Neves-Corvo massive sulfide Poisson’s equation, 23, 24, 45, 67 borehole-gravity-derived
deposit, Portugal, 183–185 polyhedron, gravity field, 104 densities, 180
New Brunswick, Canada, porosity, derived from density, density function for, 180
Heath Steele Stratmat equation, 77 gamma-gamma densities, 180
deposit, 165–166 porosity and rock density, 76–78 Michigan, 180–181
Newton, Isaac, 2, 6 Portuguese pyrite belt, Neves- reference ellipsoid, 10–11, 12, 13
Newtonian physics, 2 Corvo Group, 184 and earth’s topography, 11
Newtonian potential, 43 positivity constraints on model relative to normal and
Newton’s inverse square law, 25 densities, 164 theoretical gravity, 10
Newton’s law, 92 potential fields, analyses of, reference model, 159
Newton’s second law, 6 21–24, 125 regularization, 159
Newton’s universal law of Gauss’ theorem, 21–22 regularization parameter, λ, 161,
attraction, 5 Laplace’s and Poisson’s 162, 163
Newton-type techniques, 167 equations, 22–24 and generalized cross-
nil zone, 76 Pratt system, 99 validation curve, 163
noise, 160 Pratt theory, 99 relative gravity, 57
noise contribution to observed Pratt-Hayford model, 99 relative-gravity instruments,
gravity value, 87 Precambrian basement, 173 54–62, 88
nonuniqueness, 45 prestack depth migration, and meters, 88
normal fault, density contrasts gravity data, 176 satellite-derived gravity, 60–62
and gravity anomaly, 178 primary lines, 66 spring gravimeters, 54–58
“normal” gravity, and latitude, prism, horizontal, infinite superconducting gravity
11–12 length, gravitational meter, 59–60
normal gravity, defined, 10 attraction, 36–38 vibrating-string gravimeters,
“normal” gravity field, effects, 90 prism, truncated, gravitational 58–59
normalization of variables, 167 attraction, 34 relative-gravity meters, and
nulling of instruments, 57 prism interpretations, relative to absolute-gravity meters,
Nyquist frequency, 70 a single anomaly, 48 compared, 189
and antialiasing filter, 70 proof mass, 64 reoccupation rate, 58
Prudhoe Bay, Alaska, 4D reservoir monitoring, 65, 140
ocean loading, 89 gravity maps, 190 residual map, 164
oceans, and balance by mass Prudhoe Bay, Alaska, 4D resolutions, in mGal and μGal,
excesses, 90 microgravity surveys, and various gravity surveys,
ocean surface, mean, as absolute-gravity meters, 189 68–69
equipotential surface, 10 Prudhoe Bay reservoir, Alaska, reverse fault, 177, 179
offshore gravity surveys, 103 time-lapse gravity surveys, densities, density contrasts,
Index  217
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

and gravity anomaly, 179 boundaries, 191 shale, density, 74, 75


rhyolite, 73 and definition of plate shale, velocity-density
ribbon device, 68 boundaries, 191 relationship, 83
rock density, 3, 73–86 and definition of sediment shallow and deep targets,
and gravity anomalies, 73–86 loads, 191 gravitational anomalies,
constituent densities, 78 and definition of transform decrease with height, 121
density and porosity, 76–78 faults, 191 shallow sources, and
gravity anomaly, cause, 84–86 basin modeling and attenuation with height, 120
methods for deriving, hydrocarbon thermal shallow targets, and energy in
measuring, and evaluating maturity, 193 shorter wavelengths, 118
density, 78–84 categories of satellite short-wavelength noise, 114
near-surface rock densities, systems, 191 simple Bouguer anomaly, 97,
typical, 73–76 insensitivity to short- 98
rock densities in sedimentary wavelength variations, 191 simple Bouguer plate, 96, 104
basin, 171 satellite-derived gravity, 60–62 simple Bouguer plate value, 93
rock salt, velocity-density satellite-measured gravity simulated annealing, 167
relationship, 83 gradiometry, 65 sinx/x methods, 46
rollover, seismic reflections, 177 satellite radar altimetry, 61 slim-hole borehole gravity
rotational potential of earth, 9 satellite systems for gravimetry, system, 67
rotation of earth, gravitational categories, 191 “slope” model, 94
effect, 9 Scintrex CGS gravity meter, 58 soil and alluvium, density,
Scintrex instruments, 89 range, 75
salt, 73, 74, 75, 78, 79, 167, Scintrex slim-hole borehole solid angle, 19, 20
171, 172, 175–176, 183 gravity meter, 60 solid-angle cone, 21
and crossover zone, 183 screw “backlash,” 57 Somigliana’s equation, 11
and density contrast with sea-bottom gravity surveys, 65 source layer, constant density,
adjacent rocks, 183 SEASAT, gravimetry satellite uneven topography, top and
and positive gravity anomaly, system, 61, 191 bottom, 107–109
175 sea surface, variation in height, source rocks, elements
anomaly, associated, 191 necessary to produce
components, 176 sea-surface surveys, and free- observable anomalies, 170
base of, inversion for, in air correction, 97 source rocks, location of, 109
exploration, 167 sea-surface topography, spatial-convolution filters, and
compressibility, 73 measurement of, 61 second vertical derivatives,
constituent density, 78 seawater, density, 73 114
density, 75, 78, 79 second-derivative technique, 173 spectral analysis, 118–125
incompressibility, 171, 172 second vertical derivative, 114, analytic signal, 123–124
with caprock, 175–176 116, 121, 124 directional and second-order
salt dome, densities of and measurement of derivatives, 122–123
sedimentary rocks as curvature, 121 downward continuation,
function of depth, 75 as edge-detection tool, 114 119–120
salt domes, 63, 111, 113, filter, 124 first vertical derivative,
172–173, 183 seismic and magnetic data, 120–121
interpretation, Gulf of integrated with gravity data, matched filtering, 124
Mexico, 172–173 182–183 second vertical derivative,
salt keel, 183 seismic data, projected by use 121–122
salt stock, 183 of gravity data, 182 upward continuation, 118–119
salt structures, 4 Wyoming overthrust belt, wavelets, 124–125
sandstone, density, 74, 75 182–183 sphere, gravitational anomaly
sandstone, velocity-density seismic pitfalls, examples, of, 28–29
relationship, 83 177–180 sphere, gravitational attraction,
Santos Basin, offshore Brazil, seismic velocities and rock calculations for, 27–28
193 density, 176 sphere, solid, uniform,
satellite gravity, basin self-leveling gravimeters, 59 gravitational attraction at
modeling, thermal semi-infinite horizontal sheet, external point, 16
calculations, and gravity anomaly, 128 sphere, solid, uniform,
hydrocarbon thermal semi-infinite slab, gravitational gravitational attraction at
maturity, 193 anomaly, 41, 42, 44 internal point, 16
satellite gravity and satellite- higher-order derivatives, 41, sphere anomaly, 109
derived gravity, 190–194 42, 44 spherical cap, and infinite
and definition of basins, 191 properties, 41, 42, 44 Bouguer plate, 105
and definition of crustal shadow zone, under fault, 177 spherical shell, attraction of,
218  
Fundamentals of Gravity Exploration
Downloaded 06/20/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

15–17 equal vertical attraction, 143 survey, 54


spheroid, described relative to horizontal 2D lamina, loci, universal gravitational constant,
geoid and ellipsoid, 10 equal vertical attraction, 143 5, 6–7
spring gravimeters, 54–58 infinite strike length, 142 upward-continuation filter,
“spring hysteresis,” 57 unit half-width circle, properties, 119
spring length, difference, and 144–145 “upward” flux, 150
difference in gravity, 56 unit half-width ellipse, U. S. Gulf Coast, 63
stacking velocity, 83, 84 145–146 U. S. Navy, 63
errors resulting from Tikhonov regularization
dispersion of energy, 83 approach, 164 variable Bouger density, 96
station gravity, 97 Tikhonov’s curve, 161, 162 variable density, 170
stations, separation, based on time-lapse (4D-gravity) vertical component of gravity,
depth of target, 66 surveys, 189–190 calculating, 25
steradians, 19 time-lapse gravity surveys, vertical motion, airborne and
“still” reading, 57 Prudhoe Bay reservoir, 140 marine surveys, effects, 66
Stokes’ theorem, 7 topographic density, variable, 110 vertical prism, finite depth,
strike length of geology, total gradient, and 3D analytic gravity calculations, 32–34
relative to cross-sectional signal, 124 vibrating-string gravimeters,
dimensions, 43 3D analytic signal, and total 58–59
string device, 68 gradient, 124 advantage of size, 59
structural index, simple bodies, 3D body in Cartesian system, and boreholes, 58
134 arbitrary, gravitational and submarines, 58
submarines, 58, 63, 88 attraction, 26 double-string, double-mass
and vibrating-string 3D seismic surveys, 65, 66 system, 58
gravimeters, 58 tidal gravity, 89 Vitória-Trinidade chain,
sulfur, matrix density, 76 tie lines, 66 offshore Brazil, 193
sulfur deposits, exploration for time-lapse gravity surveys volcanic rocks, gravity, and
by gravity surveys, 176 (4D), 65 magnetics, 183
and seismic velocities, 176 time variations, 88–90 volume distributions
integrated with seismic- TOPEX, gravimetry satellite represented by surface
reflection data, 176 system, 191 distributions, accuracy, 46
sun, 89 topographic cliff, gravity
superconducting gravity meter, across, 91 water, density, 73
59–60 topographic maps, and Bouguer water mounding, and gravity
and beam displacement, 59 gravity-anomaly maps, 96 anomalies, 61
differential equation of topography, 88 wavelength, and spacing of
motion, 59 torsion balance, 2, 7, 54, 62 stations, 68
iGrav SG meter, 59 tradeoff parameter, 161 wavelength, shortest observable,
sensitivity, 59 transform faults, offshore and achievable accuracy, 68
superposition, law of, 85 Brazil, 193 wavelet denoising, 125
surface density, 28 “turning point,” 148–149 wavelet transform, and
surface gravity, in combination and pipelike bodies, 149 potential-field methods, 124
with borehole gravity, and thick bodies, 149 “wedge” model, 94
181–182 and thin bodies, 148–149 West Africa, offshore, 2D gravity
2D seismic surveys, and data inversion, 193–194
Taylor series, 108 gravity-data surveys, 65 Wiener filters, and amplitude
tears, abrupt step function 2D targets, gravitational filters, 124
changes, 89 attraction, 43 wollastonite, density, 79
terrain, anomalies that correlate 2D thin sheet, gravitational Worden and Scintrex spring
with, 96 anomaly, 40 gravimeter, 55
terrain correction, 93 Worden gravity meter, 58
terrain-correction compartment, uncertainty in measurement, Worden instruments, 89
94 68–69 world network, 88
Texas, 74 underground density represent- Wyoming overthrust belt,
theoretical gravity, λ, 10, 97 ed by rectangular cells, 159 182–183
thermal anomalies, lithospheric unique solutions, 46 Bouguer gravity anomaly, 182
thinning and gravity unit half-width circle (2D) and
models, 191 ellipse, 4, 141–147 Zambujal, Portugal, gravity
thin-plate model, 141–147 half-width depth rules, unit survey, 184
depth-rule diagrams, 2D and half-width circle and zero-length spring, 55
3D, 147 ellipse, 144–147
horizontal 2D cylinder, loci, United States, absolute-gravity

You might also like