Practical Geostatistics

2000

Isobel Clark
Geostokos (Ecosse) Limited,
Scotland

William V Harper
Otterbein College, Westerville,
Ohio, USA
Ecosse North America LLC, Columbus, Ohio, USA.
umpteenth Reprint, November 2008

c 2007 by Geostokos (Ecosse) Limited, Scotland.

Copyright

All rights reserved. Published simultaneously worldwide on CD.

Reproduction or translation of any part of this work beyond that permitted

by Section 107 or 108 of the 1976 United States Copyright Act without
the permission of the copyright owner is unlawful. Requests for
permission or further information should be addressed to Permissions
Department, Geostokos (Ecosse) Limited, Alloa Business Centre,
Whins Road, Alloa FK10 3SA, Central Scotland.
Tel: +44 (0)773 663 1632 Fax: +44 (0)1259 215274
e-mail: geoecosse@stokos.demon.co.uk
http://www.kriging.com http://ecossenorthamerica.com

Cataloging in Publication Data:

Clark, Isobel and Harper, William V.
Practical Geostatistics 2000 / Isobel Clark & William V. Harper.
430p. 2.5cm.
ISBN: 0-9703317-0-3 ; 0-9703317-1-1 (CD-Rom)
1. Geology Statistical methods. 2. Mining engineering. 3. Environmental
engineering I. Harper, William V. II. Title
QE33.2.S82 C43 2000
519.5024553dc21

Ecosse North America, LLC, Columbus, Ohio USA, publishers.

Typeset using Scientific Word 5.5, MacKichan Software Inc,

http://www.mackichan.com
 Bill would like to dedicate this book to
his lovely wife Paula
and
wonderful daughter Rachel

Isobels eternal gratitude goes out to

her erstwhile personal assistant, Anne Anderson

The 1979 Practical Geostatistics started with a quotation from the poem The Chorus
in Edwin Morgans anthology The New Divan, published by Carcanet/Manchester in 1977:

What is obscure is seldom clear

For this new book, longer and (we hope) much improved,
we use a longer but just as ironic quotation:

Let us all be quiet in unison

and hear what the oracle has to tell us.
The oracle says, when winter comes
the autumn is over. Are we comforted?
Special Request for our Readers

We have a favour to ask of you, the readers - we want your guidance on what changes should
be made in subsequent editions. We hope to update this book fairly frequently as we are
publishing this ourselves and have total control over what it contains, how many copies are
printed, etc. In addition to sending us questions and typos, we hope that you will take the
time to specify what new topics we should add (or what topics should be expanded). Make
sure Isobel gets a copy of all such comments. E-mail her at PG2000@kriging.com and she will
post all typos and major comments on the web page

Thanks (in advance) for all help in this eort.

 Contents
Preface

Notation

1. Introduction Page 1

1.1 Expectations
1.2 The problem to be solved
1.3 Data sets
1.4 Software

2. Why a statistical approach? Page 7

2.1 Investigating the sample data

2.2 Measures of central tendency
2.3 Measures of spread or variability
2.4 Graphical descriptions of the data
2.5 Other useful descriptive statistics
2.6 Discrete data
2.7 Into the unknown
2.8 Worked examples
2.8.1 Coal project data, calorific values
2.8.2 Iron ore example
2.8.3 Wolfcamp data
2.8.4 Scallops, total catch
2.9 Exercises

3. Normal (Gaussian) distributions Page 31

3.1 The gap between data and population

3.2 Is it a Normal distribution?
 3.3 Estimating population parameters
3.3.1 estimating the population average
3.3.2 estimating the standard deviation
3.3.3 confidence intervals for standard deviation
3.3.4 confidence intervals for mean
3.4 Selection (grade/tonnage) calculations
3.5 Summary of chapter
3.6 Worked examples
3.6.1 Coal project, calorific values
3.6.2 Iron ore example
3.6.3 Wolfcamp data
3.6.4 Scallop data, total catch
3.7 Exercises

4. Lognormal distributions (and others) Page 67

4.1 The lognormal distribution

4.1.1 estimating the mean of a lognormal population
4.1.2 confidence intervals on the population mean
4.2 The three parameter lognormal
4.3 Selection (grade/tonnage) calculations
4.3.1 two parameter lognormal reef widths
4.3.2 three parameter lognormal gold grades
4.4 More complex distributions and mixtures
4.4.1 mixtures of Normal or lognormal populations
4.5 Worked examples
4.5.1 Scallops, total caught
4.5.2 Organic matter in soil
4.5.3 Calcium in limestone
4.5.4 Geevor Tin mine, Cornwall
4.6 Exercises
5. Discrete distributions Page 103

5.1 Review of Discrete Moments

5.2 Bernoulli and Binomial Distributions
5.3 Negative Binomial and Geometric Distributions
5.4 Poisson Distribution
5.5 Mixtures of Poisson Distributions (Compound Poisson)
5.5.1 Oswego Zircon data
5.5.2 Other examples
5.6 Spatial Considerations
5.7 Solved Problems
5.8 Exercises

6. Hypothesis testing Page 135

6.1 Single sample tests

6.1.1 test on sample mean
6.1.2 test on sample standard deviation
6.2 Two sample tests
6.2.1 test on standard deviations
6.2.2 test on means
6.2.3 paired sampling
6.2.4 test for sample distribution
6.3 Worked examples
6.3.1 Heights of students
6.3.2 Geevor tin mine development versus stope
6.4 Exercises

7. Relationships Page 147

7.1 Straight line relationships

7.1.1 quantifying the strength of the relationship
7.1.2 Predicting one variable from the other
7.1.3 Calorific Value versus Ash Content
7.1.4 Calorific Value versus Sulphur Content
7.2 Other worked examples
 7.2.1 Gold grade versus reef width
7.2.2 Scallops caught
7.2.3 Application Kriges Regression Eect
7.3 Relationships involving more than two variables
7.3.1 Predicting Sulphur from Calorific Value and Ash Content
7.3.2 Application Kriges moving average template
7.3.3 Curvilinear Regression
7.3.4 Application Polynomial Trend Surface Analysis
7.4 Exercises

8. The spatial element Page 185

8.1 Including location as well as value

8.2 Spatial relationships
8.3 Inverse distance estimation
8.4 Worked examples
8.4.1 Coal project, calorific values
8.4.2 Iron ore project
8.4.3 Wolfcamp data
8.4.4 Scallops caught
8.5 Exercises

9. The semi-variogram Page 207

9.1 The experimental semi-variogram

9.1.1 Irregular sampling
9.1.2 Cautionary notes
9.2 Modelling of the semi-variogram function
9.2.1 The linear model
9.2.2 The generalised linear model
9.2.3 The Spherical model
9.2.4 The exponential model
9.2.5 Gaussian model
9.2.6 The hole eect model
9.2.7 Paddington mix model
9.2.8 Judging how well the model fits the data
9.2.9 equivalence to covariance function
 9.2.10 the nugget eect
9.3 Worked examples
9.3.1 Silver example from Practical Geostatistics 1979
9.3.2 Coal project: calorific values
9.3.3 Wolfcamp aquifer
9.4 Exercises

10. Estimation and Kriging Page 247

10.1 Estimation error

10.1.1 one sample estimation
10.1.2 another single sample
10.1.3 two sample estimation
10.1.4 another two sample estimation
10.1.5 three sample estimator
10.2 Choosing the optimal weights
10.2.1 three sample estimation
10.2.2 the general form for the optimal estimator
10.2.3 confidence levels and degrees of freedom
10.2.4 simple kriging
10.3 Ordinary kriging
10.3.1 optimal unbiassed estimator
10.3.2 alternate form: matrices
10.3.3 alternate form: covariance
10.3.4 three sample estimation
10.4 Worked examples
10.4.1 Coal project, calorific values
10.4.2 Iron ore example, (Page95)
10.4.3 Wolfcamp, residuals from quadratic surface
10.5 Cross validation
10.5.1 cross cross validation
10.6 Exercises
11. Areas and volumes Page 295

11.1 The impact on the distribution

11.1.1 Iron ore example, Normal example
11.1.2 Geevor Tin Mine, lognormal(ish) example
11.2 The impact on kriging
11.2.1 the use of auxiliary functions
11.2.2 Iron ore example, Page 95
11.2.3 Wolfcamp aquifer, quadratic residuals

12. Other kriging approaches Page 315

12.1 Universal kriging

12.1.1 Wolfcamp aquifer
12.2 Lognormal kriging
12.2.1 the proportional eect
12.2.2 the lognormal transformation
12.2.3 Geevor Tin Mine, grades
12.2.4 SA Gold Mine
12.3 Indicator kriging
12.4 Rank uniform kriging
12.5 Summary of chapter

13. Bibliography Page 325

Tables

Data Sets

Index
 Preface
This is the bit where the authors traditionally thank everyone who helped in the production
of the book and set the tone for the rest of the tome.

Isobel Clark

Isobel:

Thanks guys and girls. There are way too many people who have made contributions over the
years to expect to cover them all. The 11 years at the Royal School of Mines in London and
the 10 at the University of the Witwatersrand form the basis for the teaching material and
exercises in the bulk of these chapters. Open short courses and in-house courses to all manner
of audiences added to and enhanced both the direction of the teaching and the exercises
provided. So to all those ex-undergraduates, ex-postgraduates and course participants on
four continents, thanks.

Thanks to Dick Minnit, Fred Cawood and Hennie Greyvenstein in Johannesburg for bugging
me until the book actually got written. Thanks to Audun Clark for unconditional support.
Thanks to Gavin Lind for assistance in courses and ancillary material. And, of course, endless
thanks to my co-author who made all this possible.................

This book is intended for self-study, especially in its CD form. We hope to develop this into a
full interactive teaching medium in the future. You can also study the book and gain academic
credits at http://www.edumine.com. All comments are welcome although positive criticism
is likely to get more response!

Contact me: Dr Isobel Clark, Geostokos (Ecosse) Limited,

Alloa Business Centre, Whins Road, Alloa, FK10 3SA, Central Scotland
Tel: 0773 663 1632 Fax: 01259 215274. e-mail: drisobelclark@kriging.com
 Bill Harper

Bill: Many moons ago in the early 1980s Howard Hume and I were trying to dig into the
seemingly strange field of geostatistics. Much was not intuitive and seemed to hide behind a
strange wall of jargon. Howard and I decided to contact those individuals that we felt might
be able to shed light down the borehole. Fortunately we heard from Isobel (whose 1979 book
was one of the few readable geostatistics publications at that time) and she changed her next
U.S. trip to come to Columbus to visit us at Battelle in 1984. As they say, the rest is history.
Isobel and I collaborated on numerous articles in the 1980s. Perhaps my biggest claim to
geostatistical fame was publishing the Wolfcamp data in my first geostatistical analysis. This
resulted in many publications by others using this particular set of data.

In the late 1980s I moved away from environmental geostatistical applications after the Battelle
high-level nuclear waste program was terminated by the U.S. government. Until early 1999
most of my 1990s eorts have been focused on more classical statistical methods, operations
research, quality management, and discrete event simulation. In Spring 1999 Isobel was
staying at our place for a few days visit. After several enjoyable beverages, I raised the issue
of her many years of promising to update her book. The result of that discussion is now before
your eyes. As the Grateful Dead used to sing, What a long, strange trip its been.

I would like to recognize the invaluable support from my lovely wife, Paula, and daughter,
Rachel. I am also thankful of the continued good support of Otterbein College. Finally, I
want to thank Isobel for being Isobel.

Please send us any and all feedback. We hope to update this often to meet your requests.

Contact me: William V. Harper, Otterbein College, Mathematical Sciences

Towers Hall 139, 1 Otterbein College, Westerville, OH 43081-2006 U.S.A.
Phone: 614-823-1417 Fax: 614-823-3201
WHarper@otterbein.edu http://faculty.otterbein.edu/WHarper/
sales at: http://ecossenorthamerica.com/
 Notation
the range of influence or distance scaling parameter in a model semi-variogram
breadth of a panel or block under study
coecients estimated in a regression or trend surface analysis
cuto or pay value used in a grade/tonnage calculation
distance between a sample and a location at which an estimate is required
2.718281828..... the natural anti-logarithm function
a secondary or ancillary variable measured on a sample
the primary value measured at a sample location
the distance between any two specified locations within the study area
a subscript identifier
a subscript identifier
a subscript or integer power
length of a panel or block under study
number of local samples in an estimation
number of samples available in the current study
slope of the line in a linear semi-variogram model
estimate for the correlation coecient between two variables
unbiassed estimate for the standard deviation of a population
Students distribution for the arithmetic mean of Normal samples
Sichels estimator for the average of a lognormal population
a volume
weighting factor
standardised Normal variate, mean 0 standard deviation 1
logarithm of value
residual from a trend surface
A an area or volume whose value is to be estimated
C nugget eect 0 or sill on a semi-variogram model
F auxiliary function for the variance within a panel or block
H auxiliary function for a corner point with a 2d panel
M a specified point within an area or volume
N number of pairs of samples found in a semi-variogram calculation
P proportion of the population above a specified cuto value or pay limit
may also be used to represent percentage rather than proportion
Q an auxiliary function between two lines at right angles
S sum of squared residuals from a regression exercise
T the unknown value at an unsampled location
the distribution of Sichels estimate for the lognormal mean
V the variance of samples around their arithmetic mean
X co-ordinate for a location, usually east/west
Y co-ordinate for a location, usually north/south
Z co-ordinate for a location, usually elevation or depth
 Notation continued
lower case Greek letters
alpha power to which distance is raised in a semi-variogram model
the significance level of a hypothesis test
beta additive constant used in a three parameter lognormal distribution
gamma Sichels correction factor for the lognormal mean estimation
a model semi-variogram or function calculated from such
delta a small change
epsilon the error of estimation, actual valueestimated value
theta an angle usually expressed in degrees
lambda Lagrangian multiplier
mu arithmetic mean of values in a Normal population
nu degrees of freedom
eta damping or decay parameter in a hole eect semi-variogram model
pi 3.141592654.... the ratio of the perimeter of a circle to its diameter
rho the population correlation coecient between two variables
sigma the standard deviation of values in a Normal population
tau the arithmetic mean of values in a lognormal population
phi the standard Normal density function
chi the distribution of sample variances from Normal samples
omega the standard deviation of values in a lognormal population

Upper case Greek letters

Sigma a summation sign, e.g. =1 is sum for series = 1 2 3 1
Phi cumulative proportion under the standard Normal distribution
Psi Sichels confidence factors for the lognormal distribution
41 Lqwurgxfwlrq
4141 H{shfwdwlrqv
Ehiruh |rx vwduw uhdglqj wklv errn/ zh zrxog olnh wr pdnh lw fohdu h{dfwo| zkdw |rx fdq
+dqg fdq*w, h{shfw iurp lw dqg zkdw zh gr +dqg grq*w, h{shfw iurp wkh uhdghu1 Wklv wh{w lv
edvhg rq 5; |hduv ri frxuvhv wdxjkw wr plqlqj hqjlqhhuv/ jhrorjlvwv/ k|gurorjlvwv/ vrlo vflhq0
wlvwv/ folpdwrorjlvwv soxv wkh rffdvlrqdo jhrjudskhu/ sdwwhuq uhfrjqlwlrq h{shuw/ phwhrurorjlvw/
vwdwlvwlfldq dqg frpsxwhu vflhqwlvw1 Hyhq/ rq rqh rffdvlrq/ dq dffrxqwdqw1 Ryhu wkrvh |hduv/
zh kdyh hqghdyrxuhg wr sduh dzd| doo h{wudqhrxv pdwkhpdwlfv dqg frqfhqwudwh rq lqwxlwlyh
ghulydwlrqv zkhuh srvvleoh1 Uhdghuv lqwhuhvwhg lq uljrurxv pdwkhpdwlfdo surriv duh xujhg wr
vwrs khuh dqg wxuq wr wkh pruh wkhruhwlfdoo| edvhg errnv +fi1 Uhihuhqfh Wh{wv lq Eleolrjud0
sk|,1 Wklv errn lv qrw lqwhqghg wr wxuq rxw ixoo| hgjhg jhrvwdwlvwlfldqv1 Lw lv lqwhqghg iru
shrsoh zlwk sureohpv wr eh vroyhg zklfk fdq eh dvvlvwhg e| d jhrvwdwlvwlfdo dssurdfk1
Wr uhdg wklv errn dqg ehqhw iurp lw |rx qhhg wr eh idluo| frpiruwdeoh zlwk edvlf dojheud1
Wkdw lv/ zlwk wkh qrwlrq ri xvlqj v|perov dv vkruwkdqg iru orqjhu vwdwhphqwv1 Zh kdyh zrunhg
kdug wr eulqj |rx d frqvlvwhqw qrwdwlrq wkurxjkrxw wkh errn1 Zkhuh qrwdwlrq lv rxw ri rxu
frqwuro/ zh h{sodlq fduhixoo| zkdw hdfk v|pero vwdqgv iru dqg wu| qrw wr xvh wkdw v|pero iru
dq|wklqj hovh1 Wklv lv qrw dozd|v srvvleoh1 Iru h{dpsoh/ Vwxghqw +Zlooldp Jrvvhw, ghyhorshg
klv glvwulexwlrq iru wkh phdq ri d vhw ri vdpsohv dqg fdoohg lw wkh | glvwulexwlrq1 Khuehuw Vlfkho
ghyhorshg dq hvwlpdwru iru wkh phdq ri d orjqrupdo glvwulexwlrq dqg fdoohg lw +vxusulvh, |1
Fdofxoxv  glhuhqwldwlrq dqg lqwhjudwlrq  lv glvfxvvhg dw ydulrxv srlqwv lq wkh wh{w1 Wkh
uhdghu lv qrw h{shfwhg wr gr dq| fdofxoxv +dv vxfk, exw lv h{shfwhg wr nqrz wkdw wkh glhuhqwldo
ri % lv 2%1 Wkh rqo| rwkhu frpsolfdwlrq lv wkh iuhtxhqw xvh ri vlpxowdqhrxv htxdwlrqv1 Zh
whqg qrw wr xvh pdwul{ dojheud lq wklv errn exw zloo jlyh wkh pdwul{ irup diwhu h{sodqdwlrqv
kdyh ehhq jlyhq lq vlpsoh dojheud1 Iru h{dpsoh/ olqhdu uhjuhvvlrq lv hdvlhu wr xqghuvwdqg li
ghyhorshg zlwk dojheud/ exw yhu| vlpsoh wr lpsohphqw lq vsuhdgvkhhwv ru lq sdfndjhv vxfk dv
PdwOdeA  li pdwulfhv duh xvhg1
Li zh kdyhq*w vfduhg |rx r |hw/ eh uhdvvxuhg e| wkh idfw wkdw doo wkh dqdo|vhv duh looxvwudwhg
zlwk uhdo gdwd vhwv lq ixoo zrunhg dqvzhuv1 Li |rx kdyh wkh FG/ wkh gdwd vhwv duh lqfoxghg
dorqj zlwk vriwzduh wr uhsurgxfh wkh dqdo|vhv +iru wkh prvw sduw,1 Li |rx duh uhdglqj wkh kdug
frs|/ wkh gdwd vhwv dqg vriwzduh fdq eh grzqordghg iurp wkh Zhe1 Wkhuh duh h{huflvhv iru
|rx wr wu|1 Dqvzhuv duh dydlodeoh iru |rx wr fkhfn |rxu uhvxowv1 Prvw ri wkhvh h{huflvhv kdyh
ehhq froohfwhg dqg xvhg lq fodvvhv ru h{dplqdwlrqv dw Ilqdo +Vhqlru, \hdu dqg Pdvwhu*v ohyhov1
Lw lv rxu rzq ixqgdphqwdo uhjuhw wkdw wklv errn fdqqrw frqwdlq wkh mrnhv/ dqhfgrwhv dqg
vkhhu ixq wkdw zh kdyh rq wkh frxuvhv1 Zh gr dgylvh |rx/ krzhyhu/ wr nhhs |rxu vhqvh ri
kxprxu dqg frpprq vhqvh wr wkh iruh dw doo wlphv zkloh uhdglqj wklv errn1

4151 Wkh sureohp wr eh vroyhg

Jhrvwdwlvwlfv  dv glvfxvvhg lq wklv errn  zdv ghyhorshg lq jhrorj| dqg plqlqj1 Krzhyhu/
wkh sureohp zklfk lw zdv ghyhorshg wr wdfnoh lv pruh jhqhudo wkdq jhrorjlfdo dssolfdwlrqv1 Wklv
wh{w lv lqwhqghg dv d edvlf lqwurgxfwlrq wr vwdwlvwlfdo dqg jhrvwdwlvwlfdo dqdo|vlv ri vdpsoh gdwd
zklfk srvvhvvhv d orfdwlrq dv zhoo dv dw ohdvw rqh revhuyhg ydoxh1

Sudfwlfdo Jhrvwdwlvwlfv 5333 Sdjh 4

Wkhuh lv riwhq frqixvlrq dv wr wkh lqwhqghg remhfwlyh ri jhrvwdwlvwlfdo whfkqltxhv1 Zh ghqh
wkhp khuh dv wzrirog=

41 wr fkdudfwhulvh dqg lqwhusuhw wkh ehkdylrxu ri wkh h{lvwlqj vdpsoh gdwd>

51 wr xvh wkdw lqwhusuhwdwlrq wr suhglfw olnho| ydoxhv dw orfdwlrqv zklfk kdyh qrw |hw ehhq
vdpsohg1

Wr vhw wkh vfhqh iru wkh uhvw ri wkh errn/ ohw xv lpdjlqh wkdw wkhuh lv d +pruh ru ohvv, frqwlqxrxv
skhqrphqrq zklfk fryhuv d vwxg| duhd +ru yroxph,1

Iljxuh 414= Srwhqwlrphwulf ohyhov ryhu Zroifdps vwxg| duhd

Vrph vdpsohv kdyh ehhq wdnhq ryhu wkh vwxg| duhd dqg wkhlu orfdwlrqv qrwhg1 Phdvxuhphqwv
kdyh ehhq pdgh rq wkh vdpsohv wdnhq1 Rxu pdmru wdvn lv wr hvwlpdwh wkh olnho| ydoxh dw d
orfdwlrq zklfk kdv qrw ehhq vdpsohg1

Wkhuh duh pdq| glhuhqw zd|v wr wdfnoh wklv sureohp1 Wklv errn fryhuv mxvw rqh dssurdfk
zklfk lv edvhg rq d zhoo ghqhg vhw ri dvvxpswlrqv1 Rwkhu dvvxpswlrqv ohdg wr rwkhu phwkrgv1

D orw ri wkh fulwlflvp zklfk lv ohyhoohg dw jhrvwdwlvwlfdo hvwlpdwlrq lv irxqghg rq plvfrqfhswlrqv

derxw wkh fdsdelolwlhv dqg lqwhqwlrqv ri wkh phwkrg +fi1 vhfwlrq Vfhswlfv lq Eleolrjudsk|,1 Zh

Sdjh 5 Sudfwlfdo Jhrvwdwlvwlfv 5333

 Iljxuh 415= Dfwxdo vdpsoh gdwd dydlodeoh lq Zroifdps vwxg|

zloo wdfnoh wkrvh dv zh frph wr wkhp lq wkh wh{w1 Zh zloo dovr glvfxvv wkh vkruwfrplqjv ri wkh
whfkqltxhv zklfk zloo eh ghyhorshg dv dqg zkhq dssursuldwh1 Wkh lqwhqwlrq ri wklv errn lv wr
jlyh wkh uhdghu dq xqghuvwdqglqj ri wkh vwdwlvwlfdo dqg jhrvwdwlvwlfdo whfkqltxhv zklfk pljkw
eh xvhixo/ qrw wr od| grzq dq| odzv dqg uhjxodwlrqv rq zkdw vkrxog dqg vkrxog qrw eh xvhg1
Wkh vwdwlvwlfdo sruwlrqv ri wklv errn duh lqwhqghg wr od| wkh jurxqgzrun iru wkh jhrvwdwlvwlfdo
dqdo|vlv1 Pxfk ri wklv pdwhuldo fdq eh irxqg lq irxqgdwlrq vwdwlvwlfv errnv exw qrw lq wkh
fxuuhqw frqwh{w1 Wkh jhrvwdwlvwlfdo sruwlrqv ri wkh errn dvvxph wkdw |rx kdyh pdvwhuhg wkh
vwdwlvwlfdo whfkqltxhv ghvfulehg hduolhu1 Lw lv qrw dgylvdeoh wr cvnls dkhdg* rq wkh dvvxpswlrq
wkdw zkdw lv ehlqj glvfxvvhg kdv qr uhohydqfh wr |rxu rzq lqwhuhvwv1 Wkh ghyhorsphqw lv
h{wuhpho| olqhdu/ lq wkdw rqh vhfwlrq ohdgv lqwr dqrwkhu1 Wkhuh duh h{fhswlrqv wr wklv/ ri
frxuvh1 Iru h{dpsoh/ li |rx zloo qhyhu kdyh wr ghdo zlwk vnhzhg gdwd/ |rx fdq vnls wkh fkdswhu
rq wkh orjqrupdo glvwulexwlrq dqg lwv yduldqwv1 Li |rx zloo qhyhu ghdo zlwk pruh wkdq rqh
phdvxuhphqw shu vdpsoh/ |rx fdq vnls prvw ri wkh Uhodwlrqvklsv fkdswhu1 Li |rx qhyhu ghdo
zlwk gdwd zklfk kdv d wuhqg lq wkh ydoxhv/ |rx fdq vnls doo exw wkh uvw ihz sdjhv ri wkdw
fkdswhu1

4161 Gdwd vhwv

Wkh vruw ri dssolfdwlrqv suhvhqwhg zlwklq wkh errn duh pdlqo| jhrorjlfdo zlwk vrph k|gurorj|
dqg hqylurqphqwdo fdvh vwxglhv1 Wkh srwhqwldo dssolfdwlrqv lqfoxgh dq| irup ri phdvxudeoh
vsdwldo gdwd dqg vrph zklfk fdqqrw eh jlyhq d txdqwlwdwlyh phdvxuh/ vxfk dv urfn w|sh/ odqg

Sudfwlfdo Jhrvwdwlvwlfv 5333 Sdjh 6

xvh hwf1 Zh kdyh lqfoxghg dssolfdwlrqv ri jhrvwdwlvwlfdo whfkqltxhv lq wkh iroorzlqj hogv +vr
idu,=

 Frdo= d vlpxodwhg vhw ri gdwd edvhg rq d uhdo frdo vhdp lq Vrxwkhuq Diulfd1 Eruhkrohv
guloohg lqwr wkh frdo vhdp duh phdvxuhg iru= wklfnqhvv ri frdo +phwuhv,/ hqhuj| frqwhqw
ru cfdorulf ydoxh* ri frdo +Phjdmrxohv shu wrqqh,> dvk frqwhqw +(, dqg vxoskxu frqwhqw
+(,1 Wkuhh fr0ruglqdwhv lq phwuhv duh dydlodeoh iru wkh wrs ri wkh frdo vhdp zkhuh
lqwhuvhfwhg e| wkh gulookroh1

 JDVD= wklv gdwd vhw lv qdphg iru wkh Jhrvwdwlvwlfdo Dvvrfldwlrq ri Vrxwk Diulfd dqg
zdv xvhg lq dq looxvwudwlrq ri jhrvwdwlvwlfdo whfkqltxhv dw d phhwlqj lq Dsulo 4<;: lq
Mrkdqqhvexuj1 Wkh vdpsoh gdwd duh wdnhq iurp ghhs eruhkrohv guloohg lqwr d w|slfdo
Zlwzdwhuvudqg w|sh jrog uhhi1 Wkh phdvxuhphqwv ri lqwhuhvw duh wkh judgh ri wkh jrog
lq judpv shu wrqqh ri urfn +sduwv shu ploolrq, dqg wkh wklfnqhvv ri wkh uhhi lqwhuvhfwlrq
lq wkh eruhkroh +fhqwlphwuhv,1 Wkh 5: eruhkrohv olh dssur{lpdwho| 4 nlorphwuh dsduw dqg
frqvwlwxwh d w|slfdo gdwd vhw iru wkh sodqqlqj dqg ghvljq ri d qhz Zlwv jrog plqh1 Wkh
ydoxhv kdyh ehhq glvjxlvhg e| d idfwru exw duh rwkhuzlvh xqdowhuhg1 Fr0ruglqdwhv duh lq
phwuhv1

 Vdpsohv= wklv gdwd vhw lv edvhg rq d Zlwv w|sh jrog plqh vrph ghfdghv lqwr surgxfwlrq1
Wkh vdpsohv duh fklsshg iurp wkh idfh ri wkh uhhi lq d zrunlqj vhfwlrq ri wkh plqh
+vwrsh,1 Dv wkh idfh dgydqfhv/ qhz fkls vdpsohv duh wdnhq1 Ydoxhv zlwklq d vwrsh
duh wudglwlrqdoo| hvwlpdwhg xvlqj wkh vdpsoh ydoxhv iurp wkh idfh1 Wklv gdwd lv wrwdoo|
fwlwlrxv h{fhsw iru wkh orfdwlrqv ri wkh vdpsohv/ zklfk duh wdnhq iurp d uhdo Zlwv w|sh
jrog plqh1

 Frsshu= d vlpxodwlrq edvhg rq d vwrfnsloh ri plqhg pdwhuldo lq wkh iruphu Vrylhw Xqlrq1
Eruhkrohv kdyh ehhq guloohg lqwr wkh gxps1 Wkh guloo fruh lv fxw hyhu| 8 phwuhv dqg
dvvd|hg iru frsshu dqg fredow frqwhqw lq shufhqwdjh e| zhljkw1 Wklv lv wkh rqo| wkuhh
glphqvlrqdo vhw ri wxwruldo gdwd1 Fr0ruglqdwhv duh lq phwuhv1

 Jhhyru= wklv lv vdpsoh gdwd iurp d k|gurwkhupdo wlq ghsrvlw lq Fruqzdoo/ Hqjodqg1 Wkh
plqhudolvdwlrq dsshduv dv d frqwlqxrxv yhlq zklfk lv vxe0yhuwlfdo1 Vdpsohv ri durxqg
4nj duh fklsshg dfurvv wkh yhlq/ zklfk dyhudjhv derxw 57 lqfkhv zlgh1 Phdvxuhphqwv
duh judgh ri wlq lq srxqgv ri eodfn wlq +VqR , shu wrq ri urfn1 Wkh wklfnqhvv ri wkh
yhlq ru corgh* lv phdvxuhg wr wkh qhduhvw lqfk1 Fr0ruglqdwhv duh lq ihhw dorqj vhfwlrq dqg
hohydwlrq deryh dq duelwudu| edvh ohyho1
Fodun/ L1/ 4<:</ Grhv jhrvwdwlvwlfv zrunB/ Surf1 49wk DSFRP/ Wkrpdv M R*Qhlo/ Hg1/
Vrflhw| ri Plqlqj Hqjlqhhuv ri DLPH Lqf/ Qhz \run/ 54605581

 Zroifdps= phdvxuhphqwv ri zdwhu suhvvxuh +srwhqwlrphwulf ohyho, lq ;8 zdwhu zhoov lq

wkh Wh{dv sdqkdqgoh1 Wklv gdwd vhw zdv sduw ri d vwxg| fduulhg rxw e| wkh R!fh iru
Qxfohdu Zdvwh Lvrodwlrq lq wkh plg 4<;3v rq d srwhqwldo vlwh iru d kljk ohyho qxfohdu
zdvwh uhsrvlwru|1 Wkh Zroifdps dtxlihu xqghuolhv wkh sodqqhg uhsrvlwru|1 Rqh dvshfw
ri uhsrvlwru| sodqqlqj lv wr txdqwli| wkh ulvnv lqkhuhqw lq d euhdfk ri wkh vwrudjh idflolw|1

Sdjh 7 Sudfwlfdo Jhrvwdwlvwlfv 5333

 Vkrxog udglrqxfolghv ohdn lqwr wkh orfdo dtxlihuv/ wkh vfrsh dqg vshhg ri srwhqwldo frq0
wdplqdwlrq kdv wr eh dvvhvvhg1 Wkh suhvvxuh ri xlg zlwklq wkh dtxlihu zdv rqh ri vhyhudo
yduldeohv xvhg wr ghwhuplqh wkh wudyho sdwk dqg vshhg ri wudyho iru hvfdshg udglrqxfolghv1
Uhihuhqfh= Kdushu/ Z1Y1/ dqg Ixuu/ M1P1/ 4<;91 Jhrvwdwlvwlfdo dqdo|vlv ri srwhqwlr0
phwulf gdwd lq wkh Zroifdps Dtxlihu ri wkh Sdor Gxur Edvlq/ Wh{dv/ EPL2RQZL08;:/
Dsulo/ R!fh ri Qxfohdu Zdvwh Lvrodwlrq/ Edwwhooh Phpruldo lqvwlwxwh/ Froxpexv/ Rklr1

 Vfdoorsv= Vfdoors gdwd zhuh froohfwhg gxulqj d 4<<3 vxuyh| fuxlvh r wkh hdvw frdvw ri
Qruwk Dphulfd1 Vfdoors frxqwv zhuh rewdlqhg xvlqj d guhgjh1 Dq| vfdoors vpdoohu wkdq
:3 pp zdv whuphg d suhuhfuxlw1 Wrwdo fdwfk lv wkh vxp ri suhuhfuxlwv dqg uhfuxlwv1
Phdvxuhphqwv lqfoxghg lq wkh gdwd oh duh=

 Qdwlrqdo Pdulqh Ilvkhulhv Vhuylfh +QPIV, 7 gljlw vwudwd ghvljqdwru lq zklfk wkh
vdpsoh zdv wdnhq>
 vdpsoh qxpehu shu |hdu udqjlqj iurp 4 wr dssur{lpdwho| 783>
 orfdwlrq lq whupv ri odwlwxgh dqg orqjlwxgh ri hdfk vdpsoh lq wkh Dwodqwlf Rfhdq>
 wrwdo qxpehu ri vfdoorsv fdxjkw dw wkh vdpsoh orfdwlrq>
 qxpehu ri vfdoorsv zkrvh vkhoo ohqjwk lv vpdoohu wkdq :3 ploolphwhuv>
 qxpehu ri vfdoorsv zkrvh vkhoo ohqjwk lv :3 ploolphwhuv ru odujhu1

Uhihuhqfh= Hfnhu/ P1G1/ dqg Khowvkh/ M1I1 4<<71 Jhrvwdwlvwlfdo hvwlpdwhv ri Vfdoors
Dexqgdqfh/ Lq/ Fdvh Vwxglhv lq Elrphwu|/ Odqjh hw do1/ hglwruv1 Zloh|/ Qhz \run

 Glr{lq= D wuxfn wudqvsruwlqj glr{lq frqwdplqdwhg uhvlgxhv gxpshg dq xqnqrzq txdq0

wlw| ri wkhvh zdvwhv rqwr d idup urdg lq Plvvrxul1 Lq Qryhpehu/ 4<;6/ wkh X1V1 HSD
froohfwhg vdpsohv ri wkh vlwh1 Lq rughu wr uhgxfh wkh qxpehu ri vdpsohv uhtxluhg/ vdpsohv
zhuh frpsrvlwhg dorqj wudqvhfwv1 Wkh wudqvhfwv uxq sdudooho wr wkh kljkzd|/ dqg wklv
gluhfwlrq lv ghvljqdwhg dv wkh [0gluhfwlrq1 Wkh gluhfwlrq shushqglfxodu wr wkh kljkzd|
lv ghvljqdwhg dv wkh \0gluhfwlrq1 Gdwd duh WFGG frqfhqwudwlrq +whwudfkorurglehq}r0s0
glr{lq, lq plfurjudpv shu nlorjudp +>j2nj,1 Fr0ruglqdwhv dqg wudqvhfw ohqjwk duh jlyhq
lq ihhw1
Uhihuhqfh= ]luvfk|/ M1K1/ dqg Kduulv/ G1M1 4<;91 Jhrvwdwlvwlfdo dqdo|vlv ri kd}dugrxv
zdvwh vlwh gdwd1 Mrxuqdo ri Hqylurqphqwdo Hqjlqhhulqj/ 445=::30:;71

 Rujdqlfv= Gdwd duh Vrlo Rujdqlf Pdwwhu ydoxhv +lq judpv shu nlorjudp, ghulyhg iurp vrlo
vdpsohv wdnhq lq d uhvhdufk hog dw wkh Xqlyhuvlw| ri Qheudvnd Zhvw Fhqwudo Uhvhdufk
dqg H{whqvlrq Fhqwhu qhdu Qruwk Sodwwh/ Qheudvnd/ XVD1 Gdwd zhuh wdnhq dv sduw ri
h{shulphqwv rq yduldeoh0udwh ihuwlol}hu whfkqrorj|1 Fr0ruglqdwhv duh lq phwuhv1
Uhihuhqfh= Jrwzd|/ F1D1 dqg Khujhuw/ J1Z1 +4<<:,1 Lqfrusrudwlqj Vsdwldo Wuhqgv
dqg Dqlvrwurs| lq Jhrvwdwlvwlfdo Pdsslqj ri Vrlo Surshuwlhv1 Vrlo Vflhqfh ri Dphulfd
Mrxuqdo/ 94=5<;063<

Sudfwlfdo Jhrvwdwlvwlfv 5333 Sdjh 8

  Yhoyhwoi= Vxevdpsoh ri wkh qxpehu ri yhoyhwohdi zhhgv frxqwhg lq : phwhu duhd lq d hog
lq Qheudvnd1 Gdwd zhuh froohfwhg e| Juhjj Mrkqvrq +vhh 5qg uhihuhqfh,/ dv sduw ri d
uhvhdufk surjudp lq zhhg pdqdjhphqw dw wkh Xqlyhuvlw| ri Qheudvnd1
Uhihuhqfhv= Gdwd vhw wdnhq iurp= Jrwzd|/ F1D1/ dqg Vwurxs/ Z1Z1 4<<:1 D jhqhudol}hg
olqhdu prgho dssurdfk wr vsdwldo gdwd dqdo|vlv dqg suhglfwlrq1 Mrxuqdo ri Djulfxowxudo/
Elrorjlfdo/ dqg Hqylurqphqwdo Vwdwlvwlfv/ 5=48:04:;1
Gdwd froohfwhg e|= Mrkqvhq/ J1D1/ Pruwhqvhq/ G1D1/ dqg Jrwzd|/ F1D1 4<<91 Vsdwldo
dqg whpsrudo dqdo|vlv ri zhhg vhhgolqj srsxodwlrqv xvlqj jhrvwdwlvwlfv1 Zhhg Vflhqfh/
77=:370:431

Doo ri wkh deryh fdvh vwxglhv dsshdu vrphzkhuh zlwklq wkh wh{w1 Wkh gdwd ohv duh dydlodeoh
rq wkh FG dqg fdq eh grzqordghg iurp wkh Zhe1 Doo/ h{fhsw vdpsohv dqg srvvleo| frsshu/
duh vpdoo hqrxjk wr wdfnoh dw ghvnwrs dqg kdqg fdofxodwru ohyho1 Zh vwurqjo| uhfrpphqg wkdw
|rx fduu| rxw hdfk dqdo|vlv e| kdqg dw ohdvw rqfh wr uhlqirufh wkh zulwwhq wh{w1

4171 Vriwzduh
Li |rx kdyh wklv errn rq FG/ wkh glvn dovr frqwdlqv d cghpr* yhuvlrq ri wkh Jhrvwrnrv vriwzduh
fuhdwhg vshfldoo| iru whdfklqj1 Wklv yhuvlrq kdv voljkwo| pruh ihdwxuhv wkdq wkh HfrVVh sdfndjh
dqg udwkhu ohvv wkdq wkh ixoo Jhrvwrnrv Wrronlw1 Lw lv d Zlqgrzv edvhg sdfndjh zklfk fxuuhqwo|
rshudwhv xqghu Zlqgrzv <82<; dqg QW1 Iroorz wkh lqvwdoodwlrq lqvwuxfwlrqv vxssolhg zlwk wkh
sdfndjh1 Doo ri wkh deryh gdwd vhwv duh vxssolhg rq wkh glvn1
Li |rx kdyh wklv errn lq kdug frs|/ |rx pd| grzqordg wkh vriwzduh dqg gdwd vhwv iurp
wkh Zhe1 Fkhfn |rxu gholyhu| sdfndjh iru fxuuhqw lqvwuxfwlrqv1 Ixoo olvwlqjv ri wkh gdwd vhwv
+h{fhsw iru vdpsohv, duh jlyhq lq wkh Dsshqgl{1
Wkh vriwzduh lv lghqwlfdo wr wkh vwdqgdug Jhrvwrnrv HfrVVh dqg Wrronlw vriwzduh sdfndjhv
h{fhsw wkdw lw zloo rqo| uhdg wkh gdwd ohv vxssolhg zlwk wkh errn1

Fdxwlrqdu| qrwh rq Suhflvlrq= Lw lv sduwlfxoduo| lpsruwdqw wr uhphpehu wkdw doo ri wkh

zrunhg dqvzhuv jlyhq lq wklv errn kdyh ehhq frpsxwhg xvlqj sursulhwdu| vriwzduh/ vsuhdg0
vkhhwv dqg2ru kdqg fdofxodwruv1 Doo ri wkhvh kdyh glhuhqw ohyhov ri suhflvlrq lq wkhlu pdnhxs1
Gr qrw zruu| li |rxu dqvzhu lv rxw e| dq|wklqj xs wr d frxsoh ri shufhqw frpsduhg wr wkh
rqh lq wklv errn1 Wklv lv sduwlfxoduo| vr zkhq |rx kdyh wr vtxduh/ fxeh ru udlvh qxpehuv wr d
odujhu srzhu1
Li |rx pdnh d uhdo plvwdnh/ |rxu dqvzhu zloo eh txlwh glhuhqw iurp rxuv1 Lq prvw fdvhv/
plvwdnhv gxulqj fdofxodwlrq ohdg wr kxjh glhuhqfhv lq wkh dqvzhuv1

Sdjh 9 Sudfwlfdo Jhrvwdwlvwlfv 5333

 Practical Geostatistics was written by Isobel Clark and
published in 1979 by Elsevier Applied Science when
few books were available to assist a teacher of geostatistics.

Kriging is useful as an interpolation and estimation technique in geochemistry, rock me-

chanics, geography, hydrology, forestry and many more diverse disciplines. ....... provides
the reader with a book which can be readily understood and with fully worked examples
and illustrations Engineering Societies Library Review, 1980

It is obviously written by an experienced teacher who well knows the value of worked
examples and self-teaching exercises British Book News, 1980

a useful and easy introduction Biometrics 1980

La prsentation est claire, succinte, les dveloppements mathmatiques sont rduits leur
strict minimum. Louvrage sadresse un public non spcialis. Annales de la Socit
Gologique de Belgique, 1982

Some readers may find the text rather too chatty............ Bulletin of the Institute of
Mathematics and its Applications 1981

I think Clarks book is valuable because it makes geostatistics readily understandable.

............ I hope that Dr Clark will give us a comparison of classical statistics with geo-
statistics in a future book Economic Geology 1981

This is that future book. Jointly authored by Isobel Clark and Bill Harper, Practical
Geostatistics 2000 shows how the kriging techniques of geostatistics evolve logically
from basic statistical concepts and interpolation methods such as trend surface analysis
and inverse distance weighting.
This book is not intended for specialist mathematicians or statisticians, but for those
students and professionals to whom geostatistics is one useful tool on the workbench.
Our proposed audience includes anyone who needs to produce estimates or predictions of
quantitative values at locations where samples have not been or cannot be taken.
No prior knowledge of statistics is assumed, although the reader needs to be comfortable
with basic algebra and a little calculus. It helps if you know how to solve small sets of
simultaneous equations. All chapters include fully worked examples. Most include a list
of exercises for the reader to practice on. A separate volume of answers to the exercises
is also available. Data sets are listed and can be downloaded along with the software
from the Web.
 c

Ecosse North America, LLC

Attn: William V. Harper, wharper@otterbein.edu
Mathematical Sciences, Towers Hall 139
Otterbein College
One Grove Street
Westerville, OH 43081-2006 USA
http://ecossenorthamerica.com/