ba4e3ddd028c72570ef868df39c9fe65

DESIGNING OPTIMAL
STRATEG IES FOR

MINERAL EXPLORATION
DESIGNING OPTIMAL
STRATEGIES FOR
MINERAL EXPLORATION
J. G. De Geoffroy
Consultant, Geostatistics and Mining Exploration
Eden, New South Wales, Australia
and
T. K. Wignall
California Polytechnic State University
San Luis Obispo, California
PLENUM PRESS • NEW YORK AND LONDON

Library of Congress Cataloging in Publication Data
De Geoffroy, J. G.
Designing optimal strategies for mineral exploration.
Includes bibliographies and index.
I. Prospecting-Data processing. 2. Prospecting-Mathematical models. 3.
Mathematical optimization. I. Wignall, T. K. II. Title.
TN270.D4 1985 622'.1 85-19383
ISBN-13: 978-1-4684-1232-1 e-ISBN-13: 978-1-4684-1230-7
DOl: 10.1007/978-1-4684-1230-7
© 1985 Plenum Press, New York

Softcover reprint of the hardcover I st edition 1985
A Division of Plenum Publishing Corporation
233 Spring Street, New York, N.Y. 10013
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted
in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, without written permission from the Publisher
PREFACE
Few knowledgeable people would deny that the field of mineral exploration
is facing some difficult times in the foreseeable future. Among the woes, we
can cite a worldwide economic uneasiness reflected by sluggish and at times
widely fluctuating metal prices, global financial uncertainties, and relentless
pressures on costs despite a substantial slowing down of the rate of inflation.
Furthermore, management is forced to tum to more sophisticated and expensive
technologies and to look farther afield to more remote regions, as the better-
quality and more easily accessible ore deposits have now been revealed. This
rather gloomy outlook should persuade explorationists to cast about for a new
philosophy with which to guide mineral exploration through the challenging
decades ahead.
Once already, in the early 1960s, a call for change had been heard (Ref.
30 in Chapter 1), when it became obvious that the prospecting methods of
yesteryear, so successful in the past, could not keep up with the rapidly growing
demand for minerals of the postwar period. The answer, a massive introduction
of sophisticated geophysical and geochemical technologies backed by new geo-
logical models, proved spectacularly successful throughout the 1960s and the
1970s. But for both economic and technological reasons, the brisk pace of the
last two decades has considerably slowed down in the early 1980s, as if a new
threshold has been reached. We believe that the answer lies in a concerted effort
to make an efficient use of the technology at hand, which can be translated into
the earthy language of the man in the field as "getting more bang for the explo-
ration buck," or into the more esoteric language of management circles as "max-
imizing the efficiency of operations under budget constraints"-in other words:
optimization, which is what the new philosophy of mineral exploration should
be about.
Optimization is nothing new in itself. It has been a household word in
engineering and industry circles for quite some time. Optimization has been
instrumental in the spectacular reduction in energy demand witnessed in the past
few years, by providing a much more rational and efficient use of the resources
at hand. Unfortunately, however, the new philosophy of optimization does not
as yet seem to have made much inroad into the thinking of the mineral explo-
rationists of the 1980s. A search of the English language literature on the mineral
v
vi PREFACE
industry published during the past two decades revealed a great wealth of material
on the optimization of the production section of the business, but only precious
few journal articles on the optimization of the exploration section, although the
necessary methodology and the required computational facilities are available.
Mineral exploration, being essentially a sequential procedure, is eminently
suited for optimization, despite its inherent context of uncertainty and risk which
require the use of probabilistic methods, as opposed to the deterministic approach
that prevails in engineering. The optimization of mineral exploration as a whole
and that of the various individual stages of the exploration sequence constitutes
a rather forbiddingly large subject matter which would require several volumes
to do justice to it. Because of format restrictions we are limiting ourselves to
two practical problems of exploration planning and field investigation, including
(1) the optimization of survey designs for the detection of specified types of
mineral deposits, and (2) the optimal selection for drill testing purposes of
exploration targets outlined by field surveys.
The first part of the book, comprising Chapters 1-5, describes the meth-
odology of the optimization of the detection of ore deposits. The subject matter
is covered in a descriptive manner with a very minimal mathematical content,
so that explorationists of the older school with a modest mathematical and
statistical background should not feel discouraged by arcane language and sym-
bols. No mathematical proofs are offered, but interested readers are directed to
carefully selected references to satisfy their curiosity. Chapter 5 briefly depicts
how the theoretical methodology can be applied to the optimization of the airborne
geophysical, ground geophysical, and drilling detection of specific types of ore
deposits known to occur in North America.
The second part of the book covers the optimization of the detection of six
types of ore deposits commonly sought in North America-( 1) porphyry-
Cu-Mo, (2) contact metasomatic, (3) Ni-Cu ultramafic, (4) volcanogenic massive
sulfides, (5) Mississippi Valley-type Pb-Zn, and (6) vein gold deposits-in three
geological regions of the North American continent, namely, the Cordillera
Belt, the Precambrian Shield, and the Arctic Paleozoic Platform (Chapters 6-11).
Finally, Chapter 12 covers the crucial problem of the optimal selection of ex-
ploration targets worthy of testing.
Because we have been critical of the rather offhanded manner with which
references and selected readings have been dealt with in many recent books, we
have devoted much time and attention to this important matter. In order to assist
the more selective readers with their research into specific topics, we are sub-
mitting lists of pertinent titles at the end of each chapter. When warranted
(Chapters 1, 2, 4, and 12), the references are further segregated into clearly
labeled categories covering narrower topics.
PREFACE vii
One last word about the logo displayed on the book cover: being dissatisfied
with the perennial geologists' picks and prospectors' shovels and gold pans which
have been adorning the covers of books on mineral exploration for the past 50
years, we decided to produce a design more attuned to the times and the topic
of the book. The parabola curve aptly and concisely portrays the mathematical
optimization approach, a keystone of the book, as a proper link between modem
exploration methods (aircraft towing a "geophysical bird") of worldwide scope
(map of the world) and commercial success, which is symbolized by the outline
of a mine headframe.
A forthcoming volume titled Geomathematical Models/or Optimizing Min-
eral Exploration deals with the optimized search for the same six types of ore
deposits as in the present volume, but covers four other regions of the world.
These regions include Northern Europe, the Mediterranean, Australasia and East
Asia, as well as the Appalachian Belt and the U. S. portion of the North American
Paleozoic Platform, which could not be covered in this volume due to the lack
of adequate data.
J. De Geoffroy
T. K. Wignall
Sydney, Australia
ACKNOWLEDGMENTS
Three kinds of elements are required in order to carry out a pioneering

project of the type undertaken here: (1) inspiration, encouragement, and guid-
ance; (2) good-quality data; and (3) computational facilities for the processing
of the data and extraction of conclusions and inferences. It was our good fortune
to be able to secure the assistance of several persons with respect to each of the
three requisites mentioned above. We are very much indebted to Dr. F. P.
Agterberg, Head, Geomathematics Section, Geological Survey of Canada, Ot-
tawa, for his early interest and encouragement, and for helpful suggestions which
steered this project along its present line, away from a less fruitful earlier one.
However, we accept full and sole responsibility for any flaws, oversights, or
conceptual errors which may appear in the text. Much gratitude is also owed to
E. H. Rosenberg, Senior Editor, Plenum Publishing Corporation of New York,
for his early interest in the project, and for providing much needed encouragement
during the two arduous years which were required to complete the manuscript.
This project could not have come to fruition without the most generous
assistance of D. A. Barr, Vice President, Du Pont of Canada Exploration,
Vancouver and his staff, who provided almost one half of the whole database
required for the coverage of six types of ore deposits of the North American
continent. Much credit should be given to J. Korenic, Geologist, Du Pont of
Canada, for so ably culling the required geometric data into a readily processable
form. Among others who assisted in the arduous gathering of the necessary data,
we would be most pleased to mention Dr. E. G. Pye, formerly Director, Ontario
Geological Survey, Toronto, who most kindly provided excellent excerpts from
departmental files regarding. volcanogenic massive sulfide and nickel--copper
deposits of Ontario. Finally, much thanks are owed to W. W. Moffat, P. Eng.,
of Willowdale, Ontario, and Gordon Moffat, Geologist, Calgary, Alberta, who
generously helped us fill a number of very troublesome gaps in our database.
Because of the voracious appetite for computer time shown by pioneering
projects of this kind, we could not have succeeded without the unflagging interest
shown by Dr. B. S. Thornton, Dean, Faculty of Mathematical and Computing
Sciences, the New South Wales Institute of Technology, Sydney, Australia, who
most generously provided full use of the excellent computing facilities of his
Department. We are also indebted to Jeff Pickering, Visual Aids Officer, and
to Chas. Evans, Senior Technical Officer, both of the New South Wales Institute
ix
x ACKNOWLEDGMENTS
of Technology, for their excellent photographic and drafting work, which was
required for many of the illustrations included in the book.
Finally, we are much indebted to Julie Symonds of Eden, New South Wales,
Australia, for so ably handling the daunting task of typing the 188 statistical
tables included in this book.
It is hoped that this book will stand as an excellent example of what can
be accomplished through the fruitful collaboration of business, academe, and
government agencies in several nations separated by many miles of ocean.
CONTENTS
Chapter One Optimizing Mineral Exploration

1.1. General Statement ............................................ 1
1.2. The Mitigation of Uncertainty and Risk in Mineral Exploration .... 5
1.3. Conclusion: Scope of the Book................................ 11
References and Selected Readings .............................. 12
Chapter Two Evaluation of the Probability of Detection of

Mineral Deposits
2.1. The Detection Process ......................................... 17
2.2. Geometric Considerations Involved in the Detection of Mineral
Deposits ................................................. 22
2.3. The Theory of Geometric Probabilities as a Foundation for the
Evaluation of the Probability of Detection of
Mineral Deposits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.4. Evaluation of the Probability of Target Detection by Airborne
Geophysical Surveys ...................................... 33
2.5. Evaluation of the Probability of Target Detection by Ground
Geophysical Surveys and Drilling Programs ................. 42
2.6. Probability of "equential Detection of Ore Deposits ............. 50
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Chapter Three Cost of Detection

3.2. Costing and Cost Estimates .................................... 56
3.3. Cost Functions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.4. Tabulation of Cost Functions ................................... 68
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Chapter Four Optimizing Ore Detection

4.2. Basic Theory of Optimization .................................. 77
4.3. Methodology of Optimization .................................. 81
xi
xii CONTENTS
4.4. Application of Optimization Theory to Ore Detection ............ 88

Chapter Five Application of the Optimization Methodology to

the Search for Six Types of Ore Deposits in
North America
5.2. Construction of a Database .................................... 95
5.3. Data Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.4. Computation of Detection Probabilities ......................... 100
5.5. Design of Three Strategies for the Detection of Six Types of Ore
Deposits in North America. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
References and Selected Readings ................ . . . . . . . . . . . . .. 107
Chapter Six Designing Optimized Field Programs for the

Detection of Porphyry-Cu-Mo Deposits of the
North American Cordillera Belt
6.1. General Geological Background ................................ 109
6.2. Field Detection Methodology .................................. 111
6.3. Statistical Modeling of the Geometric Parameters of Porphyry-
Cu-Mo Deposits and their Associated Pyritic Halos ......... 114
6.4. Construction and Organization of Detection Probability Tables ... 117
6.5. Designing Three Strategies for the Search for Porphyry-Cu-Mo
Deposits in the North American Cordillera Belt.............. 118
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 139
Chapter Seven Optimized Search for Four Types of Contact

Metasomatic Deposits of the North American
Cordillera Belt
7.1. General Geological Background ................................ 141
7.3. Probability of Detection of Cu-Fe Contact Metasomatic Deposits 148
7.4. Probability of Detection of Pb-Zn-Cu-Ag Contact
Metasomatic Deposits ..................................... 161
7.5. Probability of Detection of Cu-Mo-Au Contact
Metasomatic Deposits ..................................... 175
7.6. Probability of Detection of W-Mo Contact Metasomatic Deposits.. 188
7.7. Designing Three Strategies for the Detection of Four
CONTENTS xiii
Types of Contact Metasomatic Deposits of the North

American Cordillera ....................................... 198
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 209
Chapter Eight Detection of Ni-Cu Ultramafic Deposits of the

North American Shield by Optimized
Geophysical Surveys and Drilling Programs
8.1. General Geological Background................................ 211
8.3. Statistical Modeling of Geometric Parameters of Ni-Cu Deposits.. 215
8.4. Construction and Organization of Detection Probability Tables ... 218
8.5. Designing Three Strategies for the Detection of Ni-Cu Deposits of
the North American Shield. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 219
8.6. Detection Probability and Optimization Tables .................. 223
Chapter Nine Optimized Airborne and Ground Search

for Volcanogenic Massive Sulfide
Deposits of the North American Shield
and Cordillera Belt
9.2. General Geological Background: Volcanogenic Sulfide Deposits of
the North American Shield. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 236
9.4. Statistical Modeling of Geometric Parameters of Shield
Volcanogenic Sulfide Deposits ............ -. . . . . . . . . . . . . . . .. 241
9.5. Construction and Organization of Detection Probability Tables for
Shield Volcanogenic Sulfide Deposits. . . . . . . . . . . . . . . . . . . . . .. 245
9.6. Design of Three Strategies for the Detection of Shield
Volcanogenic Sulfide Deposits ............................. 246
9.7. Detection Probability and Optimization Tables for Shield
9.8. Geological Synopsis for Volcanogenic Sulfide Deposits of the
North American Cordillera Belt. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 261
9.9. Statistical Modeling of Geometric Parameters of Cordillera
9.10. Construction and Organization of Detection Probability Tables
for Cordillera Volcanogenic Sulfide Deposits ............... 264
9.11. Designing Three Strategies for the Detection of Cordillera
Volcanogenic Sulfide Deposits ............................ 264
xiv CONTENTS
9.12. Detection Probability and Optimization Tables for Cordillera

Volcanogenic Sulfide Deposits ............................ 267
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 279
Chapter Ten Designing Optimized Field Programs for the

Detection of Mississippi Valley-Type Pb-Zn
Deposits of the North American Arctic Paleozoic
Platform
10.1. General Geological Background............................... 283
10.2. Field Detection Methodology ................................. 285
10.3. Statistical Modeling of the Geometric Parameters of North
American Arctic Deposits ................................. 288
10.4. Construction and Organization of Detection Probability Tables .. 290
10.5. Designing Three Strategies for the Detection of Arctic Mississippi
Valley-Type Pb-Zn Deposits. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 291
10.6. Detection Probability and Optimization Tables ................. 294
References and Selected Readings ............................. 303
Chapter Eleven Detection of Vein-Gold Deposits of the North

American Shield by Optimized Ground
Programs
11.1. General Geological Background ............................... 305
11.2. Field Detection Methodology ................................. 308
11.3. Statistical Modeling of Geometric Parameters of Vein-Gold
Deposits ................................................ 310
11.4. Construction and Organization of Detection Probability Tables .. 312
11.5. Design of Three Strategies for the Detection of Archean Vein-
Gold Deposits of the North American Shield ............... 314
11.6. Detection Probability and Optimization Tables ................. 316
References and Selected Readings ............................. 323
Chapter Twelve Optimal Selection of Exploration Targets for

Drill Testing
12.1. General Statement ........................................... 325
12.2. Optimal Selection of Exploration Targets Based on
Control Locations ........................................ 328
CONTENTS XV
12.3. Optimal Selection of Exploration Targets without

Control Locations ........................................ 340
References and Selected Readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 351
Appendix: Measurement Conversion Table ................... 353
Subject Index .............................................. 355
Author Index ............................................... 361
Listed Journals ............................................. 363

CHAPTER ONE
OPTIMIZING MINERAL EXPLORATION
1.1. GENERAL STATEMENT
1.1.1. Purposes and Nature of Mineral Exploration

Mineral exploration can be defined as a scientific endeavor to discover and
gain a natural resource which is difficult to find and of a nonrenewable nature.
Mineral exploration encompasses only the search for, discovery, identification,
and evaluation of the prize, but not its commercial development. The main
purpose of exploration is success to be obtained at a reasonable cost in terms of
expenditure of time, money, and skill.
Success in mineral exploration can be defined in two principal aspects;
technical success and commercial success. Technical success is highlighted by
the discovery of a mineral deposit sufficiently attractive to justify the expenditure
required to establish the commercial potential of the deposit (tonnage, grade).
The investigation leads to three possible outcomes: (1) inadequacy of the mineral
deposit to meet the criteria required for a profitable operation in the foreseeable
future; (2) qualified commercial success: the mineral deposit appears to be sub-
economic at the present time and should be put on the "shelf" for future needs
in more favorable circumstances; and finally, (3) commercial success: the mineral
deposit does meet all criteria required for a profitable operation and is now
described as an "ore body," based on the commonly accepted definition of "ore"
as mineralized material that can be mined at a profit.
Technical success depends on two critical events; occurrence and detection.
Occurrence of a specific type of mineral deposit in a restricted area is a state of
nature on which the explorationist has control only to the extent of the selection
of the area of investigation. Detection depends largely on the quality of the
search, which implies (1) the selection of the technology and methodology best
suited for the type of ore target and its environment, (2) proper planning and
organization, including scheduling and logistics, and (3) availability of risk
capital on reasonably attractive terms.
Commercial success relies on a third critical element, that of "economic
worth," whose adjunction transforms a technical success into a commercial one.
As in the case of occurrence, economic worth partly reflects a state of nature,
i.e., the intensity and extent of geological processes, on which the explorationist
1
2 CHAPTER ONE
has no control. But, again, many other factors of a geographic, economic,

financial, and political nature are involved as well.
1.1.2. The Context of Mineral Exploration: Uncertainty and Risk

1.1.2.1. Uncertainty. It is a common experience that most mineral ex-
ploration decisions such as the selection of areas, technology of investigation,
and exploration targets for testing, etc. are based on information that is limited
both in quality and quantity. Most exploration situations lie between the two
extremes of absolute certainty on the one hand, and total uncertainty on the other
hand. In practice, we have to deal with partial availability of perfect information,
or with imperfect information, or with a degree of ignorance about the quality
of information. The resulting uncertainty leads to the use of much more complex
decision models which introduce a greater possibility of occurrence of unfavor-
able outcomes than would be the case if certainty prevailed. This brings us to
the concept of risk, which reflects the degree of unfavorability with which
different possible outcomes are viewed by the decision maker.
1.1.2.2. Risk. That mineral exploration is a highly risky business is an
opinion widely shared by the mining fraternity and the public at large. While
holding the possibility of financial returns larger than that expected in most other
business ventures, mineral exploration is also beset by a high probability of
failure, which may lead to heavy financial losses. Risk may result from geolog-
ical, technological, economic, and political uncertainty, as well as intangibles
such as the quality of geological, managerial, and financial skills available in
the corporate structure of exploration companies.
Geological risk is associated with the concept of occurrence and with the
selection of areas of investigation and targets for testing. Furthermore, once a
mineral deposit has been discovered, there is a geological risk associated with
its appraisal whose validity is strongly affected by the geological variability
within the deposit.
Technological risk is associated with possible problems in mine development
due to poor rock conditions, etc., and in ore treatment due to unfavorable
characteristics of the ore material. These risks must be considered early and can
be reduced by preliminary testing, operation of pilot plants, etc. There is a much
higher level of financial risk involved in the final commitment to development
and production than in target selection for testing, because of the magnitude of
financial resources which could be wasted in the former case owing to faulty
geological advice. Economic risk is present at all stages of economic forecasting
on which the production decision hinges, including prices, costs, market con-
ditions, etc.
OPTIMIZING MINERAL EXPLORATION 3
Finally, political risk is ever present at all stages of exploration and de-
velopment of mineral deposits; it may result from unforeseen restrictions in land
tenure or access to mineral land, or changes in taxation rules, environmental
protection, as well as labor laws. Perceived high political risk favors the search
for small, high-grade, "bonanza" types of mineral deposits which are charac-
terized by low investment levels, a short life, and a high payout. On the contrary,
expected low political risk will enable the search for and development of large,
low-grade deposits featuring low profit margins and long pay-back periods.
1.1.3. Quantification of Exploration Success in a Context of

Uncertainty and Risk
1.1.3.1. The Probabilistic Concept of Exploration Success. The con-
cept of probability is the most fruitful way of expressing quantitatively the degree
of uncertainty and risk which permeates every stage of the mineral exploration
decision-making process. A probabilistic assessment of uncertainty is intuitively
understood in the context of the likelihood of occurrence of an event. The
probability of an event is expressed by a number chosen in a numerical scale
ranging from 0, standing for certainty of nonoccurrence, to unity, representing
certainty of occurrence, while the fractions of unity in between reflect the degree
of uncertainty attached to the occurrence of the event. The application of the
principle may be further stretched to express quantitatively the degree of belief
of an observer in the occurrence or nonoccurrence of an event, which leads us
to the foundation of the theory of subjective probability. Finally, the same zero
to unity scale may be used to measure the degree of knowledge about events in
the context of imperfect or partial information which affect so many of the
mineral exploration decisions.
In contrast to the previous "subjectivist" concept of probability, which is
useful to measure uncertainty as viewed by an observer, another concept of
probability, which is referred to as "objectivist" because it does not involve an
observer, proves quite useful in the quantitative assessment of risk. It is known
as the "frequentist" probability concept. Supposing that a desirable event called
"success" occurs h times out of a possible total of n times in the course of a
chance process, we can define the success ratio of the process as Rs = hln. If
the chance process is repeated many more times, so that n increases toward
infinity, the success ratio tends toward a limit which is the probability of success
P s of the event. The complement of the probability of success, which is by
definition written as 1 - P" is called the probability of failure Pf. The concept
of probability of failure provides a way of quantifying the occurrence of unfa-
vorable events, i.e., the quantification of risk.
4 CHAPTER ONE
1.1.3.2. Historical Success Ratio as an Estimate of the Probability

of Success. The concept of success ratio introduced above is of great assist-
ance to the planning explorationist, because it provides an estimate of the prob-
ability of success of a planned type of program based on the record of past
performances of similar programs in similar conditions. Both technical and com-
mercial success ratios are useful to consider.
Generally, one finds that success ratios in oil exploration are much better
documented than those of mineral exploration. (13) One of the first comprehensive
studies of the success ratio in mineral exploration was that of Koulomzine and
Dagenais, (8) published in 1959, which covered the postwar record of the Canadian
mineral exploration, with particular reference to northwestern Quebec. They
found that the overall Canadian technical success ratio is 0.116, while its com-
mercial counterpart is only 0.063. At a regional scale, the beneficial role of
geological screening based on conceptual models is well demonstrated by the
increase of the technical success ratio from 0.116 to 0.233. At a local scale, for
example in the Val d'Or region of northwestern Quebec, Koulomzine reports a
technical success ratio as high as 0.410 due to a thorough screening of the most
favorable geological areas.
1.1.3.3. Factoring the Probability of Success in Mineral Explora-
tion. It can be intuitively understood that the occurrence of some type of event
of interest could be affected by the condition that other types of events have
occurred previously. In the probabilistic language, this is tantamount to say that
the probability assigned to an event is conditional on the probability of occurrence
of other events. The concept of conditionality is the foundation of the theory of
Bayesian probabilities which we will touch on in Chapter 12 regarding the optimal
selection of exploration targets for testing.
A basic tenet of the concept of conditional probability is the following law,
which states that, if the events E and F are statistically independent, the prob-
ability of E and F occurring simultaneously equals the product of the probability
of occurrence of E by that of F.
Applying now these concepts to mineral exploration situations, we find that
the probability of technical success is conditional on the simultaneous occurrences
of two events which are statistically independent, namely, geological occurrence
and detection, so that we are justified in writing
Probability of technical success

(Probability of geological occurrence) X (Probability of detection)
Likewise, since technical success and economic worth are statistically indepen-
dent. we can write
Probability of commercial success =

(Probability of technical success) x (Probability of economic worth)
Although the expressions written above are quite justified and very useful,
it may be argued that mineral exploration is not really a one-stage procedure but
has a sequential structure, an important point which will further develop below.
Any well-planned exploration program consists of three stages, including (1)
detection and delineation of exploration targets with the probability of success
PI, (2) investigation and testing of exploration targets with a probability of
success P 2 , and finally, (3) outlining ore targets within the tested exploration
targets, with a probability of success P 3 • Since the three events are statistically
independent, we may write
Probability of commercial success =

(Probability PI) x (Probability P 2) x (Probability P 3)
1.2. THE MITIGATION OF UNCERTAINTY AND RISK IN

MINERAL EXPLORATION
1.2.1. Introduction
Uncertainty and risk are unfavorable features which are, unfortunately, an
integral part of the mineral exploration business. It is therefore the hallmark of
good management that every effort be made to minimize the magnitude of
uncertainty and risk and mitigate their harmful effects at each stage of the ore
search, whether it be planning, field execution of surveys, or final assessment
of results. There are three types of approach to uncertainty and risk reduction:
(1) repetition of trials,(2) sequential approach, and (3) optimization approach.
The first scheme, quite straightforward though potentially very costly, calls
for the repetition of trials to improve the odds of success in the long run. The
second one is more sophisticated and more resource-effective than the first one;
it is based on a structured sequence of stages, each one depending on the results
of the previous one and leading to the next one in a prescribed manner. The
third approach takes advantage of the second one; it consists of maximizing a
desired goal under cost or risk constraints at each stage of the sequential structure,
resulting in a fully optimized overall procedure within the context of a prespec-
ified corporate strategy.
6 CHAPTER ONE
1.2.2. Repetition of Trials

The concept of improving the odds of success by increasing the number of
trials is quite intuitive and well recognized in everyday life. The "spreading of
risk" over the largest possible number of independent ventures, another related
approach, is the gist of any sound investment management which is well practiced
in mineral exploration.
The theoretical justification of the repetition approach is found in the fre-
quentist concept of the probability of success, which was mentioned above in
Section 1.1.3.1. The concept is aptly illustrated by a heuristic model in the form
(Probability of success) = 1 - exp( -kN)
where k is an empirically determined constant and N is the number of trials. It

is clear that, if the number of trials N becomes quite large, the negative expo-
nential term which represents the risk factor becomes quite small, and the prob-
ability of success nears its maximum value of unity. Obviously, if the number
of trials increases, the overall cost increases in a parallel manner, quickly reaching
budget ceiling before the probability of success reaches its maximum of unity;
attempts to do so could lead to the "gambler's ruin," which plainly underscores
the serious limitations of the first approach.
1.2.3. The Sequential Approach

1.2.3.1. Summary. The sequential approach to uncertainty and risk re-
duction is an alternative avenue which is more sustainable costwise than the
previous one. The sequentiality of the structure of mineral exploration should
be considered in two main aspects, including the time domain (chronologic
sequentiality) and the space domain (spatial sequentiality). Both aspects are
integral parts of all well-planned exploration programs. Figure 1.1 summarizes
in a diagrammatic style the complex interrelationships between the two aspects
of ore search sequentiality. The ore search is divided up in a somewhat arbitrary
manner into seven stages, each affected by three types of decision making.
1.2.3.2. Sequentiality in the Time Domain. It is well recognized,
albeit in an intuitive manner only, that the mineral exploration procedure has to
be undertaken in a logical time sequence, otherwise the risk of financial loss
will markedly increase. The generally accepted sequence starts with large-scale
regional coverages first from the air, to be followed by ground geological and
geochemical reconnaissance surveys. The next stage includes more detailed geo-
logical, geochemical, and ground geophysical surveys to delineate exploration
~
DATA EVALUATION &
AREA SELECT ION DATA ACQUISITION
TARGET DELINEATION
,
STAGE
1ST
STAGE
(2)
SELECflrn OF
GEDLCGICAL REGIONS,
TARGET TIFES &
--1 (1) INITIAL DATA
ACQJISITIW, 1-- - - -4----l I
r-
srORAr.E & RETRIEVAL
EXPLORATIOO ~IETI-k)DS
IN
t (4)
INVESTIGATION OF
H(s)
• F
I0
R
2ND PROSPECfING AREAS BY
(3)SELECTla..; OF
STAGE PROSPECT ING AREAS RIMJTE SE."lSI~G & AIR- DELINEATION OF
EXPLORATI(YII AREAS ~M
BORNE GEOPHYSICAL
SJRVEYS p
J
H(s)
IT
(6j (7)
I:-''VESflGATI()II OF
I I
r---
SELECT lOr.; OF
3RD
EXPLORATIOO AREAS
EXPLORATIQ".I AREAS BY
RECCNNAISSANCE
DELINEATION OF ~l 0
•
STAGE FOR RJRTIiER
GEOWGIGA.L &
ANCMAUllS AREAS N
I~YESTrGATION
GED-O-ifMlCAL SURVEYS
• -'
H
I
,
(10) I
I
r---
INVESTIGATION OF
4TH (9)Sf:LECTIrn OF
A\lCt-IALCUS AREAS
ANO>VI.LOJS AREAS BY (11)
DELINEATIct-; OF .j
STAGE SYSTa.1ATIC GEOLOCICAL.
I
FOR RJRTI-lER EXPLORATICN TARGETS
I\'VESTIGATIO~
GECXJID.lICAL &
GEOPHYSICAL ~VEYS
•I F
E
(12)
I (13)
5TH
STAGE
SELECTION OF
EXPLORATION
I~STIGATION
OF EXPLORATIO'l
11(14)
DELlNEATIO\'
OF ORE
-..J E
TARGETS FOR TARGETS BY ID
I I
DRILL-TESTIlI.'G TARGETS
DRILLING
t P
t IA
,
(15) I C
6TH (16) (17)
SELECTIOI\' OF
ORE TARGETS !DR stST~TIC DELL'JEATIO:-: K
STAGE DETAILED DRILLING
DETAILED DRILLING OF ORE OEPOS ITS
7TH !
i
(IS)
•
LP.>.TIERGRctJl',TI
STAGE I OFH-IVESTIGA.TICN
ORE DEPOSITS
FIGURE 1.1
Flow Diagram of Mineral Exploration Sequence.
targets. The final stage consists in the detailed three-dimensional investigation

of selected exploration targets by drilling, before underground exploration is
decided on. It should be noted that each stage is, in tum, structured along a
logical time sequence in such a way as to enmesh closely with the previous stage
and the next one.
8 CHAPTER ONE
The uncertainty and ensuing risk factor may be alleviated by means of the
sequential organization of the flow of various types of information. Three main
types of information are to be considered: (1) prior information based on case
histories and acquired from data storage at the initial planning stage, (2) newly
acquired information which is gained as a result of the execution of the field
programs, and (3) feedback information arising from both success and failure
in field exploration, which is recycled to storage for future use at subsequent
stages of the same program or in other programs. As the amount and quality of
information increase progressively at each stage as compared to the previous
one, both uncertainty and risk are being gradually mitigated.
Many factors of diverse nature, including geographic, particularly climatic,
as well as economic and political ones, have to be reckoned with in the overall
planning, because of their disruptive influence on the proper time sequence of
activities. As modem ore search spreads out farther afield to investigate remote
regions affected by harsh climatic conditions, it is obvious that a special allow-
ance has to be made for these conditions when considering the logistics and
scheduling of activities. For example, airborne surveys may be carried out on a
year-round basis in most regions but the tropical ones, where the monsoon season
is quite unsuitable for systematic flying. Ground surveys should be preferably
carried out during the winter season in both sub-tropical deserts and subarctic
regions, or during the dry season in regions affected by monsoonal climates, for
reasons of greater logistics convenience and crew efficiency.
1.2.3.3. Sequentiality in the Space Domain. Sequentiality in the space

domain is a second feature of the organization of mineral exploration which
greatly assists risk reduction by means of two approaches, "scanning" and "skim-
ming." Scanning is a sequential reduction of acreage under investigation from
larger acreages of lesser geological merit to smaller acreages of upgraded po-
tential which are to be searched by methodologies of increasing resolution at an
increasing cost per unit of area. As shown by Brant,(17) the rate of acreage
reduction is higher than that of increase of unit cost, thus slowing the rate of
escalation of the overall cost.
The result of scanning is illustrated by the rightmost column of Figure 1.1,
as the acreage of the investigated area decreases from a few thousands of square
miles at stage one to only a few acres or even fractions of acre at stage six.
Scanning is also accompanied by skimming at each stage of the sequence, for
the purpose of reducing the number of targets outlined at one stage to be subjected
to more detailed investigations at the following stage.
1.2.4. The Optimizing Approach

1.2.4.1. Rationale of the Optimizing Approach. It is appropriate at
this stage to pause and briefly look back at the ground covered so far as to the
nature and structure of the mineral exploration business. Its aim is the search
for and discovery of nonrenewable resources which can be extracted at a profit.
The search requires the making of rational decisions which involve the expen-
diture of large amounts of money, time, and skills in anticipation of expected
profits. The search takes place in an environment of uncertainty and risk whose
costly effects may be alleviated by structuring the procedure of exploration in a
sequential manner.
At each stage of the sequence, decisions have to be made which require
the explorationist to weigh up and compare different courses of action. Because
of the context of uncertainty in which mineral exploration operates, the actual
consequences of the various options are not fully known in advance. The ex-
plorationist must be satisfied to make a choice, which, if not the best choice in
absolute terms, is the best achievable course of action under the prevailing
constraints of time, budget, and skills available and within the context of pre-
specified corporate policies. This is the definition of the optimization approach
in a nutshell.
Optimization is a an objective and pragmatic procedure which codifies the
selection of options in conditions of uncertainty by the use of quantified criteria
and formal rules of action in order to lead to a rational choice compatible with
corporate goals. Through the use of the Dynamic Programming approach to be
described in Chapter 4, the overall efficiency of the mineral exploration sequence
may be maximized by optimizing each stage, while taking full advantage of the
optimization of the previous stages. This is the essence of the new philosophy
of optimization which should be guiding the mineral exploration of the 1980s
and 1990s towards new records of success emulating those of the past two
decades.
1.2.4.2. Optimization of the Preliminary Planning Phase. In the

preliminary planning phase, the initial decisions of critical importance belong
to the field of subjective optimization which relies on the qualitative or semi-
quantified judgment of individuals or panels of experts. These decisions cover
the selection of regions to be searched and types of ore deposits to be targeted,
as well as methodologies to be used for the search.
Subsequent decisions are amenable to the quantitative optimization approach
which is described in Chapter 4. The first decision deals with the optimal selection
10 CHAPTER ONE
of prospecting areas within the previously selected region; this may be formally
handled by a multivariate statistical procedure based on a geostatistical model.(41,42)
The second set of decisions involves the quantitative optimization of field survey
designs with which to cover the optimally selected prospecting areas for the
detection of expected ore deposits. This problem of detection optimization is the
major topic of the present book. Finally, there arises a problem of optimal
allocation of resources including time, money, and skills between the various
stages of the exploration sequence, which is within the scope of the new science
of Operations Research. (See Chapter 4.)
1.2.4.3. Optimization of the Field Implementation Phase. The types
of decisions which have to be taken in the field implementation phase are rather
different in nature from those made by the planners. The decision maker has to
be quite flexible; at the end of each stage, decisions have to be made on the spot
and have to be tailored to field results which were unforeseen in the preliminary
planning phase.
One example is the decision whether to terminate a surveyor extend it
further in order to acquire additional data. Zero information is zero cost, but
entails a large risk factor. Complete information is unattainable because of its
prohibitive cost and the special nature and structure of geo-data. Therefore, an
optimal amount and quality of information has to be sought for as a compromise
between the two extreme cases mentioned above. The formal solution of this
optimal choice problem has been attempted by several Operations Research
workers. They use the Dynamic Programming approach combined with Bayesian
decision theory in a graphical manner (decision tree diagrams).
Another problem of optimal choice under uncertainty arises after the com-
pletion of field surveys. The sequential combination of scanning and skimming
mentioned above leads to the delineation of a number of exploration targets
which is still far too large for full investigation by drilling within the bounds of
the allocated budget. Two approaches to the optimization of the selection of
exploration targets are considered in Chapter 12 of the book. The first one,
relying on control locations which have been previously investigated, is suitable
for established mining districts. The second one, recently devised by the second
writer, does not require the availability of control locations, which is an obvious
advantage when dealing with poorly prospected regions.
Finally, once the exploration targets of maximum merit have been selected
within geological and budgetary parameters, the optimal allocation of funds
between selected targets for drill testing has to be dealt with. Celasun (Ref. 21
in Chapter 4) and other workers in the field of Operations Research have in-
vestigated the matter in a formal analytical manner.
1.2.4.4. Optimization of the Evaluation Phase. The final phase of

the surface exploration sequence covers the evaluation of results of the drill
testing of optimally selected exploration targets. Once a target has been suc-
cessfully tested and a mineral deposit has been detected and approximately
outlined, an evaluation of the commercial potential (tonnage, grade) of the
mineralization is called for, before making a decision regarding the underground
development of the deposit.
The evaluation has to be based on closely spaced sampling by surface
drilling, which is time-consuming and very expensive. Once again, we are faced
with a problem of optimal survey design as in the preliminary planning phase.
However, in this case the gist of the matter is different: instead of seeking to
optimize detection efficiency, we are now interested in optimizing sampling
precision. The degree of geological variability prevailing throughout the mineral
deposit dictates the degree of precision required for the estimates of tonnage and
grade. The desired precision level, in tum, prescribes the number of samples
required, and, therefore, the sample spacing and corresponding coverage cost.
A few workers have investigated this important optimization problem (Refs. 19
and 28 in Chapter 4).
1.3. CONCLUSION: SCOPE OF THE BOOK
Because of the prevalence of uncertainty and risk which affect every aspect
of the mineral exploration business, it is proper to view the whole ore search
procedure in a probabilistic context. The goal of mineral exploration is therefore
expressed in terms of probability of success, which, in tum, is factored into
three main components, namely, probability of geological occurrence, proba-
bility of detection, and probability of economic worth. A full coverage of the
evaluation of each of the three components could not be considered within the
one-volume format required here: for example, an adequate treatment of the
probability of occurrence alone would require a full-size book. We have therefore
decided to restrict the topic of the present volume to only one component of the
probability of success, that of the detection probability, which is fully investigated
in its theoretical aspects in Chapter 2 and in its practical implications in Chapters
6-11. Carefully selected references are provided at the end of this chapter for
the readers interested in delving further into the very important topics of occur-
rence and economic worth.
The brief coverage offered in Section 1.2.4 amply demonstrates the breadth
of scope of the optimization approach in the mineral exploration business. Despite
12 CHAPTER ONE
the rather scant attention devoted to date to the optimization of mineral explo-
ration, a full treatment of the subject, however desirable, is not possible within
the constraints of a single-volume format. Only two of seven basic types of
optimal choices arising in the course of ore search are fully covered in the present
book. These comprise (a) the optimization of field survey designs in the prelim-
inary planning phase dealt with in Chapters 2,3, and 4, and (b) the optimization
of the selection of exploration targets for testing at the end of implementation
phase, which is treated in Chapter 12. The methodology of the optimization of
detection, which is described in Chapters 2-4, is then applied to the search for
six common types of ore deposits in North America (Chapters 5-11).
REFERENCES AND SELECTED READINGS
Success in Mineral Exploration

1. BORCH, K. H., 1972, The Economics of Uncertainty, Princeton University Press, Princeton,
New Jersey.
2. BROWN, G. A., 1970, The evaluation of risk in mining ventures, Can. Inst. Min. Metall.
Bull. 63(701), 1165-1171.
3. COOPER, D.O., DAVIDSON, L. B., and REIM, K. M., 1973, Simplified financial and risk
analysis for mineral exploration, Annual Symposium on Computer Applications No. 11, Vol.
I, pp. BI-BI4, Tucson, Arizona.
4. DE GEOFFROY, J., and WIGNALL, T. K., 1974, Evaluating that exploration project, Can.
Min. J. 95(5), 42-44.
5 . GRAYSON, C. J., JR., 1960, Decisions under Uncertainty; Drilling Decisions by Oil and Gas
Operators; Harvard University, Graduate School of Business Administration, Cambridge,
Massachusetts.
6. GRIFATHS, J. C., 1967, Mathematical exploration strategy and decision making, Seventh
World Petroleum Congress Proceedings, Mexico City, pp. 599-604.
7. KNIGHT, F. H., 1921, Uncertainty and Profit, Chapter 2, Houghton-Mifflin, New York.
8. KOULOMZINE, T., and DAGENAIS, R. W., 1959, Statistical determination of the chances
of success in mineral exploration in Canada, Can. Min. J. 74(4), 107-110.
9. LEBIS, A. P., 1978, The economics of mining and other risky ventures, Can.lnst. Min. Metall.
Bull. 71(791), 169-171.
10. MACKENZIE, B. W., 1973, Corporate exploration strategies, Application of Computer Meth-
ods in the Mineral Industry Symposium, Johannesburg, pp. 1-8.
11. MACKENZIE, B. W., 1981, Looking for the improbable needle in a haystack: The economics
of base metal exploration in Canada, Can. Inst. Min. Metall. Bull. 74(830), 115-125.
12. MOORE, P. G., 1972, Risk in Business Decision, Chaps. 1,5,6,7, and 11, Longman, London.
13. PETERS, C. W., 1978, Exploration and Mining Geology, Wiley, New York.
14. PRATf, J. W., RAIFFA, H., and SCHLAIFER, R., 1964, The foundations of decision under
uncertainty: An elementary exposition, J. Am. Stat. Assoc. 59, 353-375.
15. ROSCOE, W. E., 1971, Probability of an exploration discovery in Canada, Can. Inst. Min.
Metall. Bull. 64,(707), 134-137.
Evaluation of the Probability of Ore Occurrence:

Univariate Modeling
16. BOZDAR, L. B., and KITCHENHAM, B. A., 1972, Statistical appraisal of the occurrence of
lead mines in the Northern Pennines, Trans. Inst. Min. Metall. 81, BI83-BI87.
17. BRANT, A. A., 1968, The pre-evaluation of the possible profitability of exploration prospects,
Miner. Deposita 3,1-17.
18. BROWN, D., and ROTHERY, P., 1978, Randomness and local regularity of points in a plane,
Biometrika 65, 115-122.
19. CLARK, P. J., 1956, Grouping in spatial distributions, Science 123, 373-375.
20. DREW, M. W., 1974, A deposit distribution model for uranium, Programs Analysis Unit Report
10-74, U.K. Atomic Energy Authority, London.
21. EBDON, D., 1977, Statistics in Geography: A Practical Approach, Chap. 7, Blackwell, Oxford.
22. GRIFFITHS, J. c., 1962, Frequency distributions of some natural resources materials, Bulletin
No. 63, Mineral Industry Experimental Station, University of Pennsylvania, pp. 174-198.
23. JOHNSON, N. I., and KOTZ, S., 1970, Discrete Univariate Distributions, Vol. I, Houghton-
Mifflin, Boston.
24. MACDOUGAL, E. B., 1976, Computer Programmingfor Spatial Problems, E. Arnold, London.
25. McCONNELL, M., 1966, Quadrat methods in map analysis, Discussion Paper No.3, De-
partment of Geography, University of Iowa.
26. MILLER, R. L., and KAHN, J. S., 1962, Statistical Analysis in the Geological Sciences, Chap.
\6, Wiley. New York.
27. NEFT. D. S., 1966. Statistical analysis for areal distribution, Monograph 2, Regional Science
Research Institute, Philadelphia.
28. ROGERS, A., 1974, Statistical Analysis of Spatial Dispersion: the Quadrat Method. Chap. I,
Pion, London.
29. SARMA, D. D., 1979, An exploration strategy for prospecting with a case study on copper
prospects at Ingladahl, India, Miner. Deposita 14, 263-279.
30. SLICHTER, L. B., 1960, The need for a new philosophy of prospecting, Min. Eng. (New
York) June, 570-576.
31. WILLIAMSON, E., and BRETHERTON, M. H., 1963, Tables of the Negative Binomial
Probability Distribution, Wiley, New York.

Multivariate Modeling
32. AGTERBERG, F. P., 1970, Multivariate prediction equations in geology, J. Math. Geol. 2,
319-324.
33. AGTERBERG, F. P., 1971, A probability index for detecting favourable geological environ-
ments, Can. Inst. Min. Metall. Spec. Vol. 12, 82-91.
34. AGTERBERG, F. P., 1974, Automatic contouring of geological maps to detect targets areas
for mineral exploration, J. Math. Geol. 6, 373-395.
35. AGTERBERG, F. P., 1977, Frequency distributions and spatial variability of geological vari-
ables, Proceedings of the A.P.C.O.M. Symposium No. 14. New York. pp. 287-298.
36. AGTERBERG, F. P., CHUNG, C. F., DIVI, S. R., and EADE, K. E., 1981, Preliminary
geomathematical analysis of geological, mineral occurrence and geophysical data, Southern
14 CHAPTER ONE
District of Keewatin, Northwest Territories, Canada, Geol. Surv. Can .. Open File Report No.
778, pp. 1-29.
37. CHUNG, C. F., 1978, Computer program for logistic model to estimate the probability of
occurrence of discrete events, Geol. Surv. Can. Pap. No. 78-11.
38. CHUNG, C. F., and AGTERBERG, F. P., 1980, Regression models for estimating mineral
resources from geological map data, J. Math. Geol. 12, 458-473.
39. COX, D. R., 1970, Analysis of Binary Data. Methuen, London.
40. DE GEOFFROY, J., and WIGNALL, T. K. W., 1970, Application of statistical decision
techniques to the selection of prospecting areas and drilling targets in regional exploration, Can.
Inst. Min. Metall. Bull. 63(699), 893-899.
41. DE GEOFFROY, J., and WIGNALL, T. K., 1971, A probabilistic appraisal of mineral resources
in a portion of the Grenville Province of the Canadian Shield, Econ. Geol. 66, 466-479.
42. DE GEOFFROY, J., and WIGNALL, T. K., 1973, Design of a statistical data processing
system to assist regional exploration planning; Can. Min. J. 94(11), 30-35; 94(12), 35-36.
43. HARRIS, D. P., 1965, Multivariate statistical analysis: A decision tool for mineral exploration;
Computer Applications in Mining & Exploration Symposium, Vol. I, pp. CI-C35, University
of Arizona, Tucson, Arizona.
44. HARRIS, D. P., 1969, Alaska's base and precious metals resources: A probabilistic regional
appraisal, Q. Col. Sch. Mines 64(3),295-327.
45. HARRIS, D. P., and AGTERBERG, F. P., 1981, The appraisal of mineral resources, Econ.
Geol. 75, 897-938.

Subjective Approach
46. BARRY, G. S., and FREYMAN, A. J., 1970, Mineral endowment of the Canadian Northwest:
A subjective probability assessment, Can. Inst. Min. Metall. Bull. 63(700), 1031-1038.
47. ELLIS, J., HARRIS, D. P., and VAN WIE, N., 1975, A subjective probability appraisal of
uranium resources in the State of New Mexico, U. S. Energy Research and Development
Administration, Grand Junction, Colorado.
48. GOLABI, K., and LAMONT, A., 1981, A probabilistic approach to the assessment of uranium
resources, J. Math. Geol. 13(6).
49. HARRIS, D. P., 1973, A subjective probability appraisal of metal endowment of Northern
Sonora, Mexico, Econ. Geol. 68(2), 222-242.
Evaluation of the Probability of Economic Worth:

Univariate Modeling
50. AGTERBERG, F. P., 1980, Lognormal models for several metals in selected areas of Canada,
Reprint, Mem. BRGM, No. 106, pp. 83-90.
51. AGTERBERG, F. P., and DIVI, S. R., 1978, A statistical model for the distribution of copper,
lead and zinc in the Canadian Appalachian Region, Econ. Geol. 73, 230-245.
52. ALLAIS, M., 1957, Methods of appraising economic prospects of mining exploration over
large territories, Manag. Sci. 3, 285-347.
53. DIVI, S. R., 1980, Deposit modeling and resources estimation of stratiform massive sulfide
deposits in Canada, Comput. Geosci. 6(2), 163-174.
54. DE GEOFFROY, J., and WU, S. M., 1970, A statistical study of ore occurrences in the
Greenstone Belts of the Canadian Shield, Econ. Geol. 65,496-504.
55. SANGSTER, D. F., 1980, Quantitative characteristics of volcanogenic massive sulfide deposits,
Can. Inst. Min. Metall. Bull. 73(814), 74-81.
56. SINCLAIR, A. J., 1974, Probability graphs of ore tonnage in mining camps: A guide to
exploration, Can. Inst. Min. Metall. Bull. 67(749), 71-75.
57. SINGER, D. A., COX, D. P., and DREW, L. J., 1975, Grade and tonnage relationships among
copper deposits, U.S. Geol. Surv. Prof. Pap. 907-A, AI-All.
CHAPTER TWO
EVALUATION OF THE PROBABILITY OF

DETECTION OF MINERAL DEPOSITS
2.1. THE DETECTION PROCESS
2.1.1. Introduction
The principal purpose of field programs is the acquisition of infonnation
which will lead to the detection of mineral deposits. There are two types of
approach to the problem of detection, namely, the direct and indirect methods,
which are commonly used simultaneously or sequentially to best advantage.
Table 2.1 summarizes the main aspects of the two paths with respect to the types
of target sought and detection environment, as well as the methodology and
procedures involved.
The direct approach seeks to detect the mineral deposit itself, by direct
observation. The observation is based on visual recognition, either unaided as
in ground prospecting or mapping, or aided by photogeology or remote sensing
techniques from airborne platfonns, or by mechanical probes, such as drilling,
in three-dimensional investigations. The second method requires a more elaborate
treatment to be given below, because of its indirect nature.
2.1.2. The Indirect Approach to the Detection

of Mineral Deposits
The indirect approach seeks to detect mineral deposits, or closely associated
features such as halos, in an
indirect manner, through local changes called
"anomalies," by sampling the continuum of the regional geophysical or geo-
chemical environment. There are two main types of sampling procedures; con-
tinuous and discrete sampling.
Continuous sampling surveys are feasible only from airborne platfonns.
The data are recorded as graphs, or on photographic films, or on magnetic tapes,
while sweeping the ground along regularly spaced profile lines. The data are
then subjected to the process of "digitization," which calls for the sampling of
continuous recordings at regular intervals in order to extract representative dis-
crete data for convenient processing. Discrete sampling is conducted on the
ground at various points provided by the topography (outcrops, stream beds,
17
TABLE 2.1 ......l.
C»
Methodology of the Detection of Mineral Deposits.
Detection Detection Type of Detection Procedures involved

approach method target sought environment
surface visual recognition + physical

prospecting
outcrops and "floats" sampling + laboratory analyses
subsurface and physical sampling + laboratory

trenching
bedrock analyses + visual recognition
direct
detection
underground physical sampling + laboratory
surface drilling ore deposit
bedrock analyses + visual recognition
tunneling and underground physical sampling + laboratory

underground bedrock analyses + visual recognition
drillinq
underground measurements by sensors +
bedrock mathematical data processing
and evaluation
geophysical
surveys primary mineralogic underground measurements by sensors +
halos (magnetite, bedrock mathematical data processing
indirect sulfides, etc) and evaluation
detection
primary geo- underground physical sampling + laboratory n
chemical halos bedrock analyses + statistical data I
geochemical processing and evaluation >
-C
surveys -l
m
secondary geo- surface (soil, physical sampling + laboratory ;;10
chemical halos water, plants) analyses + statistical data -l
processing and evaluation ~
o
PROBABILITY OF DETECTION OF MINERAL DEPOSITS 19
etc.) in reconnaissance surveys, or at control grid intersections, as is the case

for all systematic surveys and drilling programs.
The indirect detection approach may be used on two different types of
targets: mineral deposits themselves through their geophysical signatures and
geophysical or geochemical halos associated with the deposits, which are larger
and thus easier to detect than the deposits. Following the detection and delineation
of the halos, a second stage of investigation is carried out to locate the expected
mineral deposits, either indirectly by further geophysical surveys, or directly by
systematic drilling, within the outline of the halos.
Halos of two types, namely, primary and secondary, are known to occur
in association with many types of mineral deposits. The primary halos were
formed at the same time as the mineral deposits and show a close spatial rela-
tionship with them. There are primary halos whose mineralogical make-up in-
cludes pyrite, pyrrhotite, magnetite, or radioactive elements which are all de-
tectable by geophysical methods. Bedrock sampling followed by multielement
laboratory analysis is used to detect primary geochemical halos formed by the
syn-ore dispersion of minor elements.
The secondary halos arise from the reworking of primary halos and mineral
deposits by secondary chemical, physical, and biological processes. The resulting
halos are much larger and easier to detect than the primary ones; unfortunately,
they are not closely related to the causative body spatially, which makes the
second stage of the ore search much riskier than in the case of primary halos.
Table 2.2 presents a summary coverage of the principal techniques of in-
direct detection used in most modem exploration programs. We do not believe
it is necessary to go into further details, because the subject matter has been
well covered in many general books,(9.15.17) as well as in more specialized
ones.(4.11,12)
2.1.3. The Nature and Structure of the

Indirect Detection Process
The indirect detection of mineral deposits is a sequentially structured process
consisting of three main stages: recording, resolution and processing, and rec-
ognition and confirmation. At the initial stage, the intersection of the mineral
deposit by a sensor generates what is referred to as "messages" in information
theory parlance. (16) The structure of messages is always complex and is made
up of two main portions: (1) a portion of interest to the explorationist called
"signal," which is generated by the mineral deposit itself, and (2) a portion called
"noise," generated by the geo-environment surrounding the ore deposit, and by
extraneous bodies lying between the mineral deposit and the recording platform,
TABLE 2.2
Methodology of the Indirect Detection of Mineral Deposits.
Indirect Detection of Ore Deposits by Geophysical Sensors Indirect Detection of Ore Deposits by Sampling of Geochemical Halos
Type of Geological Techniques and sampling medium

environment processes Litho- Pedo- Hydro- 8io- Atmo-
Approach Method Techniques Parameter Application involved sampling sampling sampling sampling sampling
Earth -airborne 80i1- aerosols

magnetic atmosphere and
magnetometric field -ground interface vapors
Acceleration -airborne
"potential II
qravimetric of gravity -ground plant
Fields Biosphere
natural tissues
gamma -airborne
Natural radiometric radiation -ground
natural secondary lake
electro- trans,[X>rted water
chemical soil
nonpotential spontaneous polarization action -ground
natural Surficial
electro- stream
frequency magnetic -airborne water
AFMAGI dispersion
domain pulses -ground
deposits
resistivity
or
resistivity residual
conductivity -ground
electric soil stream
mise a la potential
silt
masse field -ground
frequency induced -airborne
applied
electro domain electro- -ground ground
Artificial magnetic time domain magnetic Lithosphere water
-airborne
(INPUT) field
primary
frequency apparent bedrock
induced ground dispersion
domain resistivity
polariza-
time transient
tion Ore deposit hydrother-
domain voltage malorrnag-
matic con-
centration
~ - -
----- --- - - - -----
Noise is a detrimental feature which may distort, mask, or suppress signals

making up the signature of mineral deposits. The nature of noise is quite varied
and includes (a) white noise of a random nature (b) thermal noise generated by
the recording instrumentation, and (c) geological noise generated by the geo-
environment.
Messages are either translated as readings visually observable and recorded
manually by an operator or are recorded automatically on graphs or more likely
on magnetic tapes. On most occasions, several messages are recorded and have
to be separated. The ability of sensors and recorders to distinguish between
messages that are closely associated either in the space or time domain is referred
to as "power of resolution. "(18) The separated messages are then subjected to a
special form of statistical processing, known as "filtering" in information theory
parlance, which separates the signal portion from the undesirable noise portion
and enhances the signal strength.(IO.16.21)
Some of the filtered signals are generated by causative bodies of no interest
to the explorationist, although they resemble and may be confused with the
signatures of mineral deposits, thus requiring discrimination. The discrimination
is achieved in an objective manner in a multivariate context by means of a
statistical method known as "pattern recognition" (Ref. 11 in Chapter 12). A
classifying model is "trained" on signals generated by known ore deposits of the
required type. The signals generated by the expected target are statistically
compared with the control signals and are classified either as target or spurious
at a prespecified level of confidence.
Since there is always some degree of uncertainty involved in the detection
of a mineral deposit, explorationists are justified in requiring that the targets be
intersected several times, at least twice, by the sensor in order to confirm the
detection.
2.1.4. Factors Involved in the Detection Process

Many complexly interrelated factors are involved in the detection process,
whether direct or indirect. Some factors are of a geological or economic nature
and should be viewed in a probabilistic context; others are technological and
methodological in nature and should be considered in a deterministic perspective.
Most of these factors interact in a sequential manner.
For the most part, geological factors involved in the detection process are
not under the explorationist's control. Among these are the characteristics of the
expected prizes. Their mineralogical, geophysical, and geochemical properties
are derived by analogy from those of known deposits of the selected type. Others,
such as geometric and economic parameters, and spatial distribution parameters
are derived by statistical modeling of known deposits.
22 CHAPTER TWO
The geological characteristics of the expected prizes will largely dictate the
choice of technology best suited to assist the detection of mineral deposits of
the selected type. The explorationist will choose the type and design of sensor
or probe with regard to performance parameters such as penetration, sensitivity
to the three types of noise mentioned above, resolution, discrimination, and
lateral coverage. (13)
The combination of geological and technological factors will, in turn, pre-
scribe the selection of the most suitable methodology. The latter is defined as
the most appropriate sequence and mix of techniques to ensure the maximization
of the probability of detection of expected ore deposits.
2.2. GEOMETRIC CONSIDERATIONS INVOLVED IN THE

DETECTION OF MINERAL DEPOSITS
2.2.1. Introduction
As indicated above, intersection of the expected target by a sensor or a
probe is the prerequisite of detection. The intersection requirement is essentially
a matter of spatial relationships which are based on geometric considerations.
These considerations involve not only the intrinsic geometric characteristics of
expected deposit, search grids, and detectors, but also their mutual configura-
tions.
The geometric characteristics of the expected deposit include size, shape,
and attitude considered in a three-dimensional context, and are probabilistic in
nature. Grid geometry is described by shape and spacing parameters which are
deterministic in nature, as are the parameters describing the detector geometry
(depth of penetration and inclination). The configuration of the expected target
and the control grid is a very important factor which can be viewed in two
aspects: a dimensional one, expressed by the ratio of the longest dimension of
the target over the grid spacing, and a directional one, expressed by the angle
between the grid and the direction of the longest dimension of the target (strike).
Another important directional consideration is the attitude of the detector with
respect to the target.
2.2.2. Geometric Characteristics of Ore Deposits

2.2.2.1. Nature of Geometric Parameters Describing Ore Depos-
its. The descriptors of the geometry of ore deposits may be grouped into two
categories: dimensional parameters and attitudinal parameters. The dimensional
Slab
Cylinder
Truncated Ellipsoid
Size parameters Shape parameters
L =length R, =B/L
T<:B<:L
B =breadth R2 .. TIL
T =thickness R3 =T/B
V = volume = kLBT
A .. cross section area =kBT
FIGURE 2.1
Size and Shape Parameters of Ore Deposits.
parameters are illustrated in Figure 2.1; they include length, breadth, and thick-
ness, which can be grouped for greater convenience of processing into ratios
such as shape ratios, or products such as areas of principal sections or volumes.
The attitudinal parameters are angular in nature and include the strike ori-
entation, dip angle, and angle of plunge. The strike orientation is measured
24 CHAPTER TWO
clockwise from the true north within 0-180 degrees. The true dip angle (ao) is
measured within 0-90 degrees below the horizontal, within the vertical plane
which is perpendicular to the strike direction. The apparent dip angle (a), mea-
sured in any other vertical plane, is related to the true dip angle by the following
relationship which is tabulated in Table 2.3.:
tan a tan ao X sin d
where d is the angle measured between the vertical plane considered and that
containing the strike line of the target. Finally, the angle of plunge measures
the inclination of the ore deposit within the vertical plane containing the strike
line, in the same manner as the dip.
TABLE 2.3
Conversion of True Dip to Apparent Dip Angles in Degrees.
Angle of True dip angle in degrees

deviation
from strike line
in degrees 5 10 20 30 40 50 60 70 80
5 0 1 2 3 4 6 9 13 26
10 1 2 4 6 8 12 17 26 45
15 1 3 5 8 12 17 24 35 56
20 2 3 7 11 16 22 31 43 63
25 2 4 9 14 20 27 36 49 67
30 3 5 10 16 23 31 41 54 71
35 3 6 12 18 26 34 45 58 73
40 3 6 13 20 28 37 48 60 75
45 4 7 14 22 31 40 51 63 76
50 4 8 16 24 33 42 53 65 77
55 4 8 17 25 35 44 55 66 78
60 4 9 17 27 36 46 56 67 78
65 5 9 18 28 37 47 58 68 79
70 5 9 19 28 38 48 58 69 79
75 5 10 19 29 39 49 59 69 80
80 5 10 20 30 40 50 60 70 80
85 5 10 20 30 40 50 60 70 80
Apparent dip angles in degrees

2.2.2.2. Factors Involved in the Definition of Geometric Parameters

of Ore Deposits. The determination of dimensional parameters of ore deposits
is affected by a complex interaction of geological, economic, and technological
factors, whereas the attitudinal parameters are essentially governed by geological
factors and are simpler to measure. The natural geometry of a mineral deposit
as expressed by dimensions and attitude is governed by geological limits which
may be stratigraphic, structural or petrographic, and geochemical in nature. The
actual dimensions of ore deposits, however, do not coincide with that of the
mineral deposits because economic considerations have to be introduced. The
"cutoff" grade, below which the mineralized material cannot be extracted at a
profit and thus is not ore, is the key economic factor affecting the determination
of the dimensional parameters of ore deposits. An unfortunate result is that the
dimensions of ore deposits will expand or shrink within the maximum geological
dimensions of the mineralization, reflecting the fluctuations of the economic
environment.
Finally, technological considerations have to be taken into account in ad-
dition to the geological and economic ones. For example, the minimum stoping
width or height requirements for easy access of the extractive machinery are an
important factor modifying the economic dimensions of ore deposits. Ore dilution
caused by weak hanging walls is a very detrimental factor affecting the dimen-
sions of ore deposits.
2.2.2.3. Statistical Modeling of Geometric Parameters. Because the

geometry of the expected prizes is not known beforehand, we have to rely on
probabilistic estimates of geometric parameters which are derived by the statis-
tical modeling of known ore deposits of the same genetic type as that of the
expected target. The purpose of model fitting is to establish on the basis of
statistical tests (x-squared goodness of fit tests) which theoretical model best
describes the observed frequency distribution of parameter measurements ob-
tained from known ore deposits. The procedure and pitfalls of statistical mod-
eling, as well as the statistical properties of the most commonly used models,
are well described in Refs. 22, 24, 26, 27, and 28.
The normal model, a cornerstone of modem Statistics, is useful when
dealing with dip angles. A related model, known as circular normal is used to
advantage to model orientated angular parameters such as strike directions. The
distributions of dimensional parameters of ore deposits are satisfactorily fitted
by lognormal models. The lognormal model is related to the normal model
through a logarithmic transformation: in other words, data are said to be log-
normally distributed if the distribution of their logarithms is normal. Figure 2.2
illustrates the fitting of lognormal models to the observed distributions of geo-
26 CHAPTER TWO
N·.
"'L' 0.73
M.·.rJI so.
'l. • 0.48
5..7
FT
~~~~~~--~~E
Qi ib 5.. 10 . . 100 (IN OJ so. ~I
MEA at' ItCIRtZONTAL PIIOoIECTION
No.
"\. °S.24
'l.0o.ze
M.o "«I'
'If . ,. .
04 OS 10 ... ~4.o III IQO
· ..
8 WIDTH FEETI
20~
No.
"\. 02.13
'l. ° 0.21 "\..~
No.
.2 'l. 0030
Me° 1178'
I Me-O. . .
4
910 -"~,a5 0 .2!1 LOG SCALE

OJ 02 Q4 OS •.o ... 36
G COPP£R IIIWIE I... '
N 0"
"\.°0().411
'l. ° 0.117
Me°S. ....
Qis •• -..... .&

-.2 0 .2 .. .II
0.' • li .. :u .6 ... &40 53
H MOL'tIIIlENITE GRADE lMoSa""
FIGURE 2.2
Histograms of Distributions of Geometric Parameters of Porphyry-Cu-Mo De·
posits of the North American Cordillera Belt and Fitted Lognormal Models.
metric parameters of porphyry Cu-Mo deposits of the Cordillera Belt of the

Americas. (25)
2.2.3. Geometry of Ore Deposits in the Context of Detection

2.2.3.1. Introduction. Since the main topic of the book is the detection
of ore deposits, it is appropriate to consider not so much the geometry of the
deposits themselves but that of the target they offer to detection. Detection is
obtained from a horizontal plane, either the ground, or the flight plane for airborne
surveys. It is restricted to a specific vertical range below the observation plane,

and involves the configuration of detector and ore deposit.
2.2.3.2. Vertical and Angled Direct Detection. We will first consider

the case of vertical detection by drilling. The target geometry for detection
purposes is defined by the dimensions of the vertical projection on the horizontal
plane of the portion of the ore deposit which lies within the range of penetration
of the probe. For practical purposes, 300 feet would be an appropriate figure for
the depth of most preliminary drilling programs.
The four parameters required for the calculation of the probability of de-
tection are the target length L" target shape ratio Rt = BrlLt, the true dip ao,
and the plunge c. Figure 2.3 illustrates how the target parameters L t , B t , and Rt
are obtained in tenns of the dimensional parameters Lh and Bh of the horizontal
FIGURE 2.3
Relationship between Target Breadth, Dip Angle, and Detection Range in the
Case of Vertical Detection of Ore Deposits.
IHOrTzontal-- - -
I ~~~t i on of
ore~posl t
=l
Surface project ion of target breadth (Bt )
Di p component
__
Surface
Depth of
I overbur4en
or capping
* (H)
I True Dip
(0)1
Vert i cal depth
ao
range of
Detect ion
I
~
Vertical Detection
True breadth of
(Airborne' ground geophysical surveys
, dr i I ling) /-ore deposit (B)
o
B -
t
...!.L
sina
+ (D-H) cota
0
o
28 CHAPTER TWO
section of the ore deposit within the range D, and the attitude parameters (true
dip and plunge). The relationships are as follows:
L, = Lh + D cot c
B, = Bh + D cot ao with Bh = Bo/sin (ao)
Bo being the "true width" in common drilling parlance, and
R, = B,IL,
If the deposits are subvertical (dip and plunge in the 85-90 degree range), the
three target parameters L" B" and R, are approximately equal to the parameters
L h, Bh, and Rh of the horizontal section of the deposits.
Let us consider now the case of angled detection by drilling to the same
vertical depth D as in the previous case. The diagram of Figure 2.4 shows that
the dip component of the target breadth B, is inflated by the addition of a term
involving the drilling angle b in the following manner:
B, = Bh + D cot ao + D cot b
This results in a much larger shape ratio R" which substantially boosts the
magnitude of the probability of detection. If we consider subvertical deposits,
the term cot ao becomes equal to zero and the dip component included in the
calculation of the target breadth is reduced to that introduced by angled drilling.
2.2.3.3. Vertical Indirect Detection. The previous section dealt with

direct detection by vertical or angled drilling. We now consider the indirect
detection of mineral deposits by geophysical sensors, which is essentially a
vertical detection situation.
The vertical distance r between the sensor platform and the mineral deposit
to be detected strongly affects the strength s of the signal generated by the
deposit, as expressed by the following relationship:
the exponent n being an integer depending on the type of detection technology

and shape of the depositY3,15) For example, n = 1 for magnetic methods, n = 2
for gravity surveys, and n = 3 for electromagnetic methods.
The nature of the spatial relationship between geophysical signatures and
the causative mineral deposits is of a critical importance in the indirect detection
Surface projection of target breadth (8 t )

Section (So) orizontal-- - - - - - - - - - --,
perpendicular
to Strike
~ section of.
or~eposlt
Dip component
_______ ~
I
Surface
Deptli of Sel41cted
overburll In drilling
or capptn~ angle b
(H) I
___ t
True Dip
ngle a o
Selected _I
I
L
Depth range
,0,
~F..:.-a ng Ie c - a0 _
+ _
b _
Angled drilling True breadth of

----->~ore deposit (8 0 )
8 - _8_ _ + (D-H)(cotao+cotb)
t 5 in a o
FIGURE 2.4
Relationship between Target Breadth, Dip Angle, and Detection Range in the
Case of Angled Detection of Ore Deposits.
process. All types of signatures essentially reflect a contrast of geophysical

properties between the ore deposit and the encasing wallrock. When "potential
field" methods (gravity, magnetics, radiometrics) are used, a large portion of
the signal intensity consists of an "edge effect" generated by the contrast, while
the remainder is attributed to a "volume effect" reflecting the intrinsic geometry
of the mineral deposit. Through mathematical processing involving the calcu-
lation of the second derivative of the potential functions, such as "continuation,"
which can be applied to all three methods, or "reduction to the Pole" for magnetic
surveys only, it is possible to ensure a very close spatial coincidence between
signature and causative body. (2,8,20) If we deal with artificially generated fields
30 CHAPTER TWO
(electromagnetics and induced polarization, in both frequency and time domains),

however, the edge effect makes up most of the signal intensity and the contri-
bution of the volume effect is quite small.
In practice, most airborne or ground geophysical surveys rely on a com-
bination of techniques of the two types, measuring natural potential or artificially
induced fields. The criterion of success becomes a multiple spatial coincidence
of various types of signatures. For example, out of 33 well-documented discov-
eries of volcanogenic massive sulfide deposits in northwestern Quebec, as many
as 29, or 83% of the total, showed good mUltiple coincidence and spatial cor-
relation with anomalies; similar results based on ten discoveries were recorded
in the new volcanogenic sulfide district of Western Ontario.
In conclusion, if we consider indirect vertical detection, we can legitimately
assume that the geometry of the target defined by combined geophysical sig-
natures closely compares with that of the target generated by the portion of the
causative mineral deposit which lies within the detection range.
2.2.4. Grid Geometry in the Context of Detection

2.2.4.1. Intrinsic Grid Geometry. An accurate recording of the location
of samples or sampling profiles is an essential requirement for all systematic
airborne or ground surveys or drilling programs in order to pinpoint the exact
location of any discovery. Airborne surveys are flown by aircraft along imaginary
lines located later from photographic recordings, while ground surveys and drill-
ing programs are carried out at the intersections of actual lines materialized by
equidistant markers.
Three parameters, (1) type of design, (2) shape, and (3) spacing, are required
to describe the geometry of control grids. The designs are of two kinds: (1)
parallel lines, mainly used for continuous sampling by airborne surveys, and (2)
orthogonally intersecting lines, chiefly used for ground surveys and drilling
programs. In several theoretical studies, however, other geometric designs such
as rhombic, triangUlar, and hexagonal point-nets are considered for ground search
programs. (34,39,45,47)
The third geometric parameter of control grids is the spacing between lines
or between points. Parallel grids are defined by the spacing s between lines. The
size of orthogonal grids is defined as s x s for square-shaped grids, and as
s x t, or sand w = sit, for rectangular grids. When considering ground sam-
pling, one should bear in mind that the number of sampling points and, therefore,
the sampling cost are inversely proportional to the square of the grid spacing;
i.e., if the grid spacing is cut by half, the sampling cost will quadruple.
2.2.4.2. Geometric Relationship Between Target and Grid. Two

main considerations are involved when selecting the spacing of a control grid
in the search for a specific type of mineral deposit. The first one is related to
the geometry of the target and can be viewed in two aspects, namely, dimensional
and directional. The second one is related to the size of the area to be covered
and the cost implications.
The dimensional aspect of the geometric relationship between target and
control grid can be simply quantified by the ratio U, of the longest dimension
of the target, L" over the grid spacing s. The ratio U, is of critical importance
in the calculation of probabilities of detection by both airborne and ground
surveys. The shape ratio of the target, R, = BIL" plays an important role in
the calculation of probabilities of detection by discrete ground sampling, but is
of lesser importance in continuous sampling airborne surveys.
Next to the dimensional aspect of the target-grid relationship, the directional
aspect has a substantial effect on detection probabilities. The directional factor
may be simply expressed by the angle between the directions of the longest
dimension of the target L, and the lines of parallel grids, or either side of square
grids, or the longest side of rectangular grids.
2.3. THE THEORY OF GEOMETRIC PROBABILITIES AS A

FOUNDATION FOR THE EVALUATION OF THE PROBABILITY
OF DETECTION OF MINERAL DEPOSITS
2.3.1. Rationale of the Application of Geometric Probabilities to

the Detection of Mineral Deposits
On the one hand, the standard theory of probabilities is concerned with
random variables, which are quantities or sets of quantities taking numerical
values in some sets of possibilities, subject to certain required conditions. On
the other hand, the theory of geometric probabilities deals with random elements
which are not quantities but geometrical objects, including points, lines, and
convex planar figures. A principal theme of the theory of geometric probabilities
is the relationship between the various geometrical objects, as expressed by
intersection, confusion, coverage, etc.
As indicated above, the detection of mineral deposits results from the in-
tersection of the target by a detector moved along a control grid; therefore, we
are faced with a geometric problem. On the one hand, the parameters defining
grid geometry and detector attitude are deterministic in nature. On the other
32 CHAPTER TWO
hand, the geometric characteristics of expected targets (size, shape, attitude, and
location) are essentially probabilistic in nature, and are expressed by expected
values and their fiducial intervals or probability distributions. Accordingly, it
seems quite appropriate to introduce the theory of geometric probabilities as a
foundation for the calculation of the probabilities of detection of mineral deposits.
2.3.2. History of the Development of the

Geometric Probability Theory
One of the earliest examples of the consideration of geometric probabilities
is the famous Buffon problem of the 18th century. A needle of length L is thrown
at random on a horizontal plane on which parallel lines are ruled a unit distance
apart. Buffon formulated the expression of the probability that the needle inter-
sects the lines.
More recently, during the last two decades, workers such as Kendall,(36)
Savinskii, (40) Singer, (45) and Solomon(50) have studied the geometric probability
aspects of some much more complex problems of intersection and coverage.
The problem of coverage involves the calculation of probabilities that certain
planar geometric figures of known shape and position are covered by other
geometric figures whose position is random. One much-studied example involves
the coverage of a planar lattice of points of known geometry by some convex
figure such as a circle, or an ellipse, whose position and orientation are random.
The probability of interest involves the number of points of the lattice lying
within the ellipse.
2.3.3. Application of the Theory of Geometric Probabilities

to the Detection of Mineral Deposits:
Survey of the Literature
There was a noticeable lack of interest in both the theory of geometric
probabilities and its applications prior to World War II. Following the impetus
of research done on practical war-time problems, interest in the application of
geometric probabilities spread to mineral exploration in the early postwar period.
In the mid 1950s, Agocs(30) was among the first workers to apply the solution
of Buffon's problem to the calculation of probabilities of detection of mineral
deposits by airborne surveys. Slichter(l9) attacked the problem of detection of
targets of varied shapes (annular, rectangular, and circular) by vertical drilling
on square grids. Since that time, most of the applied geometric probability
research concerned systematic drilling programs. Some eight years later, Celasun(31)
expanded Slichter's work by introducing grids of varied shapes (triangular, rhom-
bic, and rectangular) to guide the search for rectangular or elliptical targets of
random or preferred orientation. At about the same time, Ellis (Ref. 25 in Chapter
4) tackled the more realistic problem of detection by vertical drilling on square
grids of parallelipipedic targets in the three-dimensional space rather than in the
planar context of the previous workers.
In 1965, the first publication of Savinskii' s probability tables for the de-
tection of elliptical targets by discrete sampling grids(40) was an important step
in the application of the theory of geometric probabilities to field exploration.
A few years later, Drew(33) was among the first workers to consider the effect
of grid orientation on the detection of targets by ground surveys. Two years
later, Singer(45) published a very important set of probability tables covering the
detection of elliptical targets of random or known orientation by point-nets of
square, rectangular, and hexagonal shapes, which was much more useful for
field exploration than the earlier Savinskii tables.
More recently, in the mid and late 1970s, additional mathematical treatments
of the problem of target detection by drilling were presented by Singer,(47.48)
Shurygin,(42.43) and Drew. (34) Sinclair(44) is one of the few workers to have
touched on the detection of geochemical halos by point-net sampling in a non-
mathematical way. In 1983, the second writer developed the computer program
OPTGRID to improve on the Singer ELLIPGRID program of 1972(46) for the ground
detection of elliptical targets by square point-nets. Detection probability tables
based on the OPTGRID program are included in Section 2.5.5 of this chapter,
and in all six chapters dealing with the detection of six types of ore deposits in
North America.
In the mean time, after a gap of some 20 years which witnessed a massive
use of airborne geophysical surveys, interest in the application of the theory of
geometric probabilities to airborne ore detection was renewed. McCammon(38)
derives the expressions of the probability of detection of targets of varied shapes
by continuous sampling on parallel, orthogonal,and rhombic grids from concepts
involving conditional probabilities. Later on, Chung(32) expanded McCammon's
work by considering the grid-target orientation factor, as well as lateral coverage
along strips rather than flight lines.
2.4. EVALUATION OF THE PROBABILITY OF TARGET DETECTION

BY AIRBORNE GEOPHYSICAL SURVEYS
2.4.1. Introduction
Since the early 1950s, as most of the more easily detectable mineral deposits
had been discovered by ground prospecting, a much favored strategy for the
34 CHAPTER TWO
early stages of regional exploration programs consists of the pattern searching

for hidden mineral deposits by aircraft carrying continuously recording sensors
of various kinds. A mineral deposit is considered as detected if the sensor
intersects the target at least once, and the detection is considered as confirmed
if the target is intersected by the sensor at least twice. The matter involved in
the calculation of detection probabilities is a problem of continuous geometric
probabilities. At this stage we are interested only in selecting the grid spacing
and orientation which will maximize the probability of detection. Cost consid-
erations will be entertained later in Chapter 3.
2.4.2. Parameters of the Study

The geometric considerations involved in the calculation of the probabilities
of target detection by continuous sampling airborne surveys include (a) target
geometry, (b) target-detector configuration, (c) grid geometry, and (d) target-
grid configuration. The target geometry parameters required for the calculation
of the probability of target detection include the lengthL" shape ratioR, = BIL"
strike orientation, and true dip.
The vertical distance between sensor and target is the most important aspect
of the target-sensor configuration; it includes two components: flight elevation
above the ground and depth of the center of the target below ground. The flight
elevation component varies with the type of survey and nature of the topography.
Generally, the flight elevation is kept within a range of 700-1 ()()() feet for
magnetic surveys, but is reduced to a 300-400-foot range for radiometric surveys
and between 100 and 400 feet for electromagnetic surveys. At the Whistle Mine,
Ontario, Slichter(19) showed that the signal strength generated by a sulfide ore-
body is reduced by 400% when the flight elevation is increased by only 50%,
from 500 to 750 feet. The average depth penetration below ground is 300 feet
for most methods, with the exception of radiometric surveys.(13·15)
The most commonly used type of grid design is the parallel one, with
spacing s and orientation at right angles with the expected strike direction of the
target. However, as pointed out by Chung,(32) the lack of parallelism and equi-
distance of the flight lines causes difficulties in detection and interpretation. The
matter of three-dimensional navigational control is of the utmost importance.
When the likely orientation of the expected target is not known, square grid
designs may be used to advantage despite the greater coverage cost, because it
reduces the probability of missing the target. If it proves too costly to fly a
regular square pattern, a compromise rectangular design with spacing s and shape
ratio w = sit smaller than unity may prove advantageous. A rectangular grid
design can be usefully considered as intennediate between the parallel design

(w = 0) and the square grid design (w = 1).
The target-grid configuration is expressed by dimensional and directional
parameters. The dimensional parameter is the ratio U, of the longest dimension
of the expected target L, over the grid size s. Agocs(30) illustrates the marked
effect of the variation of spacing, from 1/4 to 1 mile, on the geometry and
structure of a large aeromagnetic anomaly outlining the Marmora, Ontario, mag-
netite deposit. Henderson(8) shows similar results for the aeromagnetic coverage
of a portion of the State of Pennsylvania. The directional parameter is the angle
between the direction of the longest dimension of the target L, and the flight
lines of a parallel grid or sides of orthogonal grids.
2.4.3. Calculation of the Probabilities of Target Detection by

Airborne Surveys
2.4.3.1. Scope of Study. The three following cases are covered by the
present study:
1. calculation of probability of single and confirmed detection of randomly
oriented targets of varied shapes by airborne surveys on grids of diverse
designs;
2. calculation of probability of single detection of oriented targets of varied
shapes by airborne surveys on parallel grids;
3. calculation of the probability of single detection of randomly oriented
linear targets by airborne surveys covering parallel strips.
The expressions of detection probabilities pertaining to the three cases listed
above are assembled in Table 2.4 for the convenience of the reader. Originally,
Agocs(30) listed many of them without mathematical proof; this was later provided
by McCammon(38) based on the concepts of conditional probabilities. Later,
Chung(32) fonnulated the expressions related to the detection of linearly oriented
targets by airborne surveys on parallel grids and along parallel strips.
2.4.3.2. Probabilities of Detection and Confirmed Detection of Ran-

domly Oriented Targets of Varied Shapes by Airborne Surveys on Par-
allel Grids. The right half of Table 2.5 lists the probabilities of detection of
randomly oriented targets with three shape ratios by airborne geophysical surveys
on parallel grids for values of the ratio U, = Longest target dimension (L,) over
grid spacing s that vary between 0.1 and 2.0, based on the expressions listed in
the upper tier of Table 2.4. As noted from Tables 2.4 and 2.5, there is a
TABLE 2.4
Expressions of Probabilities of Detection of Targets of Varied Shapes by Airborne Geophysical
Surveys on Varied Types of Grids. w
a-
Probability of at Least One Intersection of Randomly Oriented Targets of Various Shapes
by Parallel or Orthogonal Grids of Lines
Linear shape Circular shape Elliptical shape

Grid char- Target7grid
"Linear" probability component "Circular" probability component Composite probability
acteristics geometry
parallel o <u ~ 1 p1=;ohr 1 1 Pc=U Pe=(1-R)P1+RP c

spacing (5) U> 1 P
U(1 - =[- ACOS-J II - -'-\. P = 1 R "" target breadth/target length
1 11 u2) U c
I - = . , . - , . - - + - - - - - + - - - c ; - - - - - - - - - - - + - - - - - - - - - - - - + u = target length/grid size
I 0 <u < 1 P = !:!(4 - u) P = U(2 - u) Pi - "linear" probability component
S1ze
I ~quar(eSXS) 1 TT C
1 - - - , - - . . . , . . - - + - - , - - - + - - - c ; - - - - - - - - - - - + - - - - - - - - - - - - - - 1 p = "circular" probability component
Ire~tangular I W = ~ "1 P = !:![2(W + 1) - UW] p "" U[(W + 1) _ UW] c
I sue (SXT) T 1 11 C
Probability of at Least One Intersection of a Linear or Elliptical Target

with Preferred Orientation (et) by a Parallel Grid
Grid char- Target/grid
Linear shape Elliptical shape
acteristics ~e_~
parallel
0< U ~ P '" 3£ _ ~ cos2a Pe = (l - R) P 1 + RU
spacing (5) 1 'If 3'1f
Probability of at Least Two Intersections of Randomly Oriented Tarf)ets

of Various Shapes by a Parallel Grid
Grid char-
Circular shape Elliptical shape
acteristics
parallel
p = 1 P (l-R)Pl+R
spacing (S) c e
Probability of at Least One Intersection of Randomly Oriented Targets

of Various Shapes by Parallel Strips
Grid char-
Linear shape Circular shape Elliptical shape
acteristics (')
strip central Target/grid
"linear" probability component I "circular" probability component compoSite probability
line spacing geometry :r:
(5) )-
2U W W
strip width O<U~l-§ P =-+- P =U+- P e = (1 - R)P l + RPc
1 n S c S
(W)
Probability of at Least One Intersection of a Linear or Elliptical Target

"-I
m
;lO
with a Preferred Orientation (a) hy Parallel Strips
-I
Grid char-
Linear shape Elliptical shape ~
acteristics
strip central Target/grid o
"linear" probability component composite probability
line spacing geometry
(5)
2U 2U W
strip width O<U~l-~ P 1 = -;:;- - h X eos2et + S p '" (1 - R)P + R/u + ~)
5 e 1 ~ 5
(W)
TABLE 2.5
Probabilities of Detection of Randomly Oriented Targets of Varied Shapes by
Airborne Surveys.
Target length L feet, and breadth = B feet

R is defined as the ratio B/L in the interval O(R(l
U is defined as the ratio Length/Grid spacing (S)
Detection Detection
Square grid (S)x(S) feet Parallel grid with spacing(S)ft
U=L/S R=0.10 R=0.50 R=1.0 R=0.10 R=0.50 R=1.0
0.100 0.131 0.157 0.190 0.067 0.082 0.100

0.200 0.254 0.301 0.360 0.135 0.164 0.200
0.300 0.369 0.432 0.510 0.202 0.245 0.300
0.400 0.477 0.549 0.640 0.269 0.327 0.400
0.500 0.576 0.654 0.750 0.336 0.409 0.500
0.600 0.668 0.745 0.840 0.404 0.491 0.600
0.700 0.753 0.823 0.910 0.471 0.573 0.700
0.800 0.829 0.887 0.960 0.538 0.655 0.800
0.900 0.898 0.939 0.990 0.606 0.736 0.900
1.000 0.965 0.978 1. 000 0.673 0.818 1.000
Confirmed detection
Parallel grid with spacing(S)ft.
1.100 0.116 0.509 1.000 0.714 0.841 1. 000

1.200 0.144 0.525 (*) 0.743 0.857 (* )
1.300 0.179 0.544 (*) 0.766 0.870 (*)
1.400 0.217 0.565 (*) 0.785 0.881 (*)
1.500 0.259 0.588 (*) 0.801 0.889 (*)
1.600 0.302 0.612 (*) 0.814 0.897 (* )
1.700 0.348 0.638 (*) 0.826 0.903 (*)
1.800 0.395 0.664 (*) 0.836 0.909 (* )
1.900 0.443 0.691 (*) 0.845 0.914 (*)
2.000 0.492 0.718 (*) 0.854 0.919 (* )
N.B. (*) indicates that all probabilities 1.000

38 CHAPTER TWO
discontinuity for L, = s, or V, = 1, beyond which another expression of prob-

ability of detection has to be used. It can be readily noted that, for a given value
of V" the probabilities of detection increase substantially when the shape changes
from linear to circular.
The lower left corner of Table 2.5 lists the probabilities of confirmed de-
tection (at least two target intersections by the sensor) of randomly oriented
targets of three kinds of shape by airborne surveys on parallel grids. The variation
of the V, ratio is restricted to the 1.0-2.0 range, because it is intuitively under-
stood that the probability of confirmed detection is always equal to zero when
the ratio V, is smaller than unity.
A comparison of the lower left and right halves of Table 2.5 illustrates the
drastic reduction of magnitude of the probability of detection of linear targets
for specified values of V" when confirmation is required. For example, the
reduction is 85% when V, = 1.1 and 43% when V, = 2.0. If we consider
elliptical targets, the reduction is not so great: it is only 40% for V, = 1.0 and
22% when V, = 2.0.
Since most airborne geophysical sensors, and particularly the radiometric
ones, respond to the effect of a narrow strip of land on either side of the actual
flight lines instead of that of the flight lines alone,(13.32) it is appropriate to
consider the effect of lateral coverage on the detection probabilities. Chung shows
that if w is the width of strips and s the spacing between the edges of adjoining
strips, the probabilities of detection are inflated by the ratio wis. (See the fourth
tier of Table 2.4.)
2.4.3.3. Probabilities of Detection of Randomly Oriented Targets
of Varied Shapes by Airborne Geophysical Surveys on Orthogonal
Grids. When the expected orientation of the target is not known, it may be
preferable to use an orthogonal grid design rather than a parallel one. The
probabilities of detection of targets of three kinds of shape for values of V,
ranging between 0.1 and 1.0 are listed in the upper left corner of Table 2.5 for
square grid designs.
A comparison between the probabilities of detection for parallel grids and
those for square grids for similar target shapes and values of the ratio V, shows
a substantial improvement of probabilities when using square control grids. For
example, the improvement is 95% for V, = 0.1 and down to 43% for V, = 1.0
for linear targets. In the case of circular targets the figures are respectively, 90%
and 0%.
When V, lies in the interval of 1 to square root of 2, the expression used
for the calculation of probability of detection of a linear target by airborne surveys
on square grids is the same as that used for parallel grids when V, is greater
than unity. (See the left comer of the first tier of Table 2.4.) When (V), is greater
than the square root of 2, the probability of detection is always unity. Similarly,
the probability of detection of a circular target by a survey on a square grid is
always unity when V, is greater than 1.0, while the probabilities of detection of
elliptical targets are bracketed by those of the other two types of shape. The
expressions of the probabilities of detection of linear and circular targets by
rectangular grids defined by the spacing s and ratio w = sit are shown in the
bottom portion of the first tier of Table 2.4.
2.4.3.4. Generalized Aerial Search on Various Types of Grids. As
mentioned above, the probabilities of detection of an elliptical target can be
interpolated by considering any ellipse as intermediate between a line and a
circle. It was also indicated at the end of Section 2.4.2 that an airborne search
flown on a rectangular grid may be considered as intermediate between a search
on a parallel grid and one on a square grid. We now have four limiting cases
including two for target shape and two for grid shape. The most general case,
that of a search for an elliptical target (R in the interval 0-1) on a rectangular
grid with shape factor w in the interval 0-1, will be bracketed by the four extreme
cases, as illustrated by Figure 2.5.
The probabilities of detection can be calculated by combining the two
interpolation expressions listed below for each value of the ratio V,. If Ph Pc.
O.B
i
"-
l;
~ 06
~
ffi
FIGURE 2.5 ~
o 0.4
Generalized Search: Detec- ....
I&J
tion of Elliptical Randomly Ori-
ented Targets by Airborne Sur- ; PrIII·~t
veys on Orthogonal Grids.
[From McCammon, R. B.(38)
Figure 4, by permission of
Plenum Publishing Corpora-
tion, New York.]
40 CHAPTER TWO
and P e are the probabilities of detection of linear, circular, and elliptical targets
for a specified type of grid, the relationship between the three is as follows:
R being the target shape ratio in the interval 0-1. If Pp' P s, P r are the probabilities
of detection of a target of specified shape R by surveys on parallel, square, and
rectangular grids, the expression relating the three is written as
Pr = (l - w)Pp + wP s
where w is the grid shape ratio in the interval 0-1.
2.4.4. Probabilities of Detection of Oriented Targets of Varied

Shapes by Airborne Surveys on Parallel Grids
Agocs' and McCammon's studies on which the above is based considered
only the case of randomly oriented targets. Chung(32) was the first worker to
consider the effect of target orientation on the probabilities of detection of linear
targets by continuous sampling surveys on parallel grids. The second tier of
Table 2.4 shows the expression of the probability of detection of an oriented
linear target, as derived by Chung from Marriot's work,(37) when V, lies in the
0-1 interval. The angle a is that between the flight lines and the expected strike
direction of the target. In practice, however, the direction of the strike is known
only within a confidence interval of, say, + or -10 degrees. The correct way
of calculating the probability of detection corresponding to the confidence interval
for the strike direction is to integrate the expression
f +'!Ti18
-'!T/18 2V13-rr cos (20) da
If the angle a is taken as 90 degrees (flight lines perpendicular to the expected

strike within its confidence interval), the correct expression of the probability
of detection of a linear target is
The diagram shown in Figure 2.6 illustrates the effect of varying target-grid
orientations on the magnitude of the probability of detection of a linear target.
I'robabi Ii ty of detection Angle of flight

lines with expected
strike of target
1.0~----------------------------------~
0.9 -
0.820 90 degrees
0.8
0.775 70
0.7 0.671 SO
0.6 0.549 30
0.5 0.458 o
0.4
0.3
0.2
0.1
0.05 0.25 0.45 0.65 0.85

Ratio U ; target length/flight line spacing
Target shape ratio R ; E/L ; 0.10
FIGURE 2.6
Probabilities of Detection of Linear Targets by Airborne Surveys on Parallel
Grids Oriented at Different Angles with Respect to Targets.
For example for (U,) = 1 the probability of detection increases by 80% when
the grid is rotated from 0 to 90 degrees with respect to the expected strike.
Circular targets do not exhibit any preferred orientation because of the
symmetry, so that the probability of detection is always equal to the ratio U,.
The probability of detection of an elliptical target whose long axis is making an
angle of a degrees with the grid is obtained by combining the expressions of the
probability of detection of linear and circular targets as indicated in the previous
Section 2.4.3.4. For example, the increase in the probability of detection of an
elliptical target with shape ratio Rb = 0.50 is 30% when the grid is rotated from
o to 90 degrees with respect to the long axis of the target if (U,) = 0.50.
42 CHAPTER TWO
2.5. EVALUATION OF THE PROBABILITY OF TARGET

DETECTION BY GROUND GEOPHYSICAL SURVEYS AND
DRILLING PROGRAMS
2.5.1. Introduction
The widespread use of discrete sampling programs on ground grids preceded
that of continuous sampling by airborne surveys along parallel profiles. Most of
the work done on the application of the theory of geometric probabilities to
ground exploration during the past three decades pertained to drilling programs
rather than geophysical or geochemical surveys, because of the incentive to
reduce the very high cost of systematic drilling.
2.5.2. Parameters of the Study

2.5.2.1. Target Geometry and Target-Detector Configura-
tion. Targets are assumed to be elliptical in shape with a major axis L, and a
minor axis B,. The parameters required for the calculation of detection proba-
bilities are L, and the shape ratio R, = B,IL" as well as the true dip and the
direction of the major axis with respect to true north. A vertical detection range
of 300 feet is assumed for geophysical surveys and for vertical or angled drilling
programs alike. Note that, for practical reasons, surface angled drilling is re-
stricted to the 35-90 degree range of inclination.
2.5.2.2. Grid Geometry and Grid-Target Configuration. Ground grids

for discrete sampling are considered as point-nets made up of contiguous cells
which have a regular geometric shape. The three parameters defining the grid
are the shape and size of the cells and the orientation of the net with respect to
the expected strike direction of the target. The sampling is carried out at the
nodes of the net. Much research work has been done on drilling grid shapes in
order to determine the most cost-effective designs. The shapes considered vary
from rhombic or lozenge(31,39,42,43) to hexagonal and derived equilateral trian-
gular.(34,45,46) However, in practice, only orthogonal, and particularly square
grids, are considered for field work.
The dimensional relationship between target and grid, which is an important
consideration in the calculation of the probabilities of detection, is simply ex-
pressed by the ratio U, = longest dimension of the target L, over the grid spacing
s. Drill hole spacing requirements are a function of the geology and the nature
of the program. The spacing should be subjected to a continuing statistical
analysis in order to reduce progressively the size of the net and adjust it to the
succession of objectives of the search.
In the case of ground geophysical surveys, the spacing between readings
should be adjusted to the anticipated depth and size of the target sought for.
This is particularly important in the case of the high-cost induced polarization
surveys, where the spacing of the electrodes as well as their configuration are
of critical importance.
The directional aspect of the target-grid relationship has to be considered.
It is advantageous to lay a geophysical grid at right angles to the geological
grain of the survey area, because there is a greater geological contrast across
the strike than along it, thus enhancing the magnitude of the geophysical re-
sponse. (12) However, we will find that the probability of detection of an elliptical
target by a ground sampling survey on a square grid is maximized if the grid is
orientated at angles varying from 18 to 45 degrees with respect to the long axis
L, of the target, depending on the shape ratio R, of the target. If we consider
rhombic or rectangular grids, the detection probabilities are maximized when
the longest dimensions of both grid-cell and target coincide.
2.5.3. Methodology of the Calculation of the Probabilities of

Detection of Elliptical Targets by Discrete Sampling on
Square Grids
2.5.3.1. Introduction. Because of the discrete nature of the sampling,
the mathematical aspects of the methodology of detection probability calculation
are vastly more complicated than in the case of detection by continuous sampling
airborne surveys. The 1969 Singer tables(45) and ELLIPGRID computer
program(46)considerably improved on Savinskii's contribution to the problem of
detection of elliptical targets by point-net sampling.(40) However, in 1982, the
second writer of this book perfected the OPTGRID computer program which, while
based on same general geometric foundation as Singer's, is more efficient and
accurate and better attuned to field exploration requirements than the fonner
programs.
2.5.3.2. Detection of Oriented Elliptical Targets. The OPTGRID ap-
proach is more efficient than previously published programs, because it makes
use of infonnation implicitly contained in the previous studies but not taken
advantage of, which results in the maximization of detection probabilities. In
the first stage of the overall optimization of target detection based on the dynamic
programming approach (see Chapter 4), it was found that detection probabilities
can be maximized depending on the target shape ratio R, = B,IL, by orienting
44 CHAPTER TWO
the grid with respect to the longest dimension of the target L, in the following
manner:
• for R, less than 0.15, the "maximal" angle is 18-20 degrees;
• for R, in the 0.15-0.30 interval, the "maximal" angle is 30 degrees;
• for R, greater than 0.30, the "maximal" angle is 45 degrees.
If we consider the confirmed detection requirement (at least two target intersec-
tions by the detector), the "maximal" angle is zero degrees since such a config-
uration makes the best use of the full length of the target.
The OPTGRID approach is also more realistic than the previous ones, which
assume that the strike direction of the target is accurately known. However, this
is not the case in practical field exploration situations. Generally, the strike
direction is approximately known within a confidence interval of, say, ± 10
degrees about a mean orientation. In the OPTGRID program, the grid is orientated
at the optimal angle with the target mean strike direction according to the target
shape, and probabilities of detection are calculated within the full 20 degree
confidence interval by integration.
Unfortunately, the integration by standard analytical methods is intractable
because of the very complex nature of the expression of the probability of
detection in terms of grid and target parameters. A graphical method based on
Simpson's rule provides an approximation to the integration. In the Simpson
method, any small arc (P(,P 2 ), however complex in shape, which is limited by
two ordinates corresponding to an elementary increment h of the variate x, is
replaced by an elemental arc of parabola passing through the points PI and P 2 •
Thus the area comprised between the two ordinates, the parabola arc and the x
axis, is a numerical approximation of the value of the integral of the very
complicated function F(x) in the interval (x, x + h). This can be easily extended
to any finite arc of curve of any complicated shape corresponding to the (a, b)
interval for the x variate, which can be divided into n intervals of width h equal
to (b - a)/n. This leads to the Simpson expression
lab F(x) dx #h/3[F(a) + 4F(a + h) + 2F(a + 2h) + 4F(a + 3h) +
+ 2F(a + (n - 2)h) + 4F(a + (n - l)h) + F(b) ]
which can be very easily programmed on the computer.

The introduction of the Simpson rule to take into account the confidence
interval for the target-grid orientation improves greatly the accuracy and prac-
ticality for field use of the detection probability calculations. This is illustrated
by the following example considering the detection of an elliptical target of shape
R, = 0.25. According to previous results, we find that the optimal orientation
of the grid is 30 degrees plus or minus 10 degrees with respect to the long axis
(L,) of the target. Thus the interval of variation of the orientation angle is from
20 to 40 degrees, or 'Tr/9 to 2'Tr/9 if expressed in radians. The 'Tr/9 interval is
divided into 100 elemental intervals, each 'Tr/9oo wide. The computer is pro-
grammed to calculate the area under the curve between the x axis and the 'Tr/9
and 2'Tr/9 ordinates by means of the Simpson's approximation formula. The
probability of detection, which represents the mean height of the probability
function within the interval 'Tr/9 = 0.348, is then easily obtained by dividing
the Simpson result by the coefficient 0.348.
2.5.3.3. Detection of Randomly Oriented Elliptical Targets. Since

the long axis L, of the target is randomly oriented within the horizontal plane,
there is no maximal orientation of the grid which would enhance the probability
of detection, as was the case in Section 2.5.3.2. Because of the symmetry
introduced by the use of a square grid, it is sufficient to restrict the scope of the
study to the 0-'Tr/4 range of grid-target orientations.
Instead of applying the Simpson rule to the interval from + 10 to - 10
degrees about the optimal grid orientation, we apply it to the full interval ex-
tending from 0 to 'Tr/4. The elemental increment h is equal to 'Tr/4oo. The value
of the probability of detection is obtained by dividing the area calculated by
means of the Simpson rule by 'Tr/4 = 0.785.
2.5.4. Application of the Methodology to the Construction of

Detection Probability Tables
2.5.4.1. Detection Probabilities for Oriented and Randomly Ori-
ented Targets of Varied Shapes. The probabilities of detection of linear
(R, = 0.10), elliptical (R, = 0.50), and subcircular (R, = 0.90) targets with
longest dimension L, by surveys on square grids of size s are calculated by the
method described above for varying values of the grid-target dimensional ratio
A, = L,l2s in the 0-2 range. The probabilities of detection of oriented and
randomly oriented targets are displayed in the left portion of Tables 2.6 and 2.7,
respecti vel y .
When A, is less than 0.5, i.e., L, is less than s, the probabilities of detection
of oriented and randomly oriented targets are equal for similar values of R, and
A,. However, when A, is greater than 0.5, the probabilities of detection of oriented
targets are substantially larger than that of randomly oriented targets of similar
46 CHAPTER TWO
TABLE 2.6
Probabilities of Detection of Oriented Targets of Varied Shapes by Ground
Surveys or Vertical Drilling Programs.
Survey design: square grid with spacings S by S feet

Elliptical targets with expected major axis L feet, and
expected minor axis = B feet
R is defined as the ratio B/L in the interval O(R(=l
A is defined as the ratio semi-major axis/grid size = L/2S
in the interval 0.5(A(2
Detection Confirmed detection

A=L/2S R=O.lO R=0.50 R=0.90 R=O.lO R=0.50 R=0.90
0.50 0.079 0.393 0.707 0.000 0.012 0.016

0.55 0.095 0.475 0.839 0.000 0.040 0.053
0.60 0.113 0.565 0.919 0.002 0.079 0.109
0.65 0.133 0.662 0.969 0.005 0.128 0.235
0.70 0.154 0.756 0.997 0.009 0.252 0.398
0.80 0.201 0.911 (*) 0.021 0.411 0.908
0.90 0.254 0.993 (*) 0.040 0.622 (*)
1. 00 0.314 (* ) (*) 0.067 0.974 (*)
1.10 0.380 (* ) (* ) 0.100 (*) (*)
1. 20 0.452 (* ) (* ) 0.136 (* ) (*)
1. 30 0.531 (* ) (*) 0.174 (* ) (*)
1. 40 0.613 (* ) (*) 0.214 (*) (* )
1. 50 0.695 (* ) (* ) 0.254 (* ) (* )
1. 60 0.777 (*) (* ) 0.295 (*) (*)
1. 70 0.850 (* ) (*) 0.335 (* ) (* )
1. 80 0.910 (*) (* ) 0.375 (*) (* )
1.90 0.955 (*) (* ) 0.413 (* ) (* )
2.00 0.984 (* ) (* ) 0.449 (* ) (*)
(*) Probabilities = 1.000

N.B. (1): If A (0.5 all probabilites are the same as randomly
orientated targets, and may be read from that table
N.B. (2): The detection probability is maximized when the grid is laid
out at the following angles with the expected target strike:
18 degrees + or - 10, when R ( 0.15
30 degrees + or - 10, when 0.15 ( R ( 0.50
45 degrees + or - 10, when 0.50 ( R ( 1.00
N.B. (3): The probability of confirmed detection is maximized when the
grid. is laid out at an angle of 0 degrees + or - 10 with the
expected target strike.
TABLE 2.7
Probabilities of Detection of Randomly Oriented Targets of Varied Shapes by
Ground Surveys or Vertical Drilling Programs.
5urvey design: square grid with spacings 5 by 5 feet

Elliptical targets with expected major axis L feet, and
expected minor axis = B feet
R is defined as the ratio B/L in the interval O(R(=l
A is defined as the ratio semi-major axis/grid size = L/25
in the interval O(A(=2
Detection Confirmed detection

A 1/25 R=0.10 R=0.50 R=0.90 R=i0.l0 R=0.50 R=0.90
0.10 0.003 0.016 0.028 0.000 0.000 0.000

0.20 0.013 0.063 0.113 (**) (**) (**)
0.30 0.028 0.141 0.254 (**) (**) (**)
0.40 0.050 0.251 0.452 (**) (**) (**)
0.50 0.079 0.393 0.707 0.000 0.003 0.022
0.60 0.113 0.553 0.916 0.000 0.013 0.102
0.70 0.152 0.713 0.994 0.002 0.057 0.391
0.80 0.196 0.854 1. 000 0.005 0.151 0.817
0.90 0.244 0.947 (* ) 0.011 0.473 (* )
1. 00 0.295 0.986 (* ) 0.019 0.875 (*)
1.10 0.349 1. 000 (*) 0.030 0.994 (*)
1.20 0.406 (*) (* ) 0.045 (*) (*)
1. 30 0.463 (*) (*) 0.064 (*) (*)
1.40 0.520 (*) (* ) 0.088 (*) (*)
1. 50 0.576 (*) (* ) 0.118 (*) (*)
1.60 0.631 (* ) (*) 0.153 (*) (*)
1. 70 0.683 (*) (*) 0.195 (*) (*)
1. 80 0.731 (*) (*) 0.245 (*) (*)
1. 90 0.773 (*) (*) 0.303 (*) (*)
2.00 0.807 (*) (*) 0.417 (*) (*)
N.B. (*) Probabilities 1. 000 (**) Probabili ties 0.000
shape for similar At values. For example, the improvement is 6% for At = 1,

rising to 20% for At = 1.5, and to as much as 25% for At = 2 in the case of
linear targets R t = 0.10. For elliptical targets R t = 0.50, however, the maxi-
mum increase is only 7%, obtained when At = 0.8, and there is hardly any
improvement when we deal with subcircular targets, as could be intuitively
appreciated.
48 CHAPTER TWO
2.5.4.2. Confirmed Detection Probabilities for Oriented and Ran-

domly Oriented Targets of Varied Shapes. The probabilities of confinned
detection (at least two intersections of the target by the detector) are tabulated
in the right-hand portions of Table 2.6 (oriented targets) and Table 2.7 (randomly
oriented targets). Obviously, when At is less than 0.5, or L t less than s, the
probabilities of confinned detection are always nil in both cases.
A comparison of both sides of each table shows that the probabilities of
confinned detection are always much smaller than their single detection coun-
terparts, as could be intuitively expected. For example, the probability of con-
finned detection is only 2% of that of single detection for At = 0.60, increasing
to a maximum of 45% for At = 2, when we consider oriented linear targets
R t = 0.10. In the case of randomly oriented targets, the probability of confinned
detection is only 2.5% of that of single detection when At = 0.8, reaching a
maximum of 52% for At = 2.
We can conclude that the grid size required to obtain a specified level of
probability of confinned detection should be much smaller than in the case of
single detection for specified values of R t and At, therefore considerably boosting
coverage costs. This could be easily expected: extra assurance costs extra money.
2.5.5. Influence of Orientation Errors on Detection Probabilities

In practical field exploration situations, it is usually not possible to specify
exactly the strike orientation of an expected target. The best we can do in
statistical tenns is to specify a confidence interval about an expected orientation
angle, as stated in Section 2.5.3.2. A topic of considerable interest is to find
out how the probabilities of detection are affected by errors of varying magnitudes
on target-grid orientation, if we assume that the other target parameters R t and
L t are known without error.
The results of Mickey and Jespersen' investigation of the topic(39) are sum-
marized by a set of curves graphed in Figure 2.7. The curves were obtained by
varying the angular error from 0 to 90 degrees against the shape ratio R t , for
specified values of the probability of detection. These curves may be considered
as contours representing a "sensitivity surface," i.e., the sensitivity of detection
probability to orientation errors for various target shapes.
Mickey's conclusion was that the detection probability is affected only to
a minor degree by even large errors of target orientation specification for a wide
range of conditions. For example, the orientation of the long axis of an ellipse
of shape ratio R t = 0.50 has to be in error by as much as 48 degrees for the
probability of detection to fall from 0.75 to 0.70. But the detection probability
becomes increasingly, though moderately, sensitive to orientation errors, as the
0.2
-0
;lJO
o
OJ
)-
OJ
r-
0.25
~
o..,.,
o
~
m
.2 Cl
! 0.33 <5
~ z
o..,.,
.75
~
Z
m
0.5 s;:r-
o
m
-0
o
<Jl
~
<Jl
1.0
o 10 20 l) 40 50 60 70 IKl !l
AIItjt of orientation .or in degrees
FIGURE 2.7
Contours of Sensitivity of Detection Probability to Errors in Target Orientation. ~
[From Drew, L. ).(34) Figure 5, by permission of Plenum Publishing Corporation, 1.0
New York.]
50 CHAPTER TWO
target becomes more elongated. For example, if the shape ratio R, = 0.20, the
orientation error has to be 30 degrees in order to reduce the probability of
detection from 0.75 to 0.70, instead of 48 degrees for the more rounded target
shape.
2.6. PROBABILITY OF SEQUENTIAL DETECTION OF ORE DEPOSITS
2.6.1. Introduction
In the previous sections of this chapter, we considered only single-stage
detection procedures, such as airborne geophysical surveys, ground geophysical
surveys, or drilling programs. It would be fruitful, however, to consider a com-
monly used two-stage sequential approach to the field detection of ore deposits.
The procedure requires the detection of halos of the kinds known to be associated
with specified types of ore deposits, as a first stage, to be followed by the
detection within the halo of the causative ore deposit.
The methodology of the calculation of the probability of sequential detection
is divided into three steps: (1) calculation of the probability of detection of the
halo, (2) calculation of the probability of detection of the ore deposit within the
halo, and (3) calculation of the overall probability of success based on (1) and
(2).
2.6.2. Probability of Detection of Halos

There are several types of primary halos which surround mineral deposits
as three-dimensional envelopes. Some of them are detectable by airborne or
ground geophysical surveys, such as pyritic and magnetite halos associated with
porphyry-Cu-Mo deposits, magnetic halos surrounding contact-metasomatic de-
posits, magnetic and radioactive halos associated with syenitic intrusions or
diatremes. Primary geochemical halos of minor elements can be easily detected
by systematic bedrock sampling. The procedure of calculation of the probability
of detection of these two types of halos is similar to that used in relation to the
detection of mineral deposits themselves, as described in Sections 2.5.3 and
2.5.4.
The detection of secondary geochemical halos is not considered here, be-
cause halos and causative bodies are not closely related spatially, so that an
intermediate stage, generally a geophysical survey, has to be introduced before
the final drilling stage.
2.6.3. Probability of Detection of Mineral Deposits within Halos

If an ore deposit of area At lies entirely within a larger area Ah which is
limited by any type of convex perimeter, the probability that a sampling location
randomly chosen within the area Ah will intersect the ore deposit is calculated
according to the theory of geometric probabilities as
Assuming that both areas are elliptically shaped, their extent is easily calculated
as the product of the lengths of the major and minor axes by 'lT/4.
It should be noted that the probability of detection calculated in the manner
described above is a minimum value which can be easily exceeded at a substan-
tially higher cost by systematic grid sampling within the perimeter of the halo.
In the latter case the probability of detection is calculated as shown above in
Sections 2.5.3 and 2.5.4.
2.6.4. Overall Probability of Sequential Detection

One of the basic laws of the standard theory of probabilities states that if
events E and F are statistically independent, the probability of E and F occurring
simultaneouly is the product of the probabilities of occurrence of E and F. (See
Section 1.1.3.3 of Chapter 1.) By applying this law to our sequential situation,
we can express the overall probability of sequential detection as the product of
the probability of detection of the halo by that of detection of the ore deposit
within the halo, both being calculated as shown in Sections 2.5.3 and 2.5.4.
Detection: General Topics

I. BIRDSALL, T. G., 1965, The theory of signal detectability in Information Theory in Psy-
chology, pp. 391-402, Free Press, Glencoe, Illinois.
2. CRIBB, J. L., and VICKERS, I. R., 1973, Experience with transformation of aeromagnetic
data to the Pole, Bull. Austr. Soc. Expl. Geophys. 4 (September), 1-8.
3. CUMING, J. D., and WICKLUND, A. P., 1978, DiamondDrillHandbook, J. K. Smit, Toronto,
Canada.
4. ELLIOT, I. L., and FLETCHER, W. K. (Ed.), 1975, Geochemical Exploration 1974, Elsevier,
Amsterdam.
52 CHAPTER TWO
5. GREGORY, A. F., 1967, Remote sensing in the search for metallic ores: A review of current
practices and future potential, Geol. Surv. Can. Econ. Geol. Rep. 26, pp. 511-526.
6. GOETZ, A. F. H., ROCK, B. N., and GOWAN, L. c., 1983, Remote sensing for exploration:
An overview, Econ. Geol. 78, 573-590.
7. GREENWOOD, J. E. W., 1965, Air photographs in economic mineral exploration, Geol. Surv.
Can. Pap. 65-6.
8. HENDERSON, R. G., and ZIETZ, I., 1949, The upward continuation of anomalies in total
magnetic intensity fields, Geophysics 14, 517-533.
9. KUZWART, M., and BOHMER, M., 1978, Prospecting and Exploration of Mineral Deposits.
Elsevier, Amsterdam.
10. LEE, Y. W., 1960, Statistical Theory of Communication, Wiley, New York.
II. LEVINSON, A. A., 1974, Introduction to Exploration Geochemistry, Appl. Publ. Maywood,
Illinois.
12. PARASNIS, D. S., 1974, Mining Geophysics. Elsevier, Amsterdam.
13. PATERSON, N. R., 1971, Airborne electromagnetic methods as applied to the search for sulfide
deposits, Can. Inst. Min. Metall. Bull. 64(705), 29-38.
14. PEMBERTON, R. H., 1962, Airborne electromagnetics in review, Geophysics 27,691-713.
15. PETERS, W. C., 1978, Exploration and Mining Geology. Chap. 8, Wiley, New York.
16. RAISBECK, G., 1963, Information Theory: An Introductionfor Scientists and Engineers; MIT
Press, Cambridge, Massachusetts.
17. REEDMAN, J. H., 1979, Techniques in Mineral Exploration, Applied Science Publishers,
London.
18. ROSENBERG, P., 1971, Resolution, detectability, and recognizability, Photogramm. Eng 37,
1255-1258.
19. SLICHTER, L. B., 1955, Geophysics applied to prospecting for ores, Econ. Geol. Jubilee
Volume 50, 885-969.
20. SPECTOR, A., 1971, Aeromagnetic map interpretation with the aid of the digital computer,
21. ZURFLUEH, E. G., 1967, Applications of two-dimensional linear wavelength filtering, Geo-
physics 32, 1015-1033.
Statistical Modeling of Geometric Parameters of Ore Deposits

22. AGTERBERG, F. P., 1974, Geomathematics, Chaps. 7, 11, and 15, Elsevier, Amsterdam.
23. AITCHISON, J., and BROWN, J. A. C., 1966, The lognonnal distribution with special ref-
erences to its use in economics; Cambridge University Press, Cambridge.
24. DAVIS, J. C., 1973, Statistics and Data Analysis in Geology, Wiley, New York.
25. DE GEOFFROY, J., and WIGNALL, T. K., 1973, Statistical models for porphyry Cu-Mo
deposits of the Cordillera Belt of North and South America, Can. Inst. Min. Metall. Bull.
66(735), 84-90.
26. JOHNSON, N. I., and KOTZ, S., 1970, Continuous Univariate Distributions, Vol. 2, Hough-
ton-Mifflin, Boston.
27. KOCH, G. S., and LINK, R. F., 1970, Statistical Analysis of Geological Data. Wiley, New
York.
28. KRUMBEIN, W. C., and GRAYBILL, F. A., 1965, Introduction to Statistical Models in
Geology, McGraw-Hill, New York.
29. WALPOLE, R. E., 1974, Introduction to Statistics. Macmillan, New York.
Evaluation of the Probability of Detection of Ore Deposits

30. AGOCS, W. B., 1955, Line spacing effect and determination of optimum spacing illustrated
by Marmora, Ontario, magnetic anomaly, Geophysics 20(4), 871-885.
31. CELASUN, M., 1964, The allocation of funds to reconnaissance drilling programs, Q. Col.
Sch. Mines 59(4), 169-186.
32. CHUNG, C. F., 1981, Application of the Buffon needle problem and its extensions to parallel
line search sampling schemes, J. Math. Geol. 13(5), 371-390.
33. DREW, L. 1., 1967, Grid-drilling exploration and its application to the search for petroleum,
Econ. Geol. 62, 698-710.
34. DREW, L. 1., 1979, Pattern drilling exploration: Optimum pattern types and hole spacing when
searching for elliptical targets, J. Math. Geol. 11(2), 223-254.
35. GRIFFITHS, 1. c., 1966, Grid spacing and success ratios in exploration for natural resources,
Mineral Industry Experimental Station, Pennsylvania State University, Special Publication
No I.
36. KENDALL, M. G., and MORAN, P. A. P., 1963, Geometrical Probability, Chas. Griffin,
London.
37. MARRIOT, F. H. C., 1969, Associated directions, Biometrics 25(4),775-776.
38. McCAMMON, R. B., 1977, Target intersection probabilities for parallel lines and continuous
grid types of search, J. Math. Geol. 9(4), 369-382.
39. MICKEY, M. R., and JESPERSEN, H. W., 1954, Some statistical problems of uranium
exploration, U. S. Atomic Energy Commission Report RME-3105.
40. SAVINSKII, I. D., 1965, Probability Tables for Locating Elliptical Underground Masses with
a Rectangular Grid, Consultants Bureau, New York.
41. SCHUENMEYER, 1. H., and DREW, L. 1.,1977, An exploratory drilling exhaustion sequence
plot programme, Comput. Geosci. 3(4), 617-632.
42. SHURYGIN, A. M., 1976, Discovery of deposits of given size in boreholes with preselected
probability, J. Math. Geol. 8(1), 85-88.
43. SHURYGIN, A. M., 1976, The probability of finding deposits and some optimal search grids,
J. Math. Geol. 8(3), 323-330.
44. SINCLAIR, A. 1., 1975, Some considerations regarding grid orientation and sample spacing,
in Geochemical Exploration 1974, pp. 133-140, Elsevier, Amsterdam.
45. SINGER, D. A., and WICKMAN, F. E., 1969, Probability tables for locating elliptical targets
with square, rectangular and hexagonal point nets, Pennsylvania State University Mineral In-
dustry Experimental Station, Spec. Pub!. No 1-69.
46. SINGER, D. A., 1972, ELLlPGRID, a FORTRAN IV program for calculating the probability of
success in locating elliptical targets with square, rectangular and hexagonal grids, Geocom.
Programs No.4, 1-16.
47. SINGER, D. A., 1975, Relative efficiencies for square and triangular grids in the search for
elliptically shaped resources targets, J. Res. U. S. Geol. Surv. 3, 163-167.
48. SINGER, D. A., 1976, RESIN, a FORTRAN IV program for determining the area of influence of
samples or drill holes in resources target search, Comput. Geosci. 2, 249-260.
49. SINGER, D. A., and DREW, L. 1., 1976, The area of influence of an exploratory drill hole,
Econ. Geol. 71, 642-647.
50. SOLOMON, H., 1978, Geometric Probability, Soc. Ind. App!. Math. (S.I.A.M.), Pennsyl-
vania.
CHAPTER THREE
COST OF DETECTION
3.1.1. Introducing Cost Considerations in Ore Detection

In the previous chapter our main concern was the calculation of the prob-
ability of detection of expected targets in terms of the geometry and relative
configuration of targets, detectors, and control grids. The emphasis was on the
maximization of detection, but no consideration was given to the cost required
to achieve the set goal.
Since mineral exploration is a business enterprise and has to be considered
in its economic context, it is obvious that cost considerations will have to be
introduced before any course of action is taken. In most cases, cost will be a
major factor restricting the range of available options, if we are to stay within
prespecified budget limits.
Cost may be directly quantified and expressed in monetary units, as is the
most common practice, or in time units. Cost may also represent the quantification
of concepts of error, risk, or any other events which are detrimental to the
acquisition of a reward.
3.1.2. Definitions and Terminology

In the common usage, terms such as price, cost, and expenditure are often
rather loosely used in an interchangeable manner, at the risk of some confusion
which should be clarified.
Price is the number of monetary units which has to be provided in exchange
for specified amounts of goods or services or reward. The rate is theoretically
fixed by the law of supply and demand, but is more often than not administratively
set to suit specific policies. Price is neutral: it does not imply any connotation
of activity or effort to acquire a reward.
Cost carries the implication of planned activity. Cost is the product of the
monetary price per unit by the number of units of goods or services which the
planner expects to be required to carry out a specified activity. Any complex
activity such as mineral exploration can be broken into a number of component
activities, each carrying a component cost. Program costing or cost estimation
is an exercise of major importance in planning.
55
56 CHAPTER THREE
Expenditure belongs to the execution stage. Expenditure is the number of

monetary units actually expended in carrying out a planned activity for which a
preliminary cost estimate had been made. Expenditure is strictly limited by the
allocation of resources decided on at the planning stage.
3.2. COSTING AND COST ESTIMATES
3.2.1. Introduction
Most of the planning stage of any mineral exploration program is taken up
by the organization of activities and the cost estimations of actual expenditures.
Costs cannot be determined exactly beforehand: they have to be estimated,
because mineral exploration functions in a context of uncertainty affecting the
execution and outcome of planned activities.
Cost estimation is generally based on the "analogy" principle. Initial esti-
mates are made by comparison with well-documented case histories of programs
which were carried out in economic and geographic environments similar to that
of the planned project. More finely tuned estimates follow. They take into account
local conditions and are indexed for time and geographic variations. There are
two well-recognized methods of costing projects: unit costing, and ratio costing.
3.2.2. Cost Estimates

Cost estimation follows a well-established sequence starting with "order of
magnitude" estimates, followed by "preliminary estimates," and ending with
more refined "definitive estimates." Each stage has a different degree of precision
requirement attached to the cost estimates. At the initial stage, one could be
satisfied with a 30%-40% error tolerance, to be cut down to 20% for the inter-
mediate stage and further down to the 5%-10% range at the final stage.
Order of magnitude and preliminary estimates are essentially based on the
analogy principle with a relatively minor share of local input. They are docu-
mented mainly from "outside" sources of information such as the publications
of State or national Bureaus of Mines, Departments of Trade and Commerce,o 1)
buyer's guides, trade journals, etc., which report on projects all over the world.
Initial estimates are mainly based on "unit cost" information obtained from as
wide a range of sources as possible, in order to provide an average figure
bracketed by low and high estimates.
The definitive estimates should have a much larger local content than the
initial estimates. They depend mainly on "inside" sources within the same com-
pany or administration, or on information obtained from other companies op-
COST OF DETECTION 57
erating in the same district or adjoining regions. Final estimates may be further
refined by indexation of unit costing to take into account time and geographic
variations, thus leading to the ratio costing approach.
3.2.3. Unit Costing

At the initial stage, as we are dealing with "order of magnitude" estimates,
overall unit costing is generally favored, covering a project as a whole or in-
dividual targets within the project. More accurate preliminary estimates are based
on the breaking down of a project into a sequence of basic component activities.
Each component is characterized by a specific unit cost attached to it, and the
contribution of the component to the overall cost is obtained by multiplying the
unit cost by the magnitude of the activity.
Unit costs may be considered in a spatial context: as cost per unit of distance
or area covered or traveled, or as cost per unit of distance or volume worked.
Unit costs may be considered in a time context, based on hours (flying), days,
or months (salaries, charter transportation, etc.). Finally they may be considered
in a task context, such as cost per station, cost per sample, etc. A schedule of
unit costs expressed in 1982 U.S. dollars covering airborne and ground surveys
and physical exploration as it applies to remote regions of North America was
compiled from various public and private sources and is displayed in Table 3.1.
3.2.4. Indexation of Unit Costs

Uncertainty associated with cost estimation based on the analogy principle
is well known, but it may be mitigated by the practice of indexing unit costs
for time and geographic variations. Time variations generally result from the
superimposition of short-term business cycles on longer-term underlying trends
such as inflation, etc. Unit costs are usually reported in current monetary units
(dollar) for a specified year or period of years. If we want to apply the unit cost
to today's estimation problem, an adjustment must be made so that the cost is
reported in real or constant dollars for a base year. An index most commonly
used for that purpose is the Gross National Product Implicit Deflator, or Inflator,
depending of the base year chosen. The index is based on the total value of
goods and services constituting the Gross National Product of a nation during a
specified period. An alternative is the Wholesale Price Index, which is based on
the changes in wholesale prices of a large range of products and commodities
within a specified period. It should be noted that the two indices do not, as a
rule, behave in the same manner. (4)
Geographic cost variations reflect many factors including local economic
and geographic conditions. For instance, the cost multiplier for a specified type
58 CHAPTER THREE
TABLE 3.1
Schedule of Exploration Unit Costs for Remote North American locations
(U .S. Dollars).
Airborne surveys
Reconnaissance work:
Remote Sensing: Colour photo (including Infra-red) $150/sq. mile
Infra-red Scanner $520/sq. miles
Magnetic $ 50/line mile
Magnetic + Radiometric $ 75/line mile
Detailed work: fixed wing aircraft:
E.M. $ aO/line mile
E.M. + magnetic $IOO/line mile
Helicopter:
E.M. $lOO/line mile
E.M. + Magnetic $130/line mile
Ground surveys
Reconnaissance:
Geological mapping: $150/sq. mile
Geochemical drainage survey $lOO/sq. mile
Gravity $300/sq. mile
Detailed Work:
Gridding: open country $ 50/line mile
dense bush $l20/line mile
Geological survey $500/sq. mile
Geochemical (Rock or soil sampling) $IOOO/sq. mile
Laboratory analysis: $4 per sample
Geophysical surveys:
Magnetic $200/line mile
E.M. $250/line mile
Radiometric $l20/line mile
Self Potential $l20/line mile
Gravimetric $500/line mile
I.P. $600/line mile
Bedrock exploration
Trenching: $3/cubic foot
Drilling: percussion: reconnaissance $IO/linear foot
systematic $6/linear foot
Diamond Drilling: reconnaissance $35/linear foot
systematic $20/linear foot
Geophysical logging of drill hole: $5/linear foot
COST OF DETECTION
of survey coverage of a unit of area in Alaska or on the Arctic Coast may be,
respectively, 3 times or 5 times the cost of a similar coverage in the western
U.S.A. The practice of indexation makes it possible to compare meaningfully
unit costs for similar activities carried out in different areas and in different
periods of time.
3.2.5. Ratio Costing

The more finely tuned final cost estimates are greatly assisted by the ratio
costing approach, which gets around the difficulties caused by time and geo-
graphic variations without resorting to indexation. In the previous approach of
unit costing, the analogy principle was applied to the unit costs themselves for
each component of the whole program. In the ratio costing approach, the analogy
principle is applied to the percentage of the total cost which is attached to each
component of the program. The rationale of ratio costing is that, while individual
unit costs may considerably vary over time and geographically, the mix of types
of costs and their relative contributions to the total cost of the project can be
expected to vary very little over long periods of time, unless some drastic
technological change occurs.
The very detailed cost data provided by Metz(ll) for 23 exploration projects
carried out during the past decade in three regions of Alaska provided the base
material for an example of ratio costing which follows. The 23 programs were
grouped into three categories: 3 regional reconnaissance, 6 localized reconnais-
sance, and 14 localized systematic coverage programs. The cost of each type of
program was broken down into four main elements: salaries, transport, camp
and provisions, and exploration. Transport was subdivided into air and ground,
and exploration into four types of activities comprising geological and geochem-
ical surveys, geophysical surveys, drilling, and miscellany, including staking.
If we consider the reconnaissance programs, we find that the overall cost
of regional coverages is about 10 times that of localized coverage, while the
cost mix is about the same: infrastructure cost share of 90% against actual
exploration share of 10%. The overall cost of localized systematic coverage is
about 25 times that of localized reconnaissance, but the cost structures are quite
different.
The infrastructure share of the overall cost falls from 90% to 65% when
moving from localized reconnaissance to localized systematic coverages. On an
itemized basis, we find that the salaries share falls from 36% of the total to 23%,
the air transport share remains at 24% in both types of coverages, while that of
camp and provisions falls from 25% to only 12%. In contrast, the share of total
60 CHAPTER THREE
exploration rises from 10% (Q 35%, including a rise of the drilling share from
0% to 25% of the total expenditures.
Providing that similar data bases are available for other regions of North
America, ratio costing of the type described above for Alaska should greatly
assist exploration management in establishing accurate definitive estimates for
their projects.
3.3. COST FUNCTIONS
3.3.1. Introduction: Concept of Cost Function

The concept of cost function is based on the recognition that each type of
exploration activity has a cost structure based on two main components; one is
distance-related, the other is sampling-related. The structure of each component
may be rather complex: for instance, the distance-related component may cover
such items as traveling to and within the survey area, as well as the construction
of control grids. The principles of ratio-costing are combined with unit-costing
in order to construct a simplified model of a specific type of exploration activity
in terms of the two components mentioned above. The model is called a cost
function. Three types of models are considered in this chapter: (1) overall regional
exploration, (2) airborne geophysical surveys, and (3) discrete ground sampling
programs including drilling. They were assembled on Table 3.2 for the readers'
convenience.
3.3.2. Cost Function Models for Regional Exploration

The distance factor directly or indirectly affects all stages of regional ex-
ploration from the initial reconnaissance stage to the target drill-testing stage.
As a result, the overall cost function for regional exploration may be expressed
as a linear function of the distance from the center of the prospecting area to
the nearest transport facilities (railhead, road, airbase, etc.). This type of model
has been used by the writers for various regional programs in the eastern Canadian
Shield, and is written as follows:
C = c(1 + kD)
where C is the cost function, c is the estimate of the cost of exploration of a
unit of area, D is the distance from the center of the unit area to the nearest
transportation facility in miles, and k is a coefficient of proportionality determined

empirically. For instance, in the central portion of the Superior Province, we
had c = 0.635 $ million for a 100-square-mile cell, and k = 0.0182.
Another type of model, more flexible than the previous one because it
incorporates two coefficients instead of one, was used in the Grenville region
and is written as follows:
where c = 0.265 $ million to cover a lOO-square-mile cell, d 1 = distance from

the center of the cell to the nearest road, d 2 = distance from the center of the
whole prospecting area to the nearest railhead. The coefficients kl and k2 were
determined empirically as equal to 0.0056 and 0.0275, respectively.
Finally, another type of model of an exponential nature has been used also
in the Superior Province, and is made up as follows:
c = cO + kef')
where c = 0.165 $ million for the coverage of a 36-square-mile cell, k is a

coefficient in the range 0-1, in this case 0.086, and a is an exponent in the range
of 1-2, determined empirically in our program as 1.12.
3.3.3. Cost Function for Airborne Geophysical Surveys

The unit cost of airborne geophysical surveys depends on the type of aircraft
used, helicopters being much more expensive to run and maintain than fixed-
wing aircraft, and on the survey technology. Combined surveys are generally
favored; the most popular combination includes magnetic, electromagnetic, and
radiometric instrumentation.
Since geophysical airborne surveys are run along flight lines on a continuous
sampling basis, the sampling- and distance-related components are confounded.
There are, however, additional distance-related cost components which reflect
(a) the distance from aircraft base to local operation base (ferrying), (b) distance
from local operational base to survey area, and (c) the cost of surveying-in the
fiducial points required for the control of the flight lines.
There are two main types of grids commonly used as control for airborne
geophysical surveys: parallel grids with spacing s, and orthogonal grids, either
square with dimensions s by s, or rectangular with dimensions s by t, with t
greater than s.
TABLE 3.2
Tabulation of Expressions of Cost Functions for Various Stages of the Mineral Exploration Sequence.
Stage of Methods and

Exploration Techniques of Parameters of Cost Type and Geometry Cost Function Model for
Sequence Coverage Functions of Survey Control Cost Function Model for Area (A) One Mile Square (L = 1 = 1)
area covered =A sq. parallel lines

C = Co (~+~) = (~
mI. or strips (5) S 2 C1 C0 5 + 2)
Airborne area length (L) and
Geo12h:lsical width(n
Regional Surveys perimeter P = 2 (LH)
control grid spacing
(Magnetic, square grid
= Sf (parallel lines, c=c (2A+~) = (~
Electromagnetic) (5 X S) o 5 2 C1 2C 0 5 + 1)
Coverage square grid)
INPUT, or S,T, (rectangular
grid)
(Radiometric)
rectangular
C = unit cost ($/line grid C = C + ~ c = c (~ + ~ +2)
o S
h! T + ~)J loS '1'
emile) (5 X T)
Ground Reconnais- Co = overhead cost

c::ance: Cd = cost of travelling no systematic
Geological. to station
Ge~che;ical: bedrock Cs of observa-
= cost grid c=c Hlfic +nkC (not considered)
Sampling-; Stream tion at station ods
Geophysical: n = number of stations
- (gra:vIrn;t;ic)
Systematic area covered == A = square grid
Ground Surveys LX ~ C= Cs(~ + ~ [~ + 20L + 1J
(S S)
perimeter = 2 (L + 9,)
X C1 = Cs(~ + 1)(~ + 21)
(geophysical, grid spacings: S,T
Selective Cd = "distance" unit
geochemical) rectangular
cost
grid C= Cs(~ + 1)[~ + 20L + 1J
C = sampling cost C1 = Cs(~ + 1)(~ + 21)
(S T)
X
and Sat station
Cd = 20 Cs
Local
area covered = A = square qrid
LX ~
Systematic perimeter = 2 (L + 9,)
Coverage (S X S)
C= Cs(§ + 1)[~ + 0.02L + 1J C1 = Cs(~ + 1) (~ + 1.02)
grid spacings : S,T
Drilling Cd = "distance" unit
cost
rectangular
C = sampling cost
grid
Sat station C= C/~ + 9[~ + 0.02L + 1J C1 = cs(~ + 1)(~ + 1.0~
(S X T)
Cs =50 Cd
D = distance from
centre of explora-
Overall tion area to near- C=C XAX (1 + K.D~)
a
est transportation
Exploration facilitieso (not considered) o< 1K " (not considered)
A = area in sqo miles
Sequence C = overall cost of 1 ( a ( 2
0
coverage of unit
area in easily
accessible regions
64 CHAPTER THREE
Parallel Grids. Considering a rectangular area of A square miles with di-

mensions L and I, the total mileage of flight lines with spacing s miles required
to cover the area is expressed as
M = Ais + PI2
in terms of the area A and perimeter P, with A = Lt, and P = 2(L + I); the
cost function is then written as:
C = c(Als + P12)
with c the unit cost for the combined geophysical surveys.

If we choose a survey area covering one mile square, we have L = I = I
mile, and A = 1 mile square, so that the cost function becomes more simply
C 1 = c(1/s + 2)
Rectangular Grids. The total mileage of flight lines run in a rectangular

pattern with spacings sand t miles which is required to cover an area of A square
miles with a perimeter P miles is
M = A(lIs + lit) + PI2
and the cost function becomes
C = c[A(lIs + lit) + PI2l
The simplified expression for the coverage of a one mile square area is
C1 = c(1/s + lit + 2)
Square Grids. Since both spacing sand t are equal, the expression of the
cost function becomes
C = c(2AIs + Pis)
for the coverage of an area A square miles, and
C1 = 2c(lIs + 1)
for the coverage of a one mile square area.

3.3.4. Cost Functions for Ground Surveys and Drilling Programs

3.3.4.1. Introduction. All systematic ground surveys require the laying
out of a permanent control grid for the accurate location of each sampling point.
The structure of the cost function is thus quite different from that pertaining to
the continuous sampling airborne surveys of the previous section 3.3.3. We have
to deal with a sampling-related component and a distance-related component.
The latter, in tum, comprises grid-related and traveling-related subcomponents.
All three terms of the cost structure are weighted by different unit costs.
Supposing we wish to cover an area of A square miles rectangular in shape
with dimensions L and I miles by means of a discrete sampling rectangular grid
with spacing sand t miles; the expression of the distance-related cost component
is
L(lIt + 1)
and that of the sampling-related component is
(lIt + l)(Lls + 1)
If Cd and Cs are the unit costs for the distance- and traveling-related com-
ponents, respectively, the most general expression of the cost function for discrete
ground sampling becomes
C = (lIt + 1) [LCd + (Lis + I)c s ] (1)
3.3.4.2. Cost Function for Ground Surveys. Based on past experience,

it is generally agreed that the distance-related component of the cost structure
is a significant item, the largest share of which is taken by the cost of establishing
the control grid for the survey. We estimate that the $/mile unit cost Cd of the
distance-related component is 20 times the $/station unit cost of the sampling-
related component C s •
Therefore the general cost expression (I) becomes
C = cAlIt + l)(Lls + 20L + 1) (2)
for the coverage of an A square-mile area of dimensions L and I mile. For the
coverage of a 1-mile-square-area, the expression becomes
C1 = cAlIt + l)(1/s + 21) (3)

66 CHAPTER THREE
The expressions (2) and (3) are further simplified when we deal with square
grids where s = t miles. The general cost expression for the coverage of an
area of A square miles becomes
C = cAlis + 1)(Lls + 20L + 1) (4)
and for the coverage of a I-mile-square area, we now have
C1 = cs (lIs + I)(lIs + 21) (5)
3.3.4.3. Cost Function for Drilling Programs. The structure of the cost
function for drilling programs is quite different from that of the ground surveys,
because of the marked predominance of the sampling-related cost component
over the distance-related one.
The sampling component is a function of the grid spacing, as illustrated by
Figure 3.1, and of the depth of holes, as well as being affected by the unit cost
(cost per linear foot) which varies with the type of drilling. If we take the unit
cost for "chum" drilling, the cheapest of all drilling methods, as unity, then the
unit cost for noncoring percussion drilling is about 2.5 and that for the coring
diamond drilling method is about 7.
The distance-related cost component of the function includes two subcom-
ponents. One reflects the traveling cost between setups (moving), and the second
one that of establishing the control grid for the drilling program and the surveying
of collar locations. Generally the unit cost for moving is substantially larger than
that pertaining to the control grid.
Based on past experience, we can simplify the expression of the cost function
for drilling programs by assuming that the $/hole unit cost for the sampling
component Cs is about 50 times the $/mile unit cost for the whole distance
component Cd, so that the most general expression for a rectangular grid (1)
becomes
C = cAllt + I)(Lls + 0.02L + 1) (6)
for an area of A square miles with dimensions L and I miles. The expression (6)
is further simplified if we deal with a I-mile-square area, as L = I = 1 mile,
and becomes
cs (lIt + I)(lIs + 1.02) (7)

COST OF DETECTION
Hole spacing (equivalent units)
FIGURE 3.1
Graph of Drilling Cost versus Hole Spacing when Holes Cost $4000 Each.
[From Drew, L. J. (34) (of Chapter 2), Figure 8, by permission of Plenum Publishing
Corporation, New York.]
If we choose a square grid with spacings s by s miles instead of a rectangular

one, then expression (7) is further simplified and becomes
C1 = csO/s + 1)(1/s + 1.02) (8)
So far we have considered only vertical drilling programs. However, angled

drilling should be favored when searching for dipping targets, providing that the
68 CHAPTER THREE
likely direction of the dip can be ascertained. The shallowest surface drilling
angle which can be efficiently used in the present state of the technology is 35
degrees. If we drill a hole with an inclination of b degrees frorn the horizontal,
the unit cost is affected by the angle factor: lIsin b, which is greater than unity.
A 35-degree angled-hole will cost 75% rnore to drill than a vertical hole to the
same vertical depth. For 55-degree angled-drilling, the relative cost increase is
22%, and for 65-degree drilling, it is only 10%.
3.4. TABULATION OF COST FUNCTIONS
3.4.1. Foreword
For the readers' convenience, we tabulated five cost functions covering the
three rnain types of field prograrns, including airborne surveys, ground surveys,
and drilling prograrns. Tables 3.3 and 3.4 cover airborne geophysical surveys
on parallel and square grids, based on cornbined rnagnetic, electrornagnetic, and
radiornetric technologies, which are costed at 70 1982 U. S. dollars per line rnile
for fixed-wing craft, and $100 per line rnile for helicopter surveys.
Table 3.5 covers cornbined geophysical surveys including the following
commonly used cornbinations: rnagnetic + electrornagnetic, rnagnetic +
gravity, rnagnetic + induced polarization which are costed at 6 1982 U.S. dol-
lars per station. Sirnilarly, we are using the $6 station cost for cornbined rnultiele-
rnent geochernical surveys including processing laboratory costs. Table 3.6 deals
with vertical drilling on square grids to a depth of 300 feet. Percussion holes are
costed at 2000 1982 U.S. dollars and diarnond drill holes at $6000. The costs
shown in Table 3.6 for vertical drilling are adjusted by rneans of the angle fac-
tor: lIsin 55 when considering 55-degree angled drilling (Table 3.7).
3.4.2. Cost Function Tables

The cost functions pertaining to airborne geophysical surveys are to be
found on pages 69 and 70. The cost function for ground geophysical and geo-
chernical surveys is tabulated on page 71, while those dealing with vertical and
angled drilling appear on pages 72 and 73.
TABLE 3.3
Cost Function for Airborne Surveys on Parallel Grids in U.S. Dollars per Mile
Square.
Survey design: parallel flight lines with spacing S feet = smiles

Cost function: C = (l/s + 2)c, where c is unit cost in $/line mile
Grid spacing C in $ when C in $ when

S in feet s in miles c = $70/1ine ml. c = $lOO/line ml.
200 0.038 1988 2840

400 0.076 1064 1520
600 0.114 756 1080
800 0.152 602 860
1000 0.189 509 728
1200 0.227 448 640
1400 0.265 404 577
1600 0.303 371 530
1800 0.341 345 493
2000 0.379 324 464
2200 0.417 308 440
2400 0.455 294 420
2600 0.492 282 403
2800 0.530 272 389
3000 0.568 263 376
3200 0.606 255 365
3400 0.644 248 355
3600 0.682 242 347
3800 0.720 237 339
4000 0.758 232 332
4200 0.795 228 326
4400 0.833 224 320
4600 0.871 220 315
4800 0.909 217 310
5000 0.947 213 306
5200 0.985 211 302
70 CHAPTER THREE
TABLE 3.4
Cost Function for Airborne Surveys on Square Grids in U.S. Dollars per Mile
Square.
Survey design: square grid with spacings S by S feet = s by smiles

Cost function: C = 2 (l/s + l)c, where c is unit cost in $/line mile
Grid spacing C in $ when C in $ when

S feet smiles c = $70 c = $100
200 0.038 3836.00 5480.00

400 0.076 1988.00 2840.00
600 0.114 1372.00 1960.00
800 0.152 1064.00 1520.00
1000 0.189 879.20 1256.00
1200 0.227 756.00 1080.00
1400 0.265 668.00 954.29
1600 0.303 602.00 860.00
1800 0.341 550.67 786.67
2000 0.379 509.60 728.00
2200 0.417 476.00 680.00
2400 0.455 448.00 640.00
2600 0.492 424.31 606.15
2800 0.530 404.00 577 .14
3000 0.568 386.40 552.00
3200 0.606 371.00 530.00
3400 0.644 357.41 510.59
3600 0.682 345.33 493.33
3800 0.720 334.53 477.89
4000 0.758 324.80 464.00
4200 0.795 316.00 451. 43
4400 0.833 308.00 440.00
4600 0.871 300.70 429.57
4800 0.909 294.00 420.00
5000 0.947 287.84 411.20
5200 0.985 282.15 403.08
TABLE 3.5
Cost Function for Ground Geophysical or Geochemical Surveys on Square
Grids in U.S. Dollars per Mile Square.

Cost function C = (l/s + 1) (l/s + 21)c, where c is the cost in $ at
each station
Grid spacing C when

S feet smiles c = $6
100 0.019 23822.64

150 0.028 12206.64
200 0.038 7792.56
250 0.047 5590.17
300 0.057 4307.76
350 0.066 3482.79
400 0.076 2913.84
450 0.085 2500.83
500 0.095 2189.00
550 0.104 1946.16
600 0.114 1752.24
650 0.123 1594.15
700 0.133 1463.03
750 0.142 1352.65
800 0.152 1258.56
850 0.161 1177.47
900 0.170 1106.91
950 0.180 1044.98
1000 0.189 990.23
1050 0.199 941.49
1100 0.208 897.84
1150 0.218 858.53
1200 0.227 822.96
1250 0.237 790.62
1300 0.246 761.10
1350 0.256 734.05
1400 0.265 709.17
1450 0.275 686.22
1500 0.284 664.98
1550 0.294 645.28
1600 0.303 626.94
1650 0.313 609.84
1700 0.322 593.86
1750 0.331 578.88
1800 0.341 564.83
72 CHAPTER THREE
TABLE 3.6
Cost Function for Vertical Drilling Programs on Square Grids in U.S. Dollars
per Mile Square.

Cost function C (l/s + 1)' (l/s + 1.02)·c, where c is the cost in
thousands of $ of a 300 foot vertical drill-hole
Grid spacing C (percussion) C (diamond)

S feet smiles c = 2 ($thou. ) c = 6 ($thou. )
100 0.019 5791.00 12373.00

200 0.038 1502.62 4507.85
400 0.076 403.85 1211. 54
600 0.114 192.47 577.42
800 0.152 115.82 347.47
1000 0.189 79.l3 237.38
1200 0.227 58.54 175.61
1400 0.265 45.72 l37.17
1600 0.303 37.15 111. 46
1800 0.341 31.10 93.30
2000 0.379 26.64 79.93
2200 0.417 23.26 69.77
2400 0.455 20.61 61.82
2600 0.492 18.49 55.48
2800 0.530 16.77 50.31
3000 0.568 15.35 46.04
3200 0.606 14.15 42.45
3400 0.644 l3.14 39.41
3600 0.682 12.27 36.80
3800 0.720 11. 51 34.54
4000 0.758 10.86 32.57
4200 0.795 10.28 30.84
4400 0.833 9.77 29.30
4600 0.871 9.31 27.94
4800 0.909 8.90 26.71
5000 0.947 8.54 25.61
5200 0.985 8.20 24.61
TABLE 3.7
Cost Function for 55-degree Angled Drilling Programs on Square Grids in U.S.
Dollars per Mile Square.

Cost function C {lis + 1)0 {lis + 1.02).c/sin (55), where c is the
cost in thousands of $ of a 300 foot vertical drill-hole
Grid spacing C (percussion) C (diamond)

S feet smiles c = 2 ($thou. ) c = 6 ($thou.)
200 0.038 1834.36 5503.07

400 0.076 493.01 1479.02
600 0.114 234.96 704.89
800 0.152 141. 39 424.18
1000 0.189 96.60 289.79
1200 0.227 71.46 214.38
1400 0.265 55.82 167.46
1600 0.303 45.35 136.06
1800 0.341 37.97 113.90
2000 0.379 32.53 97.58
2200 0.417 28.39 85.17
2400 0.455 25.16 75.47
2600 0.492 22.57 67.72
2800 0.530 20.47 61. 42
3000 0.568 18.73 56.20
3200 0.606 17.28 51.83
3400 0.644 16.04 48.11
3600 0.682 14.98 44.93
3800 0.720 14.06 42.17
4000 0.758 13.25 39.76
4200 0.795 12.55 37.65
4400 0.833 11. 92 35.77
4600 0.871 11. 37 34.10
4800 0.909 10.87 32.61
5000 0.947 10.42 31. 26
74 CHAPTER THREE
I. BAILLY, P. A., 1964, Methods, costs, land requirements and organization in regional explo-
ration for base metals, Preprint, A.I.M.E. Fairbanks meeting, Alaska, March 18-21.
2. CALLAWAY, H. M., 1950, Expense of exploration, Econ. Geol. 45, 328-330.
3. COX, J. L., 1961, Cutting costs through operations research, Min. Congr. 1.47,45-46.
4. CRANSTONE, D. A., and MARTIN, H. L., 1973, Are ore discovery costs increasing? Can.
Min. 1. 94(4), 53-64.
5. DE GEOFFROY, J., and WIGNALL, T. K., 1973, Design of statistical data processing system
to assist regional exploration planning, Can. Min. 1. 94(11), 30-35; 94(12), 35-36.
6. DE GEOFFROY, J., and WIGNALL, T. K., 1974, Evaluating that exploration project, Can.
Min. 1. 95(5), 42-44.
7. DERRY, D. R., 1970, Exploration expenditure, discovery rate and methods, Can. [nst. Min.
Metall. Bull. 63(694), 362-366.
8. DERRY, D. R., and BOOTH, J. K. B., 1978, Mineral discoveries and exploration expenditures
(1966--1976), Min. Mag. 138(5),430-433.
9. GRANT, F. S., 1971, Some thoughts on the next decade in mineral exploration, Geoexploration
9,63-77.
10. KUZW ART, M., and BOHMER, M., 1978, Prospecting and Exploration of Mineral Deposits,
Elsevier, Amsterdam.
11. METZ, P. A., and CAMPBELL, B. W., 1982, Cost of exploration for metallic minerals in
Alaska, 1982, M.I.R.L. Report No. 56, Department of Commerce and Economic Development,
State of Alaska, Fairbanks.
12. PERRY, A. J., 1968, Organization and costs of mineral exploration in the Southwest U.S.A.,
Pacific S.W. Mining Industry Conference, A.I.M.E., May 1968.
13. PETERS, C. W., 1959, Cost of exploration for mineral raw material, Cost Eng. 4(3), July.
14. PETERS, C. W., 1967, Cost and value of drill hole information, Min. Congr. 1. 53(1), 56-59.
15. SARMA, D. D., 1979, An exploration strategy for prospecting with a case study on copper
prospects at Ingladahl, India, Miner. Deposita, 14, 263-279.
16. SLICHTER, L. B., 1955, Geophysics applied to prospecting for ores, Econ. Geol. 50th An-
niversary Volume 50, 885-969.
CHAPTER FOUR
OPTIMIZING ORE DETECTION
4.1.1. Rationale of the Optimization of Ore Detection

In the previous chapters, we first introduced the concept of probability of
detection (Chapter 2) and showed how the theory of geometric probabilities
could be put to use to compute the detection probabilities for various types of
field programs in terms of target, control grid, and detector geometry and mutual
configurations. Obviously, in the economic context in which mineral exploration
operates, we have to take into account the amount of effort required to obtain
detection of the expected prizes. There has to be a balancing of the detection
reward against its cost.
An inspection of the detection probability tables exhibited in Chapter 2
(Tables 2.3, 2.4, and 2.5) will readily confirm the state of affairs described
above. We can see in the lower portion of the tables that, as the grid spacing
increases, or, conversely, as the coverage cost per unit of area decreases, the
probability of detection falls very gradually toward zero. But if we look at the
upper portion of the tables, we find that, as the grid spacing diminishes, and
conversely as coverage costs rise rapidly, the probability of detection improves
only very slowly until it reaches its peak value of unity.
This is well described graphically by the heuristic model shown in Figure
4.1, which relates in a simple manner the probability of detection of any type
of mineral resources to the search expenditure, as follows:
Pd = 1 - exp( -Kx)
where Pd is the probability of detection, x is the expenditure in dollars for the

saturation coverage of a unit of area, and K is a parameter to be determined
empirically. It can be seen from the left portion of the graph that the probability
of detection rises rapidly to 0.50 or more for relatively modest increments of
exploration expenditures. In the right portion of the graph, however, when the
probability of detection exceeds 0.80, we find ourselves in a situation of dimin-
ishing returns, where large increments of expenditures result in ever-decreasing
75
76 CHAPTER FOUR
Pd Probability of detection
, .00
0.80
0.60
Fitted
Pd = , model:
0.40
0.20
0.00 x
0 0.25 0.50 0.75 , .0 , . 25
Ex Plora t ion expenditure in US !; million for the cov~rage
of a 36 square mile cell
FIGURE 4.1
Graph of Probability of Detection versus Exploration Expenditure.
improvements in probability of detection, until its maximum value of unity is

reached.
It is obvious that a tradeoff has to be sought between the legitimate desire
to maximize the detection of the prizes on the one hand, and the harsh realities
of budget stringency on the other hand. If the budget allocation is too low, the
mesh of the net will be too coarse by necessity and the prize will fall through
and go undetected. Ifthe mesh is too fine, the cost of the search will be so great
as to lead possibly to the "gambler's ruin," or to a deplorable but not unknown
situation when the cost of the investigation exceeds the value of the prize. The
practical recognition of these exigencies intuitively leads to the concept of the
optimization of ore search.
4.1.2. Survey of Literature on Detection Optimization

The optimization of detection is a critical step in the overall optimal planning
of mineral exploration. It follows optimal choices covering ore deposit types to
be targeted, regions to be covered, and prospecting areas to be selected for
OPTIMIZING ORE DETECTION 77
investigation within the regions, and methodology of coverage. However, if one

uses the volume of publications to gauge the level of awareness and interest in
specific topics, it appears that relatively little consideration has been given to
date to the topic of detection optimization in mineral exploration.
The results of a diligent search of commonly available English language
literature proved rather disappointing. Among the enormous mass of papers
related to mineral exploration published during the past three decades, only a
handful, perhaps not more than 25, deal with the topic of field exploration
optimization. The majority are about drilling, leaving only very few dealing with
geophysical or geochemical survey optimization. Most of the 25 are included in
our list of references and selected readings.
The topic of airborne survey optimization has been very briefly touched on
by Agocs (Chapter 2, Ref. 22) and McCammon (Chapter 2, Ref. 30), in the
context of the calculation of the probabilities of detection. Otherwise, the matter
does not appear to have been investigated in the literature. The optimization of
ground geophysical surveys has been mentioned by Brant (Chapter 1, Ref. 15)
and by Cobb (Chapter 4, Ref. 13). The optimization of geochemical surveys
has been touched on by Sinclair (Chapter 2, Ref. 36), and treated by Kelley
(Chapter 4, Ref. 30) and by the first writer (Chapter 4, Ref. 22).
The optimization of systematic drilling programs has received the bulk of
what little attention has been devoted to date to field exploration optimization,
because of the incentive provided by the mitigation of the very high expenditure
required by the search for "blind" ore targets. Slichter(32) was among the first
workers to consider this problem, in 1955. At least six more workers published
on the same topic during the following two decades. The most recent papers are
those by Shurygin(31) in 1976 and by Drew(24) in 1979.
4.2. BASIC THEORY OF OPTIMIZATION
4.2.1. Fundamental Concepts and Definitions

Most business situations, including mineral exploration, require manage-
ment to choose the best possible course of action from a rather large number of
options without perfect knowledge of the possible outcomes. In order to deal
with the complex problems of optimal choice under uncertainty, several important
concepts have to be established and defined. The three main concepts of interest
are goals, criteria, and strategies.
Goals are most often of a qualitative and subjective nature. They are defined
within the framework of corporate policies and resources. Goals may call for
78 CHAPTER FOUR
the acquisition of some type of reward that can be directly translated into mon-
etary units, for example, gross value, gross profit, etc. But in other cases,
different kinds of rewards are sought such as effectiveness, efficiency, precision,
etc., which are indirectly translated into monetary units at the outcome.
The quantification of the qualitative goals into criteria is a necessary step
of the optimization of management's decision making. Criteria may be simple
expressions, or composite ones such as products and ratios or differences which
are better suited to deal with more complex situations.
The third basic tenet of the optimization of business decision making is the
concept of strategy. After defining qualitative goals and quantifying them into
criteria of choice, management has to establish a set of rules to follow, which
will unambiguously tell the decision maker how to deal with the criteria in order
to choose that course of action which is optimal in the prespecified context of
corporate policies and resources. This set of rules is what is referred to as
"strategies. "
4.2.2. Criteria
4.2.2.1. Structure of Criteria. Criteria may be simply structured by
representing a single quantity such as gross value, cost, probability of success,
risk of error as expressed statistically by the variance or coefficient of variation,
or risk of failure as expressed by the complement of the probability of success.
Criteria may be structured as composite expressions such as the product of two
of the quantities listed above. For example, the expected monetary value (EMV)
criterion proposed by Grayson (Chapter 1, Ref. 5) is the product of gross value
times the probability of success; the expected loss criterion is the product of cost
times the probability of failure (complement of the probability of success). Ratios
of two single quantities are also commonly used as criteria, for example, the
effectiveness ratio, which is the ratio of the number of targets actually detected
over the total number of expected targets in the area under investigation. Another
ratio, used by Slichter(32) as an exploration criterion, is his "prospecting success
ratio," which compares the gross value of the prizes to the cost of their detection.
Finally, more complex and flexible structures, such as differences between
some of the composite expressions listed above, are generally more useful when
dealing with actual business situations. The general design is that of a difference
between a reward function and a cost function, both expressed in terms of
common variates and units. Two useful difference criteria are referred to as the
"efficiency criterion" and the "payoff criterion" in the present study.
4.2.2.2. Efficiency Criterion. The efficiency criterion is a convenient

and flexible type of criterion which is quite suitable for most business, engi-
neering, and mineral exploration situations. It is a difference criterion based on

a heuristic model which implies a comparison between a reward and the cost
required to acquire it.
The efficiency criterion may be simply expressed in an analytical manner
as follows:
Efficiency function (x,y,z,) = Reward function (x,y,z)

- K . Cost function(x,y,z)
(x,y,z,) being common variates expressed in common units such as dollars,

spacing in feet, etc., and K being a scaling factor. The reward function may be
expressing gross value, probability of success, precision, etc.
The precision reward function may be used to advantage in engineering
design situations as well as in mineral exploration. We can cite the design of
optimal geochemical sampling programs, or the design of drill sampling plans
for newly discovered ore deposits as good examples of the application of the
precision reward concept. The reward function is expressed in terms of the
number of samples per unit area required to obtain a prespecified level of pre-
cision, which is translated into sample spacing. Likewise, the cost function is
expressed in terms of sample spacing. Kelley(30) gives an example of the appli-
cation of the criterion to the optimization of geochemical programs, while Hew-
lett(28) describes the optimization of the sampling of a porphyry copper deposit
in the Southwest U.S.A., and Roubens(l9) optimizes the sampling of a coal bed
in the U.K.
The detection reward function is that used throughout this study to optimize
the search for six types of ore deposits in North America, by means of airborne
geophysical, ground geophysical, and drilling programs. The criterion pits the
probability of detection expressed in terms of grid spacings, or grid orientation
with respect to target, or detector orientation with respect to target, against a
cost function also expressed in terms of grid spacings or angular intervals.
4.2.2.3. Payoff Criterion. The payoff criterion is the difference expres-
sion which is most commonly used in general business situations. It is also useful
in certain aspects of mineral exploration planning. The reward function is the
expected monetary value already mentioned. The EMV is expressed as the
economic value of an event (gross value, gross profit) weighted by the probability
of occurrence of that event (probability of success). The EMV reward function
is matched against a "loss function" which is another product-type expression
referred to above as "expected loss." It is expressed as the cost of failure weighted
by the probability of failure Pf' which is the complement of the probability of
success P s (Pf = 1 - Ps).
80 CHAPTER FOUR
In the context of mineral exploration situations, the payoff criterion is not

so widely useful as the efficiency criterion, particularly when we are interested
in the optimization of individual stages of the exploration sequence. It is not
rewarding to match the expected gross value of exploration success against the
cost of failure of an airborne or ground geophysical surveyor even a drilling
program of average size, because of the imbalance between the orders of mag-
nitude of the two terms of the comparison. However, if we pit the expected
gross profit, say 20% of the gross value of the prize, instead of the whole gross
value, against the expected cost of a whole exploration sequence, we obtain a
much more realistic and useful criterion which has been used in the literature
during the past two decades (see Chapter 1, Ref. 4, 17, 30, 41, and 42).
4.2.3. Strategies
After quantifying the corporate goals as criteria, we need set rules in order
to enable us to select the optimal course of action among a variety of available
options. The guiding rule is referred to as "strategy" in the terminology of
statistical decision theory. In that sense, an optimizing strategy is not a value
judgment; it is only a uniform rule of action which is established in accordance
with prespecified corporate policies. A strategy is meant to guide a choice of
action but is not a particular action at a given time.
The role played by strategies in the methodology of decision theory is as
follows: each option available to the decision maker is assigned a numerical
score which is the value taken by the criterion for that particular option. All
options are then ranked in order of the magnitude of their scores for the decision
maker to make his selection according to the prespecified strategy. There are
three main categories of strategies: (1) those seeking extrema of one type of the
other (maxima or minima) of the criterion without constraints, (2) those seeking
extrema of one type or the other under constraints, and (3) hybrid strategies
seeking to combine extrema of two types, such as the "minimax" strategy so
favored by many Operations Research workersY6)
Strategies calling for maximization will lead to the maximization of a cri-
terion which implies favorable consequences, such as gross value, profit, pre-
cision, detection, occurrence, success, etc. The strategy requires simply the
ranking of the various options based on their scores, i.e., the numerical values
of the criterion, in order of decreasing magnitude, leading to the selection of
the top-scoring option as the most desirable course of action. In the opposite
case, the selected strategy will be that of minimization, based on a criterion
which implies unfavorable circumstances, such as cost, loss, risk, error, etc.
Again, the numerical values of the criterion will be used as scores for the various
available options; the scores will be ranked in order of increasing magnitude,

and the option with the smallest score will point to the most favored course of
action.
However, most realistic business situations are much more complex than
the ones described above and require more sophisticated strategies with built-in
constraints. These are known as "optimization strategies." The two most favored
kinds are (a) maximization of a "favorable" criterion under the constraint of cost
or risk minimization, and (b), conversely, the minimization of an "unfavorable"
criterion such as cost or risk, under a constraint requiring a specified acceptable
level of reward. Both are very useful in mineral exploration situations and are
the cornerstones of the optimization of the search for six types of ore deposits
in North America.
4.3. METHODOLOGY OF OPTIMIZATION
4.3.1. Introduction
In the previous section, the general concepts of optimization were described
in a qualitative manner outside any mathematical context. However, since the
coming of age of the new science of "Operations Research" born from research
into World War II resources allocation problems, analytical methods of systems
optimization based on a mathematical apparatus have rapidly grown in use and
sophistication. The remarkable development of computer capabilities in the past
two decades has greatly assisted the inroad of the mathematical approach, because
of the very large computational requirements of analytical solutions, particularly
when dealing with probabilistic situations.
4.3.2. Fundamentals
4.3.2.1. Introduction of a Mathematical Model. The methodology
of maximization of favorable criteria and minimization of unfavorable criteria
can be translated mathematically into the study of the analytical behavior of an
"objective function." The function is represented by a dependent variate z whose
value is uniquely determined by a number of independent variates (Xi>
X2, . . • ,X;, . . . ,xn ). Optimization in a mathematical context is turned into the
search for extrema of the objective function which represent the criterion. The
extrema may be either maxima or minima, depending on the type of criterion
and strategy chosen. Extrema may be of a "local nature," i.e., occurring within
specified intervals of the value of the independent variates. They may be of a
82 CHAPTER FOUR
"global" nature if they occur within the full range of variation of all independent
variates. Optimization is achieved when we find the numerical values of the
independent variates for which the objective function reaches an extremum of
the type required by the chosen strategy.
However, we should appreciate that in realistic business situations most
systems operate under some kind of constraint, whether physical, monetary, or
time-related. Therefore, our concern should be to translate the concept of con-
straint into mathematical language. Harbaugh(6) very aptly illustrates the quan-
tification of the concept of constraint by means of a geometric analogy. (See
Figure 4.2.) In order to be more general, he considers two types of objective
functions z of two independent variates XI and X2. One model is linear in nature,
but the other one is quadratic.
The linear objective function is simply represented by the expression
z = axl + bX2 (1)
The geometric representation is that of a plane of height z (Figure 4.2, upper

left comer) which is represented by straight line contours restricted to the upper
right quadrant because of the initial restrictions on the independent variates. If
there are no other constraints, there will be no point on the plane for which the
objective function will reach a maximum, because as XI and X2 increase indef-
initely to the right and upward, so will z. If we follow the convention of shading
the permissible area, we find that all points of the surface yield feasible solutions
because there are no constraints.
In order to be more realistic, we should consider adding constraints to the
objective function. If the constraints are linear, they can be represented by straight
lines which describe linear inequalities, as follows:
(2)
(3)
The three expressions (1), (2), and (3) are graphically represented in the middle-
left portion of Figure 4.2 by the contoured plane and two oblique lines which
define a permissible (shaded) region yielding all feasible solutions. Among them,
the optimum solution, which is maximal in this example, lies at the intersection
of the two inequality lines. At that point, z attains its maximum without going
beyond the constraints. The reasoning would be the same if the inequalities are
not linear (lower left portion of Figure 4.2).
In many instances, the objective function is of a quadratic nature and may
ObjlCtive function Objective function

lillll/' nonlil18lr
No
constraints
(II) (t.)
Two lin....
inequalities
(c) (d)
Two nonlin....
Inequalities
FIGURE 4.2
Geometric Representation of Optimization Under Constraints. [From
Harbaugh, S. W., and Bonham-Carter, G. R.,<6) Figure 8.3, by permission of
John Wiley, New York.)
84 CHAPTER FOUR
be represented by the following expression, which is a general second-degree

polynomial, where a,b,e,d,e are coefficients:
z = axl + bxz + eXT + dxlXz + erz (4)
The geometric representation of the objective function is a parabolic surface

shown in the upper right portion of Figure 4.2. In this case a local maximum
exists at point M whose location is specified by appropriate values of the two
independent variates XI and Xz. If we apply the constraints described by the two
lines (2) and (3), we find that the permissible region does include the local
maximum, which then becomes a unique solution for the optimization problem
(center left portion of Figure 4.2).
It would not be difficult to imagine other cases where the coefficients e,d,e,j
are such that the two constraint lines intersect below the local maximum M. In
such a case, the actual maximum under specified constraints will lie at the
intersection of the two lines, being lower in absolute value than the original
local maximum, which is now out of bounds. In many complex problems, the
constraints could themselves be represented by portions of circles, parabolas,
ellipses, etc., instead of straight lines, and there could be more than two inde-
pendent variables, which would preclude any geometric analogy.
4.3.2.2. Methods of Optimization. Optimization is now redefined as
the search for extrema of an objective function under constraints and guidance
of a strategy. Within this mathematical context, optimization may be conducted
by a variety of methods which are reviewed in detail by Wilde(IO·II) and aptly
summarized by Harbaugh. (6) These authors distinguish two broad classes of
optimizing methods, direct search and indirect search for extrema, in two dif-
ferent contexts: deterministic and probabilistic.
In the deterministic context, which is that of most engineering optimization
situations, the independent variates can be determined in a unique manner through
a specified functional model. In most general business situations and in mineral
exploration in particular, the deterministic definition of the variates does not
apply. The independent variates are represented by probability distributions or
by expected values within confidence intervals. As a result, the dependent variate
of the objective function and the extrema are similarly expressed in a stochastic
manner.
4.3.3. Indirect Search for Extrema Based on Calculus

In its simplest form involving only one dependent variate Z(x) and one
independent variate X within the constraints a :::;; X :::;; b, the indirect search based
on calculus calls for the setting of the first derivative Z'(x) to zero. The next
step consists of finding the roots of the equation Z' (x) = 0 which lie within the
required interval (a,b). Among these, only real roots should be retained, the
complex ones being discarded. Depending on the prespecified strategy, either
the roots that result in the maximization of the objective function (maximal roots)
or the minimal roots are of interest in the problem at hand. The screening is
done by calculating the second derivative Z"(x) and examining its sign for each
extremal value of the independent variate: a negative sign indicates a maximal
root and a positive sign a minimal one.
The method can be easily extended to the more complex and general n-
dimensional case, where we have to deal with n independent variates rather than
only one, such as Xl> X2, Xi, . . . ,xn • The problem is handled by setting the first
partial derivative of Z with respect to each of the n independent variates to zero,
one at a time. The result is a system of n simultaneous equations. The system
is solved in order to determine the values of the independent variates which
satisfy the constraints and generate extremal values of the objective function Z
of the type required by the chosen strategy.
The calculus approach to optimization, straightforward as it may seem, is
fraught with difficulties, many of which are. listed by Bellman. (I) First, the form
and structure of the objective function, even in the simpler uni- or bivariate
cases, may be such as to involve products of transcendental functions which
would make the derivation and finding of extremal roots an intractable com-
putational problem, beyond the capability of even the most modem high-speed
computers. A second difficulty pointed out by Bellman is the ambiguity which
occurs when several extremal roots of the required type are found within the
permissible area. Furthermore, we may have to contend with artificial extrema
resulting from discontinuities of the objective function within the permissible
area, or from the presence of boundaries. Finally, in the multidimensional case,
there may be too many variates and too many constraints to make the calculus
method workable at all.
4.3.4. Direct Search Approach

The direct search approach is heuristic and sequential in nature, leading to
successively improved solutions by series of iterations determined by an algo-
rithm. When no further improvement can be obtained, the solution reached,
although an approximate one, is considered as the best solution to the problem
at hand. Several types of algorithms are commonly used, including the "hill
climbing," and the "elimination" algorithms, both leading to the extrema sought
by direct ascent.
86 CHAPTER FOUR
The direct search approach may provide the only feasible solution to op-
timization in two very common types of situations: (1) when the form of the
objective function is too complex and proves intractable for the calculus ap-
proach, and (2) when the nature of the objective function is not sufficiently well
known to be translatable into pure mathematical terms. The direct search ap-
proach can be successfully handled graphically when dealing with one or two
independent variates and a similarly small number of constraints.
4.3.5. Indirect Search Approach Based on

Mathematical Programming
The mathematical programming approach may be used to advantage in
complex business and engineering situations, when dealing with n independent
variates and m constraints, a type of situation which is beyond the capabilities
of the indirect search based on calculus and the direct search approach. The
mathematical programming approach is based on recursive algorithms, which
are well suited for computerization. Readers should note that mathematical pro-
gramming and computer programming are not to be confused; although the former
requires the latter, the two terms have quite different meanings. The two best-
known methods of mathematical programming are "linear programming" and
"dynamic programming."
Linear programming is the most favored method when dealing with single
stage optimization problems involving n independent variates and n linear con-
straints, such as are most of the complex allocation problems dealt with in
engineering. The SIMPLEX algorithm, which is well suited for computer ap-
plication, is one of the best known. Harbaugh(6) describes its structure and its
application in some detail by means of a numerical example. If the objective
function and the constraints are not linear in structure, other types of algorithms
have to be used.
Dynamic Programming is a more recently developed approach designed to
handle optimization problems quite different from that tackled by linear pro-
gramming. Instead of dealing with one stage and numerous variates and con-
straints, we may wish to optimize multistage sequential situations involving only
a few independent variates and constraints. Although lumped with linear pro-
gramming as part and parcel of mathematical programming, dynamic program-
ming is of a radically different nature. Dynamic programming is not an algo-
rithmic method as are the direct search and linear programming approaches, but
it should be described as a technique of problem formulation.
DynamiC programming parcels out an initially complex problem into a
sequence of simpler and more tractable ones which can then be optimized either
by the indirect calculus-backed approach or by the direct search method. It is a

very versatile and flexible approach which may be applied to any type of problem,
whether linear or nonlinear, deterministic or probabilistic, provided that the
problem at hand can be reformulated as a multistage decision process. The
recursive nature of the procedure allows the application of the results of the ith
stage to facilitate the optimization of the (i + l)th stage of the sequence.
The very flexibility and versatility of dynamic programming is just one
aspect of a more general philosophy of optimization of complex systems, referred
to as Operations Research, whose application is spreading into many fields of
business and engineering.
4.3.6. Operations Research

The term "operations research" (OR) was coined during World War II by
British scientists who pioneered it as a systematic approach to solve very complex
decision problems involved in optimal allocation of scarce resources of war
material and manpower in the context of war situations. OR has since been
considerably broadened to encompass problems of optimal allocation in any type
of business or engineering situation. A relatively new and very useful addition
to OR is the field of sensitivity analysis. Sensitivity analysis is the study of likely
effects of changes introduced to the structure or mode of operation of complex
systems. Its purpose is to differentiate and isolate the most important factors
which are responsible for the changes to the systems from the ones with the
least or no effect. A simple example of sensitivity analysis has already been
described in Section 2.5.6 of Chapter 2.
As a problem-solving approach, OR has to be regarded as both an art and
a science. The science aspect pertains to the use of a mathematical apparatus
and algorithms appropriate for the objective solving of complex decision prob-
lems. It is an applied science devoted to observing, analyzing, and predicting
the behavior of man-machine systems based on "models." The art part of the
definition of OR is justified because most complex decision problems include
important elements which cannot be translated into mathematical terms, one of
them being the human element, which is an integral part of every decision
environment.
Much of the interest in OR originates from business and engineering circles.
With the exception of mining engineering production optimization problems,
OR has had only a small impact on the mineral industry. A survey of the available
literature in the English language regarding the application of OR to the mineral
industry was done by Coyle in 1969.(15) Out of a total of 75 papers reviewed,
60 dealt mainly with production optimization, mine design, particularly open pit
88 CHAPTER FOUR
optimization, and production scheduling, using the linear programming ap-

proach. Only 15 papers dealt with mineral exploration optimization, based on
direct and indirect search approaches, but none on the dynamic programming
method. A diligent search conducted by the writers during the past two years
led to the conclusion that the imbalance between mining engineering and mineral
exploration applications of OR is persisting, if not worsening.
4.4. APPLICATION OF OPTIMIZATION THEORY

TO ORE DETECTION
4.4.1. Introduction
By bringing together detection probabilities and cost functions for each type
of field program as the two key building blocks we can now construct a model
for the optimization of ore detection which takes full account of the basic theory
of optimization, as described in Sections 4.2 and 4.3 of the present chapter.
The "efficiency" model, already mentioned in Section 4.2.2, is the one
selected for the study of ore search optimization, because it is well suited for
the optimization of individual stages of the mineral exploration sequence. The
diagram shown in Figure 4.3 illustrates the makeup of the model in terms of the
probability of detection as a reward function, cost function, and scaling factors.
It should be pointed out that we are interested in optimizing detection, i.e.,
technical success, only, not commercial success, as they are defined in Section
1.1.1 of Chapter 1, so that no consideration of economic worth of the expected
target is introduced in the model.
4.4.2. Practical Application of the Efficiency Model

4.4.2.1 Introduction. Since we are dealing with the application of the
theory of linear optimization, the objective function should be linear in structure
with all terms expressed in terms of a common variate, in the general form
F(x) = KR(x) - MC(x)
where R(x) and C(x) are the reward function and the cost function, respectively,
K and M are scaling factors, and x is the independent variate.
In the context of the application of the efficiency model to control grid
optimization, the grid spacing s is the independent variate, the probability of
detection P is the reward function, and the objective function is written as
F(s) = KP(s) - MC(s)

Statistical Deterministic
modelling of modelling of
geometric parameters cost factor
of ore deposits (C)
, EFFICIE N CY
, MODEL
- -
Detection Detection Efficiency

probability
KP(s)
cost
mC(s) = function
F(s)
I
N.B. k & m = scaling factors
s = grid size
FIGURE 4.3
Construction of an Efficiency Model for the Optimization of Ore Detection.
On the one hand, the range of variation of the reward function P(s) is by definition
restricted to the 0-1 interval. On the other hand, that of the cost function C(s)
extends from 0 to infinity, theoretically at least. In order to deal with this
disparity, we have to choose the values of the scaling factors K and M in such
a way as to keep the terms KP(s) and MC(s) within compatible domains. This
can be achieved by defining M as the reciprocal of the cost required to ensure
a 0.95 probability of detection of a target with expected parameters L, and R"
and K as the reciprocal of 0.95, equal to 1.05. As a result, the efficiency function
will start out in the negative region when s is small and reach a peak when s
increases, heading again for negative values when the grid spacing becomes
quite large, as illustrated by the graphs shown in Figure 4.4. If we wish to
optimize the grid orientation or drilling angle, the independent variate is an
angular interval instead of the spacing s.
The actual determination of optimal survey parameters, including grid spac-
ing and orientation and detector orientation, results from the sequential appli-
cation of either the indirect calculus-based or the direct search approach within
the framework of dynamic programming. The two main types of surveys have
to be handled separately because of the fundamental differences between con-
90 CHAPTER FOUR
Probability of det~ction = p(s) Cost function = C(s}
Efficiency function
Fls) = klPls) - k 2 C(s)
Grid spacing _ 5
Optimal
t
grid spacing
FIGURE 4.4
Determination of Optimal Grid Size Based on the Efficiency Criterion.
tinuous sampling (airborne surveys), on the one hand, and discrete sampling
(ground surveys), on the other hand.
4.4.2.2. Optimization of Airborne Geophysical Surveys. Two main
types of grids are used as control for airborne geophysical surveys for the purpose
of detecting elliptical targets with long axis L, and shape ratio R, = B,IL,. The
grids are of the parallel design with spacing s or of the orthogonal design with
spacings sand t, t being greater than or equal to s.
If we use the parallel design, the probabilities of detection are initially
maximized by orientating the grid at right angles with the expected direction of
the long axis of the target (See Chapter 2, Section 2.4.4). The variate V, = L,ls
describes the relative grid-target geometry. As noted in Chapter 2, Section 2.4.3,
the probability of detection (reward function) shows a discontinuity for V, = 1,
i.e., when the grid spacing equals the long axis of the target. When V, lies in
the 0-1 interval, the structure of the reward function is rather simple (See Table
2.4, Chapter 2). However, when V, is greater than 1, the makeup of the function
becomes more complex, involving radicals and inverse trigonometric terms which
would unduly complicate the differentiation and the finding of extremal roots,
if the indirect calculus approach is envisaged. Furthermore, the discontinuity for
U, = I mentioned above introduces an artificial maximum without practical
value. Under these circumstances, the direct search approach is obviously a more
rewarding avenue for the optimization of airborne surveys.
The computer is programmed to calculate the numerical value of the effi-
ciency function in terms of the target parameters L, and R" with the grid spacing
s as variate. An iteration algorithm guides the search for the value of s which
maximizes the efficiency function when s varies. The output of the program
includes (a) the optimal value of the spacing s, (b) the corresponding coverage
cost per unit of area, (c) the associated optimal detection probability, and finally,
(d) a printed graph of the efficiency function.
If we consider the square design s by s, the structure of the efficiency
function is a rather simple quadratic function of lis, which may be easily dif-
ferentiated, leading to the extraction of the maximal root by the calculus ap-
proach. When dealing with rectangular grids with spacing s and shape parameter
w = sit, we find from Chapter 2, Section 2.4.3.4 that the probability of detection
may be calculated as a weighted average of the probabilities associated with the
parallel and square designs, as follows:
Probability (rectangular grid) = w . Probability (square grid) +

(l - w) Probability (parallel grid)
As the cost function may also be expressed in terms of s and w in a similar

manner (See Chapter 3, Section 3.3.3), it is possible to calculate the optimal
grid size parameters for the airborne detection of a target of specified parameters
L, and R,.
4.4.2.3. Optimization of Ground Surveys and Drilling Programs.
The probabilities of detection of an elliptical target with parameters L, and R,
by discrete sampling on a square grid of size s by s are calculated in terms of
L" R" and s by means of the OPTGRID program designed by the second writer
(see Chapter 2, Section 2.5.3). The structure of the expression of the probability
of detection (reward function) is a very complex combination of transcendental
functions including radicals and inverse trigonometric terms. Not only is the
expression difficult and cumbersome to differentiate in terms of s, but the search
for extremal roots proves quite intractable, so that the indirect calculus approach
has to be rejected in favor of the direct search approach, as was the case for the
optimization of parallel grids for airborne surveys.
We are faced with four considerations: two of them involve the target
92 CHAPTER FOUR
orientation. These are random orientation and expected orientation within a

confidence interval of ± 10 degrees. The other two involve the relative config-
uration of target and detector. These are vertical detection by geophysical surveys
and drilling, and angled detection by drilling.
The simplest case is that of vertical detection of a randomly oriented elliptical
target L" R, with a mean dip of a degrees. The optimization of the grid is a one-
stage computerized procedure based on the iterative algorithm mentioned above.
The output consists of the optimal spacing, corresponding coverage cost per unit
of area, probability of optimal detection, and a graph printout of the efficiency
function.
The most complex case involves the detection by drilling to a specified
vertical depth at an angle of b degrees searching for an elliptical target with
dimensional parameters L, and R" mean dip a degrees, and with a preferred
orientation within a stated confidence interval. The procedure is completed se-
quentially by the dynamic programming approach, as follows: (1) optimization
of the orientation of the grid with respect to the expected target direction, (2)
optimization of the spacing of the grid which has been previously optimally
oriented, and (3) optimization of the drilling angle within the acceptable range
of 35-90 degrees, when the holes are optimally spaced along a grid which has
been, in tum, optimally oriented in stage (1).
We are now ready to consider the application of the methodology described
above in Chapters 2-4 to the optimization of the search for six types of ore
deposits in three regions of the North American continent, which will be covered
in the following Chapters 5-1l.
Methodology of Optimization
1. BELLMAN, R.E., and DREYFUS, S. E., 1962, Applied Dynamic Programming, Princeton
University Press, Princeton, New Jersey.
2. BROOKS, S. H., 1959, Comparison of maximum-seeking methods, Oper. Res. 7, 430--457.
3. DANTZIG, G., 1963, Linear Programming and Extensions, Princeton University Press, Prince-
ton, New Jersey.
4. DIXON, L. C. W., 1972, Non-linear Optimization, Chap. 1, The English University Press,
London.
5. DIXON, L. C. W. (Ed.), 1976, Optimization in Action, Proceedings of the Optimization
Conference, Univ. of Bristol, January 1975, Chap. 1, Academic, London.
6. HARBAUGH, S. W., and BONHAM-CARTER, G. R., 1970, Computer Simulation in Geo!ogy,
Chap. 8, Wiley, New York.
7. PETERSON, E. L., 1961, Statistical Analysis and Optimization of Systems, Wiley, New York.
8. ROSENBROCK, H. H., 1960, An automatic method for finding the greatest or least value of
a function, Comput. J. 3, 175-184.
9. SIVAZLlAN, B. D., and STANFEL, L. E., 1975, Optimization Techniques in Operations
Research, Prentice-Hall, Englewood Cliffs, New Jersey.
10. WILDE, D. J., 1964, Optimum-Seeking Methods, Prentice-Hall, Englewood Cliffs, New Jersey.
II. WILDE, D. J., and BEIGHTLER, C. S., 1967, Foundations of Optimization, Prentice-Hall,
Englewood Cliffs, New Jersey.
12. WILLIAMS, K. B., and HALEY, K. B., 1959, A practical application of Linear Programming
in the mining industry, Oper. Res. Q. 10, 131-137.
Operations Research
13. COBB, H., 1960, Operations Research: A tool in oil exploration, Geophysics 25, 1009-1022.
14. COX, J. L., 1961, Cutting costs through operations research, Min. Congr. J. 47,45-46.
15. COYLE, R. G., 1969, Review of the literature on operations research in the mining industry,
Trans. Inst. Min. Metall. London 78, AI-AI9.
16. DUCKWORTH, E., 1962, A Guide to Operations Research, Methuen, London.
17. HAZEN, S. W., Jr., 1968, Operations Research: A growing force in the mineral industries,
Min. Eng. 20, 88-90.
18. PRUSS, D. E., and FREEMAN, G. W., 1961, Mining exploration: An operations research
and simulation approach, Min. World 14, 42-43.
19. ROUBENS, M. (Ed.), 1977, Advances in Operations Research, North-Holland, Amsterdam.
Optimization of Ore Detection

20. BARNES, H. L., and LAVERY, N. G., 1977, Use of primary dispersion for exploration of
Mississippi Valley-type deposits; J. Geochem. Expl. 8, 105-115.
21. CELASUN, M., 1964, The allocation of funds to reconnaissance drilling programs; Q. Col.
Sch. Mines 59, 169-185.
22. DE GEOFFROY, J., and WV, S. M., 1970, Design of a sampling plan for regional geochemical
surveys, Econ. Geol. 65, 340-347.
23. DREW, L. J., 1967, Grid drilling exploration and its application to the search for petroleum,
Econ. Geol. 62, 698-710.
24. DREW, L. J., 1979, Pattern drilling exploration: Optimum pattern types and spacings when
searching for elliptical targets, J. Math. Geol. 11,(2), 223-254.
25. ELLIS, R. M., and BLACKWELL, J. H., 1959, Optimum prospecting plans in mineral ex-
ploration, Geophysics 24(2), 344-358.
26. FAYINI, G., and ASSAD, R., 1979, An optimized decision model for area selection in massive
sulfide exploration, Can. Inst. Min. Metall. Bull. 72(804) 118-126.
27. GRIFFITHS, J. c., and DREW, L. J., 1966, Grid spacing and success rations in exploration,
A.P.C.O.M. Symposium No 6, Vol. I, pp. QI-Q24.
28. HEWLETT, R. F., 1965, Design of drill-hole grid spacings for evaluating low-grade copper
deposits, U.S. Bur. Mines Rep. Invest. No. 6634.
29. HODGSON, W. A., 1962, Optimum spacing for soil sample traverses, A.P.C.O.M. 10th
Symposium, Johannesburg, South Africa, pp. 75-78.
CHAPTER FOUR
30. KELLEY, 1. c., and McMANUS, D. A., 1969, Optimizing sediment sampling plans, Marine
Geol. 7,465-471.
31. SHURYGIN, A. M., 1976, The probability of finding deposits and some optimal search grids,
J. Math. Geol. 8(3), 323-330.
32. SLICHTER, L. B., 1955, Geophysics applied to prospecting for ore, Econ. Geol. Jubilee
Volume SO, 885-969.
CHAPTER FIVE
APPLICATION OF THE OPTIMIZATION

METHODOLOGY TO THE SEARCH FOR
SIX TYPES OF ORE DEPOSITS IN
NORTH AMERICA
The four previous chapters described the methodology of optimization of

ore detection without reference to any specific type of ore deposit or geological
region of occurrence. The six following chapters will illustrate the application
of the methodology to the search for six genetic types of ore deposits in three
geological regions of North America. The necessary data were obtained from
private sources and available publications and were assembled in a computerized
data bank.
For the sake of practicality, each of the six following chapters is designed
to be self-contained in order to avoid constant references to other chapters, despite
the risk of being repetitious. The standard sequence for each chapter comprises
five sections:
(i) General geological background;

(ii) Synopsis of field detection methodology;
(iii) Statistical modeling of the geometric parameters required for the cal-
culation of detection probabilities;
(iv) Construction of detection probability tables;
(v) Design of three detection strategies, including the optimization of search
grids.
5.2. CONSTRUCTION OF A DATABASE
The initial stages of the project are (a) selection of ore deposit types and
regions to be covered, (b) acquisition of the necessary data, and (c) preparation
of the data for computerized storage and statistical processing.
95
96 CHAPTER FIVE
Various interrelated considerations are involved in the construction of a

database. A principal consideration is the availability of data. Because the main
emphasis of the book is on deposit detection, the main endeavor was to secure
the required geometric data, which are usually much more difficult to obtain
than the economic ones. Fortunately, we were able to secure about 50% of the
required data from private sources. The second half was culled from many of
the references listed at the end of the chapter, and from numerous issues of
various North American journals.
Another important consideration deals with practical computational and
statistical requirements, such as the availability of space in the computer memory,
allocation of computer time for computation and outputting, type of output,
statistical quality of samples, etc.
A summary of the database is presented in Table 5.1. The study covers a
total of 392 commercial and subeconomic ore deposits which belong to six main
genetic types and four subtypes in three regions of North America. Five of the
TABLE 5.1.
Database for the Optimal Detection of Six Types of North American Ore
Deposits.
Region Type of deposit Number of deposits
Cordillera Belt Porphyry-Cu-Mo 57

Contact-metasomatic 113
Volcanogenic sulfides 29
Sub-total: 199
Precambrian Volcanogenic sulfides 102

Shield Ni-Cu-ultramafic 25
Vein-gold 52
Sub-total: 179
Arctic Paleozoic Mississippi Valley Pb-Zn 14

Platform
Grand total: 392

SEARCH FOR ORE DEPOSITS IN NORTH AMERICA 97
deposit types, namely, porphyry-Cu-Mo, contact metasomatic, Ni-Cu ultra-

mafic, volcanogenic massive sulfides, and Mississippi Valley-type Pb-Zn are
important sources of base metals, while the sixth is a main source of precious
metals. The contact metasomatic type is divided into four sUbtypes and the
volcanogenic type covers two regions; otherwise, the study deals with one type
of deposit and one region at a time.
The three geological regions covered by the study are the North American
Precambrian Shield, the North American Cordillera Belt, and the North American
Paleozoic Platfonn. The fourth main geological unit of North America, the
Appalachian Foldbelt, was not included because adequate geometric data could
not be obtained for a sufficient number of its volcanogenic sulfide deposits.
The geometric parameters measured on all 392 deposits of the database are
those required for the calculation of detection probabilities. The parameters
belong to two types: dimensional and attitudinal. The first category includes the
length and breadth of the horizontal section of the portion of ore deposits which
lies within the average range of present day detectors (300 feet), and the shape
ratio R = breadth/length. The second category includes the unorientated true
dip of the deposits and the strike direction measured in degrees from true North
in the right half circle. All five parameters were available for four types of
deposits, but strike direction data were inadequate for the porphyry-Cu-Mo and
Mississippi Valley types which were consequently assumed to be randomly
orientated. The total number of numerical data bits covering the 392 deposits
which were stored in the computerized data bank amounts to 2025.
5.3. DATA PROCESSING
5.3.1. Introduction
The statistical processing of the data was carried out in three steps: (1)
preliminary processing, (2) construction of detection probability tables, and (3)
design of optimal field surveys for the detection of specific types of ore deposits.
Two types of computers and two types of word processors available at the
New South Wales Institute of Technology, Sydney, Australia, were used by the
second writer to store, summarize, and process the 2025 data bits. They include
a main frame Honeywell computer, level 66/60, for storage and computations,
a Hewlett-Packard HP9816 computer and HP7580 plotter, an APPLE II for
simpler computations and table printing, and a LISA for composition and printing
of more complex synoptic tables.
98 CHAPTER FIVE
5.3.2. Preliminary Processing

5.3.2.1. Sampling. The foundation of statistical processing lies in the
requirement of adequate sampling of a homogeneous population of objects, in
the present case all ore deposits of a specific genetic type, whether discovered
or undiscovered within a specific region. Two aspects of sample specification,
namely nature and size, are of crucial importance in judging the validity of the
conclusions drawn from the samplesyl)
According to statistical theory, the samples should be statistically repre-
sentative of the whole population and be randomly selected. These requirements
create difficulties when dealing with populations of ore deposits, because the
definition of ore (material which can be mined at a profit) immediately introduces
a bias in the sampling. A further bias of a geographic nature has to be taken
into account, because some portions of regions are better documented than others.
Furthermore, restrictions in access to data can be expected to result in additional
bias.
The sample size requirement is also quite important. In most practical cases,
we find that a sample size of 50 is more than satisfactory to meet the most
exacting requirements of the theory of Statistics. Such is the case in the present
study for the porphyry-Cu-Mo deposits of the Cordillera, volcanogenic sulfide
deposits, and vein gold deposits of the Shield. Satisfactory sample sizes ranging
from 25 to 50 were obtained for the Ni-Cu ultramafic deposits of the Shield,
volcanogenic sulfide deposits of the Cordillera, pyritic halos associated with
porphyry-Cu-Mo of the Cordillera, and contact metasomatic deposits of the
Pb-Zn-Cu-Ag and W-Mo types. A smaller but still adequate sample, in the
15-25 range, was collected for the contact metasomatic Cu-Fe type. A less than
satisfactory sample, in the 10-15 range, had be be accepted for the Arctic
Mississippi Valley-type of Pb-Zn deposits. Any sample size below 10 would
not be acceptable for standard parametric statistical processing: one should tum
to nonparametric statistical methods which are specially devised to handle small
samples.
5.3.2.2. Statistical Summarization of Data. Once a population of ore
deposits of a specific type has been sampled, however imperfectly, the range of
measurements of each parameter is calculated (largest-smallest), and the mea-
surements are grouped into classes and counted as frequencies. The frequency
distributions are summarized graphically by means of histograms and numerically
by means of statistics which represent the concepts of centrality, dispersion, and
shape of distributions.
The concept of centrality is expressed numerically by the arithmetic mean
(average) which is particularly suitable for the nonnal and other types of sym-
metrical distributions. The geometric mean is more suitable when dealing with
asymmetric distributions such as the log-nonnal type. If the data are ordered by
magnitude, another statistic called median may be used. The median divides the
area under the histogram into two equal portions and is referred to as the 50th
percentile of the distribution. By extending this idea, we can think of other
statistics which divide the area under the histogram into four equal portions and
are called "quartiles." The first quartile or 25th percentile corresponds to 25%
of the area, the second quartile or 50th percentile is the median, and the third
quartile covers 75% of the area. Finally, the mode is another expression of cen-
trality which corresponds to the highest frequency, that is the peak of the histo-
gram. The mode may not be unique, as is the case in multimodal distributions.
The concept of dispersion or spread of the data about the mean has to be
introduced to describe the variability of the data. The range, already mentioned
above, is a very simple expression of the data spread. A more sophisticated one
is the variance or its square root, the standard deviation. In many instances, it
is more convenient to express the dispersion by a dimensionless coefficient such
as the ratio standard deviation/mean.
Finally, the concept of shape has to be considered in order to describe fully
the nature of frequency distributions. Mean, mode, and median coincide in
symmetrical distributions such as the nonnal model. Most distributions are asym-
metrical, being skewed either on the left or the right side. The coefficient of
skewness expressed as the ratio (mean - mode)/standard deviation is a very
convenient way of describing numerically the concept of shape in a dimensionless
manner. It can be seen that the skewness is equal to zero when the distributions
are symmetrical since mean and mode coincide. Mathematical transfonnations
of the observed data may be used for the "deskewing" of observed distributions
into more symmetrical and more tractable ones. In the common case of lognonnal
distributions of observed data, a logarithmic transfonnation will result in a nonnal
distribution for the logarithms of the data.
5.3.2.3. Model Fitting. The fitting of various theoretical statistical models

to the frequency distributions of observed data is an important next step in the
statistical processing. Among the many advantages derived from the fitting of
nonnal and other associated models, the computation of confidence intervals for
the mean of the fitted distribution counts as a major one. The limits of the fiducial
interval are a function of standard deviation, sample size, and a Student-t statistic
provided by tables: the larger the sample size and the smaller the standard
deviation, the narrower the fiducial interval of the mean. Most commonly, 95%
100 CHAPTER FIVE
confidence intervals are used to indicate that we are confident that the true mean
of the population will fall within the interval in 95% of the cases, the probability
of error, that is of the mean falling outside either limit, being 2.5%.
A "goodness of fit" test is used to test the degree of fit between the theoretical
model and the actual distribution of observed data at a specified level of con-
fidence, generally 0.05. The chi-squared test of a nonparametric nature, which
is most commonly used, is easily calculated in terms of the difference between
observed and expected frequencies for each class. If the calculated expression
is less than the critical value provided by tables for the required level of confidence
and degree of freedom, then we can accept the model as a good fit at the 0.05
confidence level. In the present study, lognormal models were successfully fitted
to the distributions of all dimensional parameters of the six types of deposits.
As far as the attitudinal parameters are concerned, normal and circular normal
models were found to be the best fits for the distributions of dip angles and strike
directions, respectively, for all but the contact metasomatic types of ore deposits.
5.4. COMPUTATION OF DETECTION PROBABILITIES
5.4.1. Summary
As indicated in Sections 2.2.3 and 2.2.4 of Chapter 2, the computation of
detection probabilities for the six types of ore deposits covered by the present
study involves the following geometric considerations:
(i) Intrinsic Target Geometry. As described by the dimensional parameters
of the horizontal sections of the portions of deposits within a 300-foot detection
range, the shape of the horizontal section of the six types of deposits is generally
elliptical, varying from nearly linear (vein gold deposits) to subcircular (pyritic
halos of porphyry deposits).
(ii) Relative Configuration of Detector and Target. The deposits covered
by the study belong to several categories on the basis of the dip parameter. Two
types of deposits including porphyry-Cu-Mo and Mississippi Valley Pb-Zn are
either subhorizontal or subvertical so that the target dimensions for detection
purposes are the same as that of the deposits. The other four types are dipping
deposits for which the target breadth is obtained by adding a dip component to
the breadth of the horizontal section of the deposit, as illustrated by Figures 2.7
and 2.8 of Chapter 2, according to the following expressions:
B, = Bh + (D - H) . cot a for vertical detection

and
B, = Bh + (D - H)[(cot a + cot b)] for angled detection
where B, and Bh are the target and deposit horizontal breadths, D is the detection
range taken as 300 feet, and H is the overburden thickness which is taken as 30
feet based on an average of 40 measurements in the Cordillera and 50 mea-
surements in the Shield. Initially, we use 55 degrees as an "all-purpose" drilling
angle. Later, as a part of an optimization study by dynamic programming, the
drilling angle is optimized under cost constraints.
(iii) Grid Geometry and Relative Configuration of Grid and Target. Parallel
and square grid designs are used for the airborne detection of all types of deposits,
except vein gold and contact metasomatic (W-Mo), which are not detectable by
airborne surveys. Square grids are used for the detection of all types of deposits
by ground surveys and drilling programs. The relative configuration of grid and
target involves a dimensional element, the ratio V, = target length/grid spacing,
and a directional element which is the orientation of the grid with respect to the
expected strike direction of the target. Optimization studies show that the de-
tection probabilities for airborne surveys on parallel grids are maximized when
the grid is laid at 90 degrees to the expected target strike. The optimal ground
grid orientation with respect to the expected strike direction lies in the 20-45
degree range, depending on the shape ratio of deposits. No optimal grid ori-
entation is considered for the porphyry-Cu-Mo deposits and their associated
pyritic halos and for the Mississippi Valley Pb--Zn deposits, which are assumed
to be randomly orientated owing to the lack of adequate information.
5.4.2. Construction of Detection Probability Tables

5.4.2.1. Description of Tables. The probability tables dealing with the
detection of six types of ore deposits in North America are of two designs. The
first design provides the probabilities of single detection (at least one target
intersection by the grid) and of confirmed detection (at least two intersections)
for varying grid spacings. The second design, which should prove very useful
for planning purposes, provides the grid spacings and associated coverage cost
per unit of area required to obtain prespecified levels of probability of detection
by various types of field programs.
Table 5.2 illustrates the first design, taking the detection of por-
phyry-Cu-Mo deposits of the Cordillera by ground geophysical surveys or ver-
tical drilling as an example. The head of the table displays all pertinent infor-
102 CHAPTER FIVE
TABLE 5.2.
Example of Probability Table for the Detection of Porphyry-Cu-Mo
Deposits of the North American Cordillera Belt by Ground Geo-
physical Surveys or Vertical Drilling to a Depth of 300 Feet.
Survey design: square grid with spacings S by S feet.

the expected shape of the ore deposits is elliptical
with major axis = L in the 95% confidence interval:
L.c.1. ( geometric mean ( U.c.1., and minor axis = B feet,
R is defined as the ratio BIL with geometric mean = 0.46.
Probability of detection
Spacing S L.c.1. G. mean U.c.1.
in feet L = 2820 3110 3820ft
800 1. 000 1. 000 1.000

1000 0.986 1. 000 1.000
mx 1200 0.914 0.991 1. 000
1400 0.788 0.942 0.999
1600 0.660 0.852 0.973
1800 0.547 0.745 0.925
2000 0.455 0.643 0.847
1m 2200 0.383 0.552 0.761
2400 0.324 0.475 0.677
2600 0.277 0.411 0.600
2800 0.239 0.358 0.530
3000 0.208 0.313 0.470
3200 0.183 0.276 0.419
4000 0.117 0.176 0.273
4800 0.081 0.123 0.189
5600 0.060 0.090 0.139
6000 0.052 0.078 0.121
N.B. Mx: Grid spacing range for 0.980 detection probability.

Lm: Grid spacing range for 0.500 detection probability.
mation required for the calculation of detection probabilities, induding (a) grid
design, (b) target mean shape ratio and mean length within its 95% confidence
interval, as derived from model fitting and, (c) optimal grid orientation with
respect to target strike within a ± 10 degree confidence interval. The latter
statement is omitted in Table 5.2, because porphyry-Cu-Mo deposits are con-
sidered as randomly orientated. The body of the table displays the numerical
values of detection probabilities for grid spacings varied within a realistic range
of 8~OOO feet in steps of 200-800 feet. For each spacing, the table provides
TABLE 5.3.
Example of Table Displaying Grid Sizes and Associated Coverage Cost
per Mile Square for Specific Probability Levels in the Detection of Ni-Cu
Ultramafic Deposits of the North American Shield by Vertical Drilling to
a Depth of 300 Feet.
Survey Design: Square Grid with Spacings S by S Feet

Unit Cost = $6000 Per Diamond Drill-Hole
Probability Level Grid Size Drilling Cost Per Mile Square

Of Detection In Feet In U.S.$. Thousands
0.05 1850 89.58

0.15 1150 188.25
0.25 900 283.73
0.35 750 388.81
0.45 700 438.91
0.55 600 577.42
0.65 550 675.43
0.75 500 803.19
0.85 450 974.35
0.95 300 2077.99
the 95% confidence interval for the detection probability based on the 95%
fiducial interval for the target length.
Table 5.3 displays an example of the second design of probability tables
which covers the detection of Ni-Cu ultramafic deposits of the Shield by vertical
drilling to a depth of 300 feet. Since this type of table is always inserted between
those of the first design, the target information is not repeated in the heading of
the table. The specified levels of probability of detection are varied within a
range of 0.05-0.95 with 0.05 increments.
5.4.2.2. Organization of Tables. Since in practical situations individual

exploration programs deal with one specific type of deposit at a time, all detection
probability tables are organized in a standard sequence which is repeated for
each of the six types of deposits, resulting in a total of 118 tables interspersed
within the next six chapters. The standard sequence is divided into three sections
of three or four tables each. The first section covers airborne geophysical detection
and includes four tables. Since there are no known airborne geophysical methods
to assist the detection of contact metasomatic (W-Mo) type or vein-gold, the
first section is omitted in the sequence of tables displayed in Chapters 7 and 11.
104 CHAPTER FIVE
The second section deals with vertical ground detection by geophysical surveys
or drilling, and the third section covers angled detection by drilling.
5.5. DESIGN OF THREE STRATEGIES FOR THE DETECTION OF

SIX TYPES OF ORE DEPOSITS IN NORTH AMERICA
5.5.1. Introduction
An inspection of Tables 5.2 and 5.3 immediately shows that the probability
of detection rapidly decreases to a fraction of percentage as the grid spacing
increases and corresponding coverage cost decreases. Conversely, as the grid
spacing decreases and the corresponding coverage cost rises very rapidly, the
probability of detection reaches it maximum value of unity. It seems intuitive
that three strategies may be considered, including two options covering each end
of the range, and a "middle of the road" option bracketed by the other two. They
are referred to as the "liminal" strategy for the lower end of the range, "maximal"
strategy for the upper end, and "optimal" strategy as the compromise option.
5.5.2. Liminal Strategy

At the lower end of the range, one may seek to minimize the coverage cost
provided that the probability of detection exceeds a minimum acceptable level,
referred to as "threshold" level-hence the term "liminal" derived from the latin
"limen" for threshold. The 0.500 probability level is an obvious choice for the
threshold: below this value, the probability of failure of detection which is
complementary to the probability of detection would be greater than the prob-
ability of success, thus removing any incentive to carry out the survey.
This point is illustrated in Table 5.2. The three grid spacings corresponding
to the liminal probability level of 0.500 for the mean target length and its 95%
confidence limits are shown by short lines; they are, respectively, 1900, 2325,
and 2900 feet. It can be seen from the table that 1900 feet is the grid spacing
resulting in the cheapest coverage cost which provides a probability of detection
of at least 0.500 in 95% of the cases. Any smaller spacing, say 1800 feet, will
produce a detection probability greater than 0.500 for a higher coverage cost;
any larger spacing, say 2200 feet, will result in a cheaper coverage cost but the
probability of detection may fall below the required threshold of 0.500, if the
true target length falls at the fiducial limits or outside the fiducial interval. The
probability that the target length is smaller than the lower fiducial limit of 2820
feet resulting in a probability of detection lower than 0.500 is one half of
5% = 2.5%. The same reasoning may be applied to the confirmed detection
situation, resulting in a liminal grid spacing which guarantees a 0.500 level of

confirmed detection at least, with a probability of error of only 2.5%.
We feel that the liminal strategy should prove attractive to explorationists
faced with the problem of covering large areas of little known potential within
strict budget constraints.
5.5.3. Maximal Strategy

At the other end of the range, explorationists may be interested in the
coverage of small areas of very high potential in well-known mining districts.
They feel justified in requiring a very high level of probability of detection, as
close as practically possible to certain detection, say 0.980, despite the much
higher coverage cost. Since the probability of detection nears its maximum value
of unity, the second strategy is referred to as "maximal."
Based on the same Table 5.2, it is found that the three grid spacings
corresponding to the required probability level of 0.980 for the mean target
length and its 95% fiducial limits are, respectively, 1000, 1300, and 1500 feet.
The lO00-foot spacing corresponding to the lower fiducial limit for the target
length is the smallest grid spacing, resulting in the cheapest coverage cost which
can guarantee a detection probability of at least 0.980 in 95% of the cases. A
cheaper coverage resulting from a larger grid spacing, say 1200 feet, may not
produce a detection probability of at least 0.980 in 95% of the cases, because
the true length of the target may fall near the lower end of the fiducial interval;
as a result, the probability of detection would be 0.914, instead of the required
0.980 level.
5.5.4. Optimal Strategy

5.5.4.1. Methodology. The optimal strategy is a compromise option
which should suit the majority of exploration programs requiring the coverage
of moderately large areas of fair to good potential within strict budget constraints.
The strategy calls for the maximization of the probability of detection subject
to the constraint of cost minimization. The methodology of field program op-
timization is based on a sequential procedure comprising several stages, each
one requiring the application of the efficiency criterion, as described in Chapter
4 (Dynamic Programming).
In the most general case of oriented dipping deposits, the procedure is
broken into three steps. When dealing with randomly oriented deposits such as
porphyry-Cu-Mo or associated pyritic halos or Mississippi Valley Pb-Zn de-
posits, the first step is omitted. The initial step is the optimization of the ori-
106 CHAPTER FIVE
entation of survey grids with respect to the expected strike direction of the target,
which is estimated from previous surveys, photogeological studies, or the results
of statistical modelling. As mentioned in Section 5.4.1, the optimal orientation
of parallel airborne grids is 90 degrees; for ground surveys, it ranges between
18 and 45 degrees, depending on the type of deposit.
The second step is the optimization of grid spacings for optimally oriented
grids. The optimization is carried out analytically for airborne surveys on square
grids as described in Section 4.3.3 of Chapter 4. In the case of parallel grids
for airborne surveys, the analytical approach runs into difficulties because of a
discontinuity of the efficiency function when the grid spacing equals the target
length. When considering ground surveys and drilling programs, the expression
of the efficiency function is very complex and its derivation intractable, ruling
out the analytical approach. The direct search approach has to be used in the
two latter cases. (See Section 4.3.4 of Chapter 4.)
·9
---. -.~
'"
uO. 825 ,/'
-....
~ .75
/
u
I
/
II>
~
.675
/
°
;::,
!!! .525
.6
/
£l.
• /
c:
.-0
.45
/
/
t; .375
c:
"
lJ..
/
.3
>-
u
....u-
c: .225
/
-
;;::
lLI
.15
.075 /
/
....
Lll
N
°....
II>
II>
N
~
°° II>
....
II>
Lll
°
II>
<.t>
II>
....
N
t~ .... '"°
II>
00
II>
Square grid spacings S by S feet
FIGURE 5.1.
Graph of EffiCiency Function for the Detection of Volcanogenic Massive Sulfide
Deposits of the North American Cordillera Belt by Optimally Angled (40 de-
grees) Drilling to a Vertical Depth of 300 Feet. The Optimal Grid Spacing is
Indicated by an Arrow.
The third step is the optimization of the drilling angle which is omitted for
subhorizontal deposits such as porphyry-Cu-Mo and Mississippi Valley Pb-Zn
types. The optimization is based on the direct search approach using the efficiency
criterion. This presupposes some knowledge of the likely direction of the dip,
as may be obtained from previous surveys or photogeological studies, to avoid
the risk of drilling down-dip. The graph shown in Figure 5.1 illustrates the three-
stage optimization of drilling programs for the detection of volcanogenic massive
sulfide deposits of the Cordillera.
5.5.4.2. Construction and Organization of Optimization Tables.

All tables are constructed on the basis of the statistical concept of 95% confidence
interval for the mean target length, which is translated into 95% fiducial intervals
for the parameters of optimized grids. The headings of tables provide additional
pertinent information such as type of grid and unit costs. The parameters of
optimized grids are (a) optimal grid spacing, corresponding coverage cost per
unit of area, and associated optimal detection probability, and (b) optimal ori-
entation of grid and optimal drilling angle when it applies. Altogether 30 such
tables have been constructed to summarize the results of the optimization of field
programs in the search for six types of deposits in North America. The appropriate
optimization tables are inserted at the end of sections of detection probability
tables for each type of deposit and each type of field program. For the reader's
convenience, a total of 11 detection summary tables have been constructed to
assemble the detection characteristics pertaining to the liminal, maximal, and
optimal options.
1. ANNIS, R. c., CRANSTONE, D. A., and VALLEE, M., 1978, A survey of known mineral
deposits in Canada that are not being mined, Mining Bulletin MR 181, Department of Energy,
Mines, and Resources, Ottawa.
2. BATEMAN, A., 1942, Economic Mineral Deposits, Chap. 23, Wiley, New York.
3. BROBST, D. A., and, PRATT, W. P. (Ed.), United States mineral resources, U.S. Geol. Surv.
Prof. Pap. No. 820.
4. CANADIAN MINES HANDBOOK, 1978-1979, The Northern Miner Press, Toronto.
5. CRANSTONE, D. A., and WHILLANS, R. T., 1979, Canadian reserves of copper, nickel,
lead, zinc, molybdenum, silver, and gold as of January 1st, 1978, Mining Bulletin MR 185,
Department of Energy, Mines, and Resources, Ottawa.
6. DERRY, D. R., 1980, World Atlas of Geology and Mineral Deposits, Mining Journal Books,
London.
7. DOUGLAS, R. J. W., (Ed.), 1970, Geology and economic minerals of Canada, Geol. Surv.
Can. Econ. Geol. Rep. No. I, Ottawa.
108 CHAPTER FIVE
8. GEOLOGICAL SERVICES, PROVINCE OF QUEBEC, 1967, Annotated bibliography on

metallic mineralization of the region of Noranda, Mattagami, Val d'Or and Chibougamau,
Special paper No.2, Department of Resources, Quebec City.
9. GILBERT, C. (Ed.), 1957, Structural geology of Canadian ore deposits, Congress Volume,
Canadian Institute of Mining and Minerals, Montreal.
10. GILL, I. E. (Ed.), 1972, Section 4, Mineral Deposits, 24th International Geology Congress,
Ottawa, pp. 107-160.
II. GRIFFITHS, J. c., and ONDRICK, C. W., 1968, Sampling a geological population, Kansas
Geological Survey Computer Contribution No 30.
12. LANG, A. H., 1970, Base metals, Canadian Shield, Geol. Surv. Can. Econ. Geol. Rep. No.
I, Ottawa.
13. LAZNICKA, P., 1972, The University of Manitoba File of World's non-ferrous metal deposits
(MANIFILE), University of Manitoba, Department of Earth Science, Winnipeg, Manitoba.
14. LINDGREN, W., 1933, Mineral Deposits, McGraw-Hill, New York.
15. MARTIN, H. L., CRANSTONE, D. A., and ZWARTENDYK, J., 1976, Metal mining in
Canada, 1976-2000, Mining Bulletin MR 167, Department of Energy, Mines, and Resources,
Ottawa.
16. NIGGLI, P., 1954, Rocks and Mineral Deposits, W. H. Freeman, San Francisco.
17. NOBLE, J. A., 1955, The classification of ore deposits, Econ. Geol. Jubilee Volume SO,
155-169.
18. PARK, C. F., and MACDIARMID, R. A., 1975, Ore Deposits, W. H. Freeman, San Francisco.
19. RIDGE, J. D. (Ed.), 1968, Ore deposits of the United States, 1935-1967, A.I.M.E., New
York.
20. RIDGE, J. D., 1972, Annotated bibliographies on mineral deposits in the Western Hemisphere,
Geol. Soc. Am. Mem. 131.
21. ROSS, C. S., 1935, Copper deposits of the Eastern United States; Copper Resources of the
World; 16th International Geological Congress Vol. I, pp. 158-162.
22. STOCKWELL, C. H., 1957, Geology and economic minerals of Canada, Geological Survey
of Canada, Econ. Geol. Ser. No. I.
23. THOMPSON, J. E., 1957, Metal Resources Circular No.2; Ontario Department of Mines,
Toronto.
24. WILSON, H. D. B., and LAZNICKA, P., 1972, Copper belts, lead belts and copper-lead lines
of the World, 24th International Geological Congress, Montreal.
25. WOLF, K. H. (Ed.), 1976, Handbook of Stratabound and Stratiform Ore Deposits, Elsevier,
Amsterdam.
CHAPTER SIX
DESIGNING OPTIMIZED FIELD

PROGRAMS FOR THE DETECTION OF
PORPHYRY-Cu-Mo DEPOSITS OF THE
NORTH AMERICAN CORDILLERA BELT
6.1. GENERAL GEOLOGICAL BACKGROUND
6.1.1. Main Geological Features

The term "porphyry--copper-molybdenum" deposit was first coined some
70 years ago to describe a type of base metal occurrence in the North American
Cordillera Belt. Since that time, geologists have applied the term to a broad
variety of disseminated sulfide deposits associated with differentiated plutons
that occur within well-defined orogenic belts. The deposits show many gradations
between the "porphyry--copper," "molybdenum-stockwork" and "contact met-
amorphic" types. The latter is described in more detail in Section 7.5 of Chapter
7.
Intrusive rocks hosting the porphyry--copper-molybdenum deposits vary in
composition from felsic to intermediate. Although the quartz monzonite type is
the most common, there are many variations extending from syenite to diorite.
The former class is known in the British Columbia section of the North American
Cordillera, while the latter is found outside North America, mainly in the South-
west Pacific Island Arcs. A conspicuous development of porphyritic phases and
breccia pipes is a common feature of plutons hosting copper-molybdenum
deposits.
Lowell and Guilbert(l8) describe the development of a roughly concentric
pattern of hydrothermal alteration zones as another outstanding feature of the
environment of porphyry-Cu-Mo deposits. Variations of the zoning pattern
reflect mineralogic, lithologic, and structural features of the pre-ore environment,
as well as the effect of post-ore geological events such as faulting and erosion. (10)
In some orogenic belts such as the North American Cordillera, the zoning pat-
tern is much better developed than in other regions and greatly assists field
exploration.
109
110 CHAPTER SIX
6.1.2. Geographic and Chronologic Distribution

Since their first recognition in North America, porphyry-Cu-Mo deposits
have been discovered along a limited number of orogenic belts of diverse ages.
This peculiar geographic distribution has been recently explained in the context
of global tectonics which emphasizes the role of zones of subduction along the
edges of tectonic plates as a leading control for the occurrence of the deposits. (23)
The model appears to explain rather well the distribution of deposits of the
northern section of the Cordillera Belt, which extends over a length of some
2200 miles and width of 300 to 400 miles through British Columbia, the Yukon,
and southern Alaska, along the structural zone separating the Pacific plate from
the American plate. But it fails to account satisfactorily for the distribution of
porphyry-Cu-Mo deposits within the U. S. section of the Cordillera Belt, which
are scattered over a width of more than 1000 miles through the States of Montana,
Utah, Colorado, New Mexico, and Arizona.
Recent measurements of absolute age by the potassium-argon dating method
have shown that the age span of the porphyry-Cu-Mo deposits of the world
extends over almost 400 million years, from the early Paleozoic times to late
Miocene. In the Cordillera Belt of the Americas, the South American deposits
are the more recent ones, with a modal age of 40-50 millions years. The North
American age distribution is bimodal, with an early peak in the Permo-Triassic
times, mainly in British Columbia, and a later peak in Cretaceous-Eocene times,
chiefly in the U.S. section of the Cordillera.
6.1.3. Metallogenic and Economic Features

Two main types of porphyry deposits are covered in the present chapter:
porphyry--copper and molybdenum stockwork. The first type may be subdivided
into two categories: copper-molybdenum and copper-gold. The latter is char-
acterized by unusually high ratios of precious metals/copper, generally greater
than 0.01 oz of gold and silver per percentage point of copper. A frequent feature
of the copper-gold category is an unusual abundance of magnetite, which assists
the geophysical detection of the deposits from the air and on the ground. (25)
Both categories of porphyry deposits commonly exhibit a crudely concentric
pattern of distribution of major and minor metals which is superimposed on the
hydrothermal zoning pattern already mentioned. As many as four, but generally
only three zones may be identified: they are termed as core, intermediate, and
outer zones. Generally copper, molybdenum, gold, and silver are associated in
the core zone, while zinc and lead appear in the intermediate zone. The outer
zone features an association of lead, zinc, silver, and manganese.
The overall tonnage of porphyry deposits being presently mined in the North
DETECTION OF PORPHYRY DEPOSITS 111
American Cordillera Belt varies from a few million tons of high-grade ore as in
some British Columbia deposits to several hundred million tons of low-grade
material as in many U.S. deposits. The tonnage of deposits is controlled by
physical and chemical factors. The former include (a) the volume of the intrusive,
(b) attitude and lithology of wallrock, and (c) spacing and size of cracks shattering
the host rock. The latter reflect the nature of the mineralizing solutions and that
of the metal precipitating environment. Post-ore geological events such as uplift-
ing and subsequent erosion may result in the removal of substantial tonnages of
ore, while downfaulting may preserve large tonnages and promote upgrading
through supergene alteration.
Current grades of porphyry-copper-molybdenum ore mined in the North
American Cordillera Belt vary from 0.5% to 1.5% copper, accompanied by
0.02% to 0.06% molybdenite, with gold generally under 0.05 ozlton and silver
under 0.1 oz/ton. Conversely, in molybdenum-stockwork deposits, the molyb-
denite grade ranges from 0.1 % to 0.3%, while the copper grade varies in the
0.15% to 0.25% only. Secondary enrichment of copper and gold in the upper
portions of deposits is of little importance in the northern portion of the Cordillera
Belt because of unfavorable climatic conditions. But in the southwestern U.S.
section of the Cordillera, the extent and intensity of supergene enrichment have
influenced production decisions in many cases, particularly in earlier days.
The increasing importance of porphyry-type base metal deposits as world
sources of copper, molybdenum, and valuable minor metals such as gold, silver,
selenium, tellurium, and rhenium justifies the prominent place given to the
porphyry-target in present-day exploration planning. The large size of the por-
phyry deposits, their shallow occurrence, favorable attitude and even metal
distribution make it possible to use large-scale, low-cost surface production
methods that are amenable to computerized mining and scheduling, thus off-
setting the low grade of the ore. However, porphyry projects are susceptible to
economic and financial risks because of their long preproduction time require-
ments, high capital costs, and low profit margins resulting in long payback
periods.
6.2. FiElD DETECTION METHODOLOGY
6.2.1. Detection Approaches

As indicated in Section 2.1 of Chapter 2, there are two approaches to the
detection of any type of mineral deposits: direct and indirect methodologies. The
first one is based on geological recognition either by visual means from the
112 CHAPTER SIX
ground (unaided) or from airborne platform (assisted), or by mechanical means

(drilling) from the surface. The second one relies on the recognition of the
geophysical signature of mineral deposits on the one hand, or on the geophysical
or geochemical detection of halos associated with the mineral deposits on the
other hand. Ideally, the two approaches are used together, either simultaneously
or sequentially, in order to boost their effectiveness.
6.2.2. Direct Geological Approach

During the past two decades, explorationists have developed an increasing
interest in the use of LANDSAT satellite imagery to assist the detection of
outcropping or near-outcropping mineral deposits. The combination of syn-ore
alteration of a pyrometasomatic or hydrothermal nature with post-ore weather-
ing may produce intense color anomalies. Multiple Spectral Sensing (MSS)
techniques can be used to detect the color anomalies from satellite or aircraft
platforms.
Abrams(l) describes the typical "red thumb" signatures of known por-
phyry-Cu-Mo deposits in the southwest Cordillera of the United States. Among
the most conspicuous are the Bisbee, San Manuel, and Pima-Mission deposits.
The color anomalies, reaching one square mile in extent, consist of brownish
and greenish discoloration accompanied by extensive iron and manganese stain-
ing. The intensity and extent of the anomalies depend on diverse factors, in-
cluding (a) lithology, (b) abundance of sulfides, (c) intensity of pyrometaso-
matism, and (d) intensity of weathering.
Unfortunately, the field of application of the remote sensing method is
severely restricted to regions with thin vegetation or alluvial cover, or both.
Success can be expected in regions of arid climate, whether subtropical or arctic.
The heavy vegetation and alluvial cover prevailing in tropical or temperate
regions preclude the effective use of the method. Detailed ground mapping of
lithologic and alteration features remains the most effective direct method in
most regions.
6.2.3. Indirect Geophysical Approach

The indirect geophysical approach to the detection of porphyry-Cu-Mo
deposits may be handled in two ways. The first one calls for the airborne or
ground detection of the geophysical signature of the deposits themselves, fol-
lowed by the drill testing of the anomaly. The second one relies on the detection
of associated halos to be followed by the systematic drilling investigation of the
halos in the search for the deposits. (See Chapter 2, Section 2.6.)
As many as one-half of the known porphyry-Cu-Mo deposits of the North

American Cordillera Belt exhibit well-developed pyritic halos containing 5% or
more pyrite by volume disseminated throughout the intrusive and wallrock.
Magnetite-rich halos commonly surround copper-bearing intrusives, particularly
the gold-rich type.(25) Both kinds of halos are good targets for airborne geo-
physical detection. A combination of electromagnetic technique in the time
domain (INPUT) and magnetrometric technique appears to be the most effective
for airborne halo detection.
The ground detection of porphyry-Cu-Mo deposits themselves by geo-
physical techniques is a rather difficult problem. The geophysical environment
is not very favorable because of the relatively low abundance of sulfides (5%-15%
by volume) which occur in a discrete manner within large volumes of intensely
fractured rocks. The standard techniques available at present are not always
suitable to attack the problem.
The standard electromagnetic techniques are at a severe disadvantage, except
for the detection of massive sulfide pods within intrusive aureolae. The self-
potential technique, so attractive because of its simplicity, rapidity, and low
coverage cost has proved rather disappointing owing to the great complexity of
the electrochemical phenomena prevailing in the porphyry environment. Mixed
success is the record of the gravimetric technique in the southwestern Cordillera
of the United States. Gravity lows are recorded over the zones of greatest al-
teration in the intrusive bodies, while gravity highs occur in the vicinity of
massive sulfide bodies within the intrusive aureolae.
In most circumstances, the winning ground geophysical approach requires
the combination of the fast and low-cost magnetometric technique with the slower
and costlier induced polarization method, which is noted for its ability to detect
disseminated sulfide material. A recently developed hybrid technique, known as
magnetic induced polarization (M.I.P.), is described by Seigel. (22) M.I.P. appears
to provide a capability of detection of both disseminated and massive sulfides,
while avoiding the cumbersome features of the induced polarization technique.
6.2.4. Indirect Geochemical Approach

A frequently used indirect approach relies on the detection of the geochem-
ical halos associated with porphyry-Cu-Mo deposits which are of two kinds.
One type, referred to as primary halos, is syn-ore and occurs as a three-dimen-
sional envelope in the bedrock surrounding the deposits. The second type, known
as secondary halos, occurs in the surficial weathered environment of deposits
and does not necessarily surround the deposits. Detection of porphyry-Cu-Mo
114 CHAPTER SIX
deposits through their primary halos has gained increasing attention in recent
times.
The success of the method relying on the detection of primary halos depends
on the presence of sufficient rock exposure or the availability of drill core to
supplement poor outcropping. The technique is keyed on the distribution of
minor metal associations which are used as proximal indicators. For example,
molybdenum-rhenium or copper-selenium associations denote the core zone of
the mineralized environment; the cadmium-gallium-germanium association with
zinc characterizes the outer zone of deposits.
Secondary halos are the product of the weathering and transport of material
from the mineral deposits themselves or from their primary halos. Sampling
media are residual topsoil, stream silt, plant matter, or ground water. Sillitoe(25)
reports on the merits of using gold as a tracer element in residual soils because
gold abundance in soils appears to reflect closely that of the hypogene gold in
the underlying protore.
6.3. STATISTICAL MODELING OF THE GEOMETRIC PARAMETERS

OF PORPHYRY-Cu-Mo DEPOSITS AND THEIR ASSOCIATED
PYRITIC HALOS
6.3.1. Scope of Study

A sample of 57 commercial and subeconomic porphyry-Cu-Mo and mo-
lybdenum-stockwork deposits was assembled to represent statistically the whole
population of known and undiscovered porphyry-type deposits of the North
American Cordillera Belt. The deposits are listed by names and geographic
regions in Table 6.1. The geographic breakdown is as follows: 25 deposits of
the sample occur in the U.S. section of the Cordillera Belt and 32 were found
in the British Columbia-Yukon section of the belt. Pyritic halo data of a quality
adequate for statistical processing were available for 43 of the 57 deposits in-
cluded in the data base.
The three geometric parameters required for the calculation of detection
probabilities and the design of optimized field programs were measured for each
of the 57 deposits. They consist of the length, breadth, and shape ratio
(breadth/length) of the horizontal section of the portion of deposits which lie
within the 300-foot maximum range of the commonly used types of field detec-
tors. The parameter measurements were grouped into frequency distributions,
to which statistical models were fitted.
[ DETECTION OF PORPHYRY DEPOSITS 115
TABLE 6.1
List of Porphyry-Cu-Mo Deposits of the North American Cordillera Included
in the Database.
British Columbia (Canada) united States
Red Mountain Gibraltar Castle Dome Questa

Boss Mountain Brenda Copper City Butte
Sileurian Chieft. Hudson Bay Mtn. Mineral Park San Manuel
Alwin Schaft Creek Esperanza Morenci
Granisle Lornex Yerrington Twin Butte
Cariboo Bell Endako Glacier Peak Urad-Henderson
Polyanna Gem Silver Bell Santa Rita
Newman Island Island Copper Tyrone Climax
Ingerbelle Valley Copper Bagdad Bingham
Similkameen Casino Safford
B.C. Molybdenum Serb Creek Mission
Huckleberry Salal Miami-Inspir.
Stikine Galaxy Bisbee
Adanac Sheba Ely
Bethlehem Trojan Ray
Highmont Lytton Ajo
6.3.2. Statistical Modeling

The frequency distributions of parameter measurements are summarized by
means of four statistics which are listed in the lower half of Table 6.2 for the
deposits themselves and of Table 6.3 for the associated pyritic halos. The statistics
are as follows: the arithmetic mean of the observed data, mode, standard de-
viation, and coefficient of skewness. Their makeup and significance have been
described previously in Section 5.3.2 of Chapter 5.
The asymmetry of the distributions of deposit parameters is indicated by
the magnitude of the skewness coefficient. The skewness is rather slight for the
length and shape ratio (less than 0.01), but is rather strong for the breadth (greater
than unity). The skewness of the distributions of all three parameters for the
pyritic halos is moderate (0.1-1.0 range).
TABLE 6.2
Summary of Statistical Modeling of Geometric Parameters of Porphyry-Cu-Mo
Deposits of the North American Cordillera Belt (Sample Size: 57 Deposits).
Statistic of Length of horiz. Breadth of horiz. Shape ratio

fitted model section in feet section in feet R = BIL
25%ile 2800 1140 0.35

95%L.c.l. 2820 1550 0.38
G. mean 3110 1640 0.46
95%U.c.l. 3830 1950 0.55
75%ile 6000 5300 0.78
Disp. coef. 0.06 0.05 0.94
Statistic of
observed data
Arith. mean 3510 3240 0.52

Mode 3500 2000 0.50
Stand. dev. 320 260 0.25
Skewness 0.03 4.8 0.08
TABLE 6.3
Summary of Statistical Modeling of Geometric Parameters of Pyritic Halos As-
sociated with Porphyry-Cu-Mo Deposits of the North American Cordillera Belt
(Sample Size: 43 of the 57 Deposits Included in Table 6.1).
Statistic of Length of horiz. Breadth of horiz. Shape ratio

25%ile 4591 2449 0.50

95%L.c .l. 4870 2920 0.54
G. mean 6374 3829 0.62
95% U.c.l. 8344 5022 0.71
75%ile 10,564 5000 0.77
Disp. coef. 0.39 0.51 0.21
Statistic of
observed data
Arithm. mean 7660 44~6 0.65

Mode 5450 4000 0.60
Stand. dev. 4355 2500 0.19
Skewness 0.51 0.17 0.26
Lognormal models were successfully fitted to the observed distributions of

all three parameters for the deposits and their halos, as indicated by the results
of X2 goodness of fit tests run at the 0.05 confidence level (see Chapter 5, Section
5.3.2). The statistics describing the fitted models are listed in the upper half of
Table 6.2 for the deposits and Table 6.3 for the halos. They include (a) geometric
mean, (b) the upper and lower limits of the 95% confidence interval of the mean,
(c) the 25th and the 75th percentiles, and (d) the coefficient of dispersion. The
latter indicates the degree of variability of the data about the mean of the fitted
data in a dimensionless manner. A small variability (less than 0.1) affects the
length and breadth parameters of the deposits, but a larger variability, close to
1.0, is indicated for the shape ratio. The variability of the three parameters is
moderate (0.1-1. 0 range) for the pyritic halos.
6.4. CONSTRUCTION AND ORGANIZATION OF DETECTION

PROBABILITY TABLES
6.4.1. Introduction
For detection purposes, porphyry-Cu-Mo deposits are described as either
tabular subhorizontal or pipe like subvertical bodies with horizontal cross sections
of elliptical shape. The calculations of detection probabilities therefore apply to
plane elliptical targets whose dimensions are those of the horizontal section of
the deposits without the addition of dip components (see Section 2.2.3 of Chapter
2). Because of the lack of accurate orientation data, both porphyry deposits and
their associated halos are considered as randomly oriented plane targets.
6.4.2. Description of Tables

There are three kinds of tables for both porphyry deposits and their associated
halos. The first two cover the cases of single detection (at least one target
intersection by the detector) and confirmed detection (at least two intersections)
and are constructed in a similar manner. The third kind of table displays grid
spacings and corresponding coverage costs per unit of area for specified levels
of detection probability.
The heads of the first two types of tables display the statistics of the fitted
models as obtained from Tables 6.2 and 6.3, including the mean length and its
95% confidence limits and the mean shape ratio. The probabilities of detection
or confirmed detection are calculated for grid spacings increased in steps of
50-250 feet, as required by the type of field program under consideration, first
for the mean length, and then for the upper and lower limits of the 95% confidence
118 CHAPTER SIX
interval of the mean. As a result, the tables display the probability of detection
or confirmed detection within its 95% confidence interval for each grid spacing.
6.4.3. Organization of Tables

Since most readers will have to consider only one specific type of field
program at one time, the tables have been grouped into blocks, each covering
one kind of program. The first block covers airborne geophysical surveys for
the detection of porphyry deposits based on parallel and square grids (Tables
6.6-6.9). The second block deals with the detection of halos by airborne geo-
physical surveys also on parallel and square grids (Tables 6.11-6.14). The third
block relates to ground geophysical surveys and vertical drilling programs to a
depth of 300 feet (Tables 6.16-6.18). Ground detection tables for the pyritic
halos were not considered because the halos are commonly detected from the
air, following which the search for the deposits themselves is carried out on the
ground within the halos. (See section 6.2.3 of this chapter.)
6.5. DESIGNING THREE STRATEGIES FOR THE DETECTION OF

PORPHYRY-Cu-Mo DEPOSITS IN THE NORTH AMERICAN
CORDILLERA BELT
6.5.1. Search Strategies

Three kinds of single-stage strategies aimed at the detection of porphyry
deposits and their associated halos are described below. They are referred to as
the liminal, maximal, and optimal options, which are designed to deal with three
different kinds of exploration situations. The detection characteristics corre-
sponding to the three options are listed in Table 6.4. for the porphyry deposits
and in Table 6.5. for the pyritic halos. The listed parameters include probability
of detection, corresponding grid spacing, and associated coverage cost per unit
of area.
A two-stage variation on the optimal option is introduced to cover the
sequential procedure described previously in Section 6.2.3. The first stage con-
sists of the optimal detection of pyritic halos. The second one calls for the
optimal detection of the deposit within the halos by means of systematic vertical
drilling programs. The sequential optimal option is based on the dynamic pro-
gramming approach described in Section 4.3.5. of Chapter 4.
o
~
m
TABLE 6.4
o
oz
Summary of Detection Characteristics of Porphyry-Cu-Mo Deposits of the North
American Cordillera Belt for Three Types of Detection Strategies. o
"TI
~
Liminal detection Optimal detection Maximal detection o
;C
Coverage cost: Coverage cost: ~
US$/mile sq. US$/ mile sq.

:::c
-<
Type of survey Grid size: feet Coverage cost Detection Grid size: feet
and Confirmed US$/Illile sq. proba- Confirmed
~
grid geometry Detection detection Grid size: ft bility Detection detection om
Airborne a
VI
geophysical $235 $375 $268 0.79 $560 $585 =i
parallel 3900 ft 1600 ft 2900 ft 930 ft 850 ft VI
grid
Airborne $258 - $328 0.86 $412 -

geophysical 6750 ft 3960 ft 2750 ft
square grid
Ground $540 $858 $595 0.80 $898 $1040

geophysical 1900 ft 1225 ft 1700 ft llOO ft 960 ft
square grid
Vertical $86,000 $205,000 $102,000 0.80 $205,000 $262,500

drilling 1900 ft 1225 ft 1700 ft llOO ft 960 ft
square grid
~
~
I..C
~
N
o
TABLE 6.5
Summary of Detection Characteristics of Pyritic Halos Associated with Porphyry-Cu-Mo
Deposits of the North American Cordillera Belt for Three Types of Detection Strategies.
Liminal detection Optimal detection Maximal detection

I
Coverage cost: Coverage cost:

US$/mile sq. US$/mile sq.
and Confirmed uS$/mile sq •. proba- Con;:irmed
grid geometry Detection detection Grid size: ft bility Detection detection
Airborne $185 $220 $200 0 0 86 $275 $371

geophysical 8000 ft 4600 ft 6120 ft 2700 ft 1600 ft
parallel grid
Airborne $152 - $242 0.93 $295 -

geophysical 12,000 ft 7230 ft 4800 ft
--
n
:::r
»
."
...,
m
;;>:l
Vl
X
6.5.2. Liminal Option

The liminal option is based on a grid design which provides a single or
confirmed detection probability of at least 0.500, as described in Section 5.5.2
of Chapter 5. The option is attractive only when explorationists face the problem
of ensuring a reasonably effective coverage of large areas of little known potential
within strict budgetary constraints. The detection parameters pertaining to the
liminal option are listed in the leftmost portions of Table 6.4 for the porphyry
deposits and Table 6.5 for the halos. A comparison of the two tables immediately
shows that the coverage costs per unit of area required to ensure liminal detection
are substantially lower for the halos than for the porphyry deposits, which offer
much smaller targets for detection.
The cost multiplying factor required to ensure confirmation of detection
instead of single detection is about 1.6 for the detection of porphyry-Cu-Mo
deposits by airborne and ground geophysical surveys. However, for the asso-
ciated pyritic halos, the cost required to obtain confirmation is only 1.2 times
that of single detection; this emphasizes the benefit of initially requiring confirmed
detection for halos rather than single detection. The cost mUltiplying factor
escalates to 2.4 when confirmed detection of porphyry deposits by vertical drilling
is considered.
Two kinds of airborne survey control grids are considered in the present
study, the parallel design and the square design. A comparison of the merits of
the two designs is a matter of interest for exploration planning. The coverage
cost per unit of area provides a simple yardstick for the comparison. An inspection
of the leftmost portions of Tables 6.4 and 6.5 shows that the selection of the
square grid design is justified only for the search for halos because it leads to a
reduction of 18% of the coverage cost. On the contrary, the selection of the
square grid design is not recommended for porphyry deposit detection, because
it would result in a 10% coverage cost increase.
6.5.3. Maximal Option

If, instead of minimizing coverage cost in return for a specified detection
probability level (0.50), we wish to maximize the probability of detection to the
near certainty level, say 0.980, we will then select the maximal option. Explo-
rationists will find the maximal option quite attractive when considering the
coverage of small areas of high potential in well prospected regions. The detection
characteristics pertaining to the maximal option are listed in the rightmost portions
of Tables 6.4 and 6.5 covering single and confirmed detection of porphyry
deposits and their associated halos.
122 CHAPTER SIX
The cost multiplier required to obtain confirmation of detection of porphyry

deposits by airborne and ground geophysical surveys is only 1.1 instead of 1.6
for the liminal option. In the case of halo detection, however, the cost multipliers
are 1.4 for the maximal option against 1.2 for the liminal strategy. As a result,
explorationists choosing the maximal option should be planning for the confirmed
detection of porphyry deposits, while being satisfied with the single detection
of the pyritic halos. When planning drilling programs for porphyry deposits, the
confirmed detection approach of the maximal option is the most attractive one,
because the cost multiplier is only 1.3 against 2.4 for the liminal option.
When faced with the choice between parallel and square grid designs for
the airborne geophysical detection of porphyry deposits and associated halos,
the coverage cost per unit of area provides a convenient criterion. One may see
from Table 6.4 that the selection of the square grid design leads to a coverage
cost saving of 26% for porphyry deposit detection under the maximal option.
On the other hand, it can be seen from Table 6.5 that the square grid design is
not recommended for halo detection because the corresponding coverage cost is
10% higher than that of the parallel design.
6.5.4. Optimal Option

The optimal option is based on a compromise between the low cost of the
liminal option which guarantees at least a 0.500 detection probability and the
high cost maximal option which provides nearly certain detection. The optimal
option should prove most attractive in the majority of exploration situations
which call for the coverage of moderately large areas of good potential within
strict budgetary constraints.
The optimization of the field program designs was carried out by a com-
puterized direct search method based on the efficiency criterion, as described in
Section 4.4 of Chapter 4. Optimization tables are constructed for each of the
three main types of field programs, namely, airborne geophysical surveys, ground
geophysical surveys, and vertical drilling, in a manner described in Section 5.5.3.
of Chapter 5. All tables provide the optimized grid parameters, including prob-
ability of optimal detection, optimal grid spacing, and corresponding coverage
cost per unit of area, within 95% confidence limits which are based on the
fiducial limits of the mean target length.
Tables 6.10. and 6.15 display the parameters of optimized grids for the
detection of porphyry deposits and associated halos by airborne geophysical
surveys on parallel and square grid designs. Optimization results for the detection
of porphyry deposits by ground geophysical surveys and vertical drilling are
listed in Tables 6.19a and 6.19b, respectively. No ground optimization results
are provided for the halos because ground detection is not considered for them.
The detection characteristics pertaining to the optimal option are culled from
the tables listed above and assembled into the central portions of Tables 6.4
(porphyry deposits) and 6.5 (halos). An inspection of both tables shows that
detection parameters corresponding to the optimal option are bracketed by those
associated with the liminal and maximal options, as could be expected. Table
6.4 shows that the optimal probabilities of detection of porphyry deposits by all
types of geophysical surveys lie within a range of 0.79-0.86 with corresponding
coverage cost per unit of area only marginally higher than of the liminal option.
The optimal probability of detection by drilling programs (0.80) compares fa-
vorably with that of the maximal option, while the corresponding coverage cost
is only one half of that of the maximal cost. The optimal probabilities of detection
of halos by airborne surveys are close to that provided by the maximal option,
while coverage costs are substantially lower (Table 6.5).
If we wish to compare the detection performances of parallel and square
grids for airborne geophysical surveys searching for porphyry deposits or halos
under the optimal option, we cannot rely on the coverage cost per unit of area
as a criterion, because the probability of detection is not held at a specified level.
The "expected loss" criterion combining coverage cost and detection probability
can be used to advantage. It is written as a product as follows:
Expected Loss = (Coverage cost) x (Probability of failure)
The latter term is the complement of the optimal probability of detection written
as (1 - Pd ). Obviously, the choice of design will be based on the minimization
of the expected loss due to the failure of detection. We find from Table 6.4 that
the expected loss per unit of area is $46 for the optimal square grid design against
$56 for the optimal parallel design, which should be rejected. Similarly, the
square grid design should be our choice for the optimal detection of pyritic halos
by airborne surveys because the expected loss is only $17 as compared with $28
for the second design.
6.5.5. Sequential Optimal Strategy

After describing three single-stage strategies referred to as liminal, maximal,
and optimal options, we should consider a two-stage optimal approach which
illustrates the use of the dynamic programming method covered in Chapter 4.
The purpose of the first stage of the sequence is to detect and outline pyritic
halos associated with porphyry deposits by optimized airborne geophysical sur-
veys, because they are much larger and more easily detectable targets than the
deposits themselves. The following step is the optimized detection of the deposits
by systematic drilling within the halos.
CHAPTER SIX
The theory of geometric probabilities provides that the probability of a

randomly chosen sample point falling on a target of area at which is entirely
included within a larger area A is equal to the ratio a/ A. Taking the porphyry
deposits of the North American Cordillera Belt and their pyritic halos as an
example, we find that the probability mentioned above equals only 0.15, as
derived from the ratio of the mean deposit area (2.85 million square feet) over
the mean halo area (18.6 million square feet).
As seen in Section 1.1.3 of Chapter 1, the probability of a compound event
is the product of the component probabilities, provided that the corresponding
events are statistically independent. Based on the results listed in the central
portions of Tables 6.4 and 6.5, we find that the compound optimal probability
of detection of porphyry deposits of the Cordillera Belt by optimized airborne
geophysical surveys on parallel grids followed by optimized drilling is equal to
0.86 x 0.80 = 0.69. The corresponding result would be equal to
0.93 x 0.80 = 0.74, if we choose the optimal square grid design instead of
the parallel one at the initial stage of the program. These results are equal to
nearly five times the probability of detecting a prophyry-Cu-Mo deposit by
random sampling within its halo.
6.5.6. Detection Probability and Optimization Tables

The reader will find in the following section a total of ten tables that display
the probabilities of detection and optimal designs in the search for por-
phyry-Cu-Mo deposits of the Cordillera Belt and their associated pyritic halos
of airborne geophysical surveys (pages 125-134). A further five tables display
the probabilities of detection and optimal designs pertaining to the search for
porphyry-Cu-Mo deposits by ground geophysical surveys and by vertical drilling
(pages 135-138).
TABLE 6.6
Probabilities of Detection of Porphyry-Cu-Mo Deposits of the North American
Cordillera Belt by Airborne Geophysical Surveys (Continuous Readings) on
Parallel Grids.
Survey design: parallel lines with spacing S feet,

Randomly orientated elliptical targets with expected major axis L feet
in the confidence interval: l.c.l. ( geom. mean ( u.c.l.
and minor axis = B feet
R is defined as the ratio B/L with geometric mean = 0.46
Probabili ty of detection
Grid spacing l.c.l. geom. mean u.c.l.
S feet L=2820 3110 3840 feet
800 1. 000 1. 000 1. 000

1000 0.920 1. 000 1. 000
1200 0.903 0.922 1. 000
1400 0.886 0.908 1. 000
1600 0.869 0.895 0.916
1800 0.851 0.881 0.905
2000 0.832 0.866 0.894
2200 0.811 0.851 0.883
2400 0.789 0.836 0.871
2600 0.752 0.820 0.859
2800 0.699 0.803 0.847
3000 0.652 0.784 0.835
3200 0.611 0.751 0.822
3400 0.575 0.707 0.808
3600 0.543 0.668 0.794
3800 0.515 0.632 0.778
4000 0.489 0.601 0.747
4200 0.466 0.572 0.711
4400 0.445 0.546 0.679
4600 0.425 0.522 0.649
4800 0.408 0.501 0.622
5000 0.391 0.481 0.597
5200 0.376 0.462 0.574
5400 0.362 0.445 0.553
5600 0.349 0.429 0.533
5800 0.337 0.414 0.515
6000 0.326 0.401 0.498
126 CHAPTER SIX
TABLE 6.7
Determination of Grid Size for Specified Probability levels in the Detection of
Porphyry-Cu-Mo Deposits of the North American Cordillera Belt by Airborne
Geophysical Surveys on Parallel Grids.
Survey design: parallel lines with spacing S feet

Unit cost =US$70 / line mile
Specified detection Required grid spacing Corresponding

probability level S in feet cost in $ per mile square
0.10 20100 $ 158.39

0.15 14200 $ 166.03
0.20 11000 $ 173.60
0.25 9000 $ 181.07
0.30 7600 $ 188.63
0.35 6500 $ 196.86
0.40 5800 $ 203.72
0.45 5200 $ 211.08
0.50 4700 $ 218.64
0.55 4300 $ 225.95
0.60 3900 $ 234.77
0.65 3600 $ 242.67
0.70 3400 $ 248.71
0.75 3200 $ 255.50
0.80 2600 $ 282.15
0.85 2000 $ 324.80
0.90 1300 $ 424.31
0.95 500 $ 879.20
TABLE 6.8
Probabilities of Confirmed Detection of Porphyry-Cu-Mo Deposits of the North
American Cordillera Belt by Airborne Geophysical Surveys (Continuous Read-
ings) on Parallel Grids.

in the confidence interval: 1.c.1. ( geom. mean ( u.c.1.
Probabili ty of confirmed ratio

S feet L=2820 3110 3840 feet
800 1. 000 1. 000 1.000

1000 0.838 1. 000 1.000
1200 0.685 0.860 1.000
1400 0.581 0.725 0.923
1600 0.507 0.627 0.795
1800 0.455 0.554 0.698
2000 0.417 0.499 0.623
2200 0.391 0.457 0.564
2400 0.375 0.425 0.516
2600 0.000 0.401 0.478
2800 0.000 0.384 0.447
3000 0.000 0.373 0.423
3200 0.000 0.000 0.404
3400 0.000 0.000 0.389
3600 0.000 0.000 0.378
3800 0.000 0.000 0.371
4000 0.000 0.000 0.000
128 CHAPTER SIX
TABLE 6.9
Probabilities of Detection of Porphyry-Cu-Mo Deposits of the North American
Cordillera Belt by Airborne Geophysical Surveys (Continuous Readings) on
Square Grids.
Survey design: square grid with spacings S by S feet,

Grid spacing 1.c.1. geom. mean u.c.l.
S feet L=2820 3110 3840 feet
2500 1. 000 1. 000 1.000

3000 0.896 1. 000 1. 000
3500 0.818 0.921 1.000
4000 0.749 0.855 0.958
4500 0.688 0.794 0.905
5000 0.636 0.740 0.852
5500 0.590 0.691 0.803
6000 0.550 0.647 0.758
6500 0.515 0.608 0.716
7000 0.484 0.574 0.679
7500 0.456 0.542 0.644
8000 0.432 0.514 0.613
8500 0.409 0.489 0.584
9000 0.389 0.466 0.558
9500 0.371 0.445 0.534
10000 0.355 0.425 0.512
TABLE 6.10
Optimal Design of Airborne Geophysical Surveys (Continuous Readings) for
the Detection of Porphyry-Cu-Mo Deposits of the North American Cordillera
Belt.
Confidence interval: all results are reported as 95% confidence intervals

respectively
Expected target length in feet: 2820 3110 3830
Expected shape ratio R : 0.46
Unit cost: $70/l.ml.
(il Grid Design:

Parallel Grid
Optimal grid spacing in feet: 2400 2900 3600
Cost in US$ per mile square: 294 267 243
Probability of detection: 0.79 0.79 0.79
(iil Grid Design:

Square Grid
130 CHAPTER SIX
TABLE 6.11
Probabilities of Detection of Pyritic Halos Associated with Porphyry-Cu-Mo
Deposits of the North American Cordillera Belt by Airborne Geophysical Sur-
veys (Continuous Readings) on Parallel Grids.

in the confidence interval: I.c.1. ( geom. mean ( u.c.1.
R is defined as the ratio B/L with geometric mean =0.62
S feet L=4870 6374 8344 feet
2500 1.000 1.000 1.000

3000 0.920 1.000 1.000
3500 0.905 1.000 1.000
4000 0.889 0.918 1.000
4500 0.871 0.906 1.000
5000 0.810 0.894 0.921
5500 0.737 0.881 0.913
6000 0.675 0.866 0.904
6500 0.623 0.812 0.894
7000 0.579 0.754 0.884
7500 0.540 0.704 0.874
8000 0.506 0.660 0.859
8500 0.477 0.621 0.808
9000 0.450 0.586 0.764
9500 0.427 0.555 0.723
10000 0.405 0.528 0.687
TABLE 6.12
Determination of Grid Size for Specified Levels of Probability in the Detection
of Pyritic Halos Associated with Porphyry-Cu-Mo Deposits of the North Amer-
ican Cordillera Belt by Airborne Geophysical Surveys on Parallel Grids.

Unit cost = US$70 / line mile

0.10 45800 $ 148.07

0.15 32400 $ 151.41
0.20 25000 $ 154.78
0.25 20400 $ 158.12
0.30 17200 $ 161.49
0.35 14900 $ 164.81
0.40 13100 $ 168.21
0.45 11700 $ 171.59
0.50 10600 $ 174.87
0.55 9700 $ 178.10
0.60 8900 $ 181.53
0.65 8200 $ 185.07
0.70 7700 $ 188.00
0.75 7200 $ 191.33
0.80 6700 $ 195.16
0.85 6200 $ 199.61
0.90 4100 $ 230.15
0.95 1600 $ 371. 00
132 CHAPTER SIX
TABLE 6.13
Probabilities of Confirmed Detection of Pyritic Halos Associated with Por-
phyry-Cu-Mo Deposits of the North American Cordillera Belt by Airborne
Geophysical Surveys (Continuous Readings) on Parallel Grids.

in the confidence interval: I.c.1. ( geom. mean ( u.c.1.
Probability of confirmed detection

Grid spacing I.c.1. geom. mean u.c.l.
S feet L=4870 6374 8344 feet
1500 1. 000 1. 000 1. 000

2000 0.861 1. 000 1. 000
2500 0.761 0.883 1. 000
3000 0.699 0.794 0.929
3500 0.660 0.734 0.845
4000 0.636 0.693 0.784
4500 0.622 0.663 0.739
5000 0.000 0.642 0.704
5500 0.000 0.628 0.678
6000 0.000 0.621 0.658
6500 0.000 0.000 0.642
7000 0.000 0.000 0.631
7500 0.000 0.000 0.624
8000 0.000 0.000 0.000
TABLE 6.14
Probabilities of Detection of Pyritic Halos Associated with Porphyry-Cu-Mo
veys (Continuous Readings) on Square Grids.

S feet L=4870 6374 8344 feet
4500 1.000 1. 000 1. 000

5000 0.963 1.000 1.000
5500 0.927 1. 000 1.000
6000 0.890 1. 000 1. 000
6500 0.852 0.963 1. 000
7000 0.816 0.936 1.000
7500 0.782 0.908 1. 000
8000 0.749 0.879 0.980
8500 0.719 0.850 0.962
9000 0.691 0.823 0.941
9500 0.664 0.796 0.920
10000 0.639 0.770 0.898
134 CHAPTER SIX
TABLE 6.15
the Detection of Pyritic Halos Associated with Porphyry-Cu-Mo Deposits of
the North American Cordillera Belt.

respectively
Expected shape ratio R = 0.62
Unit Cost: $70jl.ml.
(i) Grid Design:

Parallel Grid
(ii) Grid Design:

Square Grid
TABLE 6.16
Probabilities of Detection of Porphyry-Cu-Mo Deposits of the
North American Cordillera Belt by Ground Geophysical Surveys
(Discrete Readings) or Vertical Drilling to a Depth of 300 Feet.

The expected shape of the ore deposits is elliptical
L.c.1. ( geometric mean ( U.c.l., and minor axis = B feet
in feet L = 2820 3110 3820ft
800 1. 000 1.000 1.000

1000 0.986 1. 000 1. 000
1200 0.914 0.991 1. 000
1400 0.788 0.942 0.999
1600 0.660 0.852 0.973
1800 0.547 0.745 0.925
2000 0.455 0.643 0.847
2200 0.383 0.552 0.761
2400 0.324 0.475 0.677
2600 0.277 0.411 0.600
2800 0.239 0.358 0.530
3000 0.208 0.313 0.470
3200 0.183 0.276 0.419
4000 0.117 0.176 0.273
4800 0.081 0.123 0.189
5600 0.060 0.090 0.139
6000 0.052 0.078 0.121
136 CHAPTER SIX
TABLE 6.17
Determination of Grid Size for Specified Probability Levels in the Detec-
tion of Porphyry-Cu-Mo Deposits of the North American Cordillera Belt
by Vertical Drilling to a Depth of 300 Feet.
Survey Design: Square Grid With Spacings S by S Feet


0.05 6900 18.91

0.15 4300 30.05
0.25 3300 40.87
0.35 2900 48.08
0.45 2500 58.48
0.55 2300 65.56
0.65 2000 79.93
0.75 1800 93.30
0.85 1600 111.46
0.95 l400 137.17
TABLE 6.18
Probabilities of Confirmed Detection of Porphyry-Cu-Mo De-
posits of the North American Cordillera Belt by Ground Geo-

Randomly orientated elliptical ore bodies
with major axis = L feet in the 95% confidence interval:
1.c.l ( geometric mean ( u.c.l., minor axis = B feet
R is defined as ratio B/L with geometric mean = 0.46
Probabili ty of confirmed detection

Spacing L.c.1. G. mean U.c.1.
S feet L = 2820 3110 3820ft
900 0.951 1. 000 1. 000

1000 0.926 0.995 1.000
1100 0.793 0.982 1.000
1200 0.383 0.943 0.998
1300 0.253 0.873 0.995
1400 0.166 0.570 0.989
1500 0.109 0.352 0.938
1600 0.071 0.251 0.884
1700 0.046 0.178 0.739
1800 0.030 0.126 0.418
1900 0.019 0.089 0.319
2000 0.012 0.063 0.243
2200 0.004 0.031 0.140
2400 0.000 0.015 0.080
2600 0.000 0.006 0.045
2800 0.000 0.002 0.026
3000 0.000 0.000 0.014
138 CHAPTER SIX
TABLE 6.19(a)
Optimal Design of Ground Geophysical Surveys on Square Grids for the De-
tection of Porphyry-Cu-Mo Deposits of the North American Cordillera Belt.
Survey Design: Square Grid with Spacing S by S Feet.

Confidence Interval: All results are reported as 95% confidence intervals
respectively.
Expected Target Length in Feet: 2820 3110 3830
Expected Shape Ratio R = 0.46
Unit Cost: $6/Station
Optimal Grid Spacing in Feet: 1350 1700 2100

Cost in $ Per Mile Square 734 594 496
Probability of Detection 0.82 0.130 0.88
TABLE 6.19(b)
Optimal Design for the Detection of Porphyry-Cu-Mo Deposits of the North
American Cordillera Belt by Vertical Drilling to a Depth of 300 Feet.
Survey design: square grid with spacing S by S feet

respectively
(i) Percussion drilling:

unit Cost = $2000/hole
Cost in $thou. per mile square: 48 34 25
(ii) Diamond drilling:

unit Cost = $6000/hole
Probability of detection: 0.82 0.80 O.BB
1. ABRAMS, M. J., BROWN, D., and LEPLEY, L., 1983, Remote sensing for porphyry-copper
deposits in Southern Arizona, Econ. Geol. 78, 591-604.
2. CLARK, K. F., 1972, Stockwork molybdenum deposits in the Western Cordillera, Econ. Geol.
67,731-758.
3. CREASEY, S. c., 1959, Some relations in the hydrothermally altered rocks of porphyry -copper
deposits, Econ. Geol. 54, 354-373.
4. DE GEOFFROY, J., and WIGNALL, T. K., 1972, A statistical study of the geological char-
acteristics of the porphyry-copper-molybdenum deposits of North and South America, Econ.
Geol. 67, 656-668.
5. DE GEOFFROY, J. and WIGNALL, T. K., 1973, Statistical models for porphyry-
copper-molybdenum deposits of the Cordilleran Belt of North and South America, Can. Inst.
Min. Metall. Bull. 66(735), 84-90.
6. ENERGY, MINES, AND RESOURCES DEPARTMENT OF CANADA, 1981, Canadian min-
eral deposits not being mined in 1980, Report M.R. I. 80-7.
7. FOUNTAIN, D. K., 1968, Geophysics applied to the exploration and development of Cu and
Mo deposits of British Columbia, Can. Inst. Min. Metall. Bull. 61(676), 1199-1206.
8. FOUNTAIN, D. K., 1972, Geophysical case histories of disseminated sulfide deposits in British
Columbia, Geophysics 37, 142-159.
9. GODWIN, C. I., 1976, Casino porphyry-Cu-Mo deposit; Porphyry deposits of the Canadian
Cordillera, Can. Inst. Min. Metall. Special Volume 15, 344-354.
10. GUILBERT, J. M., and LOWELL, J. D., 1974, Variations in zoning patterns in porphyry ore
deposits, Can. Inst. Min. Metall. Bull. 67, 99-109.
11. HOLLISTER, V. F., 1973, Characteristics of porphyry copper deposits of South America, Min.
Eng. 25, 51-56.
12. HOLLISTER, V. F., POTTER, R. R., and BARKER, A. L., 1974, Porphyry-type deposits of
the Appalachian Orogen, Econ. Geol. 69, 618-630.
13. HOLLISTER, V. F., ANZALONE, S. A., and PRICHTER, D. H., 1975, Porphyry-copper
deposits of Southern Alaska and contiguous Yukon Territories, Can. Inst. Min. Metall. Bull.
68(755), 104-111.
14. HOLLISTER, V. F., 1978, Geology of the Porphyry-Copper Deposits of the Western Hemi-
sphere, A.I.M.E., New York.
15. JAMES, A. H., 1971, Hypothetical diagrams of several porphyry-copper deposits, Econ. Geol.
66,43--47.
16. KESLER, S. E., 1973, Copper, molybdenum and gold abundance in porphyry-copper deposits,
Econ. Geol. 68, 106--112.
17. LIVINGSTON, D. E., 1973, A plate tectonic hypothesis for the genesis of porphyry-copper
deposits of the Southern Basin and Range Province, U.S.A., Earth Planet. Sci. Lett. 20,
171-176.
18. LOWELL, J. D., and GUILBERT, J. M., 1970, Lateral and vertical alteration-mineralization
zoning in porphyry deposits, Econ. Geol. 65, 373--408.
19. ROSE, A. W., 1970, Zonal relations of wallrock alteration and sulfide distribution at the
porphyry-copper deposits, Econ. Geol. 65,910-936.
20. ROWE, R. B., 1973, Porphyry deposits of the Canadian Cordillera, Can. Min. J. 94(11),
35-38; 94(12), 37--41.
21. SAWYER, J. P. B., and DICKINSON, R. A., 1976, Mount Nansen porphyry deposit; Porphyry
deposits of the Canadian Cordillera, Can. Inst. Min. Metall. Special Volume 15, 336--343.
140 CHAPTER SIX
22. SEIGEL, H. 0., 1974, The magnetic induced polarization (M.I.P.) method, Geophysics 39,
321-339.
23. SILLITOE, R. H., 1972, A plate tectonic model for the origin of porphyry--copper deposits,
Econ. Geol. 67, 184-197.
24. SILLITOE, R. H., 1973, The tops and bottoms of porphyry--copper deposits, Econ. Geol. 68,
799-815.
25. SILLILTOE, R. H., 1979, Some thoughts on gold-rich porphyry--copper deposits, Miner.
Deposita 14, 161-174.
26. STRINGHAM, B., 1960, Differences between barren and productive intrusives, Econ. Geol.
55 1622-1630.
27. SUMNER, 1. S., 1967, Geophysical aspects of porphyry--copper deposits, Geol. Surv. Can.
Rep. No. 26, 322-335.
28. SUTHERLAND-BROWN, A., 1969, Mineralization in British Columbia and the copper-
molybdenum deposits, Can. [nst. Min. Metall. Bull. 62(681), 26-32.
29. TITLEY, S. R., and HICKS, C. L. (Eds.), 1966, Geology of Porphyry-Copper Deposits of
Southwestern North America, University of Arizona Press, Tucson.
30. TITLEY, S. R. (Ed.), 1982, Advances in Geology of the Porphyry-Copper Deposits, University
of Arizona Press, Tucson.
CHAPTER 5 EVE N
OPTIMIZED SEARCH FOR FOUR TYPES

OF CONTACT METASOMATIC DEPOSITS
OF THE NORTH AMERICAN
CORDILLERA BELT
7.1.1. Introduction and Terminology

7.1.1.1. Contact Metamorphism. The peculiar effect of intrusive bod-
ies of acid to intermediate composition upon adjacent rock formations has been
investigated during the past 100 years and has long been favored as a ready
explanation for ore genesis. Granitic intrusions are commonly surrounded by
halos of gradually fading metamorphism which may extend for a few hundreds
of feet up to two miles, but averages half a mile in width. In contrast, effusive
rocks and basic intrusive bodies have only a moderate to strong baking action
over a few tens of feet at the most.
On the mineralogic standpoint, contact metamorphism consists essentially
in a recrystallization of minerals accompanied by a transfer of volatiles such as
H2 0 and CO 2 . The nature of the mineralogic transformation reflects mostly the
original composition of the countryrock and the intensity of metamorphism, with
a lesser influence of the nature of the intrusive. (10,22)
Generally, clastic sediments or effusive rocks of silicic composition or
metasediments show little change, unless they contain calcareous or pelitic im-
purities. Pelitic sediments (shale, slate, silstone, marl) are converted into "horn-
fels" which are made up of assemblages of biotite, hornblend, plagioclase, and
quartz, with aggregates of cordierite, staurolite, and andalusite. Carbonate sed-
iments (limestone, dolomite) undergo conspicuous recrystallization as marbles,
and being generally more reactive than other rocks, are subjected to metasomatic
alteration.
7.1.1.2. Metasomatism and the Formation of Skarn. In contrast to
contact metamorphism, metasomatism involves some marked changes in the
proportion of nonvolatile elements such as Fe, Mg, and Si, in conditions well
141
142 CHAPTER SEVEN
above the critical pressure and temperature of water. The final product, an
assemblage of new lime-rich, iron-rich minerals is called "skarn." The term was
first coined by Medieval Swedish miners and referred to the lime-silicate gangue
of certain types of iron or sulfide ores occurring in the Scandinavian Precambrian
Shield. Most skarn are coarsely crystallized and exhibit crude zonal assemblages
which reflect sedimentary features of the country rock or the geometry of the
intrusive-wall rock contact. (5)
Einaudi(5) and Smirnov(22) distinguish two main types of skarns, including
"endoskarn" and "exoskarn." The latter replaces country rock outside the intru-
sive bodies over widths of a few hundreds to more than one thousand feet. The
former extending inward, on the intrusive side, replaces the intrusive rock over
widths of a few tens of feet to a few hundreds of feet. Exoskarns are, in tum,
divided into "calcic" skarns, the most common type, made up of lime-rich
silicates, hydrosilicates, oxides, and sulfides, and "magnesian" skarns, less com-
mon, which are characterized by different types of hydrosilicates such as phlo-
gopite, tremolite, and talc.
7.1.1.3. Ore Skarn. Ore skarn occurs within the reaction skarn when
shrinkage cracks resulting from metasomatism are filled up with iron and base
metal sulfides, scheelite, barite, fluorite, and secondary quartz. Ore skarn forms
most commonly in the exoskarn zone. In a minority of cases, the ore skarn
straddles the contact and extends into the endoskarn zone (see Figure 7.1).
Ore skarns are made up of early silicates of Ca,Mg,Fe,AI, derived partly
from the wall rock, partly from the intrusive body. The silicates are followed
in the parage netic sequence by scheelite and oxides such as cassiterite and
magnetite. The sequence ends with sulfides including pyrite and pyrrhotite and
base metal sulfides (Cu,Zn,Pb,Mo).
Ore skarns often exhibit zoning patterns at differing scales. At the local
scale, crude banding results from rhythmic metasomatism, while at a larger scale
the zoning reflects stratigraphic features of the wall rock. Finally, at a semire-
gional scale, the distribution of the mineralization shows a zonal arrangement
within the metamorphic aureola of the intrusive: scheelite mineralization is nar-
rowly restricted to the immediate contact, while copper spreads outward for some
distance, and lead and zinc, farther away.
7.1.1.4. Classification of Contact-Associated Ore Deposits. Most
genetic classifications of ore deposits include a category of deposits which may
be labeled as "contact-associated," including such terms as "contact metamor-
phic" or "pyrometamorphic" or "pyrometasomatic" deposits. In this study, we
have chosen the term "contact metasomatic" to emphasize the joint role of contact
and metasomatism in the formation of the deposits.
DETECTION OF CONTACT METASOMATIC DEPOSITS 143
Metallogeny of ...
~------------------------------------~.... Pyrite
.....
contact
1
met ~Lc;oma tic ----------------------~.... Chalcopyrite
ore deposits
----------~~ ~lagnetite
I
Frequency of
contact 5%
metasomat i c
15% 70% 10%
ore deposits
granitoid endoskam exoskam marble limestone
Lithology
up to 500 ft up to 800 ft
Contact
Intrusive
"eripheral metamorphic zone..
I
up to Z500 ft
Biotite + Diopside + Calcite +

Metasomatic lVollastoni te +
Minerals Andradite +
K-Feldspar + Quartz
Tremolite +
Talc
35~o
3%
.. decreasing SiOZ
Geochemistry decreasing Al Z0 3
..
~
R% 30~o
..
increasing Fep3
5% 35%
increasing CaO
FIGURE 7.1
Environment of Contact Metasomatic Ore Deposits.
144 CHAPTER SEVEN
There are many types and sUbtypes of contact metasomatic deposits, and
their classification generally emphasizes the chief minerals of economic impor-
tance as labels for the categories. In practice, there are innumerable gradations
between four common types which are covered in this chapter. They include (a)
Fe-Cu-Au deposits, (b) complex Pb-Zn-Cu-Ag deposits, sometimes related to
small intrusives, sometimes without obvious relationship with intrusives, (c)
Cu-Mo-Au deposits which are generally associated with differentiated plutons
of the porphyry--copper-bearing type, and (d) W-Mo deposits.
7.1.2. Characteristics of Contact Metasomatic Deposits

7.1.2.1. Geometry of Contact Metasomatic Deposits. Smimov(22)
classifies the shape of contact metasomatic deposits into three main types of
increasing complexity. The least complex are the stratified deposits. Generally
tabular in gross shape, they follow favorable beds along the strike for up to 8000
feet, down-dip for up to 2000 feet, with varying thickness up to 400 feet. The
ore material terminates abruptly across the strike of the host bed and gradually
pinches out down-dip.
The second type consists of steeply inclined pipes and stocks whose di-
mensions range from a few hundreds of feet to 1000 feet. The mineralized
material often crosses the strike of the host formation and terminates abruptly.
The third type, the most complex one, consists of intricately ramified bodies
grading into stockworks, whose dimensions are generally much smaller than
those of the two previous types, ranging from a few tens of feet up to a few
hundreds.
7.1.2.2. Control of Contact Metasomatic Ore Deposits. The for-
mation of contact metasomatic ore deposits is believed to be promoted by the
conjunction of two main controlling influences: (1) a physicochemical factor
reflecting the lithology of both wall rock and intrusive and conditions of em-
placement, and (2) a structural factor reflecting the geometry of the intrusive-wall
rock contact. The role of the wall rock lithology has been abundantly documented
on a worldwide basis. An important control is exerted by stratigraphic discon-
tinuities, which appear to act as "ore traps." The role of the intrusive lithology
is less commonly stressed, though it may be quite significant. In the Cordilleran
Belt of North America, granitic intrusives, commonly exhibiting conspicuous
contact metamorphic aureolae, do not generate contact metasomatic ore deposits.
Most commonly the deposits are genetically related to intrusives of intermediate
composition (quartz monzonite, monzonite, granodiorite), more rarely to basic
intrusives (diorite, gabbro).
Smirnov(22) cites three main types of structural control for contact meta-
somatic deposits, including (a) the geometry of the intrusive-wallrock interface,
(b) geometry of the bedding and jointing of the wall rock, and (c) the geometry
of the fractures affecting both wall rock and intrusive. In the early stages of ore
deposition, (a) and (b) appear to have a dominating influence, while (c) prevails
in the later stages.
The influence of the geometry of the intrusive-wall rock contact as control
for the deposition of tungsten mineralization is well documented in the North
American Cordillera. (10) Cooling creates a corrugated contact surface by spalling,
generating troughs and grooves which act as ore traps. Kerr(\O) also shows that
the concordance of the contact surface with bedding planes of the wall rock is
far less favorable for ore deposition than discordance. Finally, the influence of
fracturing is illustrated by the accumulation or mineralization in interstrata crush
zones produced during cooling as a result of the difference in competency between
wall rock formations. Furthermore, cross-cutting and oblique faults are believed
to act as feeder channels for ore-bearing solutions.
7.1.3. Geographic Distribution of Contact

Metasomatic Deposits
Occurrence of contact metasomatic deposits is as widespread geographically
as it is chronologically. Such deposits are well known in the late Precambrian
carbonate formations of the Grenville Province in the eastern portion of the North
American Shield and in the Paleozoic fold belt of the Appalachian. They are
particularly numerous in Mesozoic foldbelts because of the favorable combination
of extensive Paleozoic carbonate formations and numerous intrusions of inter-
mediate composition and Mesozoic age, as can be found in the North American
Cordillera Belt.
In the northern half of the Cordillera, the deposits are concentrated along
two main belts. The westernmost belt consists mainly of Fe-Cu skarn deposits
and extends through the coastal and insular regions of Southeast Alaska and
British Columbia. The Intermontane Belt extends through eastern British Co-
lumbia and the Yukon and exhibits a regional zoning from Cu-Fe skarn type
on the west side to complex Pb-Zn-Ag type in the central part of the belt, and
to W-Mo skarn type along the east and northeast edge of the belt.
In the U. S. section of the Cordillera, Fe-rich skarn deposits are quite com-
mon in the States of Utah, Nevada, California, and New Mexico. Complex
Pb-Zn-Ag metasomatic deposits are mainly concentrated in the central portion
of the Cordillera, in the States of Idaho, Utah, Nevada, and Colorado. The
Cu-Mo-Au type, mainly associated with copper-bearing differentiated plutons
146 CHAPTER SEVEN
is well represented in the southeastern Cordillera, in the States of Utah, Nevada,

Arizona, and New Mexico, as well as in northern Mexico. Finally the W-rich
skarn deposits occur mainly in the southwestern portion of the Cordillera, in the
States of Utah, California, and Nevada.
7.2. FIELD DETECTION METHODOLOGY
7.2.1. Introduction
In contrast with the porphyry-Cu-Mo deposits, most contact metasomatic
deposits offer targets of modest size for detection purposes. This rules out direct
detection by remote sensing techniques from satellite platforms. Other direct
detection methods including photogeology from aircraft platforms, ground map-
ping, and drilling have proved most useful in the detection of contact metasomatic
deposits. The indirect detection approach based on the recording of the geo-
physical signature of deposits from aircraft platforms or on the ground has proved
more useful than indirect geochemical methods based on the detection of halos.
7.2.2. Direct Geological Detection

Direct geological detection has proved very valuable in the search for contact
metasomatic deposits in two cases: typical contact-associated deposits, and com-
plex metasomatic deposits that do not show any close relationship with intrusive
contacts. In the first case, photogeological mapping of the contact followed by
ground mapping is the obvious method of delineating the potentially productive
areas, based on lithologic and mineralogic clues, to be followed by geophysical
coverage and drill testing of areas of highest merit. The second case is vastly
more difficult to handle, because the complex metasomatic deposits are generally
strata-bound and have no visible surface expression. The discovery of the Burgin
No. 1 and No. 2 deposits in the Tintic District of Utah under a thick capping
of recent lava flows, in the mid-1950s, is one of the best-known cases of suc-
cessful detection of these elusive targets by direct geological methods. (16,20) The
targeting relied on the conjunction of stratigraphic, structural, mineralogic, and
geochemical clues and was followed by drill testing.
7.2.3. Indirect Geophysical Approach

The slow-flying and ground-hugging helicopter platform is to be preferred
to the cheaper fixed-wing aircraft platform for the indirect detection of contact
metasomatic deposits by airborne geophysical surveys, because of the relatively

small size of the deposits and the ruggedness of the Cordillera topography.
Airborne magnetic surveys are quite suitable for the indirect detection of Fe-Cu
and Cu-Mo types of deposits, but provide little assistance in pinpointing small
and weakly magnetized W-Mo deposits. Electromagnetic surveys in the fre-
quency domain or in the time domain (INPUT) may be of assistance in the
detection of "blind" massive sulfide deposits of the complex Pb-Zn-Cu-Ag type
or Cu-Mo type, providing that the capping is less than 200 feet in thickness.
On the ground, magnetometric surveys are a very effective and low-cost
method of detection of Cu-Fe skarn deposits, while other more expensive tech-
niques such as electromagnetic and induced polarization surveys did not prove
themselves. The combination of magnetics and electromagnetics proved quite
successful in the search for massive sulfide deposits in skarn zones in New
Mexico and Arizona. Success was also reported in the Tintic District of Utah
in the search for blind complex Pb-Zn-Cu deposits by the "down the hole"
resistivity method from drill holes and underground workings. (20) The indirect
geophysical search for W-Mo skarn deposits is most troublesome because of
the small size of the targets and the weakness of the magnetization generated
by disseminations of magnetite and pyrrhotite which surround the deposits.
Although disseminated pyrite is widespread in the immediate environment of
the W-Mo deposits, there are no records of the successful use of the induced
polarization technique for their detection in the Cordillera. Klichnikov(ll) reports
success in locating W-Mo skarn deposits in Central Asia, using a combination
of gravimetry to locate the buried apices of small intrusives, magnetics to de-
lineate contact aurolae, and induced polarization to locate the pyritic halos of
W-Mo deposits within the aurolae.

The detection of primary halos of complex Pb-Zn-Cu metasomatic deposits
based on the pioneering work of Lovering(l7) in the Tintic District of Utah is a
good example of the successful application of the indirect geochemical approach
in the search for contact metasomatic deposits. Later on, an even more sophis-
ticated approach was devised in the Tintic District, (20) relying on the delineation
of "ghost" halos caused by leakages from primary halos through thick barren
lava capping.
Detection of secondary halos by residual soil sampling proved successful
in Central Asia in the search for W-Mo-skarn deposits.(Il) However, very little
success was met with in the search for Cu-Fe skarn deposits in central British
Columbia based on soil sampling or geobotanical sampling.
148 CHAPTER SEVEN
7.3. PROBABILITY OF DETECTION OF Cu-Fe CONTACT

METASOMATIC DEPOSITS
7.3.1. Geological Synopsis

7.3.1.1. Introduction. The Cu-Fe contact metasomatic deposits are
commonly tabular or lenticular in overall shape. They crudely follow the strat-
ification of favorable beds, but often terminate abruptly across the bedding. Ore
deposits consist of an envelope of iron-rich and lime-rich silicates enclosing pods
and lenses of massive magnetite with subordinate specularite, with minor sulfides
including pyrite, pyrrhotite, and chalcopyrite occurring along the edges of the
deposits. In some cases (Alaska, British Columbia), chalcopyrite is much more
abundant and becomes the principal economic mineral. The paragenesis is as
follows: after the formation of the early silicates, pyrite magnetite and specularite
develop, replacing some the silicates. The sequence ends with the replacement
of silicates, magnetite, and calcite by chalcopyrite.
7.3.1.2. British Columbia and Alaska Section of the Cordillera.
Cu-Fe skarn deposits have been mined for many years in the coastal and insular
portion of British Columbia and Alaska. (21) Good bessemer-grade iron ore and
copper ore of moderate grade have been gained from the Kasan Peninsula, Prince
of Wales Island, Southeast Alaska.(24) The ore lies in skarnified limestone in-
tercalations in Triassic volcanics intruded by Jurassic quartz diorite stocks. Sim-
ilar deposits occur in the Queen Charlotte Islands and in Vancouver Island, off
the coast of British Columbia. A second belt of Cu-Fe deposits lies in the
Intramontane zone of mainland British Columbia. An important metallogenic
feature of the region is the crude zoning of deposit types from the west where
Fe predominates to the east where Cu prevails (example: Craigmont deposit).
7.3.1.3. U.S. Section of the Cordillera. The Cu-Fe skarn deposits of
the U.S. section of the Cordillera occur in the same general geological setting
as those of British Columbia. However, there is a gap comprising the states of
Washington, Oregon, and Idaho between the Cu-Fe skarn belts of British Co-
lumbia and those of the U.S. which stretch through the states of Utah, Nevada,
California, and New Mexico.
The mineralization is chiefly magnetite and specularite with very little or
no copper. The larger deposits occur in Utah and mostly in California, with the
principal deposit, the Eagle Mountain mine, totaling some 80 million tons of
good grade iron ore, low in phosphorus and sulfur. The deposit consists of a
string of eight lenses along a six-mile-long horizon of early Paleozoic dolostone
folded to vertical during the Laramide orogeny and intruded by a stock of
Cretaceous porphyritic quartz monzonite. Local contamination of the intrusive

by the dolostone country rock is indicated by the presence of many xenoliths
and the change in composition from monzonite to quartz diorite.
7.3.2. Statistical Modeling of Geometric Parameters of Cu-Fe

Contact Metasomatic Deposits
A total of 23 deposits were included in the database to represent the pop-
ulation of Cu-Fe deposits of the Cordillera. The deposits are listed by name and
region of occurrence in Table 7.1. The geographic breakdown is as follows: 13
deposits in British Columbia, 3 in Alaska, and 7 deposits in the rest of the United
States.
The geometry of deposits is quantitatively described by means of dimen-
sional and attitudinal parameters. The former include the length, breadth, and
shape ratio of the horizontal section of the portions of the deposits which lie
TABLE 7.1
list of Contact Metasomatic (Cu-Fe) Deposits of the
North American Cordillera Belt Included in the Data-
base.
B.C. Islands (Canada) S.W. Alaska !U 5 1\
Quatsino Mamie
F.L. mine Poor Man
Iron Hill Mt. Andrew
Prescott Utah (U~S.A.I
Paxton Iron Spring

Yellow Kid McCahill
Lake Nevada (U.S.A.)
Ford Dayton
Ridge California (U.S.A.)
Jessie Eagle Mountain
B.C. Mainland (Canada) Dale
Craigmont Iron Mountain
Merry Widow New Mexico (U.S.A.)
King Fisher Hanover
150 CHAPTER SEVEN
within the average range of detection (300 feet). The latter include the unoriented
dip and the strike direction of deposits referred to true North and measured in
degrees within the right half circle. The range of measurements of the parameters
for the 23 deposits of the sample is from 500 to 5300 feet for length, 35 to 520
feet for breadth, and 25 to 85 degrees for dip.
The measurements are grouped into classes and assembled into frequency
distributions which are summarized by means of statistics. The statistics include
arithmetic mean, mode, standard deviation, and coefficient of skewness and are
listed in the lower half of Table 7.2. The coefficient of skewness gives an
indication of the asymmetry of the distributions of observed data; the skewness
is moderately high (0.35) for the lengths, but is quite high for the four other
parameters (0.6O--().90).
The observed distributions of the dimensional parameters were fitted by
lognormal models at the 0.05 confidence level. The statistics of the fitted models,
including geometric means, and their 95% fiducial limits, first and third quartiles,
and coefficients of dispersion are listed in the upper half of Table 7.2. The
TABLE 7.2
Statistical Modeling Summary for Contact Metasomatic (Cu-Fe) Deposits of
the North American Cordillera Belt (Sample Size: 23 Deposits).
Statistic Length of Breadth of Shape Strike dir. Dip

of fitted horiz. sect. horiz. sect. ratio from T. North in
model in feet in feet R=B/L in degrees degrees
25%ile 800 150 0.12 12 0

95% L.C.L. 982 183 0.14 33 23
Mean 1273 230 0.18 52 36
95% U.c.l. 1650 288 0.23 72 48
75%ile 1700 400 0.25 95 70
Disp. coef. 0.21 0.09 0.11 0.38 0.12
Statistic
of observed
data
Arithm. mean 1667 266 0.23 56 42

Mode 1150 350 0.12 15 15
Stand. dev. 1477 128 0.18 48 34
Skewness 0.35 0.65 0.92 0.85 0.83
dispersion coefficient which gives an indication of the variability of the data is

rather small (0.1) for the breadths, and is moderate (0.2) for the lengths and
shape ratios. The observed distributions of dip and strike direction could not be
fitted by normal or circular normal models at the required 0.05 confidence level.
7.3.3. Construction and Organization of Detection

Probability Tables
For detection purposes, the eu-Fe contact metasomatic deposits are con-
sidered as orientated dipping slabs with a horizontal section of elliptical shape.
The addition of a dip component to the horizontal dimensions of ore deposits is
required in order to obtain the dimensions of the targets for detection purposes.
Since strike orientation measurements were available for all deposits of the
sample, the grid orientation factor was introduced in the calculation of proba-
bilities of detection. It was found that the probability of detection by airborne
surveys is maximized when the grid is laid at 90 degrees to the expected strike
of the target. Optimal grid orientation for vertical detection by geophysical
surveys or drilling is 30 degrees ± 10, from target strike.
Two types of tables are constructed for the display of the probability of
detection of eu-Fe deposits. One type covers the calculation of the probabilities
of single detection (at least one target intersection by the grid) and confirmed
detection (at least two intersections) by three types of field programs for grid
spacings varied between 100 and 6000 feet in steps of 100--200 feet. In both
cases, the probabilities are provided within 95% confidence intervals, which are
based on the 95% fiducial intervals for the expected target length.
The second type of table is provided for planning purposes. The tables
display grid spacings and corresponding coverage cost per unit of area for spec-
ified levels of probability of detection for each main type of field program. The
range of probability levels is from 0.10 to 0.95, and steps of 0.05 are used.
The detection probability tables are organized into three blocks, each cov-
ering one type of field program. The first block covers airborne geophysical
surveys on parallel and square grids (Tables 7.3-7.6). The second block deals
with vertical ground detection by geophysical surveys or drilling to a depth of
300 feet (Tables 7.7-7.11). The third block refers to angled detection by drilling
to a vertical depth of 300 feet (Tables 7.10-7.12).
7.3.4. Detection Probability Tables

Four tables displaying the probabilities of detection of contact metasomatic
(eu-Fe) deposits are to be found in pages 152-155. Six more tables cover the
probabilities of detection of these deposits by ground geophysical surveys and
drilling programs (pages 156-161).
152 CHAPTER SEVEN
TABLE 7.3.
Probabilities of Detection of Contact Metasomatic (Cu-Fe) Deposits of the
North American Cordillera Belt by Airborne Geophysical Surveys on Parallel
Grids.

Elliptical targets with expected major axis = L feet
Orientation of flight lines: 90 degrees with strike-line.
S feet L= 982 1273 1650 feet
._----------------------------------------------------------------------
200 1. 000 1.000 1. 000
400 0.901 0.924 1.000
600 0.849 0.885 0.912
800 0.792 0.844 0.882
1000 0.753 0.800 0.850
1200 0.627 0.750 0.817
1400 0.538 0.697 0.781
1600 0.470 0.610 0.740
1800 0.418 0.542 0.703
2000 0.376 0.488 0.632
2200 0.342 0.443 0.575
2400 0.314 0.407 0.527
2600 0.289 0.375 0.486
2800 0.269 0.348 0.452
3000 0.251 0.325 0.422
3200 0.235 0.305 0.395
3400 0.221 0.287 0.372
3600 0.209 0.271 0.351
3800 0.198 0.257 0.333
4000 0.188 0.244 0.316
4200 0.179 0.232 0.301
4400 0.171 0.222 0.287
4600 0.164 0.212 0.275
4800 0.157 0.203 0.263
5000 0.151 0.195 0.253
5200 0.145 0.188 0.243
TABLE 7.4
of Contact Metasomatic (Cu-Fe) Deposits of the North American Cordillera
Belt by Airborne Geophysical Surveys on Parallel Grids.

unit Cost = US$70 per line-mile

0.10 8180 $ 185.18

0.15 5620 $ 205.77
0.20 4280 $ 226.36
0.25 3460 $ 246.82
0.30 2900 $ 267.45
0.35 2500 $ 287.84
0.40 2200 $ 308.00
0.45 1960 $ 328.57
0.50 1780 $ 347.64
0.55 l620 $ 368.15
0.60 1480 $ 389.73
0.65 1380 $ 407.83
0.70 1280 $ 428.75
0.75 1180 $ 453.22
0.80 960 $ 525.00
0.85 740 $ 639.46
0.90 480 $ 910.00
0.95 220 $1820.00
154 CHAPTER SEVEN
TABLE 7.5
Probabilities of Confirmed Detection of Contact Metasomatic (Cu-Fe) Deposits
of the North American Cordillera Belt by Airborne Geophysical Surveys on
Parallel Grids.


S feet L= 982 1273 1650 feet
150 1.000 1.000 1.000

300 0.926 1.000 1.000
450 0.653 0.915 0.934
600 0.432 0.628 0.901
750 0.319 0.456 0.661
900 0.262 0.353 0.509
1050 0.000 0.290 0.408
1200 0.000 0.257 0.339
1350 0.000 0.000 0.293
1500 0.000 0.000 0.264
1650 0.000 0.000 0.250
1800 0.000 0.000 0.000
TABLE 7.6
Probabilities of Detection of Contact Metasomatic (Cu-Fe) Deposits of the
North American Cordillera Belt by Geophysical Surveys on Square Grids.

Grid spacing 1.c.1. geom. mean u.c.1.
S feet L= 982 1273 1650 feet
800 1. 000 1. 000 1.000

1000 0.957 1. 000 1.000
1200 0.863 1. 000 1. 000
1400 0.780 0.919 1.000
1600 0.709 0.848 1. 000
1800 0.648 0.785 0.923
2000 0.597 0.728 0.868
2200 0.552 0.678 0.816
2400 0.513 0.634 0.769
2600 0.480 0.595 0.726
2800 0.450 0.560 0.688
3000 0.424 0.529 0.652
3200 0.400 0.501 0.620
3400 0.379 0.476 0.591
3600 0.361 0.453 0.564
3800 0.343 0.433 0.540
4000 0.328 0.414 0.517
4200 0.313 0.396 0.496
4400 0.300 0.380 0.477
4600 0.288 0.365 0.459
4800 0.277 0.351 0.442
5000 0.267 0.339 0.427
5200 0.257 0.327 0.412
156 CHAPTER SEVEN
TABLE 7.7
Probabilities of Detection of Contact Metasomatic (Cu-Fe) De-

L.c.1. ( geometric mean ( U.c.1., and minor axis = S feet.
grid orientation 30 degrees + or - 10 from the expected
strike line.
in feet L = 982 1273 1650ft
300 1.000 1. 000 1. 000

400 0.964 1. 000 1.000
500 0.728 0.981 1. 000
600 0.525 0.820 0.997
700 0.386 0.639 0.934
800 0.296 0.497 0.786
1000 0.189 0.318 0.533
1200 0.131 0.221 0.371
1400 0.097 0.162 0.273
1600 0.074 0.124 0.209
1800 0.058 0.098 0.165
2000 0.047 0.080 0.134
2400 0.033 0.055 0.093
2800 0.024 0.041 0.068
3200 0.018 0.031 0.052
3600 0.015 0.025 0.041
4000 0.012 0.020 0.033
TABLE 7.8
Determination of Grid Size for Specified Probability Levels in the Detection
of Contact Metasomatic (Cu-Fe) Deposits of the North American Cordillera
Belt by Vertical Drilling to a Depth of 300 Feet.
Survey Design: Square Grid with Spacings S by S Feet.

Unit Cost = $6000 Per Diamond Drill-Hole.
Probabili ty Level Grid Size Dr illing Cost Per Mile Square

0.05 2350 63.64

0.15 1450 129.81
0.25 1150 188.25
0.35 950 258.82
0.45 850 312.92
0.55 800 347.47
0.65 700 438.91
0.75 650 500.48
0.85 600 577.42
0.95 550 675.43
158 CHAPTER SEVEN
TABLE 7.9
Probabilities of Confirmed Detection of Contact Metasomatic
(Cu-Fe) Deposits of the North American Cordillera Belt by Ground
Geophysical Surveys or Vertical Drilling to a Depth of 300 Feet.

L.c.1. ( geometric mean ( U.c.1., and minor axis = B feet.
strike line.

Spacing S L.c.1. G. mean U.c.l.
in feet L = 982 1273 1650ft
200 1.000 1. 000 1.000

250 0.981 1.000 1.000
300 0.858 1.000 1.000
350 0.619 0.943 1.000
400 0.422 0.829 1.000
450 0.293 0.639 0.947
500 0.194 0.467 0.866
550 0.124 0.355 0.749
600 0.078 0.265 0.570
700 0.028 0.136 0.375
800 0.008 0.066 0.238
900 0.001 0.030 0.142
1000 0.000 0.012 0.081
1200 0.000 0.001 0.024
TABLE 7.10
Probabilities of Detection of Contact Metasomatic (Cu-Fe) De-
posits of the North American Cordillera Belt by 55-Degree An-
gled Drilling to a Vertical Depth of 300 Feet.

the expected shape of the ore depes·i ts is elliptical
strike line.
in feet L = 982 1273 1650ft
400 1.000 1.000 1.000

500 0.963 1. 000 1.000
600 0.801 0.994 1.000
700 0.616 0.908 1.000
800 0.473 0.768 0.986
1000 0.303 0.509 0.810
1200 0.210 0.354 0.593
1400 0.155 0.260 0.436
1600 O.llB 0.199 0.334
1800 0.094 0.157 0.264
2000 0.076 0.127 0.214
2400 0.053 0.08B 0.14B
2800 0.039 0.065 0.109
3200 0.030 0.050 0.084
3600 0.023 0.039 0.066
4000 0.019 0.032 0.053
160 CHAPTER SEVEN
TABLE 7.11
Determination of Grid Size for Specified Probability levels in the Detec-
tion of Contact Metasomatic (Cu-Fe) Deposits of the North American
Cordillera Belt by 55-Degree Angled Drilling to a Vertical Depth of 300
Feet.


Of Detection In Feet In u.s.$. Thousands
0.15 1800 114.29

0.25 1400 167.46
0.35 1200 214.39
0.45 1100 248.11
0.55 1000 289.81
0.65 900 346.39
0.75 850 382.03
0.85 750 476.30
0.95 700 535.83
TABLE 7.12
(Cu-Fe) Deposits of the North American Cordillera Belt by 55-
Degree Angled Drilling to a Vertical Depth of 300 Feet.

strike line.

in feet L = 982 1273 1650ft
350 1.000 1.000 1.000

400 0.865 1.000 1.000
450 0.575 1.000 1.000
500 0.399 0.955 1.000
600 0.183 0.525 1.000
700 0.075 0.294 0.747
800 0.023 0.159 0.477
900 0.003 0.080 0.304
1000 0.000 0.034 0.191
1200 0.000 0.002 0.065
7.4. PROBABILITY OF DETECTION OF Pb-Zn-Cu-Ag CONTACT


7.4.1.1. Introduction. The Pb-Zn-Cu-Ag type of contact metasomatic
deposits of the Cordilleran Belt exhibits marked differences from the other three
types covered by the present chapter. Here, the emphasis is on metasomatism
rather than contact association. The proximal environment of the deposits is
intensely altered and skarnified, though the mineralogic context is quite different
from that of typical contact-skarn deposits. The newly formed silicates accom-
panying the complex Pb-Zn-Cu-Ag mineralization are mainly amphiboles, ep-
162 CHAPTER SEVEN
idote, and chlorite with a general absence of garnets which are so conspicuous
in the contact skarns. Finally, there is often no obvious spatial relationship
between the deposits and their metasomatic halos on the one hand, and intrusive
bodies, if any in the vicinity, on the other hand.
The complex Pb-Zn-Cu-Ag deposits are not so common nor so widely
distributed throughout the Cordillera as the Fe-Cu skarn deposits. In the Canadian
section of the Cordillera, these deposits occur in two regions including the Yukon
and the southcentral portion of British Columbia. The deposits of the central
portion of the U. S. section of the Cordillera are larger, richer, and more numerous
than in the Canadian section, and are the models of the type.
7.4.1.2. British Columbia and Yukon Section of the Cordillera. At

the northern end of the Canadian Cordillera, in the Yukon, six Cu-Au-Ag
metasomatic deposits occur in a skarnified Cambrian limestone horizon over a
strike length of about 10 miles, near the town of Whitehorse. While copper,
silver, gold mineralizations are present throughout, magnetite, and specularite
are abundant only in the central portion of the mineralized belt, and molybdenite
only at both ends.
In the Salmo area of southcentral British Columbia, two deposits of the
Pb-Zn-Ag metasomatic type, the Emerald and the Jersey, occur within a struc-
tural trough of metasomatized Cambrian limestone and dolomite, between two
small granitic stocks of Jurassic age. (19) Farther south and west, metasomatic
deposits unusually rich in gold and showing only low copper values occur near
the town of Hedley, at the Nickel Plate mine, in the general vicinity of basic
intrusive stocks ranging in composition from diorite to gabbro. At the near-by
and smaller French mine, the ore is mostly copper with minor gold and silver
values.
7.4.1.3. U. S. Section of the Cordillera. Complex Pb-Zn-Cu-Ag me-

tasomatic deposits are more common in the U. S. section of the Cordillera than
in the Canadian section, particularly in the States of Idaho, Utah, and Colorado.
In the Mackay district of Idaho, metasomatic deposits carrying mainly copper
with minor lead-zinc-silver values occur in large blocks of skarnified limestone
engulfed by a monzonitic intrusive. Similar mineralizations have been mined in
the past in the San Francisco Mountains, in Utah,(2) as well as in Nevada and
in Arizona.
However, the richest and most complex metasomatic base metal sulfide
deposits of the whole North American Cordillera are found in the Tintic district
of Utah and the Leadville district of Colorado. (20) The Tintic district had produced
prior to the mid-1950s some 15 million tons of high-grade silver-lead-zinc ore

with minor copper. The deposits occur as massive sulfide replacements in the
highly metasomatized, folded, and faulted Ophir limestone formation of Cam-
brian age, which is cut across by innumerable dykes and sills of porphyrytic
quartz monzonite. The district was considered exhausted, until some geological
rethinking in the mid-1950s led to the discovery of the high-grade Burgin No.
1 and large, low-grade Burgin No.2 deposits, under a thick capping of Missis-
sippian clastics and Eocene lava flows. This success spurred an intensified search
for similar deposits in the presently revitalized district.
Farther to the east, in the central portion of the State of Colorado, the
discovery of blanket or "manto" -type metasomatic deposits in the Leadville
district made history because of the very high grade of the silver mineralization
accompanied by moderate to low grades in base metals (lead, zinc, copper). The
mineralization is found in two limestone horizons, the Blue Limestone and the
White Limestone, separated by the barren Parting Quartzites, and cut across by
the White Porphyry and the younger Grey Porphyry. The ore extends over
considerable distances along the bedding of the Blue Limestone, either imme-
diately below the White Porphyry-Blue Limestone contact, or between tongues
of Grey Porphyry. In the lower White Limestone member, the sulfide miner-
alization extends between tongues of Grey Porphyry, or along the Parting Quartz-
ite contact.
7.4.2. Statistical Modeling of Geometric Parameters of

Pb-Zn-Cu-Ag Metasomatic Deposits
A total of 41 deposits were selected and included in the database to represent
the whole population of complex base metal sulfide metasomatic deposits of the
Cordillera. The deposits are listed by name and region of occurrence in Table
7.13. The geographic breakdown is as follows: Canadian section of the Cordil-
lera: 5 deposits; Tintic district of Utah: 12 deposits; Leadville district of Colorado:
14 deposits. The remaining 10 deposits are scattered across the States of Nevada,
Idaho, and Arizona.
The geometry of the deposits is described by means of dimensional and
attitudinal parameters. The former include length, breadth, and shape ratio of
the horizontal section of deposits, and the latter, dip and strike direction of the
deposits. The range of measurements obtained from the 41 deposits is from 450
to 7500 feet for length, 15 to 420 feet for breadth, and 0 to 75 degrees for dip.
The measurements are grouped into classes and assembled into frequency dis-
tributions whose summarizing statistics are listed in the lower half of Table 7.14.
164 CHAPTER SEVEN
TABLE 7.13
list of Contact Metasomatic (Pb-Zn-Cu)
Deposits of the North American Cordillera
Belt Included in the Database.
Utah (U.S.A.) Colorado (U.S.A.)
Horn Silver Iron Hill

Cactus Henriett-Maid
Red Warrior Cord
Tintic Standard Minnie-Helen
North Lily Sellers
Eureka Lily Imes
Burgin Giltedge
Gemini Comstock
Mammoth Chief Ibex
Plutus Rickard-Stone
Iron Blossom Golden Eagle
Godiva Mahala
Tucson
Nevada (U.S.A.) Carbonate Hill
Simon Idaho !U.S.A.)

Blue Stone
Mason Valley Empire
Casting Copper
Ludwig B.C. & Yukon (Canada)
Arizona (U.S.A.) Jersey

Emerald
Johnson Nickel Plate
Copper World French
Hardshell Whitehorse Copper
Mowry

Probability Tables
As is the case for the Cu-Fe skarn deposits of Section 7.3, the complex
Pb-Zn-Cu-Ag metasomatic deposits are considered as oriented dipping slabs
with a horizontal section of elliptical shape, for detection purposes. One type
of table is constructed to provide probabilities of detection or confirmed detection
of the complex metasomatic deposits by means of three types of field programs
for varying grid spacings. A second type of table displays the grid spacings and
TABLE 7.14
Statistical Modeling Summary for Contact Metasomatic (Pb-Zn-Cu) Deposits of
Statistic Length of Breadth of Shape Strike dir. Dip in

of fitted hor iz. sect. horiz. sect. ratio from T.North degrees
model in feet in feet R=B/L in degrees
25%ile 900 85 0.06 12 0

95%L.c.1. 949 85 0.06 35 31
Mean 1218 111 0.09 55 42
95%U.c.1. 1562 146 0.14 74 53
75%ile 1600 220 0.18 110 70
Disp. coef. 0.11 0.18 0.54 0.38 0.18
Statistic
of observed
data
Arithm.mean 1784 155 0.17 62 42

Mode 1100 210 0.18 10 0
Stand.dev. 2018 180 0.18 62 34
Skewness 0.62 0.30 0.56 0.83 1.21
corresponding coverage costs per unit of area which are required to obtain
prespecified levels of detection probabilities.
The tables are organized into three blocks, one for each type of field pro-
gram. Tables 7.15-7.18 cover airborne geophysical surveys; Tables 7.19-7.21
deal with vertical ground detection by geophysical surveys or drilling to a depth
of 300 feet. Finally, Tables 7.22-7.24 refer to angled drilling to a vertical depth
of 300 feet.

Pages 166-169 cover the probabilities of detection of contact metasomatic
(Pb-Zn-Cu) of the Cordillera Belt by airborne geophysical surveys. The fol-
lowing pages, 170-175, include six tables of probabilities of detection of these
deposits by ground geophysical surveys and drilling programs.
166 CHAPTER SEVEN
TABLE 7.15
Probabilities of Detection of Contact Metasomatic (pb-Zn-Cu) of the North
American Cordillera Belt by Airborne Geophysical Surveys on Parallel Grids.

in the confidence interval: 1.c.1. ( geom. mean ( u.c.l.
Grid spacing l.c.1. geom. mean u.c.l.
S feet L= 949 1217 1562 feet
----------------------------------------------------------------------
200 0.944 1. 000 1. 000
400 0.887 0.912 0.932
600 0.827 0.867 0.897
800 0.760 0.819 0.861
1000 0.691 0.767 0.824
1200 0.576 0.705 0.785
1400 0.493 0.633 0.741
1600 0.432 0.554 0.710
1800 0.384 0.492 0.632
2000 0.345 0.443 0.568
2200 0.314 0.403 0.517
2400 0.288 0.369 0.474
2600 0.266 0.341 0.437
2800 0.247 0.316 0.406
3000 0.230 0.295 0.379
3200 0.216 0.277 0.355
3400 0.203 0.260 0.334
3600 0.192 0.246 0.316
3800 0.182 0.233 0.299
4000 0.173 0.221 0.284
4200 0.164 0.211 0.271
4400 0.157 0.201 0.258
4600 0.150 0.193 0.247
4800 0.144 0.185 0.237
5000 0.138 0.177 0.227
5200 0.133 0.170 0.219
TABLE 7.16
Determination of Grid Size for Specified Probability Levels in the Detection of
Contact Metasomatic (Pb-Zn-Cu) Deposits of the North American Cordillera

Unit Cost = US$70 per line-mile

0.10 8160 $ 185.29

0.15 5600 $ 206.00
0.20 4280 $ 226.36
0.25 3460 $ 246.82
0.30 2900 $ 267.45
0.35 2500 $ 287.84
0.40 2200 $ 308.00
0.45 1960 $ 328.57
0.50 1760 $ 350.00
0.55 1600 $ 371.00
0.60 1480 $ 389.73
0.65 1360 $ 411.76
0.70 1200 $ 448.00
0.75 1040 $ 495.38
0.80 840 $ 580.00
0.85 640 $ 717.50
0.90 420 $1020.00
0.95 200 $1988.00
168 CHAPTER SEVEN
TABLE 7.17
Probabilities of Confirmed Detection of Contact Metasomatic (Pb-Zn-Cu) De-
posits of the North American Cordillera Belt by Airborne Geophysical Surveys
on Parallel Grids.


S feet L= 949 1217 1562 feet
500 0.486 0.736 1. 000

600 0.349 0.545 0.818
700 0.262 0.415 0.640
800 0.207 0.325 0.510
900 0.176 0.261 0.414
1000 0.000 0.216 0.341
1100 0.000 0.187 0.286
1200 0.000 0.171 0.243
1300 0.000 0.000 0.211
1400 0.000 0.000 0.189
1500 0.000 0.000 0.174
TABLE 7.18
Probabilities of Detection of Contact Metasomatic (Pb-Zn-Cu) Deposits of the
North American Cordillera Belt by Airborne Geophysical Surveys on Square
Grids.
S feet L= 949 1217 1562 feet
800 1.000 1.000 1. 000

1000 0.935 1. 000 1. 000
1200 0.833 1. 000 1.000
1400 0.747 0.886 1.000
1600 0.676 0.811 0.950
1800 0.616 0.746 0.885
2000 0.565 0.689 0.826
2200 0.522 0.640 0.773
2400 0.484 0.597 0.725
2600 0.452 0.559 0.682
2800 0.424 0.525 0.644
3000 0.398 0.495 0.610
3200 0.376 0.468 0.578
3400 0.356 0.444 0.550
3600 0.338 0.423 0.524
3800 0.322 0.403 0.501
4000 0.307 0.385 0.479
4200 0.293 0.368 0.459
4400 0.281 0.353 0.441
4600 0.270 0.339 0.424
4800 0.259 0.326 0.409
5000 0.249 0.314 0.394
5200 0.240 0.303 0.380
170 CHAPTER SEVEN
TABLE 7.19
Probabilities of Detection of Contact Metasomatic (Pb-Zn-Cu)

strike line.
Spacing S L.c.l. G. mean U.c.1.
in feet L = 949 1218 1562 i t
200 1. 000 1.000 1. 000

300 0.970 1. 000 1.000
400 0.717 0.952 1. 000
500 0.478 0.747 0.965
600 0.334 0.542 0.821
700 0.245 0.403 0.646
800 0.188 0.309 0.504
1000 0.120 0.198 0.326
1200 0.084 0.137 0.226
1400 0.061 0.161 0.166
1600 0.047 0.077 0.127
2000 0.030 0.049 0.081
TABLE 7.20
Determination of Grid Size for Specified Levels of Probability in the
Detection of Contact Metasomatic (Pb-Zn-Cu) Deposits of the North
American Cordillera Belt by Vertical Drilling to a Depth of 300 Feet.
Survey Design: Square Grid With Spacings S by S Feet.


0.05 1850 89.58

0.15 1150 188.25
0.25 900 283.73
0.35 750 388.81
0.45 700 438.91
0.65 650 500.48
0.55 600 577.42
0.65 550 675.43
0.75 500 803.19
0.85 450 974.35
0.95 400 1211.54
172 CHAPTER SEVEN
TABLE 7.21
(Pb-Zn-Cu) Deposits of the North American Cordillera Belt by
Ground Geophysical Surveys or Vertical Drilling to a Depth of
300 Feet.

strike line.

in feet L = 949 1218 1562 ft
150 1.000 1. 000 1.000

200 0.932 1. 000 1. 000
250 0.792 0.943 1. 000
300 0.469 0.844 0.968
350 0.304 0.698 0.902
400 0.197 0.422 0.816
450 0.124 0.302 0.696
500 0.077 0.215 0.454
550 0.047 0.151 0.349
600 0.029 0.105 0.269
700 0.010 0.049 0.156
800 0.003 0.023 0.088
TABLE 7.22
Probabilities of Detection of Contact Metasomatic (Pb-Zn-Cu)
Deposits of the North American Cordillera Belt by 55-Degree
Angled Drilling to a Vertical Depth of 300 Feet.

strike line.
Spacing S L.c .1. G. mean U.c.1.
in feet L = 949 1218 1562ft
300 1. 000 1. 000 1.000

400 0.998 1. 000 1. 000
500 0.853 1.000 1.000
600 0.643 0.918 1.000
700 0.476 0.754 0.983
800 0.365 0.598 0.882
1000 0.233 0.384 0.629
1200 0.162 0.267 0.439
1400 0.119 0.196 0.323
1600 0.091 0.150 0.247
2000 0.058 0.096 0.158
2400 0.041 0.067 0.110
174 CHAPTER SEVEN
TABLE 7.23
tion of Contact Metasomatic (Pb-Zn-Cu) Deposits of the North American
Feet.
Survey Design: Square Grid With Spacings S by S Feet_

unit Cost = $7325 Per Diamond Drill-Hole.
Probability Level Grid Size Dr illing Cost Per Hile Square

Of Detection Ir. Feet In U.S.$. Thousands
0.15 1550 142.87

0.25 1250 200.67
0.35 1050 267.10
0.45 950 315.98
0.55 850 382.03
0.65 800 424.21
0.75 700 535.83
0.85 650 611.00
0.95 600 704.93
TABLE 7.24.
(Pb-Zn-Cu) Deposits of the North American Cordillera Belt by
55-Degree Angled Drilling to a Vertical Depth of 300 Feet.

strike line.

in feet L = 949 1218 1562ft
300 1.000 1.000 1.000

350 0.841 1. 000 1.000
400 0.563 1.000 1. 000
450 0.396 0.839 1.000
500 0.265 0.520 1.000
600 0.111 0.345 0.780
700 0.039 0.180 0.473
800 0.010 0.088 0.29i
900 0.,001 0.038 0.179
1000 0.000 0.014 0.103
1100 0.000 0.003 0.056
1200 0.000 0.000 0.027
7.5. PROBABILITY OF DETECTION OF Cu-Mo-Au CONTACT


7.5.1.1. Introduction. Differentiated plutons, some of them of the cop-
per-bearing porphyry type, are known to produce intense contact metasomatism
in carbonate and other kinds of "reactive" wall rock formations, resulting in the
formation of typical gametiferous skams. Many examples have been described
by Einaudi(5) in the southwest United States Cordillera, and by others in the
southwest Pacific Island Arcs, and in the USSR. (22)
176 CHAPTER SEVEN
7.5.1.2. Canadian Cordillera. At the northern end of the Canadian Cor-

dillera, in the Yukon, the large low-grade W-Mo deposit referred to as Lotung,
consists of scheelite mineralization in garnet-<iiopside skarn formed by the me-
tasomatic alteration of Pennsylvanian argillite and limestone by a differentiated
quartz monzonite stock of Cretaceous age. In central British Columbia, pyrite-
chalcopyrite mineralization with or without magnetite and specularite occur in
Paleozoic limestone formations intruded by differentiated grano-<iiorite stocks
of Mesozoic age.
At the Red Mountain deposit, near Rossland, in southern British Columbia,
pyrite-pyrrhotite-molybdenite mineralization is hosted by "skarn-tuff' horizons
in Triassic series of argillite and andesite tuffs intruded by a Mesozoic grano-<iiorite
stock. In the Princeton region, Similkameen district, the Ingerbelle deposit is
classified as a hybrid skarn-porphyry type similar to those described by Einaudi(5)
in the southwest U.S. The Ingerbelle mineralization consists of disseminations
of pyrite and chalcopyrite in intensely skarnified and propylitized andesitic tuffs
and fragmentals of the Nicola series of Triassic age.
7.5.1.3. U.S. Cordillera. Einaudi(5) stresses the marked differences be-
tween contact metasomatic deposits associated with differentiated copper-bearing
porphyry-type plutons of the southwest Cordillera and those associated with
other types of intrusives. The porphyry-associated skarns are more massive and
finer-grained and characterized by a greater abundance of andradite and diopside
than the nonporphyry skarns. Furthermore, in the porphyry-associated skarns,
pyrite prevails over pyrrhotite and specularite over magnetite, and retrograde
alteration to clay minerals is much more intense than in the nonporphyry envi-
ronment.
Einaudi describes the following paragenetic sequence: (a) initial "contact
metamorphic" phase altering carbonates into marble and pelitic series into horn-
fels, (b) "metasomatic" phase, altering marbles into garnetiferous skarns and
introducing actinolite and biotite in the hornfels, (c) introduction of sulfides
during the main phase of the potassic-silicic alteration of the intrusive, and (d)
retrograde alteration which may be so intense that intrusive and skarn alike are
reduced to quartz-carbonate-clay assemblages.
The predominant economic minerals are chalcopyrite and molybdenite, with
subordinate amounts of zinc and lead sulfides and minor gold and silver values.
Generally, the Cu/Mo and Zn/Cu ratios and protore Cu grades are higher in the
skarn than in the intrusive itself. In some cases, such as the Ely, Nevada, deposit,
the skarns are mineralized with Wand Mo and very minor copper. Very little
supergene enrichment in copper and gold takes place because of the buffering
influence exerted by the carbonate formations on the ground water acidified by
the oxidation of pyrite.
7.5.2. Statistical Modeling of Geometric Parameters of

Cu-Mo-Au Contact Metasomatic Deposits
A total of 15 deposits were included in the database to represent the pop-
ulation of Cu-Mo-Au contact metasomatic deposits of the Cordillera. The de-
posits are listed by name and region of occurrence in Table 7.25. The geographic
breakdown is as follows: Canadian section of the Cordillera: 3 deposits; U.S.
section: 11 deposits; and Mexico section: 1 deposit.
deposits. The range of measurements obtained from the 15 deposits of the sample
is as follows: 180-4850 feet for length, 135-1250 feet for breadth, and 0-60
degrees for dip. The measurements are grouped into classes and assembled into
frequency distributions which are summarized by means of statistics, as listed
in the lower half of Table 7.26.
dimensional parameters at the 0.05 confidence level. However, the observed
distributions of attitudinal parameters could not be satisfactorily fitted by normal
TABLE 7.25
List of Contact Metasomatic (Cu-Mo-Au) De-
posits of North American Cordillera Belt In-
cluded in the Database.
Arizona (U.S.A.) New Mexico (U.S.A.)
Christmas Santa Rita

Twin Buttes Continental
Mission
Bisbee ~
Lake Shore Cananea
Silver Bell
Morenci British Columbia
(Canada)
Utah (U.S.A.) Red Mountain
Bingham Similkameen
Nevada (U.S.A.) Yukon (Canada)

Ely Logtung
178 CHAPTER SEVEN
TABLE 7.26
Statistical Modeling Summary for Contact Metasomatic (Cu-Mo-Au) Deposits
of the North American Cordillera Belt (Sample Size: 15 Deposits).
Statistic Length of Breadth of Shape Str ike dir. Dip in

of fitted horiz. sect. horiz. sect. ratio from T. North degrees
25%ile 2200 300 0.12 0 0

95%L.c.1. 2206 330 0.13 46 3
Mean 2648 480 0.18 82 22
95%U.c.1. 3180 698 0.26 105 41
75%ile 4000 1000 0.50 115 65
Disp. coef. 0.05 0.12 0.38 0.41 0.12
Statistic
of observed
data
Arithm. mean 2796 617 0.23 75 22

Mode 3000 450 0.25 110 0
Stand. Dev. 907 480 0.18 57 37
Skewness 0.01 0.37 0.08 0.61 0.62
or circular nonnal models at the required 0.05 confidence level. The statistics
summarizing the fitted models are listed in the upper half of Table 7.26.

Probability Tables
The Cu-Mo--Au contact metasomatic deposits are considered as orientated
dipping slabs with elliptical horizontal sections, for detection purposes. One kind
of probability table is constructed to provide the probabilities of detection and
confinned detection of the deposits by means of three types of field programs
for varying grid spacings. A second kind of table displays the grid spacing and
corresponding coverage cost per unit of area which are required to obtain pre-
specified levels of probability of detection for each type of field program.
The tables are organized into three blocks, one for each type of field pro-
gram. The first block covers airborne geophysical surveys on parallel and square
grids (Tables 7.27-7.30). The second block deals with ground vertical detection
by geophysical surveys or drilling to a depth of 300 feet (Tables 7.31-7.33).
The third block refers to angled drilling detection (Tables 7.34-7.36).

Four tables to be found in pages 179-182 display probabilities of detection
by airborne geophysical surveys. Six additional probability tables cover the
ground detection of the deposits by geophysical surveys and drilling programs
(pages 183-188).
TABLE 7.27
Probabilities of Detection of Contact Metasomatic (Cu-Mo-Au) Deposits of
the North American Cordillera Belt by Airborne Geophysical Surveys on Parallel
Grids.
Elliptical targets with expected major axis ~ L feet
and minor axis ~ B feet
R is defined as the ratio B/L with geometric mean ~ 0.19
S feet L=2206 1273 3180 feet
400 1. 000 1. 000 1. 000

600 0.929 0.941 0.951
800 0.905 0.921 0.934
1000 0.881 0.901 0.917
1200 0.856 0.881 0.901
1400 0.830 0.860 0.883
1600 0.803 0.839 0.866
1800 0.774 0.817 0.848
2000 0.743 0.794 0.830
2200 0.707 0.770 0.811
2400 0.678 0.743 0.792
2600 0.626 0.714 0.771
2800 0.581 0.698 0.750
3000 0.542 0.651 0.726
3200 0.509 0.610 0.726
3400 0.479 0.575 0.683
3600 0.452 0.543 0.645
3800 0.428 0.514 0.611
4000 0.407 0.488 0.581
4200 0.387 0.465 0.553
4400 0.370 0.444 0.528
4600 0.354 0.425 0.505
4800 0.339 0.407 0.484
5000 0.325 0.391 0.465
5200 0.313 0.376 0.447
180 CHAPTER SEVEN
TABLE 7.28
Contact Metasomatic (Cu-Mo-Au) Deposits of the North American Cordillera


0.10 17520 $ 161.10

0.15 12060 $ 170.65
0.20 9180 $ 180.26
0.25 7420 $ 189.81
0.30 6220 $ 199.42
0.35 5360 $ 208.96
0.40 4700 $ 218.64
0.45 4200 $ 228.00
0.50 3780 $ 237.78
0.55 3460 $ 246.82
0.60 3160 $ 256.96
0.65 2920 $ 266.58
0.70 2640 $ 280.00
0.75 2280 $ 302.11
0.80 1880 $ 336.60
0.85 1420 $ 400.28
0.90 920 $ 541. 74
0.95 420 $1020.00
TABLE 7.29
Probabilities of Confirmed Detection of Contact Metasomatic (Cu-Mo-Au)
veys on Parallel Grids.


Grid Spacing l.c.l. geom. mean u.c.l.
S feet L=2206 2648 3180 feet
350 1.000 1.000 1.000

500 0.941 1.000 1.000
650 0.923 0.936 0.947
800 0.897 0.921 0.935
950 0.690 0.906 0.922
1100 0.546 0.730 0.910
1250 0.440 0.597 0.795
1400 0.362 0.495 0.667
1550 0.304 0.417 0.566
1700 0.260 0.355 0.486
1850 0.228 0.307 0.421
2000 0.205 0.269 0.368
2150 0.192 0.239 0.325
2300 0.000 0.217 0.289
2450 0.000 0.201 0.260
2600 0.000 0.191 0.237
2750 0.000 0.000 0.218
2900 0.000 0.000 0.204
3050 0.000 0.000 0.194
3200 0.000 0.000 0.000
182 CHAPTER SEVEN
TABLE 7.30
Probabilities of Detection of Contact Metasomatic (Cu-Mo-Au) Deposits of
the North American Cordillera Belt by Airborne Geophysical Surveys on Square
Grids.

S feet L=2206 2648 3180 feet
2200 1. 000 1. 000 1. 000

2400 0.919 1.000 1. 000
2600 0.875 1. 000 1. 000
2800 0.834 0.934 1.000
3000 0.796 0.897 1. 000
3200 0.760 0.861 0.960
3400 0.727 0.828 0.928
3600 0.697 0.796 0.897
3800 0.668 0.766 0.867
4000 0.642 0.738 0.839
4200 0.618 0.712 0.812
4400 0.595 0.687 0.786
4600 0.574 0.664 0.762
4800 0.554 0.642 0.738
5000 0.536 0.622 0.716
5200 0.518 0.603 0.696
TABLE 7.31
Probabilities of Detection of Contact Metasomatic (Cu-Mo-Au)

L.c.l. ( geometric mean ( u.c.l., and minor axis = B feet,
strike line.
Spacing S L.c.l. G. mean U.c.l.
in feet L = 2206 2648 3180ft
600 1.000 1. 000 1.000

700 0.989 1. 000 1.000
800 0.933 0.997 1.000
900 0.824 0.969 1.000
1000 0.700 0.900 0.991
1200 0.501 0.700 0.901
1400 0.371 0.529 0.735
1600 0.284 0.408 0.580
1800 0.224 0.323 0.464
2000 0.182 0.262 0.377
2400 0.126 0.182 0.262
2800 0.093 0.133 0.192
3200 0.071 0.102 0.147
3600 0.056 0.081 0.116
4000 0.045 0.065 0.094
184 CHAPTER SEVEN
TABLE 7.32
tion of Contact Metasomatic (Cu-Mo-Au) Deposits of the North Amer-
ican Cordillera Belt by Vertical Drilling to a Depth of 300 Feet.


0.05 4250 30.44

0.15 2650 54.09
0.25 2050 77.14
0.35 1750 97.31
0.45 1550 117.03
0.55 1450 129.81
0.65 1250 164.37
0.75 1150 188.25
0.85 1050 218.79
0.95 950 258.82
TABLE 7.33
(Cu-Mo-Au) Deposits of the North American Cordillera Belt by
Ground Geophysical Surveys or Vertical Drilling to a Depth of 300
Feet.

L.c.1. (geometric mean (u.c.1., and minor axis = B feet,
strike line.

Spacing S L.c.1. G. mean U.C.1.
in feet L = 2206 2648 3180ft
400 1. 000 1. 000 1. 000

450 0.978 1. 000 1. 000
500 0.937 0.999 1.000
550 0.884 0.972 1. 000
600 0.820 0.937 0.999
700 0.549 0.843 0.950
800 0.385 0.698 0.879
1000 0.187 0.343 0.562
1200 0.082 0.187 0.344
1400 0.035 0.095 0.209
1600 0.014 0.047 0.120
186 CHAPTER SEVEN
TABLE 7.34
Probabilities of Detection of Contact Metasomatic (Cu-Mo-Au)
of the North American Cordillera Belt by 55-Degree Angled Drill-
ing to a Vertical Depth of 300 Feet.

L.c.l. ( geometric mean ( U.c.l., and minor axis = B feet.
strike line.
Spacing S L.c.1. G. mean U.c .1.
in feet L = 2206 2648 3180 ft
700 1.000 1. 000 1.000

800 0.999 1.000 1.000
900 0.975 1.000 1. 000
1000 0.886 0.996 1. 000
1200 0.675 0.887 0.996
1400 0.507 0.708 0.918
1600 0.388 0.557 0.766
1800 0.307 0.442 0.629
2000 0.248 0.358 0.516
2400 0.173 0.249 0.359
2800 0.127 0.183 0.263
3200 0.097 0.140 0.202
3600 0.077 0.110 0.159
4000 0.062 0.089 0.129
TABLE 7.35
of Contact Metasomatic (Cu-Mo-Au) Deposits of the North American Cor-
dillera Belt by 55-Degree Angled Drilling to a Vertical Depth of 300 Feet.


0.15 3050 55.04

0.25 2350 77 .69
0.35 2050 94.17
0.45 1850 109.37
0.55 1650 129.83
0.65 1550 142.87
0.75 1350 177.39
0.85 1250 200.67
0.95 1150 229.82
188 CHAPTER SEVEN
TABLE 7.36
Probabilities of Confirmed Detection of Contact Metasomatic Deposits
(Cu-Mo-Au) of the North American Cordillera Belt by 55-Degree Angled Drill-

L.c.l. ( geometric mean( U.c.l., and minor axis = B feet
strike line.
probability of confirmed detection

in feet L = 2206 2648 3180 feet
600 1. 000 1. 000 1. 000

700 o • 847 1. 000 1. 000
800 0.633 0.895 1. 000
900 o • 445 0.757 0.943
1000 0.323 0.545 0.856
1200 0.154 0.324 o • 546
1400 0.067 0.176 0.356
1600 0.027 0.090 0.219
1800 0.009 0.043 0.127
2000 0.002 0.109 0.071
2200 0.000 0.007 0.038
2500 0.000 o • 001 0.013
7.6. PROBABILITY OF DETECTION OF W-Mo CONTACT


7.6.1.1. Introduction. Tungsten-bearing metasomatic deposits occur within
a number of well-defined metallogenic districts of the world. The earliest known
are associated with Paleozoic foldbelts, but Mesozoic fold belts are the most
prolific setting for these deposits, and tungsten-rich skarn deposits are well
represented in the North American Cordillera.
Scheelite is the principal tungsten-bearing mineral occurring in contact me-
tasomatic deposits, whereas wolframite is the most common W-rich mineral in
pegmatitic deposits. In the contact metasomatic deposits, scheelite is generally

restricted to very fine-grained exoskarn referred to as "tactites." The tactites form
around the apex of small stocks or apophyses of larger stocks of intermediate
composition (grano-diorite, monzonite) resulting from the contamination of the
more acidic intrusive phases by carbonate country rocks.
The silicate gangue of the ore is made up of a very fine-grained admixture
of Ca-rich pyroxene and garnets, with secondary quartz and minor fluorite. The
economic mineralization consists of scheelite with minor wolframite and sulfides.
The latter include pyrite, chalcopyrite, molybdenite, galena, and sphalerite. The
oxide phase, magnetite and specularite, is usually present, but is far less abundant
than in other contact metasomatic deposits.
7.6.1.2. Canadian Section of the Cordillera. The northern end of the
Canadian Cordillera, in the Yukon Territory, is well endowed with tungsten-
skarn deposits.(3) Cathro lists a total of 22 scheelite-skarn occurrences in the
Selwyn Tungsten Belt, which extends over a length of 400 miles and a width
of 50 miles, northwest of the town of Whitehorse. The scheelite mineralization
occurs in dark green skarns made up of tremolite, actinolite, epidote, and garnet.
The skarn alteration affects Cambrian limestone formations, while overlying
Silurian to Mississippian pelitic to clastic formations are altered into hornfels in
the vicinity of small discordant stocks of Cretaceous quartz monzonite. The
scheelite deposits are generally confined to within half a mile from intrusive
contacts, and are of high grade but small size. However, the Tungsten Belt has
yielded so far two major deposits. One of them, the Cantung deposit, was
discovered in 1962;(23) it includes two scheeelite-rich horizons, one of which
was amenable to open-pitting. The second one, the Mactung deposit, some 100
miles away, was discovered more recently. It is made up of four skarn horizons
mineralized with scheelite, pyrrhotite, and chalcopyrite, separated by pyritized
hornfels.
Much farther south, lies a smaller scheelite-skarn district, in the Salmo
area of south central British Columbia, which was the main source of tungsten
ore for the Canadian war effort of the 1940-1945 periodY·19) The Emerald,
Dodger, and Feeney W-ore deposits occur along a mile-long trough occupied
by Cambrian limestone resting on Precambrian quartzites and overlain by Silurian
argillite, which dip easterly against the subvertical greizenized contact of a
Jurassic granitic stock. Some sections of the trough are also mineralized with
lead-zinc-silver ore (Jersey and Emerald zones), as already mentioned in Section
7.4.1.2.
7.6.1.3. U.S. Cordillera Section. There are as many as 30 scheelite
mineralized districts in the U.S. section of the Cordillera, which are scattered
190 CHAPTER SEVEN
across nearly all western states but Oregon. The geographic breakdown is as
follows: California: 11 districts; Nevada: 8 districts; Utah: 6 districts; the re-
maining five being shared between the states of Idaho and Washington yO) The
most productive belt extends through western California and Nevada over a
length of 700 miles and a width of 200 miles. The main commercial districts of
the tungsten belt are those of Atolia, Bishop, Fresno, and Kern in California(12)
and the Mill City district in Nevada.
The main geological setting for the scheelite deposits is that of skarnified
carbonate formations included in roof pendants which were engulfed by the large
quartz monzonite batholiths of the Sierra Nevada. Generally, the intrusive is
contaminated into quartz diorite in the vicinity of the pendants, and the contacts
are greizenized in the vicinity of mineralized skarn beds. The scheelite-bearing
skarns exhibit assemblages of andradite, diopside, and epidote with sprinklings
of pyrite, chalcopyrite, galena, and sphalerite. Minor molybdenite is known to
occur among nests of fibrous amphiboles, very close to the intrusive contact.
The Pine Creek deposits, among the largest in the world, are located in the
Kern district. (20) They produced a total of 7 million tons of ore from six ore
zones stretching through the "Tungsten Hills" of central-western California. The
Mill City district of Nevada was also an important producer during the war years.
Farther northeast, in Idaho, lies the important district of Yellow Pine, which
includes 20 known occurrences of scheelite-bearing skarns within an area 200
miles long and 100 miles wide.
7.6.2. Statistical Modeling of Geometric Parameters of W-Mo

Contact Metasomatic Deposits
A total of 37 deposits were selected and included in the database to represent
the whole population of W-Mo contact metasomatic deposits of the Cordillera.
The deposits are listed by name and by region of occurrence in Table 7.37. The
geographic breakdown is as follows: Idaho: 1 deposit; Utah: 7 deposits; Cali-
fornia: 7 deposits; Nevada: 16; and British Columbia-Yukon: 6 deposits.
deposits. The parameters are measured on all 37 deposits of the sample, and the
measurements are grouped into classes and assembled into frequency distribu-
tions. The distributions are summarized by means of statistics which are listed
in the lower half of Table 7.38.
Lognormal models were fitted to the observed distributions of the dimen-
sional parameters at the 0.05 confidence level, but the distributions of attitudinal
TABLE 7.37
list of Contact Metasomatic (W-Mo) Deposits
of the North American Cordillera Belt Included
in the Database.
Utah (U.S.A.) Nevada (Cont'd)

Old Hickory Sutton
Garnet Humbolt
Strategic Metals Oreana
Cupric victory
Copper Ranch T.N.T.
Copper King Riley
Daily Metal Kirby
Valley View
Idaho (U.S.A.) Tip Top
Yellow Pine Granite Creek
Pacific
California (U.S.A.) Alpine
Darwin Hill Tempiute
St. Charles British Columbia
Durham & Yukon (Canada)
Little Sister
Lookout Mactung
Round Valley Cantung
Pine Creek Emerald
Dodger
Nevada (U.S.A.) Feeney
St. Anthony Swakum Mountain
Rose Creek
Friedman
parameters could not be fitted satisfactorily by normal or circular normal models

at the required 0.05 confidence level. The statistics summarizing the fitted dis-
tributions are listed in the upper half of Table 7.38.

Probability Tables
As is the case for the other three types of contact metasomatic deposits,
the W-Mo contact metasomatic deposits of the Cordillera are considered as
oriented dipping slabs with elliptical horizontal section for detection purposes.
One kind of table is constructed to provide probabilities of detection or confirmed
detection of the W-Mo deposits by means of two types of ground field programs.
No tables are provided for airborne geophysical detection since no such methods
192 CHAPTER SEVEN
TABLE 7.38
Statistical Modeling Summary for Contact Metasomatic (W-Mo) Deposits of
Statistic Length of Breadth of Shape Str ike dir. Dip in

of fitted horiz. sect. horiz. sect. ratio from T. North in degrees
25%ile 500 25 0.04 0 25

95% L.c.1. 598 33 0.05 27 42
Mean 724 47 0.06 36 49
95% U.c.1. 875 66 0.09 55 57
75%ile 1050 67 0.13 70 65
Disp. coef. 0.93 0.31 0.21 0.15 0.12
Statistic of
observed
data
Arithm. mean 880 93 0.12 41 49

Mode 800 40 0.04 0 60
Stand. dev. 633 90 0.06 45 25
Skewness 0.13 0.42 1. 33 0.91 0.43
are available in the present state of the technology. A second kind of table
displays the grid spacings and corresponding coverage cost per unit of area which
are required to obtain prespecified levels of detection probabilities.
The tables are organized into two blocks. The first one covers ground vertical
detection by geophysical surveys and drilling to a depth of 300 feet (Tables
7.39-7.41) and the second block deals with angled drilling detection (Tables
7.42-7.44).

The reader will find a series of six tables displaying the probabilities of
detection of contact metasomatic (W-Mo) deposits of the Cordillera by ground
geophysical surveys and drilling programs in pages 193-197.
TABLE 7.39
Probabilities of Detection of Contact Metasomatic (W-Mo) De-

strike line.
Spacing S L.C .1. G. mean U.c.1.
in feet L = 598 724 875ft
200 1.000 1.000 1. 000

300 0.792 0.978 1. 000
400 0.474 0.679 0.897
500 0.303 0.445 0.640
600 0.211 0.309 0.451
700 0.155 0.227 0.331
800 0.118 0.174 0.254
1000 0.076 0.111 0.162
1200 0.053 0.077 0.113
1400 0.039 0.057 0.083
1600 0.030 0.043 0.063
2000 0.019 0.028 0.041
194 CHAPTER SEVEN
TABLE 7.40
tion of Contact Metasomatic (W-Mo) Deposits of the North American
Cordillera Belt by Vertical Drilling to a Depth of 300 Feet.


0.05 1400 137.17

0.15 900 283.73
0.25 700 438.91
0.35 600 577.42
0.45 500 803.19
0.55 450 974.35
0.65 410 1157.27
0.75 390 1269.95
0.85 350 1554.43
0.95 310 1953.14
TABLE 7.41
(W-Mo) Deposits of the North American Cordillera Belt by Ground

strike line.

in feet L = 598 724 875 ft
150 1. 000 1. 000 1. 000

200 0.813 1. 000 1. 000
250 0.438 0.771 0.955
300 0.236 0.449 0.779
350 0.115 0.273 0.494
400 0.052 0.154 0.333
450 0.022 0.083 0.212
500 0.007 0.042 0.130
550 0.001 0.050 0.077
600 0.000 0.008 0.044
800 0.000 0.000 0.002
196 CHAPTER SEVEN
TABLE 7.42
Probabilities of Detection of Contact Metasomatic (W-Mo) De-
posits of the North American Cordillera Belt by 55-Degree Angled
Drilling to a Vertical Depth of 300 Feet.

L.c.l. ( geometric mean ( U.c.l., and minor axis = B feet,
strike line.
in feet L = 598 724 875 ft
300 1.000 1.000 1. 000

400 0.900 1.000 1.000
500 0.628 0.863 1.000
600 0.437 0.638 0.871
700 0.321 0.470 0.681
800 0.246 0.360 0.526
1000 0.157 0.231 0.337
1200 0.109 0.160 0.234
1400 0.080 0.118 0.172
1600 0.061 0.090 0.132
2000 0.039 0.058 0.084
TABLE 7.43
tion of Contact Metasomatic (W-Mo) Deposits of the North American
Feet.
Survey Design: Square Grid With Spacings S by S feet.

Unit Cost = $7325 Per Dj.amond Drill-Hole.

0.15 1300 188.40

0.25 1000 289.81
0.35 900 346.39
0.45 800 424.21
0.55 750 480.02
0.65 700 535.83
0.75 650 611. 00
0.85 600 704.93
0.95 500 980.56
TABLE 7.44
(W-Mo) Deposits of the North American Cordillera Belt by 55-
L.c.1. ( geometric mean ( u.c.1., and minor axis = B feet,
strike line.

in feet L = 598 724 875 ft
250 1. 000 1. 000 1.000

300 0.961 1.000 1.000
350 0.384 1. 000 1.000
400 0.192 0.570 1. 000
450 0.091 0.285 0.898
500 0.035 0.159 0.432
550 0.007 0.085 0.269
600 0.000 0.039 0.166
800 0.000 0.000 0.008
198 CHAPTER SEVEN

FOUR TYPES OF CONTACT METASOMATIC DEPOSITS OF THE
NORTH AMERICAN CORDILLERA
7.7.1. Introduction
Three strategies are considered for the detection of four types of contact
metasomatic deposits in the North American Cordillera, namely, the liminal,
maximal, and optimal options, as described previously in Section 5.5 of Chapter
5.
The liminal strategy is designed to ensure a minimum level of 0.500 of
detection probability at a relatively low cost for the coverage of large areas of
little known potential. The maximal strategy ensures a near certain detection
(probability level of 0.980); but, because of its high cost, the maximal option
should be considered only for the coverage of small areas of very high potential.
The optimal strategy, a compromise option bracketed by the first two, should
be most attractive in the majority of exploration situations, when dealing with
areas of moderate size and fair to good potential.
It is most useful for planning purposes to compare the detection perfor-
mances of various types of field programs or grid designs. When the probability
of detection is held at a specified level, as is the case for the first two options,
the coverage cost per unit of area is a convenient yardstick for the comparisons.
When dealing with the optimal option, however, we have to use the expected
loss criterion, as described in Section 6.5.4 of Chapter 6, because the level of
optimal probability of detection varies from one type of field program to another.
The detection characteristics associated with the three strategies are listed
in Tables 7.45, 7.46 and 7.47, covering the four types of contact metasomatic
deposits of the Cordillera.

7.7.2.1. Cost Multiplier for Confirmed Detection. For planning pur-
poses, we need to know what additional cost is required to obtain confirmed
detection instead of single detection. For airborne geophysical surveys, the cost
multiplier is the highest (2.3) for the smaller deposits (Pb-Zn-Cu-Ag type) and
the lowest (1.7) for the larger deposits (Cu-Mo-Au type). The same range of
variation of the cost multiplier applies when dealing with ground geophysical
surveys. The cost multiplier for vertical and angled drilling programs varies from
2.2 for the larger deposits (Cu-Mo--Au type) to 2.8 for the smaller ones (W-Mo
type).
TABLE 7.45
Liminal Detection Strategy: Summary of Detection Characteristics for Four Types of Contact Metasomatic Deposits
of the North American Cordillera Belt.
Cu-Fe type Pb-Zn-Cu type Cu-Mo type W-Mo type

Coverage cost: Coverage cost: Coverage cost: Coverage cost
US$/ mile sq. US$/mile sq. US$/mile sq. US$/mile sq.
Type of survey Grid size: feet Grid size: feet Grid size: feet Grid size: feet
and Confirmed Confirmed Confirmed Confirmed
grid geometry Detection detection Detection detection Detection detection Detection detection
Airborne $388 $830 $425 $985 $255 $448 - -

geophysical 1500 f t 550 ft 1300 ft 460 ft 3200 ft 1200 ft
parallel grid
Airborne $436 - $470 - $280 - - -

square grid
Ground $1675 $3200 $2500 $4950 $825 $1408 $3198 $6690

geophysical 630 ft 375 ft 450 ft 275 ft 1200 ft 725 f t 375 ft 230 ft
square grid
Vertical $538,000 $1,380,000 $974,000 $1,815,000 $175,000 $414,000 $1,382,000 $3,723,000

drilling 630 ft 375 ft 450 ft 275 ft 1200 ft 725 f t 375 ft 230 ft
square grid
55-degree $460,000 $1,110,000 $634,000 $1,480,000 $166,400 $357,000 $898,750 $2,480,000

angled drilling 770 ft 450 ft 650 ft 400 ft 1400 ft 900 f t 550 ft 325 ft
square grid
- - - - -- -- --
TABLE 7.46
Maximal Detection Strategy: Summary of Detection Characteristics for Four Types of Contact Metasomatic Deposits
of the North American Cordillera Belt.
I
Coverage cost: Coverage cost: Coverage cost: Coverage cost:
US$/mile sq. US$/mile sq. US$/mile sq. US$/mile sq.
I
Type of survey Grid size: ft Grid size: ft Grid size: ft Grid size: ft
and Confirmed Confirmed Confirmed Confirmed
grid geometry Detection detection Detection detection Detection detection Detection detection
Airborne $1525 - $1990 - $910 - - -

parallel grid
Airborne $880 - $975 - $465 - - -

square grid
Ground $2915 $5500 $5590 $10,000 $1355 $2345 $6690 $11,105

geophysical 400 ft 260 ft 250 ft 175 ft 750 ft 475 ft 230 ft 160 ft
square grid
Vertical $1,211,000 $2,910,000 $2,940,000 $6,123,000 $388,800 $888,000 $3,723,000 $7,800,000

drilling 400 ft 260 ft 250 ft 175 ft 750 ft 475 ft 230 ft 160 ft
square grid
55-degree $900,000 $2,480,000 $1,286,000 $3,480,000 $323,000 $705,000 $1,520,000 $3,680,000

angled drilling 520 ft 350 ft 450 ft 300 ft 925 ft 600 ft 420 ft 275 ft
square grid
- ---- - ----
7.7.2.2. Comparison of Airborne Grid Designs. A comparison of the

detection perfonnance of the parallel and square designs for the control grids of
airborne geophysical surveys is of interest for planning purposes. In all three
cases (Cu-Fe, Pb-Zn-Cu, Cu-Mo-Au types) there is no incentive in switching
from the parallel design to the square design, because choosing the latter would
lead to an average increase of coverage cost of 10%.
7.7.2.3. Comparison of Drilling Programs. When dealing with dipping
targets such as the four types of contact metasomatic deposits, it is most appro-
priate to compare the merits of vertical drilling programs against angled drilling
to the same vertical depth, because of the very high costs involved in detection
by drilling. In all four cases, the 55-degree angled detection should be preferred
to vertical drilling, because it leads to a substantial saving in coverage costs.
The saving varies from 10% for the larger deposits (Cu-Mo-Au) to as much as
36% for the smaller deposits (W-Mo type).

7.7.3.1. Cost Multiplier for Confirmed Detection. The cost multi-
pliers required to obtain confinnation of single detection in the context of the
maximal strategy are nearly the same as the ones listed above for the liminal
option for the first three types of deposits. But in the case of the smaller deposits
(W-Mo type), all maximal cost multipliers are lower than their liminal coun-
terparts by some 20%-30%, thus making the confinnation requirement more
attractive in the maximal situation than in the liminal one.
7.7.3.2. Comparison of Airborne Grid Designs. Contrary to the lim-
inal situation, it is advantageous in all three cases (Cu-Fe, Pb-Zn-Cu, and
W-Mo-Au types) to shift from the parallel to the square grid design, as un-
derscored by the very substantial reduction in coverage cost per unit of area
(average 46%) attained when using the square design.
7.7.3.3. Comparison of Drilling Programs. For all four types of contact
metasomatic deposits, good planning will require the selection of angled over
vertical drilling programs because of the substantial reduction in coverage costs
incurred. The reduction is only modest (17%) for the larger deposits (Cu-Mo-Au
type), but is very attractive (60%) for the detection of the smaller deposits (W-Mo
type).

7.7.4.1. Introduction. Design optimization based on the efficiency cri-
terion was carried out for each type of field program for the detection of the
TABLE 7.47
Optimal Detection Strategy: Summary of Detection Characteristics for Four Types of Contact Metasomatic Deposits of
North American Cordillera Belt.

Type of survey Coverage cost: Detection Coverage cost: Detection Coverage cost: Detection Coverage cost: Detection
and US$/mile sq. proba- US$/mile sq. proba- US$/mile sq. proba- US$/mile sq. proba-
grid geometry Grid size: ft bility Grid size: ft bility Grid size: ft bility Grid size: ft bility
Airborne $551 0.82 $602 0.82 $345 0.82 - -

parallel grid
Airborne $599 0.84 $617 0.83 $428 0.98 - -

square grid
Ground $1946 0.92 $2190 0.75 $990 0.90 $3483 0.84

geophysical 550 ft 500 ft 1000 ft 350 ft
square grid
Vertical $675,000 0.92 $803,000 0.75 $237,000 0.90 $1,554,000 0.84

drilling 550 ft 500 ft 1000 ft 350 ft
square grid
55-degree
angled $536,000 0.92 $705,000 0.93 $247,000 0.98 $981,000 0.86
square grid
Optimal 45-
degree angled $573,000 0.96 $754,000 0.96 $265,000 0.99 $1,581,000 0.98
square grid
four types of contact metasomatic deposits of the Cordillera. The parameters of

optimized airborne geophysical surveys, ground geophysical surveys, and ver-
tical and angled drilling programs are listed in Tables 7.48-7.50 for the Cu-Fe
type; Tables 7.51-7.53 forthePb-Zn-Cu type; Tables 7.54-7.56 for the Cu-Mo-Au
type, and finally Tables 7.57-7.58 for the W-Mo type. All parameters are also
assembled for the readers' convenience in one summary tabulation (Table 7.47).
7.7.4.2. Comparison of Airborne Survey Grid Designs. Switching
from the parallel to the square grid design results in a modest reduction of the
expected loss criterion, averaging 10% for all three types of deposits, as in the
instance of the liminal option.
7.7.4.3. Comparison of Drilling Programs. Optimally angled drilling
is vastly superior to vertical drilling for all four types of deposits. The reduction
in the expected loss criterion varies from 60% for the Cu-Fe type to as much
as 86% for the smaller deposits (W-Mo type).
7.7.4.4. Optimization Tables. The reader will find below in pages 203-209
a series of eleven tables covering the optimization of the search for four types
of contact metasomatic deposits of the Cordillera by airborne and ground pro-
grams.
TABLE 7.48
Optimal Design of Airborne Geophysical Surveys for the Detection of Contact
Metasomatic (Cu-Fe) Deposits of the North American Cordillera Belt.
respectively
Unit Cost: $70/l.ml.
(i) Survey Design:

Parallel Grid
Optimal grid orientation: 90 degrees to expected strike
(ii) Survey Design:

Square Grid
TABLE 7.49
Optimal Design of Ground Geophysical Surveys for the Detection of Contact
Metasomatic (Cu-Fe) Deposits of the North American Cordillera Belt.
Survey Design: Square Grid With Spacing S by S Feet.

Confidence Interval: All results are reported as 95% confjdence intervals
respectively.

Cost in $ Per Mile Square: 2189 1946 1352
Probability of Detection: 0.73 0.92 0.86
Optimal Grid Orientation: 30 degrees + or -10 to expected strike
direction of target.
TABLE 7.50
Optimal Designs for the Detection of Contact Metasomatic (Cu-Fe) Deposits
of the North American Cordillera Belt by Drilling to a Vertical Depth of 300
Feet.

respectively
Unit Cost = $20 per linear ft.
(a) Vertical Drilling:

Optimal grid orientation: 30 degrees to expected strike of target
(b) Angled drilling at 55 degrees:

(c) Angled drilling at optimal angle = 50 degrees:

TABLE 7.51
Metasomatic (pb-Zn-Cu) Deposits of the North American Cordillera Belt.

respectively
Unit Cost: $70/1.ml.
(i) Survey Design:

Parallel Grid
(ii) Survey Design:

Square Grid
TABLE 7.52
Metasomatic (Pb-Zn-Cu) Deposits of the North American Cordillera Belt.

respectively.
unit Cost: $6/Station

206 CHAPTER SEVEN
TABLE 7.53
Optimal Designs for the Detection of Contact Metasomatic (Pb-Zn-Cu) De-
posits of the North American Cordillera Belt by Drilling to a Vertical Depth of
300 Feet.

respectively
(a) Vertical drilling:


Optimal grid orientation: 45 degrees with expected strike of target

TABLE 7.54
Metasomatic (Cu-Mo-Au) Deposits of the North American Cordillera Belt.

respectively
Unit Cost: $70/1.ml.
(i) Survey Design:

Parallel Grid
(ii) Survey Design:

Square Grid
Probability of detection: 0.95 0.98 1. 00
TABLE 7.55
Metasomatic (Cu-Mo-Au) Deposits of the North American Cordillera Belt.

Confidence Intervals: All results are reported as 95% confidence intervals
respectively.

Optimal Grid Orientation: 30 degrees + or -10 to expected strike direction
of target.
TABLE 7.56
Optimal Designs for the Detection of Contact Metasomatic (Cu-Mo-Au) De-
posits of the North American Cordillera Belt by Drilling to a Vertical Depth of
300 Feet.
respectively
Unit cost = $20 per linear ft.



TABLE 7.57
Metasomatic (W-Mo) Deposits of the North American Cordillera Belt.
Survey Design: Square Grid with Spacing S by S.Feet.

Confiddence Interval: All results are reported as 95% confidence intervals
respectively.

Cost in $ Per Mile Square 4308 3483 2914
Probability of Detection 0.79 0.84 0.90
TABLE 7.58
Optimal Designs for the Detection of Contact Metasomatic (W-Mo) Deposits
Feet.

respecti vely
unit Cost = $20 per linear ft.

(b) Angled drilling at an angle of 55 degrees:

(c) Angled drilling at the optimal angle = 50 degrees:

Probability of detection: 0.80 1.00 1. 00
I. BALL. C. W., 1951, The Emerald, Feeney, and Dodger tungsten ore-bodies, Salmo, British
Columbia, Econ. Geol. 49, 625-638.
2. BUTLER, B. S., 1913, Geology and ore deposits of the San Francisco and adjacent districts,
Utah, U.S. Geol. Surv. Prof. Pap. No. 80.
3. CATHRO, R. J., 1969, Tungsten in the Yukon, West. Miner 42(4),23-40.
4. COMPTON, R. R., 1960, Contact metamorphism in the Santa Rosa Range, Nevada, Bull.
Geol. Soc. Am. 71, 1383-1416.
5. EINAUDI, M. T., 1982, General features, description and origin of skarns associated with
porphyry--<:opper plutons, in Advances in the Geology of Porphyry-Copper Deposits of South-
western North America, pp. 139-210, University of Arizona Press, Tucson.
6. EVANS, A. M., 1980, An Introduction to Ore Geology, Chap. II, Blackwell, Oxford.
210 CHAPTER SEVEN
7. GHOSE, N. C., 1966, Behaviour of trace elements during thermal metamorphism and/or gran-
itization of metasediments and basic igneous rocks, Geol. Rundsch. 55, 608-617.
8. GHOSE, N. c., 1970, Geochemistry of thermal metamorphic and granitization processes in
aureole rocks around Richughuta, Palamua Dis!., Bihar, India, Geol. Rundsch. 59, 686-724.
9. HOLSER, W. T., 1947, Metasomatic processes, Econ. Geol. 42, 384-395.
10. KERR, P. F., 1946, Tungsten mineralization in the United States, Geol. Soc. Am. Mem. No.
15.
II. KLICHINOV, V. A., and SEGALOVICH, V. I., 1979, The application of geophysics to
exploration forcbromite and tungsten, Geol. Surv. Can. Econ. Geol. Rep. No. 31, pp. 476-484.
12. KNOPF, A., 1917, Tungsten deposits of northwestern Inyo County, California, u.S. Geol.
Surv. Bull. No. 640, pp. 229-249.
13. LINDGREN, W., 1933, Mineral Deposits, Chap. 28, McGraw-Hill, New York.
14. LITTLE, H. W., 1959, Tungsten deposits of Canada, Geol. Surv. Can. Econ. Geol. Ser. No.
17.
15. LOVERING, T. S., 1941, The origin of tungsten ore of Boulder County, Colorado, Econ.
Geol. 36, 229-279.
16. LOVERING, T. S., 1949, Rock alteration as a guide to ore-East Tintic District, Utah, Econ.
Geol. Monograph No. I.
17. MORRIS, H. T., and LOVERING T. S., 1952, Supergene and hydrothermal dispersion of
heavy metals in wall rocks of ore bodies, Tintic District, Utah, Econ. Geol. 47, 685-716.
18. RAMBERG, H., 1952, The Origin of Metamorphic and Metasomatic Rocks, The University
of Chicago Press, Chicago.
19. RENNIE, C. c., and SMITH, T. S., 1957, Lead-zine and tungsten ore bodies of Canadian
Exploration Ltd., Salmo, B.C., Structural Geology of Canadian Ore Deposits, Sixth Com-
monwealth Congress, Montreal, pp. 116-123.
20. RIDGE, J. D., (Ed.) 1968, Ore Deposits of the United States, Vols. I and 2, A.I.M.E., New
York.
21. SANGSTER, D. F., 1969, The contact metasomatic magnetite deposits of Southwest British
Columbia, Geol. Suv. Can. Bull. No. 172.
22. SMIRNOV, D. F., 1976, Geology of Mineral Deposits, Chap. 7 (translated by H. C. Creighton),
MIR Publishers, Moscow.
23. WHITE, L. G., 1963, The Canada Tungsten property, Flat River Area, N.W.T., Canada, Can.
[nst. Min. Metall. Bull. 56(614), 390-393.
24. YOUNG, G. A., and UGLOW, W. L., 1926, The iron ores of Canada, Vol. I, British Columbia
and the Yukon, Geol. Surv. Can. Econ. Geol. Ser. No.3.
CHAPTER EIGHT
DETECTION OF Ni-Cu ULTRAMAFIC

DEPOSITS OF THE NORTH AMERICAN
SHIELD BY OPTIMIZED GEOPHYSICAL
SURVEYS AND DRILLING PROGRAMS
8.1.1. Main Geological Settings

According to Ross and Travis, (18) there are three main tectonic settings for
the sulfide nickel deposits of the World: (1) stable platforms, chiefly of Protero-
zoic age, (2) remnants of island arcs and mobile belts, mainly of Archean age,
referred to as "Greenstone Belts," and (3) Paleozoic and Mesozoic fold belts.
The first two types of setting are well represented in the North American Pre-
cambrian Shield; the third one is found along the North American Cordillera
Belt. Ross and Travis(l8) and Naldrett(l6) recognize three genetic and lithologic
groups of sulfide nickel deposits throughout the world. The first two are associated
with layered complexes and undifferentiated intrusives, while the third one has
mafic volcanic affiliations. All three are known to occur in North America.
8.1.2. Intrusive Layered Complexes

Most of the world's primary nickel ore produced between the turn of the
century and the rnid-1950s was gained from large layered mafic-ultramafic in-
trusive complexes occurring in platforms of Proterozoic age. Mineralization
commonly consists of low-grade disseminated sulfide ore which occurs in the
basal portion of mafic layers in preference to the ultramafic layers of the com-
plexes. Notable exceptions are known in the North American Shield, including,
for examples, the famous Sudbury Basin so well described in many publications,
and the more recently discovered gabbro-peridotite layered complex at Lynn
Lake, Manitoba, which has yielded some ore of grade higher than the average
for the group.
211
212 CHAPTER EIGHT
B.1.3. Undifferentiated Mafic Intrusives

A substantial number of sulfide nickel ore deposits generally of moderate ton-
nage and grade have been discovered in the North American Shield during the 1950s
and the 1960s in association with undifferentiated mafic intrusives. The deposits
discovered in the provinces of Manitoba and Ontario and in the northwestern por-
tion of the province of Quebec are of Archean age, while the Ungava deposits of New
Quebec are of early Proterozoic age. The lithology of the undifferentiated mafic in-
trusives varies from basic to ultrabasic. In relation to wallrock formations, they as-
sume two main types of geometry: conformable, either continuous (sills) ordiscon-
tinuous (lenses), and forcibly intrusive (dykes and plugs). (I)
Because of their economic importance as a major new source of sulfide
nickel ore, the large deposits of the Mystery-Moak-Thompson Lake Belt dis-
covered in eastern Manitoba in the late 1950s deserve a special mention among
the undifferentiated intrusive group. The deposits occur along a "mobile belt"
separating the western portion of the Superior Province from the southeastern
edge of the Churchill Province. The mobile belt is the seat of a strong regional
gravity anomaly which exhibits a complex structure. The anomaly consists of
two positive strips bracketing a branching negative strip which coincides with
the nickel-bearing peridotite belt. (24)
About one half of the total length of 500 miles of the mobile belt has been
prospected to date, and several nickel deposits of commercial and subeconomic
importance have been discovered within the 60-mile-Iong Mystery-Moak-
Thompson Lake Belt, along the Nelson River. One type of deposit occurs mainly
within the Mystery-Moak section of the belt. The mineralization consists of
finely disseminated pyrrhotite and pyrite with minor pentlandite occurring in
strongly serpentinized peridotite sills or lenses. The deposits are in the low-
grade, large-tonnage category and are considered as subeconomic in the present
economic circumstances.
The second type of nickel ore deposits is found in the Thompson Lake
section of the belt. The mineralization is of a much higher grade than that of
the Mystery-Moak section. The setting is strikingly different from that of other
deposits of the undifferentiated deposit group: the ore does not occur within the
peridotite sills, but appears to have been remobilized into the sillimanite gneiss
wallrock. (24,26)
B.1.4. Extrusive Mafic-Ultramafic Suites

Since the early 1960s, a new type of sulfide nickel ore deposits has begun
to make a significant contribution to the World's nickel production. These de-
DETECTION OF Ni-Cu ULTRAMAFIC DEPOSITS 213
posits occur mainly in Archean greenstone belts of the Western Australian and
North American shields and are considered to be of volcanogenic affiliation.
Coad(3) aptly summarizes the results of much of the research work done on this
type of deposits since they were first discovered in Western Australia and sub-
sequently in Canada in the 1960s.
Two types of volcanogenic Ni-Cu mafic-ultramafic deposIts are distin-
guished. The first type occurs in komatiite volcanic suites and comprises two
categories: massive sulfide ore occurring in typical extrusive settings and dis-
seminated ore found mainly in intrusive settings. The second type is found in
tholeiitic volcanic suites, which are generally far less productive than the ko-
matiite series, with a few notable exceptions in the Scandinavian Shield.
The komatiite suites, originally described in the Komati River region of the
Transvaal, are highly magnesian in composition, as expressed by a low PeOI
Mgo + PeO ratio, and show a low Ti02 content, in contrast with the tholeiitic
series. Much effort has been devoted to the study of the stratigraphic distribution
of sulfide ore within the komatiite flows. It was soon recognized that the min-
eralization of economic importance tends to concentrate as pockets, pods, and
lenses of massive sulfides occurring along corrugations, embayments, and flex-
ures of the lower contact of the extrusive bodies. The massive mineralization
commonly grades upward into disseminated sulfides within the middle portion
of the flows. In the intrusive type, however, the bulk of the mineralization is of
a disseminated nature and occurs throughout the central portion of the bodies or
even near the top contact.
B.2.1. Detection Approaches

All successful sulfide nickel exploration programs, particularly the more
recent ones in Western Australia and in the Provinces of Manitoba, Ontario, and
Quebec of Canada, have relied on a combination of direct and indirect ap-
proaches. The direct approach involves the visual detection ofthe mineralization
and associated clues either on the ground or from airborne platforms or by
mechanical means (drilling). The indirect approach, on the other hand, relies on
the ground or aerial detection of geophysical signatures of the target minerali-
zation, or on the ground detection of geochemical signatures provided by primary
or secondary halos.
214 CHAPTER EIGHT
8.2.2. Direct Detection

Mafic-ultramafic layered complexes are the only kind of prospective nickel-
bearing geological structures large enough to be detectable from airborne plat-
forms, either by photogeology techniques from aircraft or by remote multisensor
techniques from satellites. The detection is enhanced by topography, vegetation,
and color contrasts between the layered complexes and the surrounding forma-
tions. A good example, well known to the first writer, is provided by the Bay
of Islands ultrabasic complex of central western Newfoundland.
Direct visual detection from an airborne platform is not helpful in the search
for undifferentiated mafic intrusives and even less so if volcanogenic complexes
are the targets, because of their relatively small extent and lack of visual contrasts.
An integrated approach successfully used in the West Australian and Canadian
shields in the search for sulfide nickel deposits of volcanogenic affiliation calls
for reconnaissance coverage followed by detailed mapping to locate favorable
formations and pinpoint favorable horizons. Guidance is provided by the display
of "spinifex" textures, often directly recognizable in the field, which are generally
associated with environments favorable for the deposition of nickel ore. (3) Map-
ping is complemented by small-scale geophysical and geochemical coverages to
fill the gaps, and by some reconnaissance drilling to test geological targets.
8.2.3. Indirect Geophysical Detection

The indirect geophysical approach has been widely used in the search for
sulfide nickel targets in many areas of the North American and Scandinavian
shields where outcrops are scarce and glacial drift cover quite thick. The geo-
physical environment of most sulfide nickel deposits, particularly those with
volcanogenic affiliations, is quite suitable for the application of the indirect
geophysical approach. Iron sulfides including pyrite and pyrrhotite are the most
widespread mineral associated with nickel-copper mineralizations. Pyrrhotite,
for one, is a good electrical conductor of moderate magnetic susceptibility and
specific gravity. As a result, nickel-copper deposits themselves are up to 1000
times more conductive, 10 times more magnetic, and 30%-50% more dense
than surrounding host rocks. The contrast of physical properties is generally
sufficient to provide good geophysical targets.
Airborne magnetic surveys are very useful to delineate the extent and main
features of basic-ultrabasic intrusive-extrusive complexes at the initial stage of
the search for nickel-copper deposits. But ground follow-up consisting of mag-
netic surveys and geological mapping is required in order to distinguish the
favorable lithologic units from the less prospective ones. The geological inter-
pretation of the surveys is further complicated by widespread occurrences of iron
fonnations which commonly accompany volcanic flows in the greenstone belts

of Western Australia and the North American Shield.
At the next stage of the search, detailed electromagnetic surveys both in
the frequency and time domains are generally conducted from helicopter plat-
fonns in order to test the favorable lithologic units for the presence of sulfides.
Finally, induced polarization surveys are used to pinpoint on the ground and
delineate sulfide zones of possible economic interest for drill testing. However,
the induced polarization has many drawbacks, including slowness and high cost
on the one hand, and lack of discrimination on the other hand. Dowsett(5) de-
scribes the successful use of the gravimetric method to distinguish sulfides from
graphite or serpentinized fractures in areas affected by relatively light and unifonn
glacial cover such as the Chibougamau region of Northwest Quebec. He notes,
however, that marked variations in overburden thickness may generate spurious
anomalies of the same order of magnitude as what could be expected from sulfide
zones.
8.2.4. Indirect Geochemical Detection

Bedrock sampling for the detection of primary halos is a valid approach in
areas of relatively abundant outcrops such as the Abitibi greenstone belts of
northeast Ontario and northwest Quebec in the North American Shield. A quick
and low-cost field test using the chemical dimethylglioxime is of valuable as-
sistance for the screening of favorable environments for nickel mineralization.
But much more sophisticated laboratory techniques are required to detect the
variations of abundance of elements such as MgO and sulfur which point to
favorable environments for nickel deposition.
The detection of secondary halos by residual soil sampling and geobotanical
surveys in the North American Cordillera Belt was described by MillerY3)
Obial(l7) describes similar surveys in the tropical environment of the Philippines.
These techniques have not proved very useful in the glaciated environments of
the subarctic portion of the North American Shield. But their application to the
thinly covered northernmost portions of the Shield, such as the Northwest Ter-
ritories and New Quebec, should not be ruled out.
8.3. STATISTICAL MODELING OF GEOMETRIC PARAMETERS OF

Ni-Cu DEPOSITS

A total of 25 commercial and subeconomic deposits were included in the
database to represent the whole statistical population of known and undiscovered
216 CHAPTER EIGHT
TABLE 8.1
list of Ni-Cu Ultramafic Deposits of the
North American Shield Included in the
Database.
Manitoba Quebec
Rankin Inlet Marbridge
Bird River Blondeau Twp.
Lynn Lake Lake Renzy
Bowden Lake Lorraine
Thompson
Pipe
Ontario New Quebec
Alexo Cross Lake
Sothman Katiniq
Ajax Raglan
Langmuir Expo
Texmont
Trebor
Temagami Lake
Gordon Lake
Werner Lake
Kenbridge
Shebandowan
Ni-Cu ultramafic deposits of the North American Shield. The deposits are listed
by name and region in Table 8.1. An effort was made to obtain the best possible
cross section of the three main types of deposits listed in Section 8. I. The very
large Sudbury deposits were deliberately omitted from the subsample representing
the layered complexes, because their uniqueness would flaw the results of the
statistical study. The remainder of the sample is more or less evenly divided
between the undifferentiated and the volcanogenic types of deposits.

A total of five parameters are required for the calculation of detection
probabilities and the optimization of the search for Ni-Cu deposits of the North
American Shield. Three of the parameters are of a dimensional nature, including
length and breadth of the horizontal section of deposits within the range of
detectors and the shape ratio R = SIL. The other two are attitudinal parameters,
including the unoriented true dip angle in degrees and the oriented strike direction
referred to true North and measured in degrees within the right half-circle.
The range of the parameter measurements collated from the 25 deposits of

the sample is as follows: 500--6100 feet for length, 20--620 feet for breadth, and
60-85 degrees for dip angle. The parameter measurements were grouped into
frequency distributions which are summarized by means of four statistics: arith-
metic mean, mode, standard deviation, and coefficient of skewness. The values
ofthe statistics are listed in the lower half of Table 8.2. The degree of asymmetry
of the distributions of the dimensional parameters and of the strike direction is
moderate, as indicated by the skewness coefficient value of 0.5; but the skewness
is much stronger for the distribution of dip angles (coefficient close to unity).
the dimensional parameters, as indicated by the results of X2 goodness of fit tests
conducted at the 0.05 confidence level (see Chapter 5, Section 5.3.2). A normal
model was also satisfactorily fitted at the same confidence level to the observed
distribution of unoriented dip angles, and a circular normal one was selected for
the oriented strike directions.
TABLE 8.2
Statistical Modeling Summary for Ni-Cu Ultramafic Deposits of the North
American Shield (Sample Size: 25 Deposits).
Statistic Length of Breadth of Shape Str ike direct. Dip in

25%ile 750 25 0.02 32 69

95% L.c.l. 815 41 0.03 37 73
Mean 1097 62 0.06 52 77
95% U.c.l. 1475 95 0.09 67 80
75%ile 2000 120 0.14 85 86
Disp. coef. 0.11 0.25 0.62 0.16 0.05
Statistic
of observed
data
Arithm. mean 1459 114 0.12 54 77

Mode 750 35 0.03 37 85
Stand. dev. 1283 147 0.18 37 10
Skewness 0.55 0.54 0.51 . 0.46 0.82
218 CHAPTER EIGHT
The statistics describing the fitted models are listed in the upper half of
Table 8.2. They include the geometric means ofthe lognormal models, arithmetic
means for normal models, and circular means for circular normal models, all
with the limits of their 95% confidence intervals. Additional statistics include
the 25th and 75th percentiles of the fitted distributions, and the dispersion coef-
ficients. The latter gives an indication of the spread of the data about the mean
of the fitted model, which, in turn, reflects the total variability of the data. (See
Chapter 5, Section 5.3.2.) The variability is rather small for the length and
breadth parameters (range 0.1-0.25), but is substantially greater for the shape
parameter R (0.6).

PROBABILITY TABLES
8.4.1. Introduction
For detection purposes, the Ni-Cu ultramafic deposits of the North American
Shield are considered as dipping oriented slabs with a horizontal cross section
of elliptical shape. As a result, the target dimensions and shape ratios have to
be recalculated as shown in Section 5.4 of Chapter 5 by adding the appropriate
dip component to the breadth of the horizontal section of each deposit. For
example, the mean shape ratio of the actual deposits is inflated from 0.06 (see
Table 8.2) to 0.12 for the targets in the context of vertical detection to a depth
of 300 feet, and to 0.26 for 55-degree angled detection to the same vertical
depth, thus considerably boosting the probability of detection.
The orientation factor was introduced in the construction of the tables be-
cause of the resulting improvement in detection success as compared with results
obtained under the orientation randomness assumption. A dynamic programming
study showed that the probability of detection is maximized in the following
manner: (a) in the case of airborne surveys, when the parallel grid is oriented
at 90 degrees to the expected strike, (b) for ground surveys (vertical detection)
when the grid is laid at an angle of 20 degrees with the expected strike direction;
the angle should be 30 degrees when the 55-degree angled detection option is
considered. The grid is then laid out at the appropriate angle from the expected
strike direction as known from the results of previous surveys or photogeological
studies. If little is known of the likely orientation of the target, the mean of the
circular normal fitted model is a statistically valid estimate to be considered for
the design of the grids (see Table 8.2., upper half).
8.4.2. Description of the Detection Probability Tables

Three kinds of tables are constructed for the detection of Ni-Cu ultramafic
deposits of the North American Shield. The first two cover the single detection
(at least one intersection of the target) and confirmed detection (at least two
target intersections) situations and are constructed in a similar manner. The third
kind of table lists grid spacings and corresponding coverage costs per unit of
area which are required to obtain prespecified levels of detection probabilities
ranging from 0.10 to 0.95.
The heads of the first two kinds of tables display the statistics of the fitted
models as obtained from Table 8.2, including the mean target length within its
95% confidence interval and the mean shape ratio. The probabilities are calculated
first for the mean length, and then for the upper and lower fiducial limits of the
mean, for grid spacings increased in steps of 50-250 feet, depending on the type
of field program. As a result, the tables display the probability of detection or
confirmed detection within its 95% confidence interval for each grid spacing.

For the convenience of the readers, the tables have been grouped into three
blocks, each covering a specific kind of field program. The first block deals with
probabilities of detection and confirmed detection of Ni-Cu ultramafic deposits
by airborne geophysical surveys on parallel and square grids (Tables 8.4-8.7).
The second block covers the detection and confirmed detection of Ni-Cu deposits
by ground geophysical surveys or vertical drilling to a depth of 300 feet on
square grids (Tables 8.9-8.11). The third block pertains to detection and con-
firmed detection of deposits by 55-degree angled drilling to a vertical depth of
300 feet (Tables 8.13-8.15).

Ni-Cu DEPOSITS OF THE NORTH AMERICAN SHIELD
8.5.1. Search Strategies

We wish to consider three options for the detection of Ni-Cu deposits of
the North American Shield which are designed to meet the requirements of three
common exploration situations. The three options are referred to as the liminal,
maximal, and optimal strategies. The first one is tailored to deal with large
coverage of areas of little known potential within budget limits by minimizing
220 CHAPTER EIGHT
costs while ensuring at least a 0.500 level of detection probability. The second
one provides a high probability of detection, close to certainty (0.980) at a much
higher cost, which can be justified for the coverage of small areas of high merit.
Finally, the optimal option is a compromise plan based on the maximization of
the probability of detection under constraints of cost minimization. Exploration-
ists should favor the third option because it meets the requirements of the majority
of exploration situations which call for coverage of areas of moderate size and
moderate potential within strict budgetary constraints. The detection character-
istics corresponding to the three strategies are summarized and assembled into
one single table for the convenience of the readers (Table 8.3.).
8.5.2. Liminal Detection Strategy

The detection characteristics pertaining to the liminal strategy are listed in
the leftmost portion of Table 8.3. They include grid spacings and associated
coverage costs per unit of area for airborne surveys on two grid designs, ground
geophysical surveys, and vertical and angled drilling programs.
The cost-multiplying factor required to obtain confirmed detection of Ni-
Cu deposits instead of single detection averages 2.3 for airborne geophysical sur-
veys and for vertical and angled drilling programs, but falls to 1.8 for ground
geophysical surveys. Cost factor figures are not available for airborne geo-
physical surveys on square grids and for optimally angled drilling programs
because the probability of confirmed detection could not be satisfactorily
calculated.
A comparison of the merits of various grid and program designs is of obvious
interest to explorationists for planning purposes. Since the probability of detection
is held at a specified level of at least 0.500, the coverage cost per unit of area
is a suitable yardstick to use to measure the detection performance of various
designs. An inspection of the figures listed in the leftmost portion of Table 8.3
will show that the parallel grid design should be favored for airborne detection
because it reduces the coverage cost by 5% as compared to square grid coverage
cost. The preference for angled drilling over vertical drilling to the same depth
is justified by a 35% reduction of coverage cost.

Grid spacings and corresponding coverage costs for various field programs
and grid designs pertaining to the maximal option are listed in the rightmost
portion of Table 8.3, in the same manner as for the liminal strategy. The cost
multiplying factor required to obtain confirmation of detection averages 1.4 for
both airborne and ground geophysical surveys. It escalates to 1.8 for vertical
drilling programs and reaches 2.4 for angled drilling programs. A comparison
TABLE 8.3 om
Summary of Detection Characteristics of Ni-Cu Ultramafic Deposits of the North American -I
m
Shield for Three Types of Strategies. q
(5
Liminal detection Optimal detection Maximal detection Z
Coverage cost: Coverage cost: o-n
US$/mile sq. US$/ mile sq.
Type of survey Grid size: feet Coverage cost Detection Grid size: feet Z
and Confirmed US$/mile sq. proba- Confirmed T'
n
c:
$2570
c
Airborne $380 $905 $448 0.75 $1990
~
parallel grid ~
~
Airborne $398 - $494 0.76 $805 - :>
-n
geophysical 3400 ft 2090 ft 1300 ft n
square grid om
"'tI
Ground $2500 $4650 $2190 0.72 $7800 $12,000
o
VI
square grid =i
VI
Vertical $974,000 $2,280,000 $803,000 0.72 $4,507,000 $7,860,000

drilling 450 ft 275 ft 500 ft 200 ft 150 ft
square grid
55-degree
angled $632,000 $1,425,000 $621,000 0.81 $1,420,000 $3,520,000
drilling 650 415 ft 700 ft 430 ft 280 ft
square grid
Optimal 45-deg
angled - - $491,000 0
0 80 - - N
drilling 800 ft N
-l.
square grid
~-
222 CHAPTER EIGHT
of the figures on both sides of Table 8.3 shows the relative advantage of the
maximal option over the liminal strategy when specifying confirmation of de-
tection by geophysical surveys and vertical drilling; but there is little difference
between the two options when angled drilling is considered.
An inspection of the rightmost portion of Table 8.3 shows that the square
grid design should be preferred to the parallel design for airborne geophysical
surveys, because selection of the former leads to a 60% saving in coverage costs.
When comparing the merits of the two types of drilling program for field planning
purposes, angled drilling against the dip direction should be favored over vertical
drilling to the same depth because of the resulting 68% saving in coverage costs.
If the dip direction cannot be ascertained from previous surveys or photogeo-
logical studies, vertical drilling may be the preferred option in order to avoid
the risk of drilling down-dip.

Design optimization based on the efficiency criterion was carried out for
each kind of field program by dynamic programming, as described in Section
5.5.4 of Chapter 5. The optimal parameters include (a) optimal grid orientation
with respect to the expected target strike direction within its confidence limit,
(b) optimal grid spacing and corresponding coverage cost per unit of area, (c)
optimal drill inclination, if the angled drilling option is chosen, and (d) the
optimal detection probability. The parameters of optimized airborne geophysical
surveys on parallel and square grid designs are listed in Table 8.8, while Tables
8.12 and 8.16 cover the optimization of ground geophysical surveys and drilling
programs, respectively. All results are summarized and listed in the central
portion of Table 8.3 for the reader's convenience.
Inspection of Table 8.3 shows that the optimal probabilities of detection
for all types of field programs and grid designs range between the 0.72 and 0.81
levels, averaging 0.75. The advantage of the optimal option over the maximal
strategy is emphasized by the 66% coverage cost reduction in favor of the former,
as compared to a drop of only 23% in detection probability level.
As mentioned in the previous chapters 6 and 7, we cannot use the coverage
cost per unit of area as a criterion for the comparison of detection performances
of various field program or grid designs because the optimal probability levels
vary. We shall use instead the expected loss criterion, which is described in
Section 6.5.4 of Chapter 6. As a result, we find that there is no incentive in
selecting the square grid design for airborne surveys because of the 6% expected
loss increase incurred. On the other hand, the optimally angled drilling option
should be preferred to the vertical drilling option because of the resulting 56%
decrease in expected loss for the coverage of a unit of area.
8.6. DETECTION PROBABILITY AND OPTIMIZATION TABLES
The reader will find in the following section a series of thirteen tables
covering the probabilities of detection and optimal designs of the search for
Ni-Cu ultramafic deposits of the North American Shield by airborne and ground
programs (pages 223-232).
TABLE 8.4
Probabilities of Detection of Ni-Cu Ultramafic Deposits of the North American
Shield by Airborne Geophysical Surveys (Continuous Readings) on Parallel
Grids.

Flight lines at 90 degrees with strike line.
Grid spacing 1.c.l. geom. mean u.c.1.
S feet L= 815 1097 1475 feet
200 0.947 0.961 1. 000

400 0.893 0.922 0.943
600 0.836 0.882 0.915
800 0.775 0.841 0.886
1000 0.705 0.798 0.856
1200 0.656 0.751 0.825
1400 0.562 0.697 0.793
1600 0.492 0.675 0.759
1800 0.437 0.600 0.722
2000 0.393 0.540 0.740
2200 0.358 0.491 0.673
2400 0.328 0.450 0.617
2600 0.303 0.415 0.569
2800 0.281 0.386 0.529
3000 0.262 0.360 0.494
3200 0.246 0.337 0.463
3400 0.231 0.317 0.435
3600 0.219 0.300 0.411
3800 0.207 0.284 0.390
4000 0.197 0.270 0.370
4200 0.187 0.257 0.353
4400 0.179 0.245 0.337
4600 0.171 0.235 0.322
4800 0.164 0.225 0.308
5000 0.157 0.216 0.296
5200 0.151 0.208 0.285
224 CHAPTER EIGHT
TABLE 8.5
Ni-Cu Ultramafic Deposits of the North American Shield by Airborne Geo-
physical Surveys on Parallel Grids.


0.10 10020 $ 176.89

0.15 6880 $ 193.72
0.20 5260 $ 210.27
0.25 4240 $ 227.17
0.30 3560 $ 243.82
0.35 3060 $ 260.78
0.40 2700 $ 276.89
0.45 2400 $ 294.00
0.50 2160 $ 311.11
0.55 1980 $ 326.67
0.60 1820 $ 343.08
0.65 1680 $ 360.00
0.70 1360 $ 411.76
0.75 1180 $ 453.22
0.80 960 $ 525.00
0.85 720 $ 653.33
0.90 480 $ 910.00
0.95 220 $1820.00
TABLE 8.6
Probabilities of Confirmed Detection of Ni-Cu Ultramafic Deposits ofthe North
American Shield by Airborne Geophysical Surveys (Continuous Readings) on
Parallel Grids.

in the confidence interval: l.c.l. (geom. mean ( u.c.l.

S feet L= 815 1097 1475 feet
300 0.858 0.905 0.938

400 0.828 0.870 0.900
500 0.559 0.802 0.870
600 0.391 0.712 0.840
700 0.279 0.539 0.800
800 0.205 0.414 0.747
900 0.156 0.323 0.607
1000 0.127 0.255 0.498
llOO 0.000 0.204 0.412
1200 0.000 0.166 0.343
1300 0.000 0.140 0.288
1400 0.000 0.124 0.243
1500 0.000 0.000 0.206
1600 0.000 0.000 0.177
1700 0.000 0.000 0.154
1800 0.000 0.000 0.137
1900 0.000 0.000 0.125
2000 0.000 0.000 0.000
226 CHAPTER EIGHT
TABLE 8.7
Probabilities of Detection of Ni-Cu Ultramafic Deposits of the North American
Shield by Airborne Geophysical Surveys (Continuous Readings) on Square
Grids.

Grid Spacing 1.c.1. geom. mean u.c.l.
S feet L= 815 1097 1475 feet
1000 1. 000 1. 000 1.000

1200 0.972 1.000 1. 000
1400 0.921 1. 000 1. 000
1600 0.863 0.979 1. 000
1800 0.807 0.945 1.000
2000 0.754 0.904 0.994
2200 0.707 0.862 0.979
2400 0.664 0.820 0.955
2600 0.626 0.781 0.926
2800 0.591 0.744 0.895
3000 0.560 0.710 0.865
3200 0.532 0.678 0.834
3400 0.506 0.649 0.805
3600 0.482 0.621 0.777
3800 0.461 0.596 0.750
4000 0.441 0.573 0.724
4200 0.423 0.551 0.700
4400 0.407 0.531 0.677
4600 0.391 0.512 0.655
4800 0.377 0.494 0.635
5000 0.363 0.477 0.616
5200 0.351 0.462 0.597
TABLE 8.8
the Detection of Ni-Cu Ultramafic Deposits of the North American Shield.
respectively:
Unit cost: $70/1.ml.
(i) Survey Design:

Parallel Grid
(ii) Survey Design:

Square Grid
228 CHAPTER EIGHT
TABLE B.9
Probabilities of Detection of Ni-Cu Ultramafic Deposits of the
North American Shield by Ground Geophysical Surveys (Dis-
crete Readings) or Vertical Drilling to a Depth of 300 Feet.

strike line.
in feet L = 815 1097 1475ft
150 1. 000 1. 000 1. 000

200 0.962 1.000 1. 000
300 0.913 0.943 1. 000
400 0.627 0.935 0.948
500 0.419 0.723 0.978
600 0.292 0.539 0.859
700 0.215 0.403 0.701
800 0.165 0.310 0.566
900 0.130 0.245 0.457
1000 0.105 0.198 0.372
1200 0.073 0.138 0.259
1600 0.041 0.077 0.146
2000 0.026 0.050 0.093
TABLE 8.10
tion of Ni-Cu Ultramafic Deposits of the North American Shield by Ver-
tical Drilling to a Depth of 300 Feet.

unit Cost = $6000 Per Diamond Drill-Hole
Probability Level Gr id Size Drilling Cost Per Mile Square

0.05 1850 89.58

0.15 1150 188.25
0.25 900 283.73
0.35 750 388.81
0.45 700 438.91
0.55 600 577.42
0.65 550 675.43
0.75 500 803.19
0.85 450 974.35
0.95 300 2077.99
230 CHAPTER EIGHT
TABLE 8.11
Probabilities of Confirmed Detection of Ni-Cu Ultramafic De-
posits of the North American Shield by Ground Geophysical Sur-
veys or Vertical Drilling to a Depth of 300 Feet.

strike line.

in feet L = 815 1097 147Sft
100 1. 000 1. 000 1. 000

150 0.987 1. 000 1. 000
200 0.863 0.951 1. 000
250 0.627 0.829 0.944
300 0.329 0.608 0.846
350 0.201 0.348 0.701
400 0.119 0.230 0.426
450 0.069 0.150 0.305
500 0.039 0.198 0.368
550 0.022 0.060 0.153
600 0.012 0.037 0.106
TABLE 8.12
Optimal Design of Ground Geophysical Surveys on Square Grids for the De-
tection of Ni-Cu Ultramafic Deposits of the North American Shield.

respectively.

TABLE 8.13
Probabilities of Detection of Ni-Cu Ultramafic Deposits of the
North American Shield by 55-Degree Angled Drilling to a Vertical
Depth of 300 Feet.

L.c.l. ( geometric mean ( U.c.1., and minor axis = B feet,
str ike line.
in feet L = 815 1097 1475ft
350 1. 000 1. 000 1. 000

400 0.996 1. 000 1. 000
500 0.838 1. 000 1. 000
600 0.626 0.968 1. 000
700 0.466 0.815 1. 000
800 0.356 0.658 0.981
900 0.282 0.529 0.A88
1000 0.228 0.429 0.767
llOO 0.189 0.355 0.656
1200 0.158 0.298 0.559
1300 0.135 0.254 0.478
1400 0.ll6 0.219 0.412
1500 0.101 0.191 0.359
1600 0.089 0.168 0.316
1700 0.079 0.149 0.280
1800 0.070 0.133 0.249
1900 0.068 o.ll9 0.224
2000 0.057 0.107 0.202
TABLE 8.14
of Ni-Cu Ultramafic Deposits of the North American Shield by 55-Degree


0.05 2800 61. 42

0.15 1800 113.90
0.25 1400 167.46
0.35 1100 247.26
0.45 1000 289.81
0.55 900 346.39
0.65 800 424.21
0.75 750 474.68
0.85 700 535.83
0.95 650 611.00
TABLE 8.15
Probabilities of Confirmed Detection of Ni-Cu Ultramafic De-
posits of the North American Shield by 55-Degree Angled Drilling
to a Vertical Depth of 300 Feet.

L.c.l. ( geometric mean ( U.c.1., and minor axis = B feet,
strike line.

Spacing S L.c.1. c. mean U.c.l.
in feet L = 815 1097 1475ft
200 1. 000 1. 000 1. 000

300 0.941 1. 000 1. 000
400 0.543 1. 000 1. 000
500 0.279 0.735 1. 000
600 0.125 0.428 0.896
700 0.052 0.258 0.699
800 0.019 0.144 0.462
900 0.005 0.077 0.325
1000 0.001 0.039 0.219
1200 0.000 0.008 0.091
1500 0.000 0.000 0.020
TABLE 8.16
Optimal Designs for the Detection of Ni-Cu Ultramafic Deposits of the North
American Shield by Drilling to a Vertical Depth of 300 Feet.

respectively
(a) Vertical Drilling:


probability of detection: 0.84 0.81 0.77

1. CAMERON, E. M., SIDDELEY, G., and DURHAM, C. G., 1971, Distribution of ore elements
in rocks for evaluating ore potential: Nickel, copper and sulfur in ultra-mafic rocks of the
Canadian Shield, Can. Inst. Min. Metall. Spec. Vol. 11, 298-313.
2. CHAYER, T. P., 1960, Some critical differences between Alpine-type and stratiform peridotite-
gabbro complexes, 21st International Geological Congress, Pt. XIII, pp. 247-259.
234 CHAPTER EIGHT
3. COAD, P. R., 1979, Nickel sulfide deposits associated with ultramafic rocks of the Abitibi Belt
and economic potential of mafic-ultramafic intrusions, Ontario Geological Survey, Study No.
20, Toronto, Canada.
4. CORNWALL, H. R., 1966, Nickel deposits of North America, U.S. Geol. Surv. Bull. No.
1223.
5. DOWSETT, J. S., 1979, Geophysical exploration methods for nickel, Geol. Surv. Can. Econ.
Geol. Rep. No. 31, 310-321.
6. ECKSTRAND, O. R., 1972, Ultramafic flows and nickel sulfide deposits in the Abitibi Orogenic
Belt, Geol. Surv. Can. Pap. 72.-1, part A, 75-81.
7. IRVINE, T. N., and SMITH, C. H., 1967, The ultrabasic rocks of the Muskox intrusion,
N.W.T., Canada, in Ultramafic and Related Rocks, pp. 38-49, Wiley, New York.
8. JACKSON, E. D., 1961, Primary textures and mineral associations in the ultramafic zone of
the Stillwater complex, Montana, U.S. Geol. Surv. Prof. Pap. No. 358.
9. KILBURN, L. c., WILSON, H. D. B., GRAHAM, A. R., and OGURA, Y., 1969, Nickel
sulfide ores related to ultrabasic intrusions in Canada, Symposium on magmatic ore deposits,
Economic Geology Publishing Co, pp. 276-293.
10. LIEBENBERG, L., 1970, The sulfides in the layered sequence of the Bushveld igneous complex,
Geological Society of South Africa, Special Publication No.1, pp. 108-207.
11. LUSK, J., 1976, A possible volcanic-exhalative origin for lenticular nickel sulfide deposits of
volcanic association, with special references to those of Western Australia; Can. J. Sci. 13,
451-458.
12. MACKENZIE, B. W., 1968, Nickel: Canada and the World, Mining Rept No. 16, Department
of Energy, Mines and Resources, Ottawa.
13. MILLER, C. P., 1959, A comparison of plant and soil prospecting for nickel, A.I.M.E. Preprint
No. 59-L-40.
14. NALDRETT, A. J., 1966, The role of sulfurization in the genesis of iron-nickel sulfide deposits
of the Porcupine District, Can. Inst. Min. Metall. Bull. 59,489-497.
15. NALDRETT, A. J., and GASPARRINI, E. L., 1971, Archean nickel sulfide deposits in Canada:
Their classification, geological setting and genesis, with some suggestions as to exploration,
Geol. Soc. Aust. Spec. Pub. No.3, 201-226.
16. NALDRETT, A. J., 1973, Nickel sulfide deposits-Their classification and genesis with special
emphasis on deposits of volcanogenic association, Can. Inst. Min. Metall. Bull. 66(739), 45-63.
17. OBIAL, R., APELO, M., and SANTOS, G., 1972, Geochemical prospecting for nickel sul-
fides-An orientation survey, J. Geol. Soc. Philipp. 26, 1-36.
18. ROSS, J. R., and TRAVIS, G. A., 1981, Nickel sulfide deposits of Western Australia in global
perspective, Econ. Geol. 76, 1291-1329.
19. SHLANKA, R., 1969, Copper, nickel, lead and zinc deposits of Ontario, Ontario Department
of Mines, M.R.C. 12, Toronto.
20. TAYLOR, R. B., 1964, Geology of the Duluth gabbro complex near Duluth, Minnesota, Minn.
Geol. Surv. Bull. No. 44.
21. WAGER, L. R., and BROWN, G. M., 1968, Layered Igneous Rocks, Oliver & Boyd, London.
22. WAGER, L. R., VINCENT, E. A., and SMALES, A. A., 1957, Sulfides in the Skaergaard
intrusion, East Greenland, Econ. Geol. 52, 855-903.
23. WALKER, F., 1940, Differentiation of the Palisades diabase, New Jersey, Bull. Geol. Soc.
Am. 51, 1059-1106.
24. WILSON, H. D. B., and BRISBIN, W. c., 1961, Regional structure of the Thompson-Moak
Lake Nickel Belt, Can. Inst. Min. Metall. Bull. 54(596), 815-821.
25. WYLIE, P. J. (Ed.) 1967, A Review of the Geology of the Ultra-mafic and Related Rocks.
Wiley, New York.
26. ZURBRIGG, H. F., 1963, Thompson mine geology, Can. Inst. Min. Metall. Bull. 56, 451-460.
CHAPTER NINE
OPTIMIZED AIRBORNE AND

GROUND SEARCH FOR VOLCANOGENIC
MASSIVE SULFIDE DEPOSITS OF THE
NORTH AMERICAN SHIELD AND
CORDI LLERA BELT
Although volcanogenic massive sulfide deposits have been mined since

antiquity and through medieval times, the deposit type has been recognized and
defined as such only quite recently. Prior to the mid 1950s, the volcanogenic
sulfide deposits were classified as hydrothermal, genetically related to often
hypothetical granitic intrusives in favorably fractured environments. There has
been much confusion between the two terms "stratabound" and "stratiform,"
which have been used interchangeably during the past two decades, until Sang-
ster's attempt to clarify the matter. (36) He defined as "stratabound," deposits such
as the volcanogenic sulfides which occur as discontinuous bodies, generally
conformable within specific horizons of a series, but locally cutting across strat-
ification, as opposed to "stratiform" deposits, such as the Mississippi Valley
type described in Chapter 10, which are quite continuous and always concordant
with encasing strata.
Hutchinson(19) recognizes three categories of volcanogenic sulfide deposits
which appear to have gradually evolved through the geological time scale. The
first category, considered as a rather primitive and poorly differentiated type,
yields mainly zinc-copper ore with minor gold and silver and is characteristically
of early to mid-Archean age, with a resurgence in early Proterozoic and again
in Paleozoic times. The second category, considered as more differentiated than
the first one, consists of complex lead-zinc-copper-silver mineralization. It first
appeared in the early Proterozoic period and reached its peak in the Paleozoic
era, while continuing strongly through Mesozoic and Cenozoic times. The third
category, referred to as "ophiolitic," consists mainly of pyrite and chalcopyrite
mineralization genetically related to basic volcanism; it is a much later type
which appeared only in Paleozoic times and extended through Mesozoic times.
235
236 CHAPTER NINE
The geographic spread of the volcanogenic sulfide deposits is as extensive

as their chronologic span. The regions of greatest abundance of volcanogenic
sulfide deposits are the Precambrian Shields of North America, Scandinavia,
and Western Australia. The second largest concentration lies in the Paleozoic
Belts of the Appalachian of North America, Scandinavia, and eastern Australia
(Tasman Geosyncline). The third concentration is found in the Mesozoic to
Cenozoic foldbelts, including the Cordillera of the Americas, the South Pacific
Island Arcs, and the Mediterranean region of Europe. The present study covers
only two regions of occurrence, namely, the North American Shield and Cor-
dillera Belt.
9.2. GENERAL GEOLOGICAL BACKGROUND: VOLCANOGENIC

SULFIDE DEPOSITS OF THE NORTH AMERICAN SHIELD
9.2.1. Lithologic Environment

The deposits consist of one or several lenticular bodies of massive iron and
base metal sulfides grossly conformable with encasing series of acid volcanic
flows and pyroclastic of submarine volcanogenic origin. The deposits are stra-
tigraphically controlled and are believed to be genetically related to two types
of volcanic events; breaks in volcanic geochemistry marked by changes in com-
position from acidic to more basic, and changes from volcanism to sedimentation,
commonly marked by the presence of cherty iron-rich horizons. Although the
Archean volcanism produced many cycles of tholeiitic and calc-alkaline volcanic
suites, Fox(1O) points out that massive sulfide deposition occurred only in the
calc-alkaline cycles, where the sulfur content of the intermediate to acid flows
is abnormally higher than that of the basic members of the cycles. Similarly,
Riddler<33,34) draws attention to the predominance of sulfide over carbonate or
oxide facies in the exhalites intervening between volcanic-sedimentary cycles,
as a clue to the proximity of volcanogenic sulfide deposits.
Generally, volcanogenic massive sulfide deposits terminate abruptly along
the hangingwall, with only a minimal transgression of disseminated pyrite and
chloritic alteration within a few feet above the sulfide contact. (11) In relatively
little disturbed environments, the two longest dimensions of the deposits, strike
length and dip length, parallel stratigraphic features of the volcanic host for-
mation, such as flow contacts and bedding planes. On the footwall side, however,
massive sulfide ore grades into "stringer ore" and then into disseminated pyrite
within a funnel-shaped alteration pipe consisting of chlorite-sericite and epidote.
The pipe generally stands at right angles to the longest dimension of the deposits
SEARCH FOR VOLCANOGENIC MASSIVE SULFIDE DEPOSITS 237
and tapers off within a distance of 300-400 feet below the massive sulfide bodies.
Geochemically, the footwall zone of North American Archean volcanogenic
sulfide deposits shows a marked increase in Fe, Mg, and S and a decrease in
Si, Na, and K, as compared to the hanging wall formations (see Chapter 12,
Ref. 31). In many cases, the alteration pipes of Archean volcanogenic sulfide
deposits have been truncated or deformed by post-ore tectonics, or even obli-
terated by post-ore metamorphism. (36)
Because of the occurrence of repeated cycles of folding and metamorphism
which have affected Archean volcanogenic sulfide deposits since their formation,
the principal lithologic and geochemical features controlling ore deposition have
been disturbed and can be surmised only by analogy from observations made
on more recent deposits of Mesozoic and Cenozoic ages. Post-ore alteration
affecting ore controlling features belongs to three main types: regional meta-
morphism, contact metamorphism, and supergene alteration. Sangster(36) reports
that under the influence of regional metamorphism the chloritic material of the
syn-ore alteration pipes becomes unstable and is altered into an assemblage of
hornblend, biotite, and anthophyllite; if sufficient calcium is present, epidote,
actinolite, and tremolite appear. Examples of the influence of post-ore contact
metamorphism on syn-ore alteration have been described in the Noranda district
of North western Quebec. Large subcircular cordierite aggregates form at the
expense of the chlorite and give the footwall rock a peculiar spotted appearance
(dalmatianite). Supergene alteration, which is quite extensive in tropical and
subtropical climates, is not well developed in the North American region and
results only in the formation of gossans which assist direct detection from the
air and on the ground.
9.2.2. Metallogeny
Two kinds of ore material commonly make up the volcanogenic massive
sulfide deposits: massive sulfides comprising from 50% to 60% of the rock by
volume, and "stringer ore" with 20% to 30% sulfides only. Sangster 37 ) reports
that 50% of the 70 deposits of his database feature both types of ore material,
45% include massive ore only, while the remaining 5% exhibit stringer ore only,
possibly due to the truncation of the massive sulfide portion by post-ore tectonics.
A few volcanogenic sulfide deposits are nearly monomineralic, with pyrite
making up nearly 100% of the sulfides by volume (Spain, Philippines). But most
deposits are polymineralic: the ore material consists of an admixture of pyrite,
pyrrhotite, chalcopyrite, sphalerite with or without galena, minor gold and silver
values. The proportion of iron sulfide may be as high as 68% in volume (Geco,
northwest Ontario) or as low as 36% (Coronation, Manitoba); pyrrhotite is always
238 CHAPTER NINE
present in Archean deposits, but generally subordinate to pyrite. The amount of

chalcopyrite varies from as little as 10% of the sulfides by volume (Geco) to as
much 60% (Coronation). Pyrrhotite and chalcopyrite are usually predominant in
the stringer ore of the footwall zone of the deposits. Sphalerite is nearly always
of the iron-rich, high-temperature marmatite variety. Pyrite and sphalerite, with
or without galena, prevail in the massive sulfide portion of the deposits.
9.2.3. Mode of Occurrence

There are some 120 known volcanogenic massive sulfide deposits in the
Canadian portion of the Precambrian Shield, to which should be added two
deposits recently discovered in the U.S. portion ofthe shield (State of Wisconsin)
and 10 deposits occurring in Arizona, within a large Precambrian window in-
cluded in the Cordillera Belt. In his 1980 study, Sangster<37) shows that the
Canadian volcanogenic sulfide deposits are aggregated into eight large clusters
varying in size from 90 to 600 square miles. The clusters are made up of as few
as four deposits (Manitouwage cluster) or as many as 20 deposits (Noranda
cluster). The average number is 12, and the average sae of the clusters is
equivalent to a circle 20 miles in diameter. A second order of clustering, often
directional in nature, is again evident with each regional group.
Among the eight regional clusters, a total of four including the Timmins,
Noranda, 10utel-Normetal, and Mattagami districts occur in the eastern portion
of the Superior Province of the Canadian Shield. There are two districts in western
Ontario including the Manitouwage and the newly discovered Sturgeon Lake
districts, and one each in Southeastern Manitoba and in the Arctic region of the
Northwest Territories.

Volcanogenic massive sulfide deposits have been one of the most sought
for exploration targets during the past three decades, particularly in the Precam-
brian Shields of North America, Scandinavia, and Western Australia. The recent
development of the volcanogenic model for these base metal sulfide deposits has
proved very fruitful and has been a main contributing factor to the surge of
interest in volcanogenic sulfide exploration. A second factor was the steady
improvement of airborne and ground geophysical detection techniques, partic-
ularly in Canada and the Scandinavian countries, at a time when the more obvious
deposits have all been discovered after several decades of thorough prospecting.
The various field detection methods used in the search for volcanogenic
sulfide deposits can be grouped into two main categories, including a direct
approach based on visual detection on the ground or from airborne platforms
and assisted by mechanical probes (drilling). The second approach, an indirect
one, relies on the detection of the signatures of the volcanogenic sulfide deposits
by geophysical or geochemical methods.

Volcanogenic sulfide deposits are much too small targets for direct detection
by remote sensing from satellite platforms. The most effective direct approach
on the ground consists of thorough prospecting backed by detailed mapping and
complemented by some information drilling. According to Boldy, (5) the above
approach is responsible for the discovery of 47 deposits, or 43% of the total
known to date, mostly during the "pregeophysics" period (1920-1950). In the
Noranda district of Northwest Quebec the score is seven deposits or 32% of the
total known to date.
Specialized mapping at a semiregional scale backed by conceptual geolog-
ical models is presently the most successful direct approach, as the days of easy
prospecting are over. For instance, mapping based on the exhalite.concept(33,34)
seeks to define the proximal environment of volcanogenic sulfide deposits on
the basis of exhalite facies change from carbonate and oxide to sulfide. Fox(9)
describes how detailed mapping of facies changes in pyroclastic formations may
be used to locate former volcanic vents as possible sources of base metal sulfides.
The geological insight gained from the guidance by the volcanogenic model
in the greenstone belts of northeastern Ontario and northwestern Quebec has
been successfully applied to new districts such as Western Ontario and the Arctic
region of the Northern Territories, resulting in the discovery of substantial ton-
nages of sulfide ores. In the well-prospected districts of eastern Canada, the
trend of the future should be to rely on information drilling followed by systematic
drilling to search for subjacent deposits beyond the average 300 foot detection
range of present-day geophysical instrumentation, as demonstrated by the Lake
Dufault success story of recent years.

The detection ability of indirect geophysical methods largely depends on
the magnitude of the contrast of physical properties between ore deposits and
240 CHAPTER NINE
wallrock. As shown by Paterson, (28) the geophysical environment of volcanogenic

sulfide deposits is generally suitable for indirect detection, featuring a high
conductivity contrast (500-10,000 times that of wallrock), a moderately high
specific gravity contrast (2-2.5 times), accompanied by a moderate magnetic
susceptibility contrast (3-100 times), if sufficient pyrrhotite or magnetite (Mat-
tagami deposit) is present. Furthermore, the presence of pyrite disseminations
along the margin and footwall of deposits results in polarizable sulfide targets
substantially larger than the deposits themselves, which enhances the effective-
ness of airborne methods such as INPUT or ground methods such as induced
polarization surveys.
Despite the favorable characteristics of volcanogenic sulfide deposits for
geophysical detection purposes, these deposits still are elusive exploration tar-
gets. According to Patterson's summing up of the results of several years of
airborne and ground search for volcanogenic sulfide deposits in the North Amer-
ican Shield, (28) only one ground confirmed geophysical anomaly in 2000 leads
to a productive ore deposit. Hindrances to a successful detection of such deposits
are many: a relatively unfavorable geometry (small size and steep dip) is one of
them. Others include the confounding of sulfide signatures with that of graphite
or water-logged fractures, and the masking of sulfide responses by the effect of
extensive surficial conductive layers of water-logged glacial clay and tills so
prevalent in the Shield.
However, as a result of much recent research work, the discriminating power
of geophysical instrumentation has considerably improved. Recent developments
include the use of multifrequency, multicoil, and multichannel single instru-
ments, and of hybrid technologies such as pulse electromagnetic and the magnetic
induced polarization. (39) Explorationists increasingly rely on combined surveys
to eliminate the recording of spurious anomalies masking sulfide signatures. Two
of the more successful combinations are (a) electromagnetics + magnetics
+ gravimetry and (b) induced polarization + magnetics + gravimetry.
As a result, according to Boldy, (5) 54% of the volcanogenic deposits discovered
since 1950 were detected by airborne and ground geophysical surveys.

Much interest has recently developed in the detection of primary halos of
volcanogenic sulfide deposits by means of bedrock sampling as a method of
targeting "blind" sulfide deposits. Holmes, (15) Barragar, (4) and FOX(IO) showed
that major economic elements such as Cu, Zn, Pb cannot be used as proximal
indicators for sulfide deposits because they do not transgress the hanging wall
markers (hematite tuff, magnetite jasper, etc.). However, elements such as S,

As, F, and some rare earths are good proximal indicators for sulfide ore deposits,
along with ratios such as Mg + Fe/Ca + Na, which reflect the geochemical
changes occurring in the vicinity of sulfide deposits. A major drawback is the
general scarcity of outcrops which prevails over large areas of the Shield.
The detection of secondary halos is hampered in the shield by the prevalence
of transported glacial alluvium. The interpretation of soil anomalies is further
complicated in the northernmost portions of the Shield by the influence of the
permafrost. As a result, the success ratio of the indirect geochemical approach
is rather low.

SHIELD VOLCANOGENIC SULFIDE DEPOSITS

The sample selected for inclusion in the database comprises a total of 102
commercial and subeconomic volcanogenic sulfide deposits, all located in the
Canadian portion of the North American Shield. The deposits are listed by name
and province of occurrence in Table 9.1. The geographic breakdown of the
sample is as follows: greenstone belts of northeastern Ontario and northwestern
Quebec: 57 deposits (56% of total); greenstone belts of western Ontario: 10
deposits (9% of total); southeastern Manitoba: 31 deposits (30% of total); and
Northwest Territories: 4 deposits, or one half of the total discovered to date in
that region. Because of its large size and wide geographic spread, the sample
should offer a statistically satisfactory representation of the population of vol-
canogenic sulfide deposits of the North American Shield.

Measurements of the five geometric parameters required for the calculation
of detection probabilities were obtained from each of the 102 volcanogenic sulfide
deposits included in the sample. The geometric parameters are of two kinds.
The first three are of a dimensional nature; they include the length, breadth, and
shape ratio (R = BIL) of the horizontal section of the portion of the deposits
which lies within the average 300-foot range of commonly used detectors. The
statistical range of the measured parameters is as follows: from to 225 to 4850
feet for length and 20 to 600 feet for breadth. The last two parameters are
242 CHAPTER NINE
TABLE 9.1(a)
List of Volcanogenic Massive Sulfide Deposits of
the North American Shield (N.W. Quebec Sec-
tion) Included in the Database.
Noranda (Horne) Barvue

Quemont Vendome
McDonald New Formaque
Amulet "A" Belfort
Iso Barvallee
Joliet Normetal
Mobrun Poirier
Corbett Joutel
Norbec Northern Exploration
Millenbach Brouillan
Aldermac Lessard
East Waite Lemoine
West waite Scott
New Insco Jay Copper
Amulet "e" Lynx Yellowknife
Delbridge Dumagami
Vauze Pershcourt
Amulet "F" Louvicourt
Waite-Dufault Lake Berrigan
East Sullivan Mattagami
Manitou-Barvue Orchan
Louvem New Hosco
Abi tibi Copper Norita
Dunraine Phelps Dodge
Quebec Manitou Bell Allard
Garon Lake
Radiore
TABLE 9.1(b)
List of Volcanogenic Massive Sulfide Deposits of the North
American Shield (Ontario, Manitoba, and N.W.T. Sections) In-
Ontario
Ecstall willecho
Kamkotia Big Nama
Jameland Mattabi
Jamieson N.B.D.
Munro Lyon Lake
Geco Creek
Willroy Coldstream
South Bay
Manitoba
Flin-flon Chisel
Pine Bay Stall
Coronation Osborne
Centennial Freeport Reed
Schist Lake Anderson
West Schist H.B. Reed
Cuprus wim
White Lake Dickstone
Birch Lake Li ttle Stall
North Star Rail
Mandy Ghost Lake
Lost Lake
Northwest Territories
Copperman
Hackett Ruttan
Isok Fox Lake
High Lake
Taki Lake
attitudinal in nature; they include the unoriented true dips and the strike orien-
tation referred to true north and measured in degrees within the right half-circle.
The range of the measured dip angles is from 48 to 85 degrees.
The measurements of the geometric parameters are grouped into classes
and assembled into frequency distributions. The statistics summarizing the dis-
tributions of observed data are listed in the lower half of Table 9.2. Included
are arithmetic means, modes, standard deviations, and coefficient of skewness.
The latter is constructed as the ratio (mean-mode)/standard deviation to indicate
the degree of asymmetry of the frequency distributions. Skewness is slight (0.1)
244 CHAPTER NINE
TABLE 9.2
Statistical Modeling Summary for Volcanogenic Massive Sulfide Deposits of
the North American Shield (Sample Size: 102 Deposits).
Statistic Length of Breadth of Shape Strike direct. Dip in

model in feet in feet R:B/L in degrees
25%ile 530 25 0.02 62 50

95%L.c.1. 810 48 0.05 84 66
Mean 947 61 0.06 93 69
95% U.c.l. 1106 76 0.08 102 72
75%ile 1650 150 0.12 130 83
Disp. Coef. 0.12 0.31 0.34 0.21 0.13
Statistic
of observed
data
Arith.mean 1286 118 0.14 92 69

Mode 700 50 0.06 97 80
Stand.dev. 1175 152 0.20 45 16
Skewness 0.51 0.45 0.41 0.11 0.73
for the strike orientation parameter, moderate (average 0.45) for length, breadth,
and shape ratio, but is quite strong (0.75) for the dip angle parameter.
Lognormal models were successfully fitted to the observed frequency dis-
tributions of the dimensional parameters, as confirmed by the results of the X2
goodness of fit testing at the 0.05 confidence level. The statistical models fitted
to the dip angle and strike direction distributions are normal and circular normal,
respectively. The statistics of the fitted models are assembled in the upper half
of Table 9.2. They include the geometric means and their 95% confidence limits
for the lognormal models, as well as the arithmetic and circular means and their
confidence limits for the attitudinal parameters. Additional listings include the
25th and 75th percentiles and dispersion coefficients. The latter are constructed
as the ratios of standard deviations over means in order to give a dimensionless
indication of the variability of the data. The coefficient is small (0.1) for the
length parameter, and moderately large (0.33) for the breadth and shape ratio
parameters.

PROBABILITY TABLES FOR SHIELD VOLCANOGENIC
SULFIDE DEPOSITS
9.5.1. Introduction
For detection purposes, the volcanogenic sulfide deposits are considered as
oriented dipping slabs with horizontal cross sections of elliptical shape. The
target dimensions are obtained by introducing a dip component calculated in the
manner described in Sections 2.2.3 of Chapter 2 and 5.4.2 of Chapter 5 for both
vertical and angled detection situations. As a result, the mean shape ratio is
inflated from 0.06 for horizontal sections of deposits to 0.19 and 0.32 for target
sections, when vertical and 55-degree angled detection are considered, thus
considerably boosting the probability of successful detection of volcanogenic
sulfide deposits.
The latter is further increased when the control grid orientation with respect
to the expected strike direction of targets is optimized according to the results
of a dynamic programming study conducted as described in Section 5.5.4 of
Chapter 5. The optimal grid orientations for the detection of volcanogenic sulfide
deposits by various types of field programs are as follows: (a) parallel grids for
airborne geophysical surveys are to be laid at 90 degrees ± 10 to the expected
target strike, (b) ground square grids for vertical and angled detection are to be
oriented at angles of 30 and 45 degrees ± 10 with target strike, respectively.
The expected target strike direction can be estimated from previous surveys or
photogeological studies. If no prior information is available, the mean of the
fitted circular normal model is a statistically valid estimate of the strike direction
(see Table 9.2).
9.5.2. Description and Organization of Tables

Two similar sets of tables were constructed to provide the probabilities of
single detection (at least one target intersection by the grids) and of confirmed
detection (at least two intersections) of volcanogenic sulfide deposits of the North
American Shield. The grid spacings are varied from 200 to 6000 feet in steps
246 CHAPTER NINE
of 50 to 200 feet, as appropriate. For each spacing, the probability of detection

is calculated and listed within a 95% confidence interval, based on the mean
target length and its fiducial interval.
A third set of tables is constructed to provide the grid spacing and associated
coverage cost per unit of area required to ensure a prespecified level of detection
probability, which should prove most useful for planning purposes. The prob-
ability tables are organized into three blocks, one for each type of field program.
Tables 9.4.-9.7 cover detection by airborne geophysical surveys on parallel and
square grids. Vertical ground detection is covered by Tables 9.9-9.11, and 55-
degree angled detection by Tables 9.13-9.15.
9.6. DESIGN OF THREE STRATEGIES FOR THE DETECTION OF

9.6.1. Detection Strategies

Three detection strategies may be considered for the detection of volcan-
ogenic sulfide deposits in the North American Shield, each applied to a specific
exploration situation. The first one, referred to as the liminal option, ensures a
minimum level of detection probability of 0.500 at a cost low enough to carry
out large-scale coverages of area of little known potential. The second one, the
maximal option, provides near certain detection (0.980) at a much higher cost
than the liminal option, thus restricting its application to small areas of high
potential. Finally, the optimal option is an attractive compromise between the
two options listed above, which should prove suitable for the majority of ex-
ploration situations.
9.6.2. Liminal Option

The detection characteristics pertaining to the liminal option are listed in
the leftmost portion of Table 9.3, covering both single and confirmed detection
for three types of field programs. The coverage cost multiplier required to ensure
confirmed detection rather than single detection is 2.3 for airborne geophysical
surveys. It falls to 1. 95 for ground geophysical surveys and rises again to 2.5
for both vertical and angled drilling programs.
The relative detection performances of various field programs or grid designs
may be measured by means of the coverage cost per unit of area as a yardstick.
TABLE 9.3
<Jl
Summary of Detection Characteristics of Volcanogenic Massive Sulfide Deposits of the North m
American Shield for Three Types of Detection Strategies. »-
;;:0
n
::c
..,.,
Coverage cost: Coverage cost: o
;;:0
Type of survey Grid size: feet Coverage cost Detection Grid size: feet <
and Confirmed tJS$/mile sq. proba- Confirmed
o
r-
grid geometry Detection detection Grid size: ft bility Detection detection n
»-
z
Airborne o
geophysical $448 $1025 $633 0.79 $1990 $3825 Cl
parallel 1320 ft 475 ft 750 ft 200 ft lOr) ft m
Z
grid
n
Airborne $510 - $533 0.62 $975 - ~
geophysical 2050 ft 1870 ft 880 ft »-
<Jl
square grid <Jl
<:m
Ground $2620 $4950 $3483 0.95 $4410 $7900 <Jl
geophysical 430 ft 265 ft 350 ft 280 ft 180 ft C
square grid r-
..,.,
Vertical $1,120,000 $2,450,000 0.85
$1,554,000 $2,080,000 $4,508,000
om
drilling 430 ft 265 ft 350 ft 280 ft 180 ft o
square grid m
-0
55-degree
, o
<Jl
angled $572,200 $1,350,300 $705,000 0.82 $1,010,000 $1,685,000 I =i
<Jl
square grid
Optimal 40-deg
angled - - $683,000 0.83 - - N
drilling 700 ft ~
square grid '-I
L- ----
248 CHAPTER NINE
We find that the parallel grid design is preferable to the square design for airborne
geophysical searches because the coverage cost for the former is 13% cheaper.
Similarly, the 55-degree angled drilling option should be preferred to the vertical
option because the coverage cost by the former is 50% cheaper.
9.6.3. Maximal Option

The rightmost portion of Table 9.3 lists the detection parameters pertaining
to the maximal option. The cost multiplier required to ensure confirmed detection
by airborne geophysical surveys is 1.92, and for ground geophysical surveys as
well as vertical and angled drilling programs the average is 1.80. It should be
noted that the maximal mUltiplying factors are lower than their liminal counter-
parts; thus, when opting for the maximal strategy, confirmation of detection is,
relatively speaking, a more attractive requirement than in the liminal context.
If we wish to compare the relative merits of the parallel and square grid
designs for airborne surveys, we find that selection of the square design is the
more attractive option because it leads to a 54% coverage cost reduction. Angled
drilling is also much more attractive an option than vertical detection because
the coverage cost is 51 % cheaper.
9.6.4. Optimal Option

Based on the results of a three-stage dynamic programming study (see
Section 5.5.4 of Chapter 5), three grid design parameters were computed to
ensure the optimal detection of volcanogenic sulfide deposits of the North Amer-
ican Shield. They include (a) the optimal grid orientation with respect to the
expected strike direction, (b) optimal grid spacing and corresponding coverage
cost per unit of area, as well as the associated detection probability, and (c) the
optimal drilling angle when the likely dip direction may be estimated. The optimal
parameters are listed within their 95% confidence intervals in Table 9.8 for
airborne geophysical surveys on two grid designs, in Table 9.12 for ground
geophysical surveys, and in Table 9.16 for vertical and angled drilling programs.
For the reader's convenience, the optimal detection characteristics were assem-
bled and are listed in the central portion of Table 9.3.
If we wish to compare the detection performances of various field programs
or grid designs, we have to rely on the "expected loss" criterion as described in
Section 6.5.4 of Chapter 6, instead of the coverage cost criterion used above,
because the optimal probabilities of detection vary with the type of survey. We
find that the expected loss associated with the parallel grid design for airborne
surveys is 53% lower than in the square grid case; therefore, the former should
be our choice. The optimally angled (40 degrees) drilling program is a clear-cut
choice over the vertical program, because of the 50% reduction in expected loss
which can be expected.
9.6.5. Multipurpose Optimal Grids

As indicated in Sections 8.14 of Chapter 8 and 9.2.1 of the present chapter,
both Ni-Cu ultramafic deposits of volcanogenic affiliations and volcanogenic
massive sulfide deposits are known to occur in "island arc" remnants referred
to commonly as "greenstone belts" of the North American Archean Shield.
Admittedly, the two types are not spatially correlated within the Archean eu-
geosynclinal edifices of the island arcs. However, a joint coverage for both types
of deposits in little-prospected greenstone belts of unknown potential should be
considered in the context of the optimal option.
A careful comparison of Tables 8.8 and 9.8 (airborne surveys), Tables 8.12
and 9.12 (ground geophysical surveys), and finally Tables 8.16 and 9.16 (drilling
programs) shows overlaps between the paired 95% confidence intervals for the
optimal grid spacings. The regions common to the paired confidence intervals
define narrow ranges of spacings which are jointly optimal for the search for
both types of deposits within the same greenstone belts.
The overlaps of confidence intervals are listed as optimal grid spacing ranges
in Table 9.17, along with the corresponding coverage cost and associated optimal
detection probabilities for each type of field program or grid design. For example,
one can see from Table 9.17 that parallel grids with a 1300-foot spacing, and
an 1800-foot square grid size are optimal for the joint detection of the two types
of deposits by airborne surveys. Likewise, a 400-foot square grid size is jointly
optimal for ground geophysical surveys. Finally, a 700-foot spacing is jointly
optimal for the search of both deposits by drilling optimally angled between 40
and 45 degrees.
9.7. DETECTION PROBABILITY AND OPTIMIZATION TABLES FOR

The following sequence of fourteen tables deals with the probabilities of

detection and optimal designs of airborne and ground programs in the search for
volcanogenic massive sulfide deposits of the Shield (pages 250-261).
250 CHAPTER NINE
TABLE 9.4
Probabilities of Detection of Volcanogenic Massive Sulfide Deposits of the
North American Shield by Airborne Geophysical Surveys on Parallel Grids.

S feet L= 810 947 1106 feet
200 0.938 0.948 1.000

400 0.874 0.895 0.915
600 0.805 0.839 0.871
800 0.724 0.778 0.824
1000 0.616 0.732 0.773
1200 0.513 0.610 0.714
1400 0.440 0.523 0.643
1600 0.385 0.458 0.563
1800 0.342 0.407 0.500
2000 0.308 0.366 0.450
2200 0.280 0.333 0.409
2400 0.257 0.305 0.375
2600 0.237 0.282 0.346
2800 0.220 0.262 0.322
3000 0.205 0.244 0.300
3200 0.192 0.229 0.281
3400 0.181 0.215 0.265
3600 0.171 0.203 0.250
3800 0.162 0.193 0.237
4000 0.154 0.183 0.225
4200 0.147 0.174 0.214
4400 0.140 0.166 0.205
4600 0.134 0.159 0.196
4800 0.128 0.153 0.188
5000 0.123 0.146 0.180
5200 0.118 0.141 0.173
TABLE 9.5
of Volcanogenic Massive Sulfide Deposits of the North American Shield by
Airborne Geophysical Surveys on Parallel Grids.

unit cost = US$70 per line-mile

0.10 6580 $ 196.17

0.15 4520 $ 221. 77
0.20 3460 $ 246.82
0.25 2780 $ 272.95
0.30 2340 $ 297.95
0.35 2020 $ 322.97
0.40 1780 $ 347.64
0.45 1580 $ 373.92
0.50 1420 $ 400.28
0.55 1300 $ 424.31
0.60 1200 $ 448.00
0.65 1100 $ 476.00
0.70 1020 $ 502.35
0.75 860 $ 569.77
0.80 700 $ 668.00
0.85 540 $ 824.44
0.90 360 $1166.67
0.95 200 $1988.00
252 CHAPTER NINE
TABLE 9.6
Probabilities of Confirmed Detection of Volcanogenic Massive Sulfide Deposits
of the North American Shield by Airborne Geophysical Surveys on Parallel
Grids.


S feet L= 810 947 1106 feet
200 0.908 0.918 0.958

300 0.806 0.821 0.836
400 0.604 0.635 0.716
500 0.401 0.537 0.603
600 0.293 0.395 0.571
700 0.228 0.302 0.432
800 0.194 0.242 0.343
900 0.000 0.205 0.280
1000 0.000 0.000 0.236
1100 0.000 0.000 0.207
1200 0.000 0.000 0.191
TABLE 9.7
North American Shield by Airborne Geophysical Surveys on Square Grids.

S feet L= 810 947 1106 feet
800 1. 000 1. 000 1. 000

1000 0.866 0.960 1. 000
1200 0.765 0.861 1.000
1400 0.682 0.776 0.890
1600 0.615 0.703 0.816
1800 0.558 0.642 0.751
2000 0.511 0.590 0.695
2200 0.471 0.546 0.645
2400 0.437 0.507 0.602
2600 0.407 0.474 0.564
2800 0.381 0.444 0.530
3000 0.358 0.418 0.500
3200 0.338 0.395 0.473
3400 0.320 0.374 0.449
3600 0.303 0.355 0.427
3800 0.288 0.338 0.407
4000 0.275 0.323 0.389
4200 0.263 0.309 0.372
4400 0.252 0.296 0.357
4600 0.241 0.284 0.343
4800 0.232 0.273 0.330
5000 0.223 0.263 0.318
5200 0.215 0.253 0.307
254 CHAPTER NINE
TABLE 9.B
Optimal Design of Airborne Geophysical Surveys for the Detection of Volcan-
ogenic Massive Sulfide Deposits of the North American Shield.

respectively
Unit Cost: $70jl.ml.
(i) Survey Design:

Parallel Gr id
(ii) Survey Design:

Square Grid
TABLE 9.9
Probabilities of Detection of Volcanogenic Massive Sulfide Depos-
its of the North American Shield by Ground Geophysical Surveys
or Vertical Drilling to a depth of 300 Feet.

L.c.1. ( geometric mean ( u.c.1., and minor axis = B feet,
str ike line.
in feet L = 810 947 1106ft
200 1. 000 1. 000 1. 000

250 0.998 1. 000 1. 000
300 0.939 0.997 1. 000
350 0.793 0.951 1. 000
400 0.635 0.837 0.981
450 0.510 0.700 0.921
500 0.416 0.579 0.820
600 0.289 0.408 0.606
700 0.212 0.300 0.453
800 0.163 0.230 0.348
900 0.128 0.182 0.275
1000 0.104 0.147 0.222
1100 0.086 0.122 0.184
1200 0.072 0.102 0.154
1300 0.062 0.087 0.132
1400 0.053 0.075 0.114
1500 0.046 0.065 0.099
256 CHAPTER NINE
TABLE 9.10
Detection of Volcanogenic Massive Sulfide Deposits of the North Amer-
ican Shield by Vertical Drilling to a Depth of 300 Feet.

probability Level Grid Size Drilling Cost Per Mile Square

0.05 1600 111.46

0.15 1000 237.38
0.25 800 347.47
0.35 650 500.48
0.45 600 577.42
0.55 550 675.43
0.65 500 803.19
0.75 450 974.35
0.85 400 1211.54
0.95 350 1554.43
TABLE 9.11
Probabilities of Confirmed Detection of Volcanogenic Massive Sul-
fide Deposits of the North American Shield by Ground Geophysical
Surveys or Vertical Drilling to a Depth of 300 Feet.

L.c.l. ( geometric mean ( u.c.l., minor axis = B feet,
R is defined as the ratio B/L with geometric mean = 0_19
strike line.

in feet L = 810 947 1106ft
100 1. 000 1. 000 1. 000

150 0.987 1. 000 1. 000
200 0.863 0.951 1. 000
250 0.627 0.829 0.944
300 0.329 0.608 0.846
350 0.201 0.348 0.701
400 0.119 0.230 0.426
450 0.069 0.150 0.305
500 0.039 0.095 0.218
550 0.022 0.060 0.153
600 0.012 0.037 0.106
700 0.003 0.014 0.050
800 0.000 0.004 0.023
900 0.000 0.001 0.010
1000 0.000 0.000 0.004
TABLE 9.12
Optimal Design of Ground Geophysical Surveys for the Detection of Volcan-
ogenic Massive Sulfide Deposits of the North American Shield.

respectively.

Optimal grid Orientation: 30 degrees + or -10 to expected strike
258 CHAPTER NINE
TABLE 9.13
Probabilities of Detection of Volcanogenic Massive Sulfide De-
posits of the North American Shield by 55-Degree Angled Drill-

L.c.1. ( geometric mean ( U.c.l., and minor axis = B feet,
strike line.
in feet L = 810 947 1106ft
300 1. 000 1. 000 1. 000

350 0.997 1. 000 1. 000
400 0.930 1. 000 1. 000
450 0.810 0.971 1. 000
500 0.688 0.883 0.999
550 0.578 0.779 0.974
600 0.487 0.678 0.907
700 0.358 0.506 0.739
800 0.274 0.387 0.584
900 0.216 0.306 0.463
1000 0.175 0.248 0.375
1100 0.145 0.205 0.310
1200 0.122 0.172 0.260
1300 0.104 0.147 0.222
1400 0.089 0.126 0.191
1500 0.078 0.110 0.167
1600 0.068 0.097 0.146
1700 0.061 0.086 0.130
1800 0.054 0.076 0.116
1900 0.049 0.069 0.104
2000 0.044 0.062 0.094
2500 0.028 0.040 0.060
TABLE 9.14
Determination of Grid Size for Specified levels of Probability in the
ican Shield by 55-Degree Angled Drilling to a Vertical Depth of 300 Feet.


0.15 l300 188.27

0.25 1000 289.81
0.35 900 346.39
0.45 800 424.21
0.55 700 535.83
0.65 650 611.00
0.75 600 704.45
0.85 550 824.03
0.95 500 979.89
TABLE 9.15
fide Deposits of the North American Shield by 55-Degree Angled
Drilling to a Vertical Depth of 300 Feet.
Survey design: square grid wth spacings S by S feet.

strike line.

Spacing S L.c.l. G. mean U.c.1.
in feet L = 810 947 1l06ft
350 1.000 1.000 1. 000

400 0.854 1. 000 1. 000
450 0.420 0.974 1. 000
500 0.269 0.580 1. 000
550 0.169 0.375 0.980
600 0.101 0.258 0.735
700 0.025 0.114 0.325
800 0.001 0.040 0.174
900 0.000 0.007 0.085
1000 0.000 0.000 0.034
260 CHAPTER NINE
TABLE 9.16
Optimal Designs for the Detection of Volcanogenic Massive Sulfide Deposits
of the North American Shield by Drilling to a Vertical Depth of 300 Feet.
Survey Design: square grid with spacing S by S feet

respectively



TABLE 9.17
Parameters of Multipurpose Optimal Grids for the detection of Ni-Cu Ultra-
mafic and Volcanogenic Massive Sulfide Deposits of the North American Shield.
Type of survey Range of grid Range of coverage Range of

& gr id geometry size cost ($/ml. sq.) probabili ty of
detection
A.B. geophysical
parallel grid 1290 to 1350 ft $415 to $430 0.56 to 0.72
A.B. geophysical
square grid 1760 to 1990 ft $510 to $540 0.62 to 0.93
Grd. geophysical
square grid 300 to 500 ft $2200 to $4310 0.77 to 0.93
Vertical drilling $803,000 to

square grid 300 to 500 ft $2,078,000 0.77 to 0.93
Opt. angled (40

to 45 deg.) drilling $424,000 to
square grid 600 to 800 ft $705,000 0.75 to 0.98
9.8. GEOLOGICAL SYNOPSIS FOR VOLCANOGENIC SULFIDE

DEPOSITS OF THE NORTH AMERICAN CORDILLERA BELT
Volcanogenic massive sulfide deposits ranging in age from Paleozoic to

Mesozoic occur in two main regions of the North American Cordillera Belt. The
first one lies in the Canadian section of the belt, in the west central and southern
portion of the Province of British Columbia and in Vancouver Island. The second
region extends through the northern and northcentral portions of the state of
California, in the U.S. section of the belt. A third region lies in the state of
Arizona, but is not covered here because the deposits occur in a large Precambrian
window included in the Cordillera Belt (see Table 9.18). The most common
geological environment for the Cordillera volcanogenic sulfide deposits consists
of roof pendants of metavolcanics and sediments of Paleozoic to Mesozoic age
262 CHAPTER NINE
engulfed in the coastal batholits of British Columbia or the Sierra Nevada of

California.
There are some 25 known commercial or subeconomic occurrences of vol-
canogenic sulfide ores in British Columbia. Most of them are found in a
Permo-Triassic volcanic environment. The stratigraphic control for the sulfide
deposits appears to be the contact between series of lapilli-tuff, lithic tuffs, or
dacitic tuffs, often altered into quartz-sericite schists, on the one hand, and acid
pyroclastic piles, on the other hand. One deposit of possible, if doubtful, vol-
canogenic affiliation, known as "Sam Goosely," in north central British Colum-
bia, is thought to have been subjected to remobilization and redeposition in
sedimentary piles. The large Granduc deposit, presently encased in formations
mylonitized beyond recognition, is also thought to have originally formed in a
volcanogenic environment.
Two types of volcanogenic sulfide deposits are found in the Californian
section of the Cordillera Belty,21) The earlier type consists of disseminated to
massive copper-zinc mineralization occurring in middle Devonian to Permian
volcanic series of the West Shasta District. A younger and more complex type
consisting of massive Zn-Pb-Cu-Ag mineralization is found in volcanic series
of Triassic to Jurassic age which stretch southward from the East Shasta District
and along the western edge of the Sierra Nevada over a length of 300 miles.
There are some 20 known commercial or subeconomic volcanogenic sulfide
deposits in the whole Shasta District, most of them of modest size but very high
grade; a few of them are of the large tonnage and low grade type, mainly in the
West Shasta District. Deposits of the early type show the typical chlorite-sericite
footwall alteration described in Section 9.2.1 as pertaining to Archean volcan-
ogenic deposits of the Shield. Some of the younger deposits of the East Shasta
District are notable for the development of barite and anhydrite in the gangue,
which could establish a genetic link with the much more recent "kuroko" type
of volcanogenic deposits of Japan, Fiji, and the Philippines.

CORDILLERA VOLCANOGENIC SULFIDE DEPOSITS
A total of 21 commercial and subeconomic deposits were selected as a

representative sample of the whole population of known and undiscovered vol-
canogenic sulfide deposits of the North American Cordillera Belt. The British
Columbia section of the belt is represented by 13 deposits and the Californian
section by 8. The deposits are listed by name in Table 9.18.
The geometry of the deposits is described by several dimensional and at-
titudinal parameters. The former include length, breadth, and shape ratio of the
TABLE 9.18
List of Volcanogenic Massive Sulfide Deposits of
the North American Cordillera Belt Included in the
Database.
British Columbia
(Canada)
Tulsequah Lynx
Anyox H.W.
Chuchua Myra
Sam Goosly Price
Goldstream
Granduc
Kutcho
Seneca
California (U.S.A.) Arizona (U.S.A.) (*)
Shasta-King Iron King
Afterthought United Verde
Iron Mountain Copper Queen
Bully Hill + Rising Star Old Dick
Mammoth
Balaklala
Copper Hill
Keystone
(*) Precambrian window; not included in sample.
horizontal sections of deposits, and the latter, unoriented dip angle and strike
orientation with respect to true north measured in degrees within the right half-
circle. The statistical range of the length measurements for the 21 deposits of
the sample is from 300 to 7000 feet; breadth measurements vary from 25 to 220
feet, and dip angles from 28 to 85 degrees.
The parameter measurements are grouped into classes from which frequency
distributions are constructed. The statistics summarizing the distributions of
observed data are listed in the lower half of Table 9.19, including arithmetic
means, modes, standard deviations, and coefficients of skewness. The latter give
an indication of the degree of asymmetry of the distributions of observed data.
The skewness is slight (0.13) for the strike orientation, moderate for lengths,
breadths, and dips, but quite strong for the shape ratios (0.6).
the dimensional parameters, while normal and circular normal models were used
to fit the distributions of dips and strike directions. The models are summarized
264 CHAPTER NINE
by statistics which are listed in the upper half of Table 9.19, including geometric
means for the log-normal models, and arithmetic and circular means for the
attitudinal models, with their 95% confidence limits. Additional statistics are the
25th and 75th percentiles and the dispersion coefficients. The latter give an
indication of the variability of the data. Dispersion is slight for the lengths (0.12),
and moderate for the breadths and shaped ratios (0.25).

PROBABILITY TABLES FOR CORDILLERA VOLCANOGENIC
SULFIDE DEPOSITS
Volcanogenic sulfide deposits of the Cordillera are treated in the same

manner as the shield deposits for detection purposes (see Section 9.5.1). Two
sets of tables are constructed to provide the probabilities of detection or confirmed
detection of the volcanogenic sulfide deposits by means of three types of field
programs, based on grid spacings varying from 200 to 6000 feet in steps of 50
to 200 feet. A different set of tables is constructed for planning purposes, pro-
viding grid spacings and corresponding coverage costs per unit of area for spec-
ified levels of detection probability. Tables 9.21-9.24 cover airborne geophysical
surveys on parallel and square grids. Ground geophysical detection probabilities
are provided by Tables 9.26-9.28. Finally, Tables 9.30-9.32 deal with detection
by angled drilling.

9.11.1. Liminal and Maximal Strategies

Two strategies, referred to as liminal and maximal, were described in Sec-
tion 9.6.2 and are applied to the detection of volcanogenic sulfide deposits of
the Cordillera. The detection characteristics pertaining to the two strategies are
listed in the leftmost and rightmost sections of Table 9.20, respectively.
Cost multipliers required to ensure confirmation of single detection are 2.2
for airborne and ground geophysical surveys and 2.5 for vertical and angled
drilling, for the liminal option. The maximal cost multipliers are all smaller than
their liminal counterparts, making the confirmation requirement more attractive
than in the liminal context, relatively speaking.
If we wish to compare the merits of two types of grid designs for airborne
TABLE 9.19
Statistical Modeling Summary for Volcanogenic Massive Sulfide Deposits of
Statistic Length of Breadth of Shape Str ike direct. Dip in

of fitted horiz.sect. hor iz. sect. ratio from T. North degrees
25%ile 1200 35 0.02 65 50

95%L.c.1. 1201 40 0.02 72 53
Mean 1842 61 0.03 88 62
95%U.c.1. 2836 92 0.05 103 70
75%ile 3350 120 0.06 110 73
Disp.coef. 0.13 0.23 0.26 0.18 0.12
Statistic
of observed
data
AriUuTl.mean 2508 85 0.05 85 62

Mode 2000 35 0.02 93 68
Stand.dev. 1875 71 0.05 65 19
Skewness 0.31 0.28 0.62 0.13 0.33
geophysical surveys using the coverage cost per unit area as a yardstick, we find
that, as is the case for the Shield volcanogenic deposits, the square grid design
should be rejected in the liminal context because it results in a 12% cost increase.
In the maximal context, however, the square grid design is more attractive than
the parallel one, as it results in a 56% reduction of coverage cost per unit area.
An evaluation of the respective merits of angled and vertical drilling pro-
grams based on the yardstick mentioned above leads to the selection of angled
drilling over vertical detection in both the liminal and maximal contexts. Under
the former option, a 50% cost reduction is obtained by choosing angled drilling;
under the latter option, the saving in coverage cost is as much as 58%.

The optimization of control grids and field program designs for the detection
of volcanogenic sulfide deposits of the Cordillera was conducted in the same
manner as mentioned in Section 9.6.4 for the Shield deposits. The resulting
TABLE 9.20 N
Summary of Detection Characteristics of Volcanogenic Massive Sulfide Deposits of the a-
a-
North American Cordillera Belt for Three Types of Detection Strategies.
Liminal detection Maximal detection

and Confirmed US$/mile sq. proba- Confirmed
Airborne $325 $676 $424 0.80 $1525 $2905

parallel grid
Airborne $360 - $493 0.89 $670 - I

geophysical 3380 ft 2100 ft 1400 ft I
square grid
Ground $1435 $3125 $1752 0.90 $2520 $4480

square grid
Vertical $438,900 $1,092,200 $577,000 0.90 $975,000 $2,280,000

square grid
55-degree
angled $214,380 $564,2000 $424,000 0.97 $424,000 $725,000
n
:::c
drilling 1200 ft 680 ft 800 ft 800 ft 650 ft »
square grid "1:J
-j
m
Optimal 40-deg ~
angled - - $454,000 0.99 - - Z

drilling 800 ft Z
square grid m
-----~ -- --- - -_.. _ - - - - - -- --- - - -
optimal grid parameters are displayed in Tables 9.25 for airborne geophysical
surveys, 9.29 for ground geophysical surveys, and 9.33 for vertical and angled
drilling programs, and are assembled in the central portion of Table 9.20 for the
readers' convenience.
Comparisons of detection performances of various grid designs and field
programs can be based on the expected loss criterion (see Section 6.5.4 of Chapter
6). The choice of the square grid is recommended because it leads to a 35%
reduction of the expected loss which could be incurred if the parallel design was
selected. As far as drilling programs are concerned, the choice between angled
and vertical detection should be in favor of optimally angled drilling (40 degrees),
because the expected loss is 91 % lower than in the case of the vertical option.
9.11.3. Multipurpose Optimal Grid Designs

As mentioned in the previous section 9.3.5, it is advantageous to design
grids which may be jointly optimal for the detection of several types of ore
targets likely to occur within specific geological environments. An example is
the simultaneous search for volcanogenic massive sulfides and contact metaso-
matic deposits of the Cu-Fe and Pb-Zn-Cu types in the North American Cor-
dillera Belt.
If we carefully compare the following three sets of tables in Chapters 7 and
9: Tables 7.9, 7.24, and 9.24 for airborne geophysical surveys, Tables 7.13,
7.28, and 9.29 for ground geophysical surveys, and Tables 7.17, 7.32, and 9.33
for drilling programs, we find that the 95% confidence intervals of grid spacings
overlap to a small extent, thus defining narrow ranges of jointly optimal grid
spacings for the detection of the three types of ore targets mentioned above.
From Table 9.34, which displays all jointly optimal grid parameters, we
can see that optimal joint detection is provided by a ll00-foot grid spacing for
parallel airborne grids, increased to a 1700-foot spacing for square grids. For
ground geophysical surveys and vertical drilling programs the spacing is 525
feet, while optimally angled drilling (45 degrees) requires a 650-foot spacing.
9.12. DETECTION PROBABILITY AND OPTIMIZATION TABLES FOR

The following fourteen tables cover the probabilities of detection and the
optimization of the search for volcanogenic sulfide deposits of the Cordilleran
Belt by airborne and ground programs (pages 268-279).
268 CHAPTER NINE
TABLE 9.21
North American Cordillera Belt by Airborne Geophysical Surveys on Parallel
Grids.

Orientation of flight lines: 90 degrees with strike-line
S feet L=1201 1842 2826 feet
200 1. 000 1. 000 1. 000

400 0.923 0.942 0.950
600 0.883 0.913 0.925
800 0.842 0.884 0.899
1000 0.799 0.853 0.873
1200 0.751 0.821 0.846
1400 0.718 0.788 0.818
1600 0.628 0.751 0.789
1800 0.558 0.710 0.758
2000 0.502 0.671 0.723
2200 0.457 0.610 0.701
2400 0.419 0.559 0.643
2600 0.387 0.516 0.593
2800 0.359 0.480 0.551
3000 0.335 0.448 0.514
3200 0.314 0.420 0.482
3400 0.296 0.395 0.454
3600 0.279 0.373 0.429
3800 0.264 0.353 0.406
4000 0.251 0.336 0.386
4200 0.239 0.320 0.367
4400 0.228 0.305 0.351
4600 0.218 0.292 0.335
4800 0.209 0.280 0.321
5000 0.201 0.269 0.309
5200 0.193 0.258 0.297
TABLE 9.22
of Volcanogenic Massive Sulfide Deposits of the North American Cordillera


0.10 12360 $ 169.90

0.15 8500 $ 183.48
0.20 6480 $ 197.04
0.25 5240 $ 210.53
0.30 4400 $ 224.00
0.35 3780 $ 237.78
0.40 3320 $ 251.33
0.45 2960 $ 264.86
0.50 2680 $ 277.91
0.55 2440 $ 291.48
0.60 2240 $ 305.00
0.65 2060 $ 319.42
0.70 1800 $ 345.33
0.75 1560 $ 376.92
0.80 1280 $ 428.75
0.85 960 $ 525.00
0.90 640 $ 717.50
0.95 280 $1460.00
270 CHAPTER NINE
TABLE 9.23
Probabilities of Confirmed Detection of Volcanogenic Massive Sulfide Deposits
of the North American Cordillera Belt by Airborne Geophysical Surveys on
Parallel Grids.


S feet L=1201 1842 2826 feet
800 0.410 0.675 0.841

900 0.330 0.555 0.699
1000 0.271 0.462 0.587
HOO 0.228 0.389 0.499
1200 0.197 0.331 0.427
1300 0.178 0.285 0.369
l400 0.000 0.249 0.322
1500 0.000 0.220 0.283
1600 0.000 0.198 0.251
1700 0.000 0.182 0.225
1800 0.000 0.172 0.205
1900 0.000 0.000 0.189
2000 0.000 0.000 0.177
TABLE 9.24
North American Cordillera Belt by Airborne Geophysical Surveys on Square
Grids.

S feet L=1201 1842 2826 feet
1200 1.000 1. 000 1. 000

1400 0.955 1. 000 1. 000
1600 0.882 1. 000 1. 000
1800 0.816 1. 000 1. 000
2000 0.757 0.919 1. 000
2200 0.706 0.866 0.943
2400 0.660 0.817 0.895
2600 0.619 0.773 0.850
2800 0.583 0.732 0.809
3000 0.551 0.695 0.770
3200 0.522 0.661 0.735
3400 0.496 0.630 0.702
3600 0.472 0.602 0.672
3800 0.450 0.576 0.644
4000 0.430 0.552 0.618
4200 0.412 0.530 0.594
4400 0.396 0.500 0.572
4600 0.380 0.490 0.552
4800 0.366 0.473 0.532
5000 0.353 0.456 0.514
5200 0.340 0.441 0.497
272 CHAPTER NINE
TABLE 9.25
Optimal Design of Airborne Geophysical Surveys for the Detection of Volcan-
ogenic Massive Sulfide Deposits of the North American Cordillera Belt.

respectively
(i) Survey Design:

Parallel Grid
(ii) Survey Design:

Square Grid
TABLE 9.26

L.c.l. ( geometric mean ( D.c.l., and minor axis = III feet,
R is defined as the ratio E/L with geometric mean = 0.19
strike line.
probability of detection
Spacing S L.c.l. G. mean D.c.l.
in feet L = 1201 1842 2826ft
350 1. 000 1. 000 1. 000

400 0.992 1. 000 1. 000
450 0.957 1. 000 1. 000
500 0.879 0.999 1. 000
600 0.682 0.958 0.995
700 0.514 0.834 0.949
800 0.397 0.684 0.839
900 0.314 0.553 0.709
1000 0.255 0.452 0.588
1100 0.210 0.375 0.492
1200 0.177 0.316 0.416
1300 0.151 0.269 0.355
1400 0.130 0.232 0.306
1500 0.1l3 0.202 0.267
274 CHAPTER NINE
TABLE 9.27
ican Cordillera Belt by Vertical Drilling to a Depth of 300 Feet.


0.15 1700 101.64

0.25 1350 145.30
0.35 1150 188.25
0.45 1000 237.38
0.55 900 283.73
0.65 850 312.92
0.75 750 388.81
0.85 700 438.91
0.95 600 577 .42
TABLE 9.28
fide Deposits of the North American Cordillera Belt by Ground

The expected shape of the ore depsoits is elliptical
strike line.

Spacing S L.c.1. G. mean u.c.1.
in feet L = 1201 1842 2826ft
200 1.000 1. 000 1. 000

250 0.985 1. 000 1. 000
300 0.918 1. 000 1. 000
350 0.824 0.972 1. 000
400 0.686 0.919 0.973
450 0.432 0.851 0.929
500 0.321 0.767 0.873
550 0.239 0.638 0.805
600 0.176 0.435 0.718
700 0.092 0.279 0.417
800 0.047 0.177 0.284
900 0.024 0.110 0.192
1000 0.011 0.067 0.127
TABLE 9.29
Optimal Design of Ground Geophysical Surveys for the Detection of Volcan-
ogenic Massive Sulfide Deposits of the North American Cordillera Belt.

respectively.

TABLE 9.30
posits of the North American Cordillera Belt by 55-Degree An-
gled Drilling to a Vertical Depth of 300 Feet.

The expected shape of the ore deposi ts is elliptical
strike line.
in feet L = 1201 1842 2826ft
550 1. 000 1. 000 1. 000

600 0.991 1. 000 1. 000
650 0.944 1. 000 1. 000
700 0.876 1. 000 1. 000
800 0.727 0.991 1. 000
1000 0.479 0.804 0.942
1200 0.333 0.593 0.754
1400 0.245 0.436 0.576
1600 0.187 0.334 0.441
1800 0.148 0.264 0.349
2000 0.120 0.214 0.282
2200 0.099 0.177 0.233
2400 0.083 0.149 0.196
2600 0.071 0.127 0.167
2800 0.061 0.109 0.144
3000 0.053 0.095 0.126
TABLE 9.31
Determination of Grid Size for Specified Levels of Probabilities in the
ican Belt by 55-Degree Angled Drilling to a Vertical Depth of 300 Feet.

Probability Level gr id Size Drilling Cost Per Mile Square

0.15 2400 75.48

0.25 1900 105.16
0.35 1600 l36.07
0.45 1400 167.46
0.55 l300 188.40
0.65 1200 214.39
0.75 1050 267.10
0.85 1000 289.81
0.95 900 346.39
TABLE 9.32
fide Deposits of the North American Cordillera Belt by 55-Degree

strike line.

in feet L - 1201 1842 2826ft
450 1. 000 1. 000 1. 000

500 0.843 1. 000 1. 000
550 0.667 1. 000 1. 000
600 0.496 1. 000 1. 000
700 0.295 0.773 1. 000
800 0.165 0.499 0.784
900 0.087 0.343 0.530
1000 0.042 0.224 0.386
1100 0.018 0.144 0.269
1200 0.006 0.089 0.185
l300 0.001 0.052 0.124
278 CHAPTER NINE
TABLE 9.33
Optimal Designs for the Detection of Volcanogenic Massive Sulfide Deposits
Feet.

respectively

Cost in $thou. per mile square: 804 675 50]


TABLE 9.34
Parameters of Multipurpose Optimal Grids for the Detection of Contact-Me-
tasomatic and Volcanogenic Massive Sulfide Deposits of the North American
Cordillera Belt.
Type of survey Range of grid Range of coverage Range of

& grid geometry size cost ($/ml.sq. ) probability of
detection
A.B. geophysical 1050 to 1100 ft $460 to $480 0.66 to 0.82

parallel grid
A.B. geophysical 1660 to 1720 ft $615 to $635 0.86 to 0.88

square grid
Grd. geophysical 500 to 550 ft $1950 to $2200 0.73 to 0.90

square grid
Vertical drilling 500 to 550 ft $675,000 to 0.73 to 0.90

square grid $803,000
Optimally angled
(50 degr.) drilling 600 to 700 ft $564,000 to 0.97 to 0.99
square grid $705,000
l. ALBERS, J. P., and ROBERTSON, J. E., 1961, Geology and ore deposits of the East Shasta
copper-zinc district, Shasta County, California, U.S. Geol. Surv. Prof. Pap. 338.
2. ANDERSON, C. A., 1969, Massive sulfide deposits and volcanism; Econ. Geol. 64(2), 129-146.
3. ANNIS, R. C., CRANSTONE, D. A., and VALLEE, M., 1981, A survey of known mineral
deposits in Canada that are not being mined in 1980, Mineral Bulletin No. 181, Department of
Energy, Mines and Resources, Ottawa.
4. BARAGAR, W. R. A., 1968, Major element geochemistry of the Noranda volcanic belt, Can.
J. Eanh Sci. S, 773-790.
5. BOLDY, J., 1977, (Un)certain exploration facts and figures, Can. Inst. Min. Metall. Bull.
70(781), 86-95.
6. CAMERON, E. M., 1975, Geochemical methods of exploration for massive sulfide mineral-
ization in the Canadian Shield, Geochemical Exploration 1974, pp. 21-49, Elsevier, Amsterdam.
280 CHAPTER NINE
7. DESCARREAUX, J., 1973, A petrochemical study of the Abitibi volcanic belt and its bearing
on the occurrences of massive sulfide ores, Can. Inst. Min. Metall. Bull. 66, 61-69.
8. EASDON, M. M., 1970, A compilation of graphitic occurrences in the Archean part of North-
west Quebec, Ministere des Richesses Naturelles, Quebec, Open File Report G.M. 25662.
9. FOX, J. S., 1977, Rapid pyroclastic mapping in base metal exploration, Can.lnst. Min. Metall.
Bull. 70, 173-178.
10. FOX, J. S., 1979, Host rock geochemistry and massive sulfide volcanogenic ore, Can. Inst.
Min. Metall. Bull. 72(804), 127-134.
II. GILMOUR, P., 1965, The origin of massive sulfide mineralization in the Noranda district,
Northwestern Quebec, Geol. Assoc. Canada Proc. 16, 1239-1249.
12. GOODWIN, A. M., 1967, Volcanic studies in the Timmins-Kirkland Lake Noranda region of
Ontario and Quebec, Geol. Surv. Can. Pap. 67-1, Pt. A, 138-142.
13. GRAF, J. L., 1977, Rare earth elements as hydrothermal tracers during the formation of massive
sulfide deposits in volcanic rocks, Econ. Geol. 72, 527-545.
14. HODGSON, C. J., and LYDON, J. W., 1977, Geological setting of volcanogenic massive
sulfide deposits and active hydrothermal systems: Some implications for exploration, Can. Inst.
Min. Metall. Bull. 70, 95-106.
15. HOLMES, R., and TOOMS, J. S., 1973, Dispersion from a submarine/exhalative body, in
Geochemical Exploration 1972, pp. 193-202, Institute of Mining and Metallurgy, London.
16. HOPWOOD, T. P., 1975, "Quartz-eye"-bearing porphyroidal rocks and volcanogenic massive
sulfide deposits, Econ. Geol. 71, 589-612.
17. HUTCHINSON, R. W., 1965, Genesis of Canadian massive sulfides reconsidered by comparison
to Cyprus deposits, Can. Inst. Min. Metall. Trans. 68, 286-300.
18. HUTCHINSON, R. W., RIDDLER, R. H., and SUFFEL, G. G., 1971, Metallogenic rela-
tionships in the Abitibi Belt, Canada: A model for Archean metallogeny, Can.lnst. Min. Metal/.
Bull. 64, 48-57.
19. HUTCHINSON, R. W., 1973, Volcanogenic sulfide deposits and their metallogenic signifi-
cance, Econ. Geol. 68, 1223-1246.
20. JENKS, W. F., 1971, Tectonic transport of massive sulfide deposits in submarine volcanic and
sedimentary host rocks, Econ. Geol. 66, 1215-1224.
21. KINKEL, A. R. Jr., 1956, Geology and base metal deposits of West Shasta copper-zinc district,
Shasta County, California, U.S. Geol. Surv. Prof. Pap. 285.
22. LARSON, L., and WEBBER, G. R., 1977, Chemical and petrographic variations in rhyolitic
zones in the Noranda area, Can. Inst. Min. Metall. Bull. 70(784), 80-90.
23. LATULIPPE, M., 1966, The relationship of mineralization to Precambrian stratigraphy in the
Mattagami and Val d'Or districts of Quebec, Geol. Assoc. Can., Spec. Pap. 3, pp. 21-42.
24. MACGECHAN, P. J., 1981, Exploration significance of the emplacement and genesis of massive
sulfide in the Main Zone at the Norita Mine, Mattagami, Quebec, Can. Inst. Min. Metall. Bull.,
84,59-75.
25. MILLER, L. J., 1960, Massive sulfide deposits in eugeosynclinal belts, Econ. Geol. 55,
1327-1332.
26. MILLER, R. J. M., 1973, The morphology of some Canadian massive sulfide deposits, A.I.M.E.,
Annual meeting, Chicago.
27. OHMOTO, H., 1978, Submarine caldera: A key to the formation of volcanogenic massive
sulfide deposits? Min. Geol. 28, 219-231.
28. PATERSON, N. R., 1971, Airborne electromagnetic methods applied to the search for sulfide
deposits, Can. Inst. Min. Metall. Bull. 64(705), 29-38.
29. PEARSON, D. E., 1977, Volcanic suites of southwestern British Columbia, Trip No 6, Geo-
logical Association of Canada Guidebook, pp. 20-25.
30. PELTON, W. H., 1977, New I.P. method may discriminate between graphite and massive
sulfides , North . Miner 63, 27, September 15.
31. PIRIE, J. D., and NICHOL, I., 1981, Geochemical dispersion in wallrocks associated with the
Norbec deposit, Noranda, Quebec, J. Geochem. Expl. 15, 159-180.
32. PODOLSKY, G., and SLANKIS, J., 1979, lzok Lake deposit, N.W.T., Canada: A geophysical
case history, Geol. Surv. Can. Rep. 31, 641-652.
33. RIDDLER, R. H., 1971, Analysis of Archean volcanic basins in the Canadian Shield using the
"exhalite" concept, Can. Inst. Min. Metall. Bull. 64(714), 20 (abstract).
34. RIDDLER, R. H., 1973, The "Exhalite" concept: A new tool for exploration, North. Miner
November 29, 59-61.
35. SANGSTER, D. F., 1972, Precambrian volcanogenic massive sulfide deposits in Canada: A
review, Geol. Surv. Can. Pap. No. 72-22.
36. SANGSTER, D. F., and SCOTT, S. D., 1976, Precambrian strata-bound massive Cu-Pb-Zn
sulfide ores of North America, in Handbook of Strata-bound and Stratiform Ore Deposits, pp.
129-222, Elsevier, Amsterdam.
37. SANGSTER, D. F., 1980, Quantitative characteristics of volcanogenic massive sulfide deposits,
Can. Inst. Min. Metall. Bull. 73(814),74-81.
38. SAWKINS, F. J., 1972, Sulfide ore deposits in relation to plate tectonics, J. Geol. 80, 377-397.
39. SEIGEL, H. 0., 1974, The magnetic induced polarization (M.I.P.) method, Geophysics 39,
321-339.
40. SOLOMON, M., 1976, "Volcanic" massive sulfide deposits and their host rocks: A review and
an explanation, in Handbook of Strata-bound and Stratiform Ore Deposits, pp. 21-54, Elsevier,
Amsterdam.
41. SPENCE, C. D., and DE ROSEN-SPENCE, A. F., 1975, The place of sulfide mineralization
in the volcanic sequence at Noranda, Quebec, Econ. Geol. 70, 90-101.
42. STANTON, R. L., 1960, General features of the comformable pyritic orebodies; Part 1: Field
associations, Part 2: Mineralogy, Can. Inst. Min. Metall. Bull. 53, 24-29, 66-74.
43. WALKER, G. P. L., 1971, Grain-size characteristics of pyroclastic deposits, J. Geol. 79,
696-714.
44. WOLFE,W. J., 1973, Geochemical exploration in Archean metavolcanic and meta-sedimentary
belts of the Pukaskwa Region, Ontario, Can. Inst. Min. Metall. Bull. 66(735), 75-83.
45. WOLFE, W. J., 1975, Zinc abundance in early Precambrian volcanic rocks; Its relationship to
exploitable levels of zinc in sulfide deposits of volcanogenic-exhalative origin, in Geochemical
Exploration 1974, pp. 261-278, Elsevier, Amsterdam.
CHAPTER TEN
DESIGNING OPTIMIZED FIELD

PROGRAMS FOR THE DETECTION OF
MISSISSIPPI VALLEY-TYPE Pb-Zn
DEPOSITS OF THE NORTH AMERICAN
ARCTIC PALEOZOIC PLATFORM
10.1.1. General Setting

The Mississippi Valley-type Pb-Zn deposits are one of the principal sedi-
mentary sources of lead-zinc ores of the world. The deposits are generally tabular
in shape, subhorizontal in attitude, and, although occurring at rather shallow
depths, difficult to detect because they have little or no surface expression. They
are truly "stratiform" deposits, i.e., occurring as continuous and extensive blan-
kets concordant with the encasing sediments, as opposed to "stratabound" de-
posits such as the volcanogenic massive sulfide ore bodies described above in
Chapter 9.
Lead-zinc deposits of the "Mississippi Valley" type were first discovered
in several districts lying within the watershed of the Mississippi River, in the
central portion of the U.S.A., hence the name. Since the original 18th century
discovery, similar deposits have been found in other areas of the North American
Paleozoic Platform and in other regions of the world.
The Paleozoic Platform, made up mainly of little-disturbed sediments, stretches
around the North American Precambrian Shield, between the Appalachian Belt
to the east and the Cordillera Belt to the west. There are seven main districts of
occurrence of Mississippi Valley-type deposits in the North American Paleozoic
Platform. Four of these districts-viz., East Tennessee, Rosiclar, Southeast
Missouri-Tri-State, and Upper Mississippi Valley-are not covered by the present
study. The three remaining districts, all situated in the Arctic portion of the
Paleozoic Platform, include the Great Slave Lake, Selwyn Mountain, and Arctic
Islands areas, which are dealt with in this chapter.
The age of the sedimentary formations encasing the Mississippi Valley
Pb-Zn deposits of North America is Paleozoic and varies from Upper Cambrian
283
284 CHAPTER TEN
in the Southeast Missouri and in the Selwyn Mountains districts, to Ordovician

in the Rosiclar, Upper f',Iississippi Valley, East Tennessee, and Arctic Islands
districts. More recent settings are known in the Great Slave Lake area (Devonian)
and in the Tri-State area (Lower to middle Carboniferous). A special mention
should be made of the Black Angel deposit of the west coast of Greenland, in
the Arctic region, which occurs in late Precambrain metamorphosed carbonate
formations.
The Mississippi Valley-type deposits generally occur within carbonate series
composed of cherty or reefal limestones at various stages of dolomitization with
pelitic intercalations (Lower Paleozoic) or clastic horizons (middle and Upper
Paleozoic). Callahan<2 l recognizes two types of structural control which have a
regional significance: (a) underlying unconformities (Southeast Missouri, Selwyn
Mountains, and Great Slave Lake) and (b) overlying unconformities (East Ten-
nessee, Upper Mississippi Valley, and Tri-State). Local depositional controls
include biologically induced unconformities (reefs), sedimentary unconformities
(pinch outs, draping around buried ridges, and mudslides), and, finally, chem-
ically induced unconformities (solution-thinning generating collapse breccias).
The mineralogy of the Mississippi Valley Pb-Zn deposits is rather simple.
The ore material is an admixture of coarsely crystallized marcasite, rather than
pyrite, iron-poor sphalerite, and galena generally low in silver content. In a few
districts such as Southeast Missouri and the eastern margin of the Upper Mis-
sissippi Valley zinc-lead area, finely disseminated chalcopyrite accompanies the
iron and zinc-lead sulfides. The gangue is composed of coarsely crystallized
and vuggy calcite and dolomite with minor barite and fluorite. Locally, as in
Arkansas, barite is the chief economic mineral; in other areas, such as Rosiclar
in Illinois, fluorite is the main ore material. Mississippi Valley-type ore is amen-
able to low-cost concentration methods. Mining is by low-cost excavation meth-
ods including open-cut (Great Slave Lake area) and underground room and pillar
trackless mining.
10.1.2. Geological Synopsis for the North American

Arctic Deposits
The Arctic section of the North American Paleozoic Platform stretches over
a length of 1200 miles and a width of 500 miles east of the Cordillera Belt,
extending west and north of the Precambrian Shield through the Arctic Islands
to Greenland. The discovery and development of Mississippi Valley-type Pb-Zn
deposits in the North American Arctic are quite recent events. Investigations of
the Great Slave deposits in the Pine Point area started in earnest only in the early
1950s. Further discoveries were made in the Arctic Islands in the 1960s, but the
Selwyn Mountain mineralization was disclosed only in the early 1970s.
DETECTION OF MISSISSIPPI VALLEY-TYPE Pb-Zn DEPOSITS 285
In the Pine Point area of the Great Slave region, as many as 40 deposits
totalling some 100 million tons of lead-zinc ore have been discovered to date
within a roughly rectangular area extending in an east-west direction over a
length of 20 miles and a width of 3 miles. The stratigraphic setting is that of a
500-foot-thick middle Devonian carbonate formation with many vertical and
horizontal facies variations which belongs to a reefal complex of regional extent.
The carbonates are overlain by pelitic series (Buffalo River Shales) and underlain
by thick evaporites. The ore deposits are either pipelike, prismatic, or cylindrical
in shape and subvertical in attitude, or tabular and subhorizontal. The local
structural control of ore deposition appears to be a series of three parallel
southwesterly trending "hinges. "(4. 5. 7)
Farther west and north of the Great Slave region, the newly discovered
Selwyn Mountain lead-zinc district stretches along the eastern edge of the Cor-
dillera Belt into the Northwest Territories and the Yukon.(9. 10. 17) A number of
promising lead-zinc discoveries have been made and investigated in the Bonnet
Plume area of the Selwyn Mountains. So far the deposits showing the best
potential are those of Goz Creek and Cypress. The ore material consists of pyrite,
light-colored sphalerite, and galena with inclusions of boulangerite which boosts
the silver content of the mineralization. The sulfides occur in breccia zones that
extend along structural hinges in Lower Cambrian dolomite.
Because of the remoteness of the region and the rigor of the climate, more
than 10 years elapsed between the initial discoveries of promising lead-zinc
mineralization in the Arctic Islands, in the early 1960s, and the full appraisal
and development of the large Nanisivik and Polaris deposits. The Nanisivik
mineralization forms a tabular body of massive pyrite carrying varying amounts
of sphalerite and galena with attractive silver values occurring in Ordovician
dolomites. The Polaris ore consists of early marcasite followed by sphalerite and
galena in a gangue of vuggy and friable calcite with 5% ice content. On the
Greenland west coast, farther east, the large Black Angel deposit was investigated
in the late 1960s and is presently in production. The stratiform mineralization
is up to 60 feet thick and occurs in tremolite-talc marble of late Precambrian
age. The ore consists of massive sphalerite with about 20% pyrite and subordinate
galena low in silver.

Both direct and indirect approaches have been successfully used in the search
for Mississippi Valley-type Pb-Zn deposits. The direct approach relies on the
visual detection of the mineralization or associated clues on surface (prospecting
286 CHAPTER TEN
and mapping) or below surface by mechanical means (drilling). The indirect

approach seeks to detect geophysical signatures of the ore deposits themselves
or the primary or secondary halos associated with them. Commonly the two
approaches are used simultaneously or sequentially in order to reduce the financial
risk of exploration.
10.2.2. Direct Geological Detection

The direct geological detection of "blind" stratiform ore deposits of the
Mississippi Valley type is a great challenge for prospectors and geologists alike.
The task is particulary difficult inasmuch as the deposits have little or no surface
expression in areas of prevailing flat and featureless topography. Geological
detection is made easier in regions where the topography is more deeply incised
and more likely to expose the mineralization, such as in the Selwyn Range and
the Arctic Islands. Photogeologic studies from aircraft platforms may be of
assistance in detecting old karstic features in the subsurface which could lead to
ore.
10.2.3. Direct Detection by Pattern Drilling

Blind, tabular, stratiform lead-zinc deposits of the Mississippi Valley type
are prime candidates for detection by systematic vertical drilling, much as are
the "Colorado" type of uranium deposits of the Western U.S.A. Generally non-
coring drilling methods including percussion and old-style "chum" drilling are
preferred to the much more expensive coring methods. The pattern-drilling ap-
proach was developed and perfected in the older lead-zinc districts of the central
portion of the North American Platform.
The initial stage of the programs calls for drilling for stratigraphic and
mineralogic information on widely spaced lines laid at right angles to the geo-
logical grain of the region. If the regional structural trend is not apparent, the
information drilling is carried on square patterns. Subsequently, the net is tight-
ened progressively over smaller areas selected on the strength of the information-
drilling results. At the final stage, closely spaced drilling on square grids is used
to delineate and sample the deposits detected in the previous stages.

Callahan's 1967 assessment of the merits of indirect geophysical techniques
for the detection of Mississippi Valley-type Pb-Zn deposits was quite pessi-
misticY) He based his verdict on the failure of tests of various techniques
including electromagnetic, self-potential, induced polarization, and seismic sur-
veys to produce diagnostic responses to known deposits in several central U.S.

zinc-lead districts. Gravity was not tested, and the only successful technique
used was the "mise a la masse" from existing drill holes.
However, since the late 1960s, much research work has been conducted on
the geophysical environment of Mississippi Valley Pb-Zn deposits and, as a
result, the detection technology has been considerably improved. It is now ap-
preciated that such deposits do offer good geophysical targets because of the
presence of sizable masses of high specific gravity sulfides with moderate to
good conductivity and high polarizability, occurring at a shallow depth. No case
history of detection of Mississippi Valley-type deposits by airborne geophysical
techniques such as INPUT could be found in the available literature. But we
feel that, on the basis of the amount of geophysical research work presently
done, some new techniques will be perfected in the near future. Thus we are
presenting probability tables for the airborne detection of Mississippi Valley-
type deposits in Section 10.6 of this chapter.
Induced polarization appears to be the most promising ground geophysical
method for the detection of the Mississippi Valley-type deposits, as confirmed
by successful surveys combining induced polarization (I.P.) and gravimetric
work in the Pine Point area of the Great Slave Lake region. (8. 14. 15) Drawbacks
of the induced polarization technique are (a) slowness and high cost and (b)
strong interference by telluric noise in Arctic locations. Gravimetric work which
has proved useful for I.P. confirmation and tonnage calculations is hampered by
sudden variations in overburden thickness and by the presene of buried karstic
features. Modem, fully computerized seismic instrumentation could prove useful
in delineating large buried collapse structures of possible significance as ore
traps.

Advantage may be taken of the different degrees of mobility of the Zn and
Pb cations both in primary and secondary dispersion environments to design a
sequential approach for the geochemical detection of Mississippi Valley-type
deposits. Broad zinc anomalies, which are relatively easy to detect, define target
areas within which lead halos detection will pinpoint smaller target areas for
coverage by more expensive combined geophysical surveys and test drilling.
Bedrock sampling for the detection of primary halos and residual soil and ground-
water sampling have been used with varying degrees of success in the search
for lead-zinc deposits of the Mississippi Valley type in the Paleozoic Platform
of North America.
Bedrock sampling has proved useful in the Arctic region, including the
288 CHAPTER TEN
Selwyn Mountains,(IO, 17) the Arctic Islands, and the Great Slave region.(4) In
well prospected districts such as the Upper Mississippi Valley zinc-lead area,
drill core or cuttings may be used to advantage as sampling medium. (I) The
sampling of residual soil to detect secondary halos has proved quite successful
in nonglaciated areas of moderate relief (Upper Mississippi Valley) but would
be of little assistance when dealing with glacial alluvium in the Arctic region.
Groundwater sampling for zinc and copper has been used successfully by the
first writer in the Upper Mississippi Valley area (see Chapter 12, Ref. 35), and
should prove useful in other regions of moderately well dissected topography
with temperate climate and rainfall above 32 inches per year.
10.3. STATISTICAL MODELING OF THE GEOMETRIC PARAMETERS

OF NORTH AMERICAN ARCTIC DEPOSITS
10.3.1. Scope of the Study

A total of 14 commercial deposits were included in the database to represent
the population of Mississippi Valley-type deposits of the Arctic region. The
sample includes ten deposits from the Pine Point area, one from the Selwyn
Mountains, and three from the Arctic Islands, as shown in Table 10.1. The
sample size is near the lower limit of adequacy for satisfactory statistical pro-
cessing, but this is a risk which has to be accepted when dealing with newly
discovered districts.
TABLE 10.1
List of Mississippi Valley-type Pb-Zn Deposits of
the North American Arctic Paleozoic Platform In-
Northwest Territories & Arctic Islands

Pyramid No. 1 Polaris
Pyramid No. 2 Nanisivik
Coronet Goz Creek
Newconex
Greenland
X-25
R-190 Black Angel
M-40
N-8l5
A-70
K-57

Measurements of three geometric parameters including length and breadth
of the horizontal sections of deposits and shape coefficients were collected from
the 14 deposits included in the sample. The observed data range from 550 to
10,500 feet for lengths, and 225 to 1860 feet for breadths. As for the attitude
parameters of the deposits, the dips are either horizontal or vertical, and, owing
to the inadequacy of available orientation data, no strike orientation measure-
ments are included in the database.
The observed data were grouped into classes, and frequency distributions
were constructed for each parameter. The statistics summarizing these distri-
butions including arithmetic means, modes, standard deviation, and coefficients
of skewness are listed in the lower half of Table 10.2. The coefficient of skewness
which is constructed as the ratio (mean-mode)/standard deviation indicates in a
dimensionless manner the extent of the departure of observed frequency distri-
butions from the symmetry of the standard normal model. The asymmetry is
slight (coefficient near 0.1) for the length and for the shape ratio, but is quite
strong for the breadth parameter (coefficient near 0.6).
TABLE 10.2
Summary of Statistical Modeling of Geometric Parameters of Mississippi Valley-
Type Pb-Zn Deposits of the North American Arctic Paleozoic Platform (Sample
Size: 14 Deposits).
Statistic of Length of hor iz. Breadth of horiz. Shape ratio

25%ile 835 325 0.20

95%L.c.1. 927 392 0.27
G.mean 1337 539 0.40
95% U.c.l. 1927 742 0.60
75%ile 1800 1020 0.60
Disp. coef. 0.09 0.09 0.74
Statistic of
observed data
Arithm. mean 1879 662 0.49

Mode 1700 360 0.51
Stand.dev. 2388 468 0.23
Skewness 0.07 0.64 0.10
290 CHAPTER TEN
The three distributions of observed dimensional data were fitted with log-
normal models. The fit was successfully tested at the 0.05 confidence level by
means of the X square test. The statistics summarizing the fitted models are listed
in the upper half of Table 10.2. The statistics include (a) geometric means and
the upper and lower limits of their 95% fiducial intervals, (b) 25th and 75th
percentiles, and (c) dispersion coefficients. The latter is constructed as the ratio
standard deviation/mean and is used to indicate in a dimensionless manner the
spread of the data about the mean of the fitted models as a reflection of the total
variability of the data. The variability of the length and breadth parameters is
rather small (coefficient near 0.1), but that of the shape coefficient R is much
larger (coefficient near 0.8).

PROBABILITY TABLES
10.4.1. Introduction
For detection purposes, the Mississippi Valley-type deposits are considered
as either vertical pipelike or tabular subhorizontal bodies with randomly orien-
tated horizontal sections of elliptical shape, lying entirely within the average
300-foot vertical range of most commonly used detectors. Therefore, no dip
component is added to the horizontal dimensions of the actual deposits in order
to obtain the target dimensions required for the calculation of detection proba-
bilities (see Section 5.4.2. of Chapter 5).
Only two parameters are required for the construction of the probability
tables, including the mean shape coefficient, and the mean target length within
its 95% fiducial limits. The latter enable the calculation of 95% confidence limits
for the detection probabilities.
10.4.2. Construction of the Probability Tables

Two similar sets of tables were constructed for the display of detection
probabilities. One set covers single detection situations (at lest one target inter-
section by the detector) and the second one deals with confirmed detection (at
least two target intersections). The grid spacings are varied within a range ex-
tending from 200 to 6000 feet in steps of 50-250 feet, depending on the nature
of field programs. For each spacing, the probability of single or confirmed
detection is calculated and displayed in the tables within 95% confidence limits.
A third set of tables was constructed to deal with a converse situation, i.e.,
calculating the grid spacing and associated coverage cost per unit of area required
to ensure a prespecified detection probability level. These tables should prove

very useful to explorationists for planning purposes. The range of probability
levels extends from 0.10 to 0.95, using steps of 0.05.

The tables are grouped into two blocks, each covering a specific type of
exploration situation. The first block covers the detection and confirmed detection
of Mississippi Valley-type deposits by airborne geophysical surveys on parallel
and square grids (Tables 10.4-10.7). The second block deals with ground vertical
detection by geophysical surveys or drilling to a depth of 300 feet (Tables
10.9-10.12).

ARCTIC MISSISSIPPI VALLEY-TYPE Pb-Zn DEPOSITS

Three types of strategies are considered for the detection of Mississippi
Valley-type deposits of the North American Arctic to suit three types of explo-
ration situations. The liminal strategy, on the one hand, minimizes the coverage
cost while ensuring an acceptable level of detection probability of at least 0.500,
which is attractive when planning the coverage of large areas with little known
potential. The maximal strategy, on the other hand, provides near-certain de-
tection (0.980 detection probabilility level) at three to four times the liminal
coverage cost, which restricts its use to the coverage of small areas with high
potential. Finally, the optimal strategy offers a satisfactory compromise between
the two cases listed above, resulting in the highest achievable detection proba-
bility level under strict cost constraints, which fulfils the requirements of the
majority of exploration programs. The detection characteristics pertaining to the
three options are summarized and assembled into Table 10.3 for the reader's
convenience.

The detection parameters pertaining to the liminal strategy are listed in the
leftmost portion of Table 10.3. They include the grid spacings and corresponding
coverage costs associated with the 0.500 detection probability level for various
grid and field program designs. The latter cover airborne geophysical surveys
on parallel and square grids, ground geophysical surveys, and vertical drilling
on square grids.
N
\,!:)
N
TABLE 10.3
Summary of Detection Characteristics of Mississippi Valley-Type Pb-Zn Deposits of the
North American Arctic Paleozoic Platform for Three Types of Detection Strategies.

and Confirmed US$/mile sq. proba- Confirmed
Airborne $404 $756 $416 0.78 $1525 $1925

parallel grid
Airborne $438 - $504 0.77 $885 -

square grid
Ground $1465 $2720 $1353 0.89 $2750 $3285

geophysical 780 f t 460 ft 750 ft 420 ft 365 ft
square grid
Vertical $440,000 $1,090,000 $389,000 0.89 $1,092,000 $1,520,000 n

drilling 780 ft 460 ft 750 ft 420 ft 365 ft I
square grid »
"tl
--I
m
;lO
--I
m
Z
The cost multiplying factor required to confirm single detection averages

1.8 for both types of geophysical surveys but reaches 2.2 for vertical drilling.
The cost of coverage per unit of area may be used as a convenient yardstick of
detection performance for various grid designs which are considered at the plan-
ning stage. As shown on Table 10.3, there is no incentive to select the square
grid instead of the parallel design for airborne geophysical surveys, because of
the resulting 10% coverage cost increase.

The detection parameters corresponding to the maximal option are listed in
the rightmost portion of Table 10.3 in the same manner as in the case of the
liminal option. The cost multiplier required to obtain confirmation of detection
for airborne and ground geophysical surveys is 1.2. The cost factor is slightly
higher (1.5) for vertical drilling programs. It should be noted that both results
are substantially lower than for the liminal option. Thus, for the less cost-
conscious explorationist leaning toward the maximal strategy, it would be ap-
propriate to go a step further and plan for confirmed detection as well.
When comparing the merits of the two airborne survey grid designs using
coverage cost per unit area as yardstick, the results are quite different from those
of the liminal option. The selection of the square grid design becomes an attractive
proposition as it leads to a coverage cost reduction of nearly 42% as compared
to the parallel grid cost.

Optimization procedures based on the efficiency criterion were carried out
for each type of field program, in a manner fully described in Sections 4.4.2 of
Chapter 4 and Section 5.5.4 of Chapter 5. The parameters describing optimal
field program designs are (a) optimal grid spacings and corresponding coverage
costs per unit area and (b) associated probabilities of optimal detection. The
optimal parameters pertaining to airborne geophysical surveys on two grid de-
signs are listed in Table 10.8. Those regarding ground geophysical surveys are
shown in Table 10.12, and those relating to vertical drilling programs in Table
10.13. Finally, all parameters are assembled and listed in the central portion of
Table 10.3 for the reader's convenience.
In order to compare the detection performance of the two airborne grid
designs, we must use the "expected loss" criterion as described in Section 6.5.4.
of Chapter 6, instead of the usual coverage cost criterion because the optimal
detection probability levels are different. As a result, we find that the parallel
grid design should be favored over the square design because the choice of the
latter would lead to a 26% increase in expected loss per mile square.
294 CHAPTER TEN
The reader will find in the following section a series of ten tables covering
the probabilities of detection and the optimization of the search for Mississippi
Valley-type Pb-Zn deposits of the Arctic by airborne and ground programs (pages
294--302).
TABLE 10.4
Probabilities of Detection of Mississippi Valley-Type Pb-Zn Deposits of the
North American Arctic Paleozoic Platform by Airborne Geophysical Surveys of
Parallel Grids.
Randomly orientated elliptical targets with expected major axis = L feet
S feet L= 927 1337 1927 feet
200 1. 000 1.000 1.000

400 0.916 1.000 1. 000
600 0.871 0.913 1.000
800 0.821 0.882 0.920
1000 0.725 0.849 0.898
1200 0.604 0.812 0.877
1400 0.518 0.747 0.854
1600 0.453 0.653 0.829
1800 0.403 0.581 0.802
2000 0.362 0.523 0.753
2200 0.329 0.475 0.685
2400 0.302 0.436 0.628
2600 0.279 0.402 0.580
2800 0.259 0.373 0.538
3000 0.242 0.348 0.502
3200 0.227 0.327 0.471
3400 0.213 0.307 0.443
3600 0.201 0.290 0.419
3800 0.191 0.275 0.397
4000 0.181 0.261 0.377
4200 0.173 0.249 0.359
4400 0.165 0.238 0.342
4600 0.158 0.227 0.328
4800 0.151 0.218 0.314
5000 0.145 0.209 0.301
5200 0.139 0.201 0.290
TABLE 10.5
Mississippi Valley-Type Pb-Zn Deposits of the North American Arctic Paleozoic
Platform by Airborne Geophysical Surveys on Parallel Grids.

Unit Cost = US$70 / line mile

0.10 8800 $ 182.00

0.15 6200 $ 199.61
0.20 4800 $ 217.00
0.25 3900 $ 234.77
0.30 3300 $ 252.00
0.35 2900 $ 267.45
0.40 2500 $ 287.84
0.45 2300 $ 300.70
0.50 2100 $ 316.00
0.55 1900 $ 334.53
0.60 1700 $ 357.41
0.65 1600 $ 371.00
0.70 1500 $ 386.40
0.75 1400 $ 404.00
0.80 1200 $ 448.00
0.85 900 $ 550.67
0.90 600 $ 756.00
0.95 300 $1372.00
296 CHAPTER TEN
TABLE 10.6
Probabilities of Confirmed Detection of Mississippi Valley-Type Pb-Zn De-
posits of the North American Arctic Paleozoic Platform by Airborne Geophysical
Surveys on Parallel Grids.

in the confidence interval: l.c.1. ( geom. mean ( u.c.1.

S feet L= 927 1337 1927 feet
200 1. 000 1. 000 1. 000

400 0.769 1.000 1. 000
600 0.519 0.738 1. 000
800 0.421 0.556 0.801
1000 0.000 0.462 0.638
1200 0.000 0.413 0.537
1400 0.000 0.000 0.472
1600 0.000 0.000 0.431
1800 0.000 0.000 0.407
2000 0.000 0.000 0.000
TABLE 10.7
Probabilities of Detection of Mississippi Valley-Type Pb-Zn Deposits of the
North American Arctic Paleozoic Platform by Airborne Geophysical Surveys on
Square Grids.

in the confidence interval: 1.c_1. ( geom. mean ( u.c.l
S feet L= 927 1337 1927 feet
800 1. 000 1. 000 1. 000

1000 0.942 1. 000 1. 000
1200 0.855 1. 000 1. 000
1400 0.776 0.955 1. 000
1600 0.708 0.894 1. 000
1800 0.649 0.836 1. 000
2000 0.598 0.781 0.958
2200 0.554 0.732 0.916
2400 0.516 0.688 0.875
2600 0.482 0.648 0.834
2800 0.453 0.612 0.796
3000 0.427 0.580 0.761
3200 0.403 0.550 0.727
3400 0.382 0.524 0.697
3600 0.364 0.499 0.668
3800 0.346 0.477 0.641
4000 0.331 0.457 0.616
4200 0.316 0.438 0.593
4400 0.303 0.421 0.572
4600 0.291 0.405 0.551
4800 0.280 0.390 0.533
5000 0.270 0.376 0.515
5200 0.260 0.363 0.498
298 CHAPTER TEN
TABLE 10.8
Optimal Design of Airborne Geophysical Surveys for the Detection of Missis-
sippi Valley-Type Pb-Zn Deposits of the North American Arctic Paleozoic Plat-
form.

respectively
(i) Survey design:

Parallel Grid
(ii) Survey design:

Square Grid
TABLE 10.9
Probabilities of Detection of Mississippi Valley-Type Pb-Zn
Deposits of the North American Arctic Paleozoic Platform by
Ground Geophysical Surveys of Vertical Drilling to a Depth of
300 Feet.

R is defined as the ratio B/L with geometric mean = 0.40.
in feet L = 927 1337 1927ft
300 1.000 1.000 1.000

400 0.979 1.000 1.000
500 0.853 1.000 1.000
600 0.680 0.966 1.000
700 0.529 0.878 1.000
800 0.417 0.758 0.990
900 0.333 0.640 0.950
1000 0.270 0.538 0.885
1200 0.187 0.388 0.718
1400 0.138 0.287 0.565
1600 0.105 0.219 0.448
2000 0.067 0.140 0.292
2400 0.047 0.097 0.203
3000 0.030 0.062 0.130
4000 0.017 0.035 0.073
300 CHAPTER TEN
TABLE 10.10
Mississippi Valley-Type Pb-Zn Deposits of the North American Arctic Paleozoic
Platform by Vertical Drilling to a Depth of 300 Feet.

= $2000 Per Percussion Drill-Hole

Percussion Diamond
0.05 3100 14.72 44.17

0.15 1900 28.71 86.14
0.25 1500 41. 04 123.12
0.35 1250 54.79 164.37
0.45 1100 67.51 202.54
0.55 1000 79.13 237.38
0.65 900 94.58 283.73
0.75 800 115.82 347.47
0.85 750 129.60 388.81
0.95 650 166.81 500.48
TABLE 10.11
Probabilities of Confirmed Detection of Mississippi Valley-Type Pb-Zn
Deposits of the North American Arctic Paleozoic Platform by Ground
Geophysical Surveys of Vertical Drilling to a Depth of 300 Feet.

The expectes shape of the ore deposits is elliptical
R is defined as the ratio B/L with geometric mean = 0.40.

Spacing S" L.c.l. G. mean U.c.1.
in feet L = 927 1337 1927ft
300 1. 000 1. 000 1. 000

350 0.996 1.000 1. 000
400 0.865 1.000 1.000
450 0.445 1.000 1. 000
500 0.226 0.999 1.000
550 0.127 0.927 1.000
600 0.070 0.735 1.000
650 0.039 0.438 1. 000
700 0.022 0.268 1. 000
750 0.011 0.180 0.985
800 0.005 0.120 0.914
1000 0.000 0.023 0.282
1200 0.000 0.009 0.162
1500 0.000 0.000 0.016
302 CHAPTER TEN
TABLE 10.12
Optimal Design of Ground Geophysical Surveys for the Detection of Missis-
sippi Valley-Type Pb-Zn Deposits of the North American Arctic Paleozoic Plat-
form.

respectively.

Probability of Detection: 0.B3 0.B9 0.92
TABLE 10.13
Optimal Design of Vertical Drilling Programs for the Detection of Mississippi
Valley-Type Pb-Zn Deposits of the North American Arctic Paleozoic Platform.

respectively
(i) Percussion drilling:

Unit Cost = $2000/hole
Probability of detection: 0.B3 0.89 0.92
(ii) Diamond drilling:

Unit Cost = $6000/hole
Cost in $thou. per mile square: 675 3B9 219
Probability of detection: 0.B3 0.B9 0.92
I. BARNES, H. L., and LAVERY, N. G., 1977, Use of primary dispersion for exploration of
Mississippi Valley-type deposits, J. Geochem. Expl. 8, 105-115.
2. CALLAHAN, W. H., 1964, Paleophysiographic premises for prospecting for strata-bound base
metal mineral deposits in carbonate rocks, Nev. Bur. Mines Rep. No. 13, Part C, 5-50.
3. CALLAHAN, W. H., and McMURRY, H. V., 1967, Geophysical exploration of Mississippi
Valley-Appalachian type strata-bound zinc-lead deposits, Geol. Surv. Can. Rep. No. 26, 350-360.
4. CAMERON, E. M., 1969, Regional geochemical study of Slave Point carbonates, Western
Canada, Can. J. Earth Sci. 6, 247-268.
5. CAMPBELL, N. 1957, Stratigraphy and structure of Pine Point area, Northwest Territories,
Canada, in Structural Geology o/Canadian Ore Deposits, Vol. 2, Canadian Institute of Mining
and Metallurgy, Montreal, pp. 161-174.
6. JACKSON, S. A., and BEALS, F. W., 1967, An aspect of sedimentary basin evolution: The
concentration of Mississippi Valley-type ore during the late stages of diagenesis, Can. Pet.
Geol. Bull. 15, 383-433.
7. JACKSON, S. A., and FOLINSBEE, R. E., 1969, The Pine Point lead-zinc deposits, N.W.T.,
Canada: Introduction and paleoecology of the Presqu'ile Reef, Econ. Geol. 64, 711-717.
8. LAJOIE, J. J., and KLEIN, J., 1979, Geophysical exploration at the Pine Point Mines Ltd.
zinc-lead property, N.W.T., Canada, Geol. Surv. Can. Econ. Geol. Rep. No. 31,653-664.
9. MACQUEEN, R. W., 1976, Sedimentary zinc-lead deposits, Rocky Mountain Belt, Canadian
Cordillera, Geosci. Can. 3(2) 71-81.
10. MURPHY, J. D., and SINCLAIR, W. D., 1974, Lead-zinc in the Selwyn and McKenzie
Mountains, Can. Min. J. 95(4),40-44.
II. NEWHOUSE, W. H., 1933, The temperature of formation of the Mississippi Valley lead-zinc
deposits, Econ. Geol. 28, 744-750.
12. OHLE, E. L., 1959, Some considerations in determining the origin of ore deposits of the
Mississippi Valley type, Econ. Geol. 54,769-789.
13. ROEDDER, E., 1967, Environment of deposition of stratiform (Mississippi Valley-type) ore
deposits from studies of fluid inclusions, Econ. Geol. Monograph 3, 349-361.
14. SEIGEL, H. 0.,1952, Ore body size determination in electrical prospecting, Geophysics 17(4),
907-914.
15. SEIGEL, H. 0., HILL, H. L., and BAIRD, J. G., 1968, Discovery case history of the Pyramid
ore bodies, Pine Point, N.W.T. Canada, Geophysics 33, 645-656.
16. SKINNER, B. J., 1967, Precipitation of Mississippi Valley-type ores: A possible mechanism,
Econ. Geol. Monograph 3, 363-369.
17. SMITH, C. L., 1974, A newly discovered Mississippi Valley-type lead province, Yukon-
N.W.T., Canada, A.I.M.E. Preprint No. 74-1-306.
CHAPTER ELEVEN
DETECTION OF VEIN-GOLD DEPOSITS OF

THE NORTH AMERICAN SHIELD BY
OPTIMIZED GROUND PROGRAMS
11.1.1. Economic Geology Summary

Gold plays a very prominent role among the "minor" metals recovered from
varied geological sources, because of its importance on the world financial scene
and increasing industrial and ornamental uses. Two types of deposits are the
main sources of gold in the world: vein-type deposits and alluvial deposits. The
vein deposits are as widespread geographically as they are chronologically. They
are found in the Archean Shields of North America, Southern Africa, Western
Australia, India, and Brazil; along the Paleozoic foldbelts of the Appalachian
of North America, Tasman Geosyncline of eastern Australia, and the Urals of
the U.S.S.R.; and finally along the Mesozoic and Cenozoic foldbelts of the
Americas, Southwest Pacific Island Arcs, and East Asia.
The alluvial deposits consist of mechanically disseminated gold in alluvial
turbidites of recent and modem ages (Siberia, Yukon, Alaska, California, etc.)
or in fossil placers (South Africa), possibly subjected to remobilization or me-
tasomatic processes (Nevada, U.S.A). Gold is also won in substantial amounts
from various types of deposits including porphyry-copper-molybdenum and
often spatially related contact metasomatic (Cu-Mo-Au) deposits, skarn deposits
(Cu-Fe-Au) or (W-Mo-Au), volcanogenic massive sulfides, and vein deposits
(Cu-Au), (W-Au), (Sb-Au).
The most prolific period of the history of gold mining occurred on a world-
wide basis between 1845 and 1910. The ensuing decline lasted until 1936, when
a price rise from 20 to 36 U.S. dollars/oz. generated much activity up to the
early postwar years. A steady decline was again reversed in the late 1970s because
of a surge of the price of gold to the peak value of 800 U.S. dollars/oz., settling
to 400 U.S. dollars/oz. in the 1980s. According to Elevatorski,(5) the present
annual production of over 40 million ounces is geographically distributed as
follows: South Africa: 54%, U.S.S.R.: 25%, Canada: 5%, U.S.A.: 2.5%, Aus-
305
306 CHAPTER ELEVEN
tralia: 2.5%, the remainder (11%) being distributed unevenly between 20 coun-
tries throughout the world.
11.1.2. Geological Synopsis for the Archean Vein Gold Deposits

of the North American Shield
11.1.2.1. Economic and Geographic Considerations. The North
American share of world gold production comes mainly from vein gold deposits
occurring in the Canadian portion of the North American Precambrian Shield,
and from the Precambrian horst of the Black Hills in the U.S.A. (14) The Canadian
share of world gold production declined from about 5 million ounces in the
1930s (13% of world total) to about 1.6 million ounces, or about 5% of world
production, some 50 years later. However, at the present time, there is a great
surge of exploration activity taking in older districts as well as some new ones,
such as the Hemlo area near Lake Superior in central Ontario.
A total of 150 known commercial vein gold deposits occur in Canada,
including 120 in the Precambrian Shield. About 55% of the present Canadian
production is won in the Province of Ontario, about 25% is mined in the Province
of Quebec, 15% from the Northwest Territories, and 5% from the Cordilleran
Belt. The Northwest Territory deposits are mainly concentrated in the greenstone
belts of the Yellowknife region, north of Great Slave Lake, but there are scattered
deposits in the remote and hostile Arctic region (Contwoyto Lake). Very few
productive gold deposits occur in the intervening Precambrian regions of the
Provinces of Saskatchewan and Manitoba.
There are 33 known gold districts in the Precambrian portion of the Province
of Ontario. A total of 22 districts lie in the area extending between the 45th and
52nd parallels, which has been most intensively prospected because of its good
accessibility; the remaining 11 districts are scattered north of the 52nd Parallel,
a region far less accessible. The largest gold district is the Porcupine Gold Fields,
near the town of Timmins, in northeastern Ontario. In its heydays ofthe 1940s,
the Porcupine district boasted of 36 operating mines annually producing 1 million
ounces of gold, the leading mines being the Hollinger, Mcintyre, and Dome.
The gold mineralized belt extends farther east through the Kirkland Lake district,
into Northwest Quebec, where about 45 vein gold deposits are known.
11.1.2.2. Geological Considerations. The most common geological

environment for the vein gold deposits of the North American Shield consists
of greenstone belts of Archean age. These belts are made up of intricately folded
and sheared series of metasediments and metavolcanic formations showing a
well-developed "green schist" facies of regional metamorphism. The belts are
DETECTION OF VEIN-GOLD DEPOSITS 307
arcuate in shape; they are encased in granite-gneiss and migmatites and are cut
across by swarms of quartz (felspar) porphyry dykes and stocks.
The veins are generally tabular in shape: they are extensive along strike
and down dip but are very limited in their third dimension. They occur along
shears, fault zones, geological contacts, and in zones of crumpling and drag-
folding, within wide envelopes of hydrothennal alteration consisting mainly of
silicification and carbonatization.
Two main classes of vein gold deposits are generally recognized: (1) cavity-
filling and (2) replacement, while in many deposits the two types occur together.
The cavity-filling type has well-defined walls and the gold mineralization is
restricted to the vein material. In the replacement type, however, the minerali-
zation replaces the wallrock and vein material as well, showing ill-defined bound-
aries. The Kirkland Lake "Break" of northeastern Ontario and the Homestake
deposit of the Black Hills are good examples of the replacement type of vein
gold deposit. The replacement veins are generally much longer (several thousand
feet) and wider (up to 100 feet) than the cavity-filling veins, which are only
2000-3000 feet in length and 15-30 feet in width at the most.
The Porcupine Gold Fields of northeastern Ontario, being the most pro-
ductive and the most intensely investigated gold district of the Canadian Shield,
will serve as an example for the study of the structural control of gold deposition
in the North American Shield. The Porcupine structural setting is that of an
overturned, steeply dipping, tightly folded synclinal structure made up of intri-
cately dragfolded Keewatin basic flows uncomfonnably overlain by Temiskam-
ing greywacke and associated clastic sediments. The Porcupine Trough is cut
across by swarms of quartz (felspar) porphyry dykes and pipelike stocks of
Algoman age. The principal locus of the ore is the Pearl Lake porphyry stock,
which is most intensely sheared and hydrothennally altered, almost beyond
recognition. The structural controls of the deposits are as follows: (a) openings
developed along zones of weakness related to the stratigraphic succession, such
as flow contacts, bedding planes, and also along intrusive contacts; (b) openings
developed by differential movement affecting rocks of differing competencies;
basic flows, conglomerate and greywacke beds and porphyry intrusive behave
as competent fonnations in contrast with tuffs, slate, soapstone, chlorite-
carbonate schist, and pillowed lava flows.
The mineralogy of the deposits is rather simple. The gangue material consists
mainly of quartz with subordinate amounts of carbonates, mainly ankerite and,
locally, tounnaline (Porcupine), accompanied by minor chlorite, sericite, and
graphite, as well as sporadic sulfides. Among the sulfides, pyrite is the most
common and abundant, particularly in the replacement-type deposits; it is fol-
lowed in order of decreasing importance by arsenopyrite, pyrrhotite, and base
308 CHAPTER ELEVEN
metal sulfides. The latter may be quite abundant (Western Ontario), requiring
expensive smelting of the "refractory" ore. Likewise, arsenopyrite is troublesome
as it requires a costly roasting prior to the cyanadation process. In most deposits
of the North American Shield, a large proportion of the gold is associated with
the pyrite and is largely "free milling," occurring as small grains of native gold
or electrum along crystal boundaries, cleavages, or fracture planes. The rest
occurs as tellurides.

The organization of field exploration for vein gold deposits in the Archean
Shield of North America was recently reviewed by Hutson.(7) Two main kinds
of detection approaches are usually combined: (1) a direct approach based on
the visual detection and recognition of deposits or closely related features or on
the mechanical detection of the deposits (drilling), and (2) an indirect approach
relying on the detection of geophysical or geochemical signatures of the deposits
or closely related features. The general sequence of the two approaches is as
follows:
1. Direct approach: a. Airborne surveys: photogeological mapping for

structural information purposes;
b. Ground surveys: surface, prospecting, reconnais-
sance, and systematic mapping for lithological in-
formation; subsurface; trenching and drilling;
2. Indirect approach: a. Airborne surveys: aeromagnetic mapping for struc-
tural and lithological information;
b. Ground surveys: geophysical and geochemical.
As indicated by Hutson, the sequential methodology combining the two

approaches appears to be the most cost-effective search procedure for small and
elusive targets such as the vein-gold deposits of the North American Shield.

Any attempt at visual detection of vein-gold deposits or their immediate
geological environment from satellite platforms has to be ruled out because of
the small size of the targets. But photogeological surveys from aircraft platforms,
are quite useful for regional structural studies. Combined photogeological and
aeromagnetic surveys have greatly assisted the delineation of high priority areas
in the newly discovered Hemlo Gold District of Ontario.
Visual detection from the ground has proved highly effective over the past
50 years. Up to 90% of the gold deposits discovered to date in the North American
Shield were detected by a combination of prospecting, systematic mapping, and
some information drilling. In the present circumstances of high gold price, the
systematic drilling of small areas of high potential of less than a square mile in
extent should be considered in the near future as a valid and effective approach,
since the most obvious deposits have already been discovered.

The past three decades have witnessed many major discoveries of base
metal deposits with considerable assistance from airborne and ground geophysical
methods; not so for gold deposits. During the same period, only a handful of
gold discoveries were made, with geophysical methods playing only a minor
role. An obvious reason is that, whereas gold possesses high specific gravity
and electrical conductivity, its presence in a widely disseminated state in the
vein environment cannot be readily detected by most known geophysical tech-
niques relying on gravity or magnetic or electromagnetic contrast between vein
and wallrock. Therefore, two strategies have to be considered: (1) search for
structural clues using magnetics, electromagnetics (VLF), or resistivity surveys,
and (2) search for associated minerals which are more abundant and widespread
than gold itself and susceptible of easier geophysical detection, such as the iron
sulfides.
Examples of the successful application of the first strategy were described
by Kelly in the Porcupine Gold District. (8) The program combined magnetic and
resistivity surveys to take advantage of the high resistivity of the quartz material
in the cavity-filling type of vein gold deposits. A combination of magnetic and
electromagnetic (VLF) surveys is presently being used to advantage in the Hemlo
district mentioned above. The application of computerized induced polarization
equipment, both in the time and frequency (phase angle) domains, is an example
of the second strategy.(ll) Seigel warns, however, that only low amplitude and
localized responses can be expected from such small targets as vein gold deposits
usually offer; thus much more care and attention to detail are required than is
customary in the search for base metal targets.

The search for primary halos by bedrock sampling is quite difficult because
of the small extent of the halos and the general scarcity of outcrops. Some
310 CHAPTER ELEVEN
assistance could be gained from drill core sampling in well prospected areas.
Geochemical detection of secondary halos in residual soils is not applicable in
most regions of the North American Shield because of the extent and thickness
of transported glacial alluvium.

VEIN-GOLD DEPOSITS

A total of 50 vein gold deposits were selected and included in the database
to represent the statistical population of all known and undiscovered vein gold
deposits of the North American Shield. The deposits are listed by name and
province of occurrence in Table 11.1. The geographic breakdown of the sample
is as follows: the largest subsample consisting of 31 deposits covers the northeast,
northcentral, and western portions of the Province of Ontario; the second largest
subsample including 16 deposits represents the northwest portion of the Province
of Quebec, Canada's second largest producer; the third subsample, comprising
three deposits, covers the Northwest Territory portion of the shield. The size of
the sample is quite satisfactory for statistical processing purposes.

For detection purposes, the geometry of vein-gold deposits is quantitatively
described by means of three parameters of a dimensional nature and two atti-
tudinal parameters. The dimensional parameters include the length, breadth, and
shape ratio of the horizontal section of the portion of deposits which lies within
an average 300-foot detection range. The attitudinal parameters include the uno-
riented true dip of the deposits and the strike direction referred to True North
and measured in degrees within the right half-circle.
The range of the measurements of geometric parameters from the 50 deposits
included in the sample is as follows: from 355 feet to 6500 feet for length, 10
(cavity-filling type) to 200 feet (replacement type) for breadth, and 45-85 degrees
for dip. The measurements were grouped into classes and frequency distributions
were constructed. The statistics summarizing these distributions are listed in the
lower half of Table 11.2. They include arithmetic means, modes, standard de-
viations, and coefficients of skewness. The latter gives a useful indication of the
degree of departure from normal symmetry shown by the distributions of observed
data. The skewness of the distributions is rather slight for the length and shape
TABLE 11.1
List of Vein-Gold Deposits of the North American Shield Included
in the Database.
Ontario
Central Patricia Hallnor
Cochenour Willans Dome
Gold Eagle Preston East Dome
Madsen Red Lake Paymaster
McKenzie Red Lake Buffalo Ankerite
Hasaga Aunor
Little Long Lac Delnite
Jellicoe Coniaurum
Tombill Moneta Porcupine
North Empire Hollinger
Hard Rock McIntyre
Renabie Kerr Addison
Jerome Sylvanite
Leitch Wright Hargeaves
Pamour Bidgood
Broulan
N. l'i. Quebec
Omega Arntfield
Lamaque Camflo
East Malartic Canadian Malartic
Central Cadillac Granada
O'Brien Agnico
Belleterre
Northwest Territories
MacWatters
Powell Rouyn Giant Yellowknife
Wasa Cons. Discovery
Beattie Camlaren
Francoeur
ratio parameters (coefficient value = 0.01), while a moderate skewness (coef-

ficient range 0.2-0.5) is noted for the other three geometric parameters.
dimensional parameters, based on the results of the X2 goodness of fit test at the
0.05 confidence level. The models fitted at the same level of confidence to the
observed distributions of attitudinal parameters include a normal model for dips
and a circular normal model for strike directions.
Statistics summarizing the fitted models are listed in the upper half of Table
11.2. Among them are the geometric means of the fitted lognormal models,
312 CHAPTER ELEVEN
TABLE 11.2
Statistical Modeling Summary for Vein-Gold Deposits of the North American
Shield (Sample Size: 50 Deposits).
Statistic Length of Bread th of Shape Str ike direct. Dip in

of fitted hor i z • sec t. horiz. sect. ratio from T.North degrees
25%ile 1060 18 0.007 55 66

95%L.c.1. 1466 21 0.012 72 77
Mean 1753 18 0.016 81 85
95%U.c.1. 2097 38 0.022 89 93
75%ile 2800 60 0.040 95 107
Disp. coef. 0.08 0.32 0.251 0.22 0.32
Statistic
of observed
data
Arith. mean 2207 51 0.029 81 84

Mode 2200 38 0.028 70 70
Stand.dev. 1588 59 0.031 34 30
Skewness 0.01 0.21 0.032 0.33 0.50
arithmetic mean of normal model, and circular mean of circular normal model,
along with the limits of their 95% confidence intervals. Other listed statistics
include the first and second quartiles of the fitted distributions and the dispersion
coefficients, which give an indication of the spread of the data about the mean
of the fitted models. A small coefficient (0.1) indicating a low variability is
shown for the length parameter, while a moderate variability affects the breadth
and shape ratio parameters (coefficient ranging from 0.1 to 0.5).

PROBABILITY TABLES
For the purpose of calculating detection probabilities, the vein-gold deposits
of the North American Shield are considered as dipping, orientated tabular bodies
with horizontal cross sections of elliptical shape. Therefore, target dimensions
for detection purposes are not the same as that of the deposits; they have to be
recalculated for each deposit by introducing a dip component which is computed
as described in Sections 2.2.3. of Chapter 2, and 5.4.2 of Chapter 5. As a result
of the recalculation, the shape coefficients are substantially inflated, thus in-
creasing the probabilities of detection. For example, the mean shape ratio of the
horizontal sections of vein gold deposits themselves (0.016) becomes equal to
0.09 for the targets in the context of vertical detection and is increased to 0.19
when considering 55-degree angled detection.
Since strike orientation data were available for most of the 50 deposits of
the sample, the orientation factor is introduced in the construction of the tables
to improve the probability of success of the various field programs. Based on
the results of a dynamic programming study, it was found that the probabilities
of vertical detection by geophysical surveys or drilling are maximized if the grid
is laid at angle of 18 degrees ± 10 with the expected strike direction. If we
consider the 55-degree angled detection, the optimal grid orientation is at 30
degrees ± 10 with the expected strike direction.
The latter may be estimated from the results of previous surveys or from
photogeological studies. If no prior information is available, the mean of the
fitted circular normal model is a statistically valid estimate to be considered for
grid design (See Table 11.2, upper half).
11.4.2. Description of Probability Tables

Two similar table designs are used to display the probabilities of single
detection and confirmed detection of vein gold deposits by ground surveys and
drilling programs. The grid spacings are varied within a range extending from
100 to 1500 feet in steps of 50-100 feet, depending on the nature of the field
programs. For each grid spacing, the probability of single or confirmed detection
is calculated within a 95% confidence interval, which is based on the 95%
confidence interval of the mean target length. (See Table 11.2, upper half.)
A second set of tables is constructed for exploration planning purposes.
The tables display the grid spacings and corresponding coverage costs per unit
of area which are required to provide prespecified detection probability levels.
The range of probability levels extends from 0.05 to 0.95 using steps of 0.05.
11.4.3. Organization of Detection Probability Tables

No probability tables were constructed to cover the airborne geophysical
detection of vein gold deposits, because no known methods are available at the
present time. The probability tables are organized into two blocks. One block,
comprising Tables 11.4-11.6 covers vertical detection by ground geophysical
314 CHAPTER ELEVEN
surveys or vertical drilling to a depth of 300 feet. The second block covers the
55-degree angled detection by drilling (Tables 11.8-11.10).
11.5. DESIGN OF THREE STRATEGIES FOR THE DETECTION

OF ARCHEAN VEIN-GOLD DEPOSITS OF THE NORTH
AMERICAN SHIELD

Three types of detection strategies are offered to explorationists in order to
meet the requirements of three kinds of exploration situations which could arise
in the search for vein-gold deposits of the North American Shield. The first one,
the liminal option, should prove attractive when dealing with the coverage of
large areas of little known potential, because it provides a detection probability
level of at least 0.500, while minimizing the coverage cost. The maximal strategy
is a high-cost approach (three to five times that of the liminal option) providing
a high detection probability level (0.980), which should be restricted to the
coverage of small areas (less than one square mile) of high potential in well-
prospected regions. Finally, the optimal strategy is offered to exp1orationists as
a more suitable compromise between the two extreme options mentioned above,
when faced with the problem of covering moderately large areas of good potential
within strict budget limits. The detection parameters pertaining to three strategies
are assembled in Table 11.3 for the reader's convenience.

The detection parameters for ground geophysical surveys and vertical and
angled drilling in the context of the liminal option are listed in the leftmost
portion of Table 11.3. The parameters include the grid spacings and correspond-
ing coverage costs associated with the detection probability level of 0.500.
The cost multiplying factor required to ensure confirmation of single de-
tection is 2.1 for ground geophysical surveys and 2.6 for both vertical and angled
drilling to a vertical depth of 300 feet. A comparison of the detection perfor-
mances of vertical vs. 55-degree angled drilling is of interest to explorationists
for planning purposes. The most convenient yardstick is the coverage cost per
unit of area since the detection probability is held at the constant level of 0.500.
Angled drilling should definitely be favored over vertical drilling since it leads
to a 37% coverage cost reduction. But, if no prior knowledge of the likely
direction of the dip is available from previous surveys or photogeological studies,
vertical drillmg may be a safer option m order to avoid the risk of drilling down
dip and possibly failing to recognize the target.
o
~
m
q
TABLE 11.3 (5
Summary of Detection Characteristics of Vein-Gold Deposits of the North American Shield Z
for Three Types of Detection Strategies. o
..."
Liminal detection Optimal detection Maximal detection <

m
Coverage cost: Coverage cost: Z
US$/mile sq. US$/mile sq. Cl
Type of survey Grid size: feet Coverage cost Detection Grid size: feet or-
and Confirmed US$/mile sq. proba- Confirmed o
grid geometry Detection detection Grid size: ft bility Detection detection om
-,::,
Ground $1840 $3920 $1946 0.66 $4308 $10,220
ofJl
square grid
~
Vertical $605,200 $1,760,000 $675,000 0.66 $2,077,000 $6,220,000
square grid
55-degree
angled $424,200 $1,090,000 $536,000 0.90 $1,160,000 $3,490,000
square grid
Optimal 45-deg
angled - - $491,000 0.88 - -
drilling 800 ft
square grid
I
W
...I.
I.n
316 CHAPTER ELEVEN

The detection characteristics pertaining to the maximal strategy are listed
in the rightmost section of Table 11.3, in the same manner as for the liminal
option. If we require confirmation of single detection, the cost multiplier is 2.6
for ground geophysical surveys, and it escalates to 3.3 for both types of drilling
program. The maximal cost multipliers are substantially higher than their liminal
counterparts, contrary to what has been noted for all types of ore targets pre-
viously covered in this book. This difference plainly underscores the considerable
difficulties to be faced when searching for targets as small as the vein-gold
deposits are. As was the case for all other types of dipping deposits previously
covered, the choice of angled over vertical drilling to the same depth is advan-
tageous since it leads to a 52% coverage cost reduction, providing some indication
of the likely direction of dips is available.

As described in Sections 4.4.2 of Chapter 4 and 5.5.4 of Chapter 5, a three-
stage dynamic programming procedure was conducted to determine the optimal
parameters of the control grids for the ground field programs required for the
detection of vein-gold deposits. The parameters include (a) the optimal orien-
tation of the grids with respect to the expected strike direction of targets, (b)
optimal grid spacing and corresponding coverage cost per unit of area, as well
as associated probability of detection, and finally (c) the optimal drilling angle
when the likely direction of the dip is known. The optimal parameters are listed
in the following tables: Tables 11.7 for ground geophysical surveys, and 11.11
for vertical and angled drilling programs.
In order to compare the detection performances of the two types of drilling
program, we should use the "expected loss" criterion, as described in Section
6.5.4 of Chapter 6, rather than the coverage cost criterion previously used,
because of differences in optimal probability levels in the two cases. We find
that the appropriate choice is that of the optimally angled drilling (45 degrees),
because it leads to reduction of expected loss of 74% as compared to the vertical
drilling option.
The following section includes eight tables which cover the probabilities
of detection and optimal designs of ground programs in the search for vein-gold
deposits of the Shield (pages 317-323).
TABLE 11.4
Probabilities of Detection of Vein-Gold Deposits of the North
American Shield by Ground Geophysical Surveys or Vertical Drill-
ing to a Depth of 300 Feet.

strike line.
in feet L = 1466 1753 2097ft
250 1. 000 1. 000 1. 000

300 0.986 1. 000 1. 000
350 0.919 0.990 1. 000
400 0.803 0.947 0.995
450 0.685 0.860 0.974
500 0.579 0.759 0.920
600 0.420 0.576 0.756
700 0.310 0.439 0.600
800 0.237 0.339 0.478
900 0.188 0.268 0.383
1000 0.152 0.217 0.311
1100 0.126 0.180 0.257
1200 0.105 0.151 0.216
1300 0.090 0.129 0.184
1400 0.078 0.111 0.159
1500 0.068 0.097 0.138
318 CHAPTER ELEVEN
TABLE 11.5
Detection of Vein-Gold Deposits of the North American Shield by Ver-
tical Drilling to a Depth of 300 Feet.

probability Level Grid Size Drilling Cost Per Mile Square

Of Detection In· Feet In U.S.$. Thousands
0.05 1950 82.93

0.15 1200 175.61
0.25 950 258.82
0.35 800 347.47
0.45 700 438.91
0.55 650 500.48
0.65 550 675.43
0.75 500 803.19
0.85 450 974.35
0.95 400 1211.54
TABLE 11.6
Probabilities of Confirmed Detection of Vein-Gold Deposits of
the North American Shield by Ground Geophysical Surveys or
Vertical Drilling to a Depth of 300 Feet.

The expected shapeof the ore deposits is elliptical
strike line.

in feet L = 1466 1753 2097ft
100 1. 000 1. 000 1. 000

150 C.987 1. 000 1. 000
200 0.928 0.969 0.996
250 0.853 0.916 0.960
300 0.740 0.852 0.915
350 0.367 0.759 0.862
400 0.230 0.531 0.798
450 0.155 0.284 0.689
500 0.111 0.198 0.368
550 0.081 0.145 0.264
600 0.060 0.110 0.196
700 0.033 0.066 0.119
800 0.018 0.040 0.077
TABLE 11.7
Optimal Design of Ground Geophysical Surveys for the Detection of Vein-Gold
Deposits of the North American Shield.

respectively.

320 CHAPTER ELEVEN
TABLE 11.8
Probabilities of Detection of Vein-Gold Deposits of the North
American Shield by 55-Degree Angled Drilling to a Vertical Depth
of 300 Feet.

strike line.
probability of detection
Spacing S L.c.1. G. mean U .c.1.
in feet L = 1466 1753 2097 ft
400 1. 000 1. 000 1. 000

450 0.995 1. 000 1. 000
500 0.967 1. 000 1. 000
550 0.906 0.991 1. 000
600 0.820 0.966 1. 000
650 0.727 0.916 0.993
700 0.638 0.847 0.975
800 0.498 0.692 0.891
900 0.396 0.559 0.766
1000 0.321 0.457 0.640
llOO 0.265 0.379 0.537
1200 0.223 0.318 0.454
1300 0.190 0.271 0.388
l400 0.164 0.234 0.335
1500 0.143 0.204 0.292
TABLE 11.9
Detection of Vein-Gold Deposits of the North American Shield by 55-


0.15 1700 124.09

0.25 l400 167.46
0.35 1200 214.39
0.45 1000 289.81
0.55 900 346.39
0.65 850 382.03
0.75 800 424.21
0.85 700 535.83
0.95 650 611. 00
322 CHAPTER ELEVEN
TABLE 11.10
Probabilities of Confirmed Detection of Vein-Gold Deposits of
the North American Shield by 55-Degree Angled Drilling to a
Vertical Depth of 300 Feet.

R is defined as the ratio B/L with geometric mean 0.19
strike line.

in feet L = 1466 1753 2097ft
250 1. 000 1.000 1.000

300 0.977 1.000 1.000
350 0.909 0.985 1.000
400 0.818 0.933 0.996
450 0.672 0.865 0.960
500 0.455 0.774 0.910
550 0.350 0.591 0.849
600 0.268 0.451 0.771
700 0.152 0.290 0.482
800 0.081 0.182 0.333
900 0.043 0.109 0.225
1000 0.022 0.064 0.152
1100 0.011 0.038 0.098
TABLE 11.11
Optimal Designs for the Detection of Vein-Gold Deposits of the North Amer-
ican Shield by Drilling to a Vertical Depth of 300 Feet.

respectively
(al vertical drilling:

(bl Angled drilling at 55 degrees:

(cl Angled drilling at optimal angle = 45 degrees:

1. AMIRYAN, S. 0., 1960, Mineralogy of gold deposits, Akad. Nauk. Arm. S.S.R. 31(1), 43--48.
2. BOYLE, R. W., 1961, The geology, geochemistry and origin of the gold deposits of the
Yellowknife District, N.W.T., Geol. Surv. Can. Mem. No. 310.
3. COOKE, H. C., and JOHNSTONE, W. A., 1932, Gold occurrences of Canada, a summary
account, Geol. Surv. Can. Econ. Geol. Ser. No. 10.
4. DOUGHERTY, E. Y., 1939, Some geological features of the Kolar, Porcupine, and Kirkland
Lake gold fields, Econ. Geol., 34, 622-653.
5. ELEVATORSKI, E. A., 1981, Gold Mines of the World, MINOBRAS, Santa Monica, Cali-
fornia.
6. EMMONS, W. H., 1937, Gold Deposits of the World, McGraw-Hili, New York.
7. HUTSON, R. J., 1980, Exploration models for gold in the Archean, Can. Min. J. 101(4),
66-68.
324 CHAPTER ELEVEN
8. KELLY, S. F., 1957, Resistivity and magnetic surveys in 1936 on the Brou1an-Porcupine gold
prospect, South Porcupine, Ontario, Canada, in Methods and Case Histories in Mining Geo-
physics, Sixth Commonwealth Mining and Metallurgy Congress Volume, Montreal, pp. 283-291.
9. MIDDLETON, R. S., and CAMPBELL, E. E., 1979, Geophysical and geochemical methods
for mapping gold-bearing structures in Nicaragua, Geol. Surv. Can. Econ. Geol. Rep. No. 31,
pp. 779-798.
10. MOORE, E. S., 1940, Genetic relations of gold deposits and igneous rocks in the Canadian
Shield, Econ. Geol. 35, 127-139.
11. SEIGEL, H. 0., JOHNSON, I., and HENNESSEY, J., 1984, Geophysical aids to gold ex-
ploration, Northern Miner March 1st, BI7-BI8.
12. SIXTH COMMONWEALTH MINING AND METALLURGY CONGRESS VOLUME, 1957,
Structural geology of Canadian ore deposits, Canadian Institute of Mining and Metallurgy,
Montreal, Canada.
13. WHITE, W. H., 1943, Mechanism and environment of gold deposition in veins, Econ. Geol.
38, 512-532.
14. WRIGHT, L. B., 1937, Gold deposits in the Black Hills of South Dakota and Wyoming,
A.l.M.E. Trans. 126. 390-425.
CHAPTER TWELVE
OPTIMAL SELECTION OF EXPLORATION

TARGETS FOR DRILL TESTING
As outlined in Section 1.2.3 of Chapter 1, the mineral exploration sequence
starts with preliminary planning followed by field planning of surveys and data
processing, and ending with the assessment of results for further decision making.
The initial planning requires many decisions of critical importance, such as the
choice of type(s) of targeted ore deposits, selection of regions for the search and
prospecting areas within the selected regions, and finally the choice of meth-
odology of coverage of prospecting areas, including technology, and particularly,
survey design, whose optimization is the main topic of the book.
The following stage, field planning, deals largely with the scheduling and
logistics required to organize and implement in an optimal manner the field
programs that were planned in the previous stage. The purpose of the operation
is the acquisition of new data, which are assembled and processed in order to
delineate exploration targets for detailed investigation. Most field programs gen-
erate a large number of exploration targets of diverse merits from which only a
limited number are to be selected for drill testing, because of time and budget
constraints. The optimization of the screening of exploration targets is the topic
of the present chapter.
12.1.2. Data Acquisition

There are three types of geodata, which have to be collected and processed
separately because of their difference in structure and nature. The first type,
geological data, is mostly qualitative in nature; the second type, geochemical
data, is mainly quantitative and stochastic in nature, and the third one, geo-
physical data, is chiefly quantitative and deterministic.
Geological and geochemical processes are rather poorly understood and,
for the most part, are not subject to reproduction by experimentation within
present laboratory capabilities. As a result, there is a lack of measurable causative
325
326 CHAPTER TWELVE
relationships between observed features and underlying processes, which hinders

the construction of quantitative models. While geochemical data are essentially
numerical, geological data are mainly descriptive and have to be transferred from
the nominal scale of measurement to the "metric" scale. This process of "quan-
tification" is necessarily fraught with difficulties such as lack of objectivity, risk
of error, and loss of information. Certain types of multivariate statistical tech-
niques such as factor analysis and cluster analysis are favored when dealing with
rather loosely structured data of the geological or geochemical type.
Geophysical data are much more tightly structured than the two previous
types. Mathematical models of most geophysical phenomena have been con-
structed and tested either in the laboratory or by computer simulation. Geo-
physical data, always of a numerical nature, comprise (1) discrete measurements
taken at grid intersections on the ground, and (2) continuous readings along
profiles by airborne surveys. The latter are "digitized," i.e., sampled at regular
intervals, as required by statistical processing.
12.1.3. Delineation of Exploration Targets

The delineation of exploration targets is essentially sequential in nature.
The procedure calls for the investigation of areas of decreasing size to be covered
by means of technologies of increasing resolution, in order to delineate explo-
ration targets of increasing merit. At each stage of the sequence, the first step
of the procedure is the recording of the complexly interrelated geological, geo-
chemical, and geophysical signatures which are diagnostic of the environment
of deposits of specific genetic types.
The second step is the quantification of geo-signatures as "anomalies" which
are defined with respect to "thresholds." The latter are statistically determined
as a reflection of the variability of the measured property throughout the area of
interest. Thresholds are often expressed in terms of multiples of the standard
deviation of the measurements collected over the whole surveyed area. (14) A
more flexible definition of thresholds is provided by areal trend analysis, which
may be applied in three ways: (1) least-squares fitting (Chapter 12, Ref. 1), (2)
moving average (Chapter 2, Ref. 11), and (3) block averaging (Chapter 12, Ref.
8).
The third step is the actual delineation of exploration targets. The most
commonly used approach is a "univariate" one, requiring the physical super-
imposition of a number of anomaly maps, one for each type of surveyY4) The
spatial coincidence of the maximum number of anomalies of different types leads
to the delineation of exploration targets for selection and testing. The multivariate
approach, which is gaining increasing attention, was originally advocated by
SELECTION OF EXPLORATION TARGETS 327
Agterberg(1) and Culbert (7) almost a decade ago. Instead of considering each
type of measurement independently from the others, the multivariate approach
deals with "statistical patterns" based on all types of measurements at each
sampling point. The spatial characteristics of the patterns are analyzed statistically
and the targets are delineated in an objective manner.
12.1.4. Selection of Exploration Targets for Testing

During the past two decades, explorationists have increasingly turned to
large-scale integrated programs based on various combinations of sophisticated
airborne and ground survey technologies in order to pinpoint exploration targets
for test drilling. This approach obviously leads to a problem of optimal choice
under uncertainty, when the exploration staff wish to select a limited number of
drilling locations of greatest merit from a large number of potential targets.
The magnitude and difficulty of the task is well documented in the available
literature. (18) From a total of 700 ground checked geophysical anomalies within
a 30,000 square mile area of India (project Hardrock), 25 targets were selected
for drill testing, from which six emerged as potentially commercial deposits.
Similarly, only 20 drill testing targets, four of which yielded sulfide mineral-
ization of commercial potential, were selected from a total of 950 ground con-
firmed anomalies during a large-scale search for volcanogenic massive sulfide
deposits in the Shield area of Northwest Quebec during the period 1968-1970.
For a long time, the problem of target selection has been dealt with in a
subjective manner by experts "eyeballing" the various univariate maps and re-
lying on semiquantitative schemesY4. 35) However, as the ore search becomes
more complex and sophisticated, explorationists are increasingly turning to mul-
tivariate statistical decision techniques to guide objectively the selection of targets
for testing. (1.7.8. 15. 16. 19)
Two multivariate approaches to the optimal selection of exploration target
for testing are open to explorationists. The first one relies on "control locations,"
while the second one does not. In the former approach, multivariate patterns
associated with the targets to be screened are statistically compared to the patterns
associated with "control locations" known to be ore-bearing, and may be "rec-
ognized" as either potentially ore-bearing or barren. The "pattern recognition"
approach(ll) is gaining increasing attention among explorationists, as witnessed
by the number of papers devoted to the topic which are listed at the end of the
chapter (Ref. 20--39). The second approach, developed only very recently by
the second writer, tackles the more difficult problem of optimally selecting
exploration targets for drill testing in little-known regions, where no adequate
control locations are available.
328 CHAPTER TWELVE
12.2. OPTIMAL SELECTION OF EXPLORATION TARGETS BASED ON

CONTROL LOCATIONS
12.2.1. Summary of Procedure

The multivariate statistical approach described below proceeds sequentially
in four stages, as illustrated by Figure 12. 1, as follows.
Stage 1: Data Acquisition and Processing. The initial stage is carried
out in four steps: (1) acquisition of field data pertinent to the targeted ore deposit
type within the selected prospecting area, (2) acquisition of the same kind of
data from known occurrences of the selected deposit type in neighboring regions
of geological make-up similar to that of the prospecting area, (3) statistical
processing of both sets of data, and (4) delineation of targets to be screened.
Stage 2: Selection of Variates Required for the Construction of a
Geostatistical Model. First, the most appropriate economic parameter (gross
value, metal tonnage, grade) is selected as dependent variate. As a second step,
a multivariate screening of the most statistically significant independent variates
is conducted. Finally, measurements of the selected variates are obtained for all
control locations and all targets under study.
Stage 3: Construction of a Geostatistical Model. The first step is the
construction of a classifying model based on the selected geovariates, which is
"calibrated" on the ore and non ore control locations in order to provide the
probabilities that targets under study belong to either class. Secondly, a regression
model also based on the selected geovariates is constructed and calibrated on all
control locations in order to provide estimates of the economic parameters rep-
resented by the dependent variate.
Stage 4: Construction of a Payoff Criterion for the Optimal Selection
of Exploration Targets. A payoff criterion involving probabilities of success
and failure, regression estimates of economic worth and an appropriate cost
function is constructed to rank the targets under study in order of decreasing
merit, leading to the selection of a number of top-ranking targets, subject to
budget constraints.
12.2.2. Screening of Geovariates

12.2.2.1. Introduction. Dependent and independent geovariates are both
required for the construction of a geostatistical model to be used for the screening
of exploration targets. The nature of the dependent variate has to reflect quan-
t ~ [];J
surveys
-{
data storage
Economic data
1
. ( Exploration cost
on ore deposits \ data
Continuously recorded
quantitative data
( Geo-environment data
. L
••
(airborne surveys) on ore deposits
•• -
( ( ")
f\
Cluantitative Gualitative
I
J1
discrete data geological
I
(ground surveys) data
\ )
I Digitization
I I
Quantification
I Dichotonization
into binary data
I
+
1_ Data compression
by Prj~cipal component
analySIS
I Data compression
by Characteristic
analysis
1
I
Ouantification
~
-.. (i--"-----
<Selection of
dependent '\ \
Selection of
independent
-'-"\
)
gee-variates " . ___ ~-vanates ~/
:
1
Stepwise regression \
analysis
- !
~~
\
®
data
storage
signific.ant
geo-varlates
Geo-statisticaJ Model
( Multiple regression model )
........
Significant
geo-variates
( Cost function model) data from all
Prior probability
Bayesian classification model exploration
of success
targets
I Rating of exploration targets

I
FIGURE 12.1
Flow Diagram of the Optimal Selection of Exploration Targets Based on Control
Locations.
330 CHAPTER TWELVE
titatively the concept of "economic worth" in a manner which is dictated by the

purposes of the study and the availability of data. Economic worth may be
conveniently measured either by the expected gross value, or the discounted
expected gross profit, or the total metal tonnage, or the average grade.
Independent variates quantitatively describe the geoenvironment of the var-
ious control locations and targets under investigation. Generally, one starts with
a very large array of geovariates which are subjectively thought to be related to
the occurrence of targeted ore deposit types in the geological environment pre-
vailing in the prospecting area. Objective statistical procedures are then required
to screen the geovariates in two steps for relevance and statistical significance.
The first step is referred to as "data compression" and the second one as "feature
selection," according to the general "pattern recognition" terminology.
12.2.2.2 Data Compression. Data compression is an initial screening
procedure required to reduce objectively the large size of the array of subjectively
chosen independent geovariates. Principal component analysis (PCA) is a stan-
dard statistical technique applicable to geophysical, geochemical, and quantified
geological and other "metric scale" data. PCA is equivalent to a change of
coordinates in the n-dimensional space, which works best when dealing with
well-structured dataY· 4) However, when dealing with more loosely structured
data as are most quantified geological and geochemical measurements, Culbert(7)
and Klovan(16) recommend the use of the factor analysis technique, which im-
proves on PCA by adding a "rotation" of the system of reference.
Compression of descriptive (nominal scale) geological data is rather trou-
blesome because a partly subjective quantification procedure has to be carried
out before going to PCA or factor analysis. Botbol(5.6) devised the technique of
Characteristic Analysis to obviate the difficulty. First the descriptive data are
simply and objectively quantified into a binary system: presence coded as I, and
absence as O. The procedure of characteristic analysis, rather simple to handle
by computer, is illustrated by Figure 12.2. The writer successfully used the
technique to compress an array of 170 geological variates qualitatively describing
ore-bearing porphyry-type deposits of the American Cordilleras into a more
compact set of 44 most typical geological characteristics.(20)
12.2.2.3. Feature Selection. Feature selection is a crucial stage of the
procedure leading to the construction of a geostatistical model. Its principal
purpose is to select objectively among the previously compressed array of in-
dependent geovariates only those which are statistically correlated with the de-
pendent variate. This is done by stepwise regression analysis based on a linear
multiple regression function which is designed to approximate objectively and
(l) PROCESSING OF DATA

COLUMN IPOIIPHVR'I'
<;COPPER DEPOSITS)
ll141111 100 100
~;~-I :[~ g! rl ~ :!]x

100111
010010 5
2252253442]A
3.
011101 = [2 2 3 3 2 1 c
D 110 1 1 110 110111 533643 D
i!I E 0 1 101 111

01011001
101110
111110
010011
444463 E
2 3 133 4 I'
CODED DATA MATRIX II TRANSPOSE MATRIX • PRODUCT MATRIX

(6 18) (816) (616)
(2) CALCULATION OF LENGTH OF INFORMATION

VECTORS OF CHARACTERISTICS
A ./(5fl+ (2"'+(2'" +(5"'+(4"'+(2'" • 8.87

8 ./(2)2 +(5,z +(2,z +(3,z +(4,z +(3,z • 8.45
C ./ (2; +(2)2 +(3ii +(3,z+(2,z +(1,z • 5.42

D ./ (5fl +(i' +(3;Z +(6)2 +(4;Z+(3;Z • 10.20
E ./ (4)1 +(4'" +(4)2 +(4,z +.,z +(3,z • 10.65
F ./ (2,z +(31 +uf +(3f+(3)z+(4)2 • 5.82
>-
!i!~
11 E
2) D
3) A
(10.65)
(10.20)
(8.87)
I
(3) RANKING OF TYPICALITY OF CHARACTERISTICS
j MOST TYPICAL. CHARACTERISTICS
!g 4) G (8.45)-CUT-OfF
~~ 5) F (5.82)
6) C (5.42)
FIGURE 12.2
Flow Diagram of Characteristic Analysis.
quantitatively the complex relationship existing between the geoenvironment

(independent variates) and economic worth (dependent variate).(l, 4)
The linear regression equation is written in its most general form as
(1)
where y is the dependent variate, and Qi and Xi the coefficients and independent
variates representing the geo-factors i = 2 . . . . n. The theory of linear regres-
332 CHAPTER TWELVE
sion requires that the variate y be normally distributed. This can be easily attained
for y if we choose the logarithm of y as dependent variate because the frequency
distributions of most economic parameters are generally lognormal in nature. A
simple logarithmic transformation written as log (1 + x;) is usually sufficient
to "normalize" all independent variates with very skewed distributions. As a
result the most general regression equation is written as
log y = ao + al log (l + Xl) + ... + an log (1 + xn) (2)
The measurements of the independent geovariates (x;) obtained from the

control locations are used to calculate least-squares estimates of the coefficients
ai and the standard error for each coefficient. A statistical test, known as the t-
test, will show whether a coefficient is statistically significant. If not, this means
that the corresponding geo-factor is not statistically correlated with economic
worth, and thus may be dropped from the model. Figure 12.3 illustrates the
step-by-step screening of a large array of 170 qualitative geological parameters
of ore-bearing porphyry deposits of the American Cordilleras by characteristics
and regression analyses into only 12 quantified, statistically significant geovar-
iates which can be used to construct a geostatistical model.
12.2.3. Construction of a Classification Model

The outcome of the drill testing of any exploration target under study is
either success in detecting mineralization of commercial importance (population
1) or failure to do so (population 2). The objective assignment of targets to either
class can be done by using a statistical technique known as classification analysis.
In 1966, Griffiths(l3) proposed the use of discriminant functions which were
generalized later by Harris(l5) as multiple discriminant functions. More recently,
Wignall(l9) introduced a generalized Bayesian classification procedure for use in
oil and mineral exploration.
The construction of the classification model is based on the significant
independent geovariates which have been previously screened by compression
and feature selection. The function is calibrated on two sets of control locations.
One set comprises thoroughly investigated control locations which have proved
to be ore-bearing and serve as control for population 1. The second set serving
as control for popUlation 2 is composed of locations which have proved devoid
of commercial ore despite thorough investigations.
Anderson(3) describes the standards required for good classifications. There
are, first, requirements to be met by control and study data alike, such as
multi normality , and equality of the variance-covariance matrices of popUlation
FIGURE 12.3
Flow Diagram of Statistical Screening of Independent
Geovariates.
334 CHAPTER TWElVE
1 and population 2 (homoscedascity). Other requirements to be satisfied pertain

to the classification methodology itself. A satisfactory classification procedure
will minimize the probabilities of misclassification which are of two kinds. One
kind arises when a location actually beongs to population 1 but is wrongly
assigned to population 2; vice versa, a location may be wrongly assigned to 1,
though actually belonging to 2. In each case, there is a cost attached to the
misclassification, and the ratio of the two costs is what matters most to ensure
a good classification.
Earlier workers used multiple discriminant analysis and discriminant scores
for classification purposes. More recently, Bayesian classification analysis has
attracted increasing attention. The scores are transformed into probabilities that
a location belongs to a particular population, taking into account prior proba-
bilities which are derived from the historical record. The success ratio as defined
in Section 1.1.3 of Chapter 1 is a valid estimate of the prior probability of
success. The Bayesian probability of success is formulated in terms of the prob-
ability density functions and prior probabilities for populations 1 and 2 in a
manner that is illustrated by Figure 12.4.
Once the classification model is calibrated on the two kinds of control
locations mentioned above, the measurements of all significant geo-variates are
collected from each target under study and fed into the classification model to
generate the probabilities that the target belongs either to population 1 or 2. All
probabilities of success are ranked in order of decreasing magnitude, and all
locations whose probability of success is greater than 0.500 are retained for
further screening by means of the payoff criterion.
12.2.4. Design of a Criterion Optimizing the Selection of

Exploration Targets for Further Testing
The maximization of the probability of success which is mentioned in
Section 12.2.3 is one approach to the selection of targets for testing. A second
approach would be the maximization of the regression estimates of the economic
parameter represented by the dependent variate. These are obtained by feeding
the measurements of significant geo-variates obtained from the locations under
study into the multiple regression model described in Section 12.2.2. All esti-
mates within their 95% confidence intervals are then listed in order of decreasing
magnitude to enable the selection of the best targets.
However, the two previous approaches are not realistic because they do not
take into account any cost consideration. Clearly, a criterion that would provide
the maximization of economic reward in a probabilistic context (probability of
Probability density function
f. (X)
1
[(2TI)P Islr1/2 exp[-l/2 eX - x.)
1
S-1 (X - X.)']
1
p number of significant geo-variates
s = variance-covariance matrix
Isl= determinant of S
S-1 = inverse of variance-covariance matrix
eX - X.)'
1
= transpose
i for population of ore-bearing targets
i 2 for population of barren targets
Bayesian probability of occurrence of commercial ore
historical record of success for region
number of ore bodies/total number of drilled prospects
q2 1 - q1 = prior probability of failure
f (X) probability density function for population

1
f 2 (X) probability density function for population 2
FIGURE 12.4
Model of a Bayesian Classification Function.
success) under cost constraints should be much more satisfactory. The payoff
criterion which is described in Figure 12.5 fulfills these requirements. Payoff
values obtained by feeding into the model the gross economic estimates, prob-
abilities of success, and cost data for all targets under study are ranked in order
of decreasing magnitude to allow the selection of a small number of the top-
ranking targets, subject to budget constraints.
w
W
0'1
Statistically significant ~
geo-variates
~"
(geological, geophysical, geochemical) .
Multiple Bayesian Cost function

regression model Classification model model
l
Expected gross Probability that Probability of Pay-off criterion
Expected
dollar value of the target is a failure (target is for the optimal
exploration
exploration target x commercial not an ore deposita) X :. selection of
cost
(GVI deposit (Psi Pf = 1 - Ps exploration targets
n
• • • • I
Expected monetary gain Expected loss »-0
-(
m
;;:0
FIGURE 12.5 -(
Design of a Payoff Model for the Optimal Selection of Exploration Targets for Drill Testing. :E
m
r-
<
m
iiM
Montfort ,.,."
31
31e Locotion of onomoly ossocioted with .,.
known zinc deposit
'.e,o
.25 .28 3"- 33 _......3.
o 5 10 21e
.24.
. . . .--.
e3I .45 !,,,,,,I 3
I
scale of miles 2& Mineral Point'~".'.'. .4
8 .40 42 ·44
I- IOWA CO
It."town .21 ~ 3!jW LAFAYETTE CO
~.23 a:
L C)
.16 ,";".iBelmont
• • 58
.11 51
I"\~~
Tennysoh'"
FIGURE 12.6
Selection of Exploration Targets in the Upper Mississippi Valley Zinc District:
Control Locations.
12.2.5. Application of the Methodology to the Selection

of Mississippi Valley-Type Pb-Zn Prospects in the
Upper Mississippi Valley Pb-Zn District
The writers applied the method described above to the selection of Missis-
sippi Valley-type Pb-Zn prospects in the Upper Mississippi Valley District of
the North American Paleozoic Platform. (36) The dependent variate was the gross
dollar value of the deposits; the independent variates were geological and geo-
chemical in nature. The control area covering 56 control locations is shown in
Figure 12.6 and the study area including 58 exploration targets to be rated for
screening is shown in Figure 12.7.
The Bayesian classification scheme provided a clear cut separation of 22
locations showing Bayesian probabilities of success greater than 0.80 and mod-
erate to large gross value estimates, from 33 locations with probabilities of
success lower than 0.35 and low estimates. Only three of the 58 locations were
found to lie between the 0.80 and 0.35 thresholds, and are likely to have been
misclassified. A total of 14 among the 22 locations selected for testing are
concentrated within the Darlington-Blanchardville area (See Figure 12.8), where
338 CHAPTER TWElVE
043042
44
Dod~ev!y'.~ 0045 046
~~~ 30
0 49 o 47 0
0 31
500 0 48
Mineral Point 06
~~l~ 07
IOWA CO
4
0 ...5
36 If Blanchardville
038 ..
LAFAYETTE CO
Belmont
m?t:
0 27
0 21
0 53 290 ~8
0 52 0 13
0 25 0 26
0 14
12 0 Location and reference number of anomaly
in study area
o 5 10
scale J mu.. '
FIGURE 12.7
location of Exploration Targets to be Rated.
shallow lead mineralization and deeper zinc mineralization were previously de-
tected by trenching and drilling.
12.2.6. Critique of the Methodology

12.2.6.1. Merits of the Method. No claim is made that the sequential
multivariate approach described above is an exploration panacea which should
supplant the traditional role of exploration staff in the critical decision making
SElECTION OF EXPLORATION TARGETS 339
.82°·22
Dodgevil!e_ .98.
li\~
""""
C>
32
°05.08 0'
16
0.29 0·
c>07
.08° 0·28
Mineral Point .860 .09

~(~
Blanchardville
IOWA CO .060 28 oJO J19 Q73 ~
LAFAYETTE CO .98. .93 • .250 ~ 8,0.35
.240 eB2
.080 .sze.l •.90
Belmont .210 .410 2 • .99
iil .99. fi 025;3 0.13
.290 .96 f2 0.72 • .85
. ~ • .99~5
.440 DaJ~ngton 0.14
•.96 1>NJl! ~15 .eo. •.99
e.86 ()o03
~ •.89
023
0.28 Bayesian probability of succe.. for
anomaly in Itudy area
·.95 Malt favorable tarQet (Bayesian
probability .80 )
-
~
o 5
-scale of-- -
miles
10
FIGURE 12.8
Location of Exploration Targets Selected on the Basis of Bayesian Probabilities
of Success.
leading to the selection of targets for drill testing. The advantages of the statistical
procedure are fourfold: The method provides (I) an objective rating of the selected
targets, (2) the confidence intervals associated with the rating, (3) the probabilities
involved, and (4) an accurate and speedy way of processing large amounts of
data, and the possibility of updating the planning based on feedback information,
as the field program progresses.
340 CHAPTER TWELVE
12.2.6.2. Methodological Difficulties. In the present mineral explo-

ration climate, explorationists are increasingly turning their attention to remote
and litte explored regions of the world which are unlikely to provide reliable
control locations. The same problem arises in the search for new types of deposits
within regions hitherto well explored for other types of ore deposits.
A second difficulty with serious statistical implications is caused by the bias
in favor of popUlation 1 (success) as against population 2 in the makeup of the
control sample. Obviously, failure will always be less well documented than
success.
12.2.6.3 Statistical Difficulties. The geostatistical model on which the
methodology depends so much is affected by serious limitations; severe restric-
tions are imposed by the theory of the linear least-squares model on which both
multiple regression and multiple discriminant analyses are predicated. The mul-
tinormality requirement can be satisfied by means of logarithmic or power trans-
formations, which, in tum, introduce distortion, errors, and loss of information.
Anderson and Bahadur(3) have devised a remedial approach when the variance-
covariance matrices of populations 1 and 2 are not equal.
However, Agterberg (Chapter 1, Ref. 54) and Simard (Chapter 12, Ref.
24) indicate a way out of the difficulties mentioned above by the application of
logistic regression and classification models instead of the restriction-ridden linear
models used above.
12.3. OPTIMAL SELECTION OF EXPLORATION TARGETS WITHOUT

CONTROL LOCATIONS
12.3.1. Rationale and Methodology

The present approach, which was only recently developed by the second
writer, tackles the difficult problem faced by management when required to
select optimally exploration targets for testing in little-known regions where
adequate control is not available. The flow-chart shown in Figure 12.9 sum-
marizes the method.
The first stage consists of the compilation of all available geographic, to-
pographic, geologic, geophysical, and geochemical data relevant to the area to
be investigated and ore deposit type to be searched for, followed by experimental
design to quantify geologic and other qualitative relevant information. For ex-
ample, supposing there are two rock types of interest and several others of no
interest; then we could code them 1,0 for the first type, 0,1 for the second type,
and -1, -1 for the others. Each location in the prospecting area will thus have
I Compilation of data & experimental design

I
//~.
Factor analysis of all
/ relevant surveys data
/.
1
Prior classification
analysis on each Trend and trend factor
significant factor analysis and significance
and variate for each survey and factor.
The dependent variates are
the surveys and factors. and
the independent are the
geographic, geologic and
topographic variates
Probability analysis
1
for each location in Residual trend analysis to
each group: anomalies determine local anomalies
typos 1.2 •...• and significantly above or below
background the trend
Ranking of anomalies for

field-work: drill-test or
further survey diagn9Ses
FIGURE 12.9
Flow Diagram of the Optimal Selection of Exploration Targets Without Control
Locations.
a multivariate vector of survey information attached to it, and is given an iden-

tification number for greater convenience. The whole set of information is then
stored on disk file or on file in the computer memory and given an appropriate
file name. The file is then accessed and subjected to a series of optimal screening
procedures to detect the best targets. All computer programs used in the optimal
342 CHAPTER TWElVE
screening procedures, including TREND and POPMIX, are written in FORTRAN IV

and are prepared for interactive use on computers and free format in reading
from the data files.
12.3.2. Preclassification Analysis

This analysis is devised to detect anomalies on each survey variate, whether
there are control locations or not. Fisher(43) formulated the method of moments
to dichotomize sets of data into two populations. The POPMIX program devised
by the second writer is based on Fisher's theoretical considerations. Our under-
lying assumption is that there is an anomalous set (population 2) and a background
set (population 1) of locations on each quantitative survey. POPMIX provides the
following output:
1. the proportion p of locations belonging to popUlation 1 and population
2;
11. the mean ml of population 1;
iii. the mean m2 of population 2.
Thus, for each survey variate, we can select the anomalous set, since we
know the proportion, p, and the total number, N, of locations in the study area;
multiplication will yield the number of anomalous locations, and a biased division
of m2 - ml by ml will yield the value of the threshold separating the two
populations.
12.3.3. Factor Analysis

Quite commonly there is a meaningful correlation, either direct or inverse,
among the different survey variates. Factor analysis makes full use of these
interactive relationships. Factor analysis is in fact the most powerful tool to yield
linear sums and contrasts of the survey variates. Finally, factor analysis is more
powerful than principal component analysis, because it incorporates peA as a
first step, prior to the maximization of the variability for which it accounts by
"rotation. "
Our analysis is carried out using the SPSS system, which has proved to be
reliable and accurate. Among other output, the SPSS factor analysis yields the
following:
1. Factor 1, which gives the linear combination and contrast of variates
accounting for the greatest amount of variability among locations.
11. Factor 2 giving the linear sum and difference, which account for the
second largest amount of variability. Any other significant factors are
also given.
iii. The proportion of variability accounted for by each factor.

iv. The correlation matrix giving the interactive relationship coefficients
among the survey variate.
It is possible to determine which type of survey the various factors are
highlighting from the examination of the coefficients of the survey variates. In
the case of negatively correlated variates, one may determine which factor makes
use of the contrast. This exercise will help at a later stage to decide why particular
locations are to be considered as anomalous.
12.3.4. Reclassification of Locations Using Factors

Where there are two or more variates, the preclassification program, POPMIX,
is used again, this time by redefining the variates to yield factor 1, factor 2, and
any other significant factors. POPMIX now yields the anomalous set of locations
on each factor. In most field projects, at least three sets of locations become
apparent at this stage, namely, factor 1 anomalies, factor 2 anomalies, and
background.
The POPMIX output yields at this stage a vector of factor scores for each
location, the factor means, and the proportion of anomalies to background on
each factor. The factor scores enable us to separate the locations into two anom-
alous sets and a background set. These three sets are stored in the memory of
a new file along with their original information vectors.
12.3.5. Probability Analysis

The new file with three sets of locations including (1) anomalies on factor
1, (2) anomalies on factor (2), and background is now used as control set for
the classification analysis. The whole set of locations are now probabilistically
assigned to the three populations, namely, anomalies type 1, anomalies type 2,
and background. The probabilities can be used as objective rating measurements;
thus, the best targets can be selected and are identified by their geographic
coordinates.
12.3.6. Trend and Trend-Factor Analyses

One of the aims of trend and trend-factor analyses is to screen out local
anomalies from regional trends, because of the potential economic value of the
former. This stage of the analysis makes use of the TREND program designed by
the second writer and based on the multivariate general linear model (MGLM).
We find the best fit for the vectors of survey variates on the vector of geographic
344 CHAPTER TWELVE
coordinates, topographic and coded geological variates. The coefficients of the

model are found by solving the matrix equation
Y (N x p) = XB(N x m) . (m x p)
where the dimensions of the matrices are given in brackets, and where B is the
matrix of the best coefficients.
Once all coefficients are found, the data matrices Y and X are called again
and read to calculate the trend estimates at each location for each of the survey
variates. The analysis is repeated to find trend values for each factor at each
location. The main output from TREND is
1.The coefficient for each trend equation for each variate, and for each
factor;
ii. The trend and trend factor estimates for each location.
12.3.7. Residual Trends and Trend Factors

We now have the actual survey readings and factor scores on all variates
and factors, as well as the trend and trend factor estimates at each location.
Residuals are obtained at each location by subtracting the trend from the actual
survey reading. The residual trends on the factors are particularly useful for the
determination of local anomalies significantly above or below the trend. To
determine the significance, the standard error of the differences is calculated on
each factor. The residuals on each factor at each location are now screened to
find the locations with significant residuals. These locations are local anomalies
on the particular trend given in the output of the TREND program.
12.3.8. Summary-Conclusion
Since we do not have the advantage of having prior control locations to
rely on for the selection of locations of greatest merit in the prospecting area,
we proceed to construct our own set of control locations within the prospecting
area itself by means of the preclassification analysis and factor analysis. As was
the case in the first approach, we use the control set to obtain a probabilistic
classification of the locations under investigation. Their screening is further
refined in a spatial context by means of trend analysis, through which local
anomalies are separated from the global ones. The local anomalies are ranked
in order of merit, based on the magnitude of the associated residuals. It should
be noted that the second approach described here involves a sequential max-
imization, first of the probabilities of belonging to classes 1 and 2, secondly of
the magnitude of the residuals, without constraints, as opposed to the optimization

under cost constraints featured by the first approach.
12.3.9. Field Example: Mount Bundey, Northern

Territory, Australia
Our field study is an exploration project in the Mount Bundey and Mount
Guyder environs, in which the second writer was involved as field manager. We
were seeking contact metasomatic U-Ag-Au or base metal prospects.
The Mount Bundey-Annabaroo area of the Northern Territory is situated
some 70 miles ESE of Darwin. It includes most of the I-mile geological sheet,
D52 448. The rocks comprise Lower Proterozoic sedimentary formations that
have been folded and weakly cleaved before being intruded by granitoids. The
regional metamorphism is lowermost greenschist, and distinct contact aureoles
surround the granitoids. In the northern contact aureole an iron mine has been
abandoned, reportedly as uneconomic.
Previous work by the Bureau of Mineral Resources and Geopeko, Ltd. has
highlighted the prospective nature of the region. Their aerial and ground geo-
physical and geological surveys gave the following information:
I. radiometric anomalies on the granitoids and immediate environs;
11. magnetic anomalies over the northern Proterozoic outcrops, including
the iron mine;
111. correlation of the Golden Dyke Formation and Craig Creek member
with the host rock mineralization in the Alligator Rivers U-Au field;
It should be noted that the Rum Jungle Uranium mine is 30 miles to the west
of Mount Bundey, and the Alligator Rivers U-Au mines some 50 miles east.
Our party carried out a ground radiometric survey over the Mount Bundey
and Mount Guyder areas, utilizing four-wheel-drive vehicles and GIS 3 spec-
trometers. In all, 247 ground spectrometer readings were taken, and total counts
and uranium counts recorded.
The following survey data sets were then collated for each of the 247
locations: Geographical (latitude and longitude), ground and aerial radiometric
(very highly correlated), magnetomic readings, topography, and geology. Ex-
perimental designs were applied to quantify the geological data.
Our computer software package described in Sections 12.3.2-12.3.7 was
then employed to screen out 37 anomalous prospects: 31 being chosen by the
probability classification analysis and a further 6 by the residual trend as locations
significantly above the trend. The most northerly group of anomalies are on the
iron mine, and four of the southerly ones are covered by Geopeko mining leases.
346 CHAPTER TWELVE
The inner group of anomalies in the northwestern aureole of Mount Bundey have
since been drilled, and the result was a new silver mine grading up to 40 oz. to
the ton being discovered. The authors have not had any reports on the other
anomalous prospects as to whether any drilling program has been carried out.
The silver discovery was classical: the anomaly in the Mount Bundey aureole
is a radiometric high coinciding with a magnetomic low.
Optimal Selection of Exploration Targets Based on Control

Locations: General Topics
1. AGTERBERG, F. P., 1974, Geomathematics, Chap. 14, Elsevier, Amsterdam.
2. AGTERBERG, F. P., 1978, Quantification and statistical analysis of geological variables for
mineral resource evaluation, Mem. BRGM, No. 91, pp. 399-406.
3. ANDERSON, T. W ., and BAHADUR, R. R., 1962, Classification into two multivariate normal
distributions with different co-variance matrices, Ann. Math. Stat. 33, 420-432.
4. BOLCH, B. W., and HUANG, C. J., 1974, Multivariate Statistical Methods for Business and
Economics, Prentice Hall, Englewood Cliffs, New Jersey.
5. BOTBOL, J. M., 1971, An application of characteristic analysis to mineral exploration, Can.
Inst. Min. Metall. Spec. Vol. 12, 92-99.
6. BOTBOL, J. M., SINDING LARSEN, R., McCAMMON, R. B., and GOTT, G. B., 1978,
A regionalized multivariate approach to target selection in geochemical exploration, Econ. Geol.
73, 534-546.
7. CULBERT, R. R., 1976, A multivariate approach to mineral exploration, Can. Inst. Min.
Metall. Bull., 69, 39-52.
8. DE GEOFFROY, J., and WIGNALL, T. K., 1970, Application of statistical decision techniques
to the selection of prospecting areas and drilling targets in regional exploration, Can. Inst. Min.
Metall. Bull. 63(699), 893-899.
9. DE ST. JORRE, M. G. F., and WHITMAN, W. W., 1972, A probabilistic method of ranking
underground exploration proposals, Econ. Geol. 67, 789-793.
10. DUBOV, R. I., 1972, A statistical approach to the classification of geochemical anomalies,
Proceedings ofthe London Symposium of the Institute of Mining and Metallurgy, pp. 275-285.
11. FUKANAGA, K., 1972, Introduction to Statistical Pattern Recognition, Academic, New York.
12. GREEN, P. E., 1978, Analyzing Multivariate Data, Dryden Press, Hinsdale, Illinois.
13. GRIFFITHS, J. C., 1966, Application of the discriminant functions as a classification tool in
the earth sciences, Kansas Geol. Surv. Comput. Contrib. 7, 48-51.
14. HARRIS, A., 1976, Definition and priority rating of exploration targets in the Cold Spring
Pond area, Central Newfoundland, J. Geochem. Expl. 6, 91-101.
15. HARRIS, D. P., 1965, Multivariate statistical analysis: A decision tool for mineral exploration,
Symposium on Computer Application in Mining and Exploration Transactions, Vol. I, pp.
CI-C35, University of Arizona, Tucson.
16. KLOVAN, J. E., 1968, Selection of target areas by factor analysis, Western Miner 41(2),
44-54.
17. MARIOTT, F. H. C., 1974, The Interpretation of Multiple Observations, Academic, London.
18. PETERS, W. c., 1978, Exploration and Mining Geology. Wiley, New York.
19. WIGNALL, T. K., 1969, Generalized Bayesian classification functions (K classes); Application
to a regional geochemical survey in southeastern Pennsylvania, Econ. Geol. 64, 570-576.
Optimal Selection of Porphyry-Cu-Mo Prospects

20. DE GEOFFROY, J., and WIGNALL, T. K., 1972, A statistical study of geological charac-
teristics of porphyry-Cu-Mo deposits in the Cordilleran Belt of North and South America:
Application to the rating of porphyry prospects, Econ. Geol. 67, 656-668.
21. SINCLAIR, A. J., and GODWIN, C. I., 1979, Application of multiple regression analysis to
drill target selection, Casino Porphyry-Cu-Mo deposit, Yukon Territory, Canada, Trans. Inst.
Min. Metall. London 88, B93-BI06.
Optimal Selection of Ni-Cu-Ultramafic Prospects

22. BULL, A. J., and MAZUCHELLI, R. H., 1975, Application of discriminant analysis to the
geochemical evaluation of gossans, in Geochemical Exploration 1974. pp. 219-226, Elsevier,
Amsterdam.
23. CAMERON, E. M., SIDDELEY, G., and DURHAM, C. c., 1971, Distribution of ore elements
in rocks for evaluating their ore potential: Nickel, copper, cobalt, and sulfur in ultramafic rocks
of the Canadian Shield, Can. Inst. Min. Metall. Spec. Vol. 11, 298-314.
24. SIMARD, R. G., 1980, Logistic model applied to geophysical data for sulfide nickel exploration,
Optimal Selection of Volcanogenic Massive Sulfide Prospects

25. AGTERBERG, F. P., CHUNG, C. F., FABBRI, A. G., KELLY, A. M., and SPRINGER, J.
S., 1972, Geomathematical evaluation of copper and zinc potential of the Abitibi area, Ontario
and Quebec, Geol. Surv. Can .. Pap. No. 71~1.
26. DIVI, S. R., THORPE, R. I., and FRANKLIN, J. M.~ 1979, Application of discriminant
analysis to evaluate compositional controls of massive sulfide deposits in Canada, J. Math.
Geol. 11(4), 391~06.
27. FAVINI, G., and ASSAD, R., 1974, Statistical aeromagnetic and gravimetric criteria for massive
sulfide districts in Greenstone areas of Quebec and Ontario, Can. Inst. Min. Metall. Bull.
67(752), 58-63.
28. FAVINI, G., and ASSAD, R., 1979, An optimized decision model for area selection in massive
sulfide exploration, Can. Inst. M. Metall. Bull. 72(804), 118-126.
29. FOX, J. S., 1978, Interpretation of drill hole geochemical data from volcanic rocks of the
Hanson Lake Mine, Saskt., Can. Inst. Min. Meta1l71(793), 111-116.
30. FOX, J. S., 1979, Host-rock geochemistry and massive volcanogenic sulfide ores, Can. Inst.
Min. Metall. Bull. 72(804), 127-134.
31. MARCOTTE, D., and DAVID, M., 1981, Target definition of Kuroko-type deposits in Abitibi
by discriminant analysis of geochemical data, Can. Inst. Min. Metall. Bull. 74(829), 102-107.
32. PIRIE, J. D., and NICHOL, I., 1981, Geochemical dispersion in wall-rocks associated with
the Norbec deposit, Noranda, Quebec, J. Geochem. Expl. 15, 159-180.
33. WHITEHEAD, R. E. S., and GOVETT, G. J. S., 1974, Exploration rock geochemistry:
Detection of trace element halos at the Heath-Steele Mines (New Brunswick, Canada) by
discriminant analysis, J. Geochem. Expl. 3(4), 371-386.
348 CHAPTER TWElVE
Optimal Selection of Mississippi Valley-Type Pb-Zn Prospects

34. CAMERON, E. M., 1969, Regional geochemical study of Slave Point carbonates, Western
Canada, Can. J. Earth Sci. 6, 247-268.
35. DE GEOFFROY, J., and WU, S. M., 1968, Selection of drilling targets from geochemical
data in the Southwest Wisconsin zinc area, Econ. Geol. 63(8), 787-795.
36. DE GEOFFROY, J., and WIGNALL, T. K., 1970, Statistical decision in regional exploration:
Application of regression and Bayesian classification analysis in the Southwest Wisconsin zinc
area, Econ. Geol. 65,769-777.
Optimal Selection of Uranium Prospects

37. PATTERSON, D. A., PIRKLE, F. L., JOHNSON, M. E., BEMENT, T. R., STABLEIN, N.
K., and JACKSON, C. K., 1981, Discriminant analysis applied to aerial radiometric data and
its application to uranium favourability in South Texas, J. Math. Geol. 13(6).
38. PIRKLE, R., CAMPBELL, K., and WECKSUNG, G. W., 1980, Principal component analysis
as a tool for interpreting NURE aerial radiometric survey data, J. Geol. 88, 57-67.
Optimal Selection of Oil and Gas Prospects

39. WIGNALL, T. K., 1968, Oil and gas exploration: Statistical decision criteria, S.P.E. Paper
No. 223, Annual A.I.M.E.lS.P.E. meeting. Houston, Texas.
Optimal Selection of Exploration Targets Without Control

Locations
40. CONNOR, J. J., and MIESCH, A. T., 1964, Application of trend analysis to geochemical
prospecting data from Beaver Country, Utah, Computers in the Mineral Industry, Part I,
Geological Science, Vol. 9, Stanford University.
41. DE GEOFFROY, J., and WIGNALL, T. K., 1973, Design of a statistical data processing
system to assist regional exploration planning, Can. Min. J. Part 1,94(11), pp. 30-35.
42. DEMNATI, A., and NAUDY, H., 1975, Gamma-ray spectrometry in Central Morocco, Geo-
physics 40(2), 331-343.
43. FISHER, R. A., 1936, The use of multiple measurements in taxonomic problems, Ann. Eu-
genics, 7, 179-188.
44. GRAYSON, C. J., Jr., 1960, Decision under uncertainty: Drilling decisions by oil and gas
operators, Harvard Business School Division of Research, Boston.
45. GRIFFITHS, J. C., 1962, Use of computers and statistics in the exploration and development
of mineral resources, Symposium Proceedings, College of Mines, University of Arizona, pp.
E.I4-25.
46. KAUFMAN, G. M., 1962, Statistical Decision and Related Techniques in Oil and Gas Ex-
ploration, The Ford Foundation, Prentice Hall, Englewood Cliffs, New Jersey.
47. KRUMBEIN, W. C., 1959, Trend analysis of contour-type maps with irregular control point
spacing, J. Geophys. Res. 64(7).
48. MORRISON, D. F., 1967, Multivariate Statistical Methods, McGraw-Hill, New York.
49. NEEDHAM, R. S., WILKES,P. G., SMART, P. G., and WATCHMAN, A. L., 1973,Alligator
Rivers region of the Northern Territory: Results of airborne geophysical surveys conducted by
the Bureau of Mineral Resources of Australia; B.M.R. record 1973-208.
50. RAO, C. R., 1952, Advanced Statistical Methods in Biometric Research, Wiley, New York.
5!. WIGNALL, T. K., 1969, Generalized classification functions: A Bayes-theoretic approach,
Annual meeting of the Northwestern Science Association, Cheney, East Washington State
College.
CONCLUSION
After two decades of intense activity and expansion, mineral exploration is at a

crossroad. Slichter's timely call of 1960 for a new philosophy of mineral ex-
ploration has to be renewed now. The spectacular successes of the 1960s and
the 1970s resulting from the massive injection of new technology were often
accompanied by much waste of talent and resources and unnecessary duplication
which cannot be repeated in the present circumstances of economic difficulties
and financial stringency.
It is hoped that the call for a new philosophy of mineral exploration based
on the systematic optimization of the sophisticated technology at hand which is
advocated in the Preface and first chapter of this book will be heard by the
mineral exploration fraternity, and put into practice as actively as has been done
in the recent past by their mining engineering brethren. The methodology, math-
ematical and statistical apparatus, and the computational facilities are available
and ready to be harnessed by explorationists for the "good cause" of optimizing
mineral exploration.
By restricting the topic of the book to only two main items among some
of the most important aspects of the optimization of mineral exploration-namely,
the optimization of field program designs and selection of exploration targets
for testing-we were able to delve into the two subject matters in greater detail
and depth and, hopefully, demonstrate to explorationists the practical value of
optimization. We hope that other workers will take on the task where we left it
and extend the application of optimization to the many fields of mineral explo-
ration which were briefly mentioned in Section 1.2.4 of Chapter 1.
It appears that the setting up of "think tanks" of experts fully devoted to
the systematic optimization of mineral exploration, both within existing govern-
ment organizations and large multinational exploration corporations, would be
timely and highly beneficial. As a result of their efforts, the mineral exploration
business would be urged to new peaks of efficiency and success in the 1980s
and 1990s and beyond.
Looking beyond the next two decades through the proverbial crystal ball is
a notoriously risky exercise. One could expect that an extensive restructuring of
the mineral exploration methodology and technology will take place toward the
end of the century, as most ore deposits detectable within the first 300 feet of
the Earth's crust will have been located by means of the existing technology
351
352 CONCLUSION
improved by optimization. The detection of deeper subjacent ore deposits will

require some completely new technology, possibly not even heard of at the
present, which will undoubtedly be very expensive, hence a greater than ever
need for optimization. Therefore, optimization is here to stay, first as an im-
provement for the next two decades, and as a necessity beyond the end of the
century.
APPENDIX
Measurement Conversion Table
Type of Measurement Imperial to Metric Metric to Imperial
mile 1. 61 kilometers kilometer = 0.621 mile

Distance 1 yard 0.914 meter meter 1. 09 yards
1 foot 0.305 meter meter 3.28 feet
sq. mile = 2.60 sq. sq. kilometer = 0.386

kilometers sq. mile
Area acre = 0.405 hectare hectare = 2.47 acres

sq. foot = 0.093 sq. 1 sq. meter = 10.8 sq.
meter feet
1 cubic yard = 0.765 cubic meter 1.31 cubic

CUb1C meter yards
Volume cubic foot = 0.028 cubic meter 35.3 cubic

cubic meter feet
gallon = 4.55 liters liter = 0.220 gallon
short ton = 1.02 metric metric tonne = 0.980 short

tonnes ton
Mass pound 0.454 kilogram kilogram = 2.20 pounds
ounce 28.3 grams gram = 0.035 ounces
ounce/sh. ton = 27.8 gr/metr. tonne = 0.036
gr/metr. tonne oz/sh. ton
353
SUBJECT INDEX
Alaska, 59, 110, 145, 148, 149 Color anomaly, 112,237

Algorithm, 85, 92 Computer, 43, 81, 96, 97
recursive, 86 program, 43-45, 86, 91, 341-345
SIMPLEX, 86, 87 Conditional probability, 4, 5, 50, 51
Angled detection, 27, 68, 92, 100, 107 Confidence interval, 32, 40, 48, 102, 103,
Anomalous set, 343 107,249,267, 339
Anomaly, 326, 342-344 Confirmed detection, 21, 35-37, 46-48,
geochemical, 17, 19 101, 117, 151,219,245,291,313
geophysical, 17, 19, 30 Constraint, 80-84, 86, 104, 105
global,344 Contact metasomatic deposit, 96, 97, 142,
local 343, 344 detection, 146, 147
Arctic Islands, 283-288 geological setting, 141, 143-145
Control location, 10,327,332,334,337, 340
Background set, 326, 342, 343 Cordillera Belt, 25, 96, 97,109-111,114,
Bayesian classification, 329, 332, 334, 337 145, 149, 163, 176, 177, 188-190,
Bayesian decision theory, 10 236,261,262
Bayesian probability, 4, 5, 334, 335, 337, Cost, 55-57
339 analogy, 56
Black Hills, 306, 307 component, 59, 60
British Columbia, 144, 147, 148, 162, 176, estimate, 56
189 function, 60, 61, 64, 68, 79, 88-90
indexation, 57
Calculus, 84-86, 90, 91, 106 mUltiplying factor, 121, 122, 198, 201,
California, 145, 148, 190, 191,261,262, 220, 222, 246, 248, 265, 293, 314, 316
263,305 schedule, 58
Carbonate formation, 142, 148, 162, 163, Course of action, 77, 80, 87
176,284,285,287 Criterion, 77-81, 88, 89, 106,328, 334
Cavity filling, 307, 309, 310 Cutoff grade, I, 24
Centrality, 98, 99
Characteristic Analysis, 330, 331 Data
Chi-squared test, 25, 99,100,117,150, acquisition, 6-8, 10, 17-19,325,328,
217,244,311 340
Circular normal model, 25, 100,217,218, base, 95, 96
244,310 bit, 97
Classifying function, 20, 332, 334, 335, compression, 328-331, 333
337,342 processing, 10,97,98,325,328,340
Coefficient of Decision tree, 10
dispersion, 99, 116, 150,217,244,245, Degree of belief, 3
264, 289, 310 Deposit
skewness, 99, 115, 150,217,244,264, mineral, I, 17, 19,21,22,24
289, 310 ore, 1,9, 11,22,24,96,97,325,332
355
356 SUBJECT INDEX
Derivative, 84, 85, 90, 91 Exploration target, 10, 11, 325-327, 334,
Detection, 17-21,26-28,32,33,43,48 335, 340, 344
range, 21, 27-29 Extremal root, 85, 91
sensitivity, 48, 49 Extremum, 81, 82
Deterministic, 22, 31, 84
Dichotomization, 329, 330, 340, 342 Factor analysis, 326, 330, 342, 343
Digitization, 326, 329, 340 Feature selection, 330, 332, 333
Dip Fiducial interval (see also confidence
angle, 22-25, 27-29, 31, 100, 149,216, interval), 99, 102, 104, 105, 107,249,
217,242,289,310 267,279
component, 27-29,100,101,117,151, Field
218,245,290,312 artificial, 20, 28, 29
Direct approach, 90--92 potential, 28, 29
Direct ascent, 85 Filtering, 20
Direct detection, 17, 112, 146,213,214, Flight elevation, 28, 34, 35
239, 286, 308, 309 Frequency distribution, 98-100, 115, 150,
Discriminant analysis, 332, 334 163,217,243,289,310,311
Discrimination, 21, 22 Frequentist concept, 3
Dolomitization, 284, 285
Drilling, 17, 18,66-68,72,73,91 Geochemical
angle, 91, 92, 107,201-203, detection, 18,33, 113, 114, 147,215,
220--222,246-249,265-267,314, 287,288, 309, 310
316 signature, 18, 19,325
churn, 66, 286 Geographic areas
diamond, 51, 66 Central Asia, 147
optimization, 7, 11, 91, 92, 106, 107, Greenland, 284, 285
123, 124, 200, 202, 203, 222, 248, Manitoba, 211-213, 216, 237, 238
249,267,293,316 Mexico, 110, 146, 177
percussion, 66, 138, 302 Mississippi Valley, 283, 284, 337, 338
testing, 7, 10, 11, 325, 327, 339 Missouri, 283, 284
Dynamic programming, 9-11, 86, 87, 89, Ontario, 30, 34, 35, 61, 211-213,
91, 105-107,218,222,248,267,313, 237-239, 306, 309
316 Quebec, 4, 30, 212, 238, 239, 241
Selwyn Mountains, 162, 189, 283-285,
Economic worth, 1, 5, 9, 11, 24, 96, 288, 306
329-331,334,340 Southwest U.S.A., 110, 111, 145, 148,
Effectiveness, 78, 88 162, 176, 189, 190,238
Efficiency, 88 Tri-State, 283, 284
criterion, 78, 79, 89, 90, 105, 122, 201, Ungava, 212, 213, 215
222, 248, 293, 316 Geological age, 110, 145,211,235,283,
Electromagnetic survey, 20, 21, 28, 68, 113, 284,305
214,215,240,287,309 Geological detection, 17, 18, 112, 146,213,
ELLIPGRID program, 32, 33, 43, 53 214, 238, 239, 286, 308, 309
Exhalite, 236, 237, 239 Geological occurrence, I, 4, 11, 13, 14
Expected loss, 79 Geometric descriptor, 22, 23
criterion, 123, 203, 222, 248, 293, 316 Geometric factor, 24
Expected monetary value, 79, 334, 336 Geometric probability, 31-33
Expenditure, 56 Geophysical contrast, 29, 30, 61-65
SUBJECT INDEX 357
Geophysical survey cost, 58, 62, 63, 65 Linear programming, 86

Geophysical detection, 17-21, 28, 112, 113, Lithogeochemistry (see also geochemical
147,214,215,240,287,309,326 detection), 18-20, 113, 114, 147,236,
airborne survey, 34-36, 50, 61, 90, 103, 241,287,288
112, 113, 117 Logarithmic transformation, 99, 332, 340
ground survey, 42, 43, 45, 65, 91, 103, Logistic model, 340
1l3, 147,215,240,287,309 Logistics, 8, 59, 60
signature, 19, 326, 327 Lognormal model, 25, 26, 99, 100, 116,
Goal, 77, 78 117,149,163,177,190,217,243,
Goodness of fit, 99, 100 263, 289, 312
Gossan, 112, 237 Loss function, 79
Gravimetric survey, 28, 113, 147, 215, 240,
287
Great Slave Lake, 283-285 Magnetic induced polarization, 113, 240
Greenstone belts, 211, 213, 215, 236-239, Magnetic survey, 28, 65, 68, 113, 147,214,
241,249,306-308,327 215,240,308,309
Greisen, 189, 190 Manto deposit, 162, 163
Grid Marble, 141-143, 162, 163, 176, 189,285
control, 30 Mathematical programming, 84, 86
geometry, 30, 35, 36, 38, 39, 42, 101 Maximal strategy, 104, 105, 118-121,200,
Gross National Product, 57 201, 220-222, 247, 248, 265, 266,
292, 293, 315, 316
Halo, 17, 18, 113, 123, 124,240,241,287, Mean, 99,115,150,163,217,243,244,
288 263,264,289,290,310,311,342
Heuristic model, 6, 75, 78, 79, 85, 86 Median, 99
Histogram, 25, 26, 98, 99 Message, 19, 20
Homoscedascity, 234 Metasomatic deposits, 142, 144, 145
Hornfels, 141, 176, 189 Metasomatism (see also contact), 141-143
Mineral exploration
Independence, 4, 5, 124 context, 2
Indirect detection, 17-21, 112-114, 147, methodology, 1,9,10,21,22
214,215,240,241,286,287,308-310 optimization, 9-11, 75-77
Induced polarization, 30, 43, 68,113,147, purpose, I
215, 240, 286, 309 sequentiality, 6-8
Information, 10 success, 1, 4, 5
feedback,8 Minimax strategy, 81
prior, 8 Misdassification, 334, 337
theory of, 19, 52 Mississippi Valley-type deposit
Iron formation (see also exhalite), 215, 236, detection, 286-288
237,239,240,241,307 geological setting, 283, 284
statistical modelling, 289, 290
Lateral coverage, 21, 33, 35, 36, 38 Mobile Belt, 211, 212
Leadville district, 162, 163 Mode, 99, 110
Liminal strategy, 104, 121, 198, 199,220, Model
221,246,247,264,265,291-293, concept, 4, 5, 109, 143, 235, 238, 239
314, 315 fitting, 24, 25, 99, 100
Linear inequality, 82 Multiple Spectral Sensing, 17, 112, 214
Linear model, 331, 340, 343, 344 MUltipurpose survey, 249, 267
358 SUBJECT INDEX
Nickel deposit Price, 2, 3, 55

detection, 214, 215 Principal Component analysis, 330, 342
geological setting, 211-213 Probability, 3, 4, 6, 22, 84
komatiite series, 213 detection, 4, 11, 34, 35, 40, 43, 45, 48,
spinifex texture, 214 50,51,75,76,88,89,91,92, 101,
statistical modelling, 217, 218 102,117,118,151,165,178,191,
tholeiitic series, 213 218, 219, 245, 246, 264, 290, 291,
Noise, 19, 20, 22 312, 313
Normal model, 25, 99, 100,217,243,244, economic worth, 11, 14, 15
289, 310, 311 occurrence, 11, 13, 14
Northwest Territories, 215, 238, 239, 243, success, 3-6, 79, 80
284, 285, 306, 310, 311 Probe, 20, 21, 27, 28
Prospecting
Objective function, 81-83, 88-91, 105 area, 9, 10, 325
Observation platform, 17,28,34 detection, 17, 112, 146,214,238,239,
Open pit mining, Ill, 189, 284 286, 308, 309
Operations Research, 10,81,87,88 success ratio, 78, 334, 335
Ore body, I, 11, 22-24 Pyritic halo, 19, 50, 51, 110, 112-116, 123,
Ore treatment, 2 124,237,240,284,309
OPTGRID program, 43, 91, 92
Optimal choice, 9-11, 75, 77, 78 Quantification, 3, 77, 78, 326, 340
Optimal grid orientation, 10, 31, 40-44, 48, Quartile, 98, 99
49, 101, 151, 218, 245, 313
Optimal strategy, 101, 122, 123,201-203, Range, 98, 99, 114, 115, 149, 163, 177,
246-249,266,267,291,293,315,316 217,241,289,310
Optimization, v, vi, 9, 75-77, 95, 104 Ratio costing, 56, 59, 160
drilling angle, 106, 107,203,248,249, Reduction to Pole, 29
316 Refractory ore, 307, 308
survey grid, 9, 10,77, 122,202,203, Regression analysis, 330-332
222, 248, 249, 293, 316 Remote sensing, 17, 112,214
Replacement, 307, 310
Pattern Residual, 344
drilling, 32, 33, 42, 43, 65, 66, 123, 124, Resolution, 21
286,309 Reward, 78, 88, 89, 106
recognition, 21, 327, 330 Risk, 2, 5, 6, 10
Payoff criterion, 79, 330, 334
Permissible area, 82 Sampling
Pine Creek District, 190 continuous approach, 17, 18,61,90,325
Plunge angle, 23, 24 discrete approach, 17, 18,65,91,92,288
POPMIX program, 342, 343 size, 98, 288
Porphyry-Cu-Mo deposit Scaling factor, 89
economic features, 110, III Scanning, 8
detection, 112-114 Scheduling 1-8, 10, 325
geographic distribution, 110 Sensitivity analysis, 48-50, 87
geological features, 109 Sensor, 18, 19, 21, 22
statistical modelling, 114-117 Sequential approach, 6-8, 20, 50, 51, 123,
Porphyry dykes, 307 124, 326, 328, 341
Precision, 78, 79 Shape ratio, 23, 27, 28, 31, 211-213
SUBJECT INDEX 359
Shield region, 61, 96, 97, 2ll, 214, 215, Trend analysis, 326, 343, 344
235, 238, 306, 307 Threshold, 104,326,334,337,342
Signal, 19, 20 Trial repetition,S, 6
Signature, 19, 21, 326
Simpson's rule, 44, 45 Uncertainty, 2, 3, 5, 8, 10, 100
Skam, 141-143, 148, 149, 176, 189, 190, Unit costing, 56-58, 60, 65, 66
305
Variability, 99, 116, 150, 151, 165, 217,
Skimming, 8
218,289,290,312,342,343
Spatial coincidence, 30, 326
Variance, 90, 332, 335
Standard deviation, 99, ll5-ll7, 150, 165,
Variate, 81, 82, 84, 85, 328-331, 333, 342,
217,243,289, 3ll, 326
343
Statistical decision theory, 80, 81
Vein gold deposit
Statistical occurrence, 3-5, 79
detection, 308-310
Statistical pattern, 326, 327
economic geology, 305, 306
Statistical sampling, 98, 288
geological setting, 307, 308
Statistical shape, 99, ll5, 150, 151, 165,
stastical modelling, 310-312
217,243,289,310,311
Vertical continuation, 29, 30
Statistical summarization, 98, 99
Vertical detection, 27-29
Stochastic variate, 84, 325
Volcanogenic massive sulfide deposit,
Stratabound deposit, 235-237, 283
235-238
Strategy, 80, 81, 104, 105, 118, 121-123,
alteration pipe, 236, 237
198,201,202,219,220,222,246,
calc-alkaline series, 236
248,265,267,291,293,314,316
dalmatianite, 237
Stratifonn deposit, 235, 283
detection, 238-240
Strike direction, 23, 90-92, 105, 106, 148,
exhalite, 236, 239
217,243,244,286,311
geological setting, 236, 237, 261, 262
Structural hinge, 285
massive ore, 237
Structural setting, 109, 110, 144, 145, 211,
metallogeny, 237, 238
235, 238, 285, 307
occurrence mode, 238
Student-t test, 99, 344
ophiolitic type, 235
Success, I, 3-5, 88
pyroclastic series, 236, 262
ratio, 4, 329, 333, 335
stringer ore, 236, 238
Supergene enrichment, III, 176, 237
tholeiitic series, 236
Tactite, 189, 190 Wholesale Price Index, 57
Target Word Processor, 97
definition, 326
geometry, 27, 28, 31-35, 42, 43, 100, Yellow Pine District, 190
101 Yukon Territories, 110, 145, 161, 162, 176,
selection, 2, 325, 327, 328, 340, 341 189
Telluride, 307
Tintic District, 162-164 Zoning, 109, 1l0, 141-145, 148, 189, 190
AUTHOR INDEX
Abrams, M. J., 112 Elevatorsky, E. A., 305 Lovering, T. S., 146

Agterberg, F. P., 25, 77, Elliot, I. L., 19 Lowell, J. D., 109
326, 327, 330
Agocs, W. B., 32, 35, 77 Fisher, R. A., 43 Macqueen, R. W., 285
Albers, J. P., 262 Fox, J. S., 236, 239, 240 Marcotte, D., 237
Anderson, T. W., 332, 340 Fukanaga, K., 327 Marriot, F. H. c., 40
McCammon, R. B., 33, 35,
Ball, C. W., 189 Gilmour, P., 236 39
Barragar, W. R. A., 85 Griffiths, J. c., 332 Metz, P. A., 56, 59
Bellman, R. E., 85 Guilbert, J. M., 109 Mickey, M. R., 42, 48
Bolch, B. W., 330 Miller, C. P., 215
Boldy, J., 239, 240 Harbaugh, S. W., 82, 83, Morris, H. T., 147
Botbol, J. M., 330 84,86 Murphy, J. D., 285, 288
Brant, A. A., 8, 80 Harris, A., 327
Butler, B. S., 162 Harris, D. P., 327 Naldrett, A. J., 211
Henderson, R. G., 29, 35
Callahan, W. H., 284, 286 Hewlett, R. F., 11,79 Obial, R., 215
Cameron, E. M., 212, 285 Hollister, V. F., 98
Campbell, N., 285 Holmes, R., 249 Parasnis, D. S., 19,43
Cathro, R. J., 189 Hutchinson, R. W., 235 Paterson, N. R., 22, 28, 34,
Celasun, M., 10,34,42 Hutson, R. J., 308 38, 240
Chung, C. F., 33, 34, 35, Peters, C. W., 4, 19,28,
38,40 Jackson, S. A., 285 34,327
Coad, P. R., 214 Johnson, N. I., 25
Cobb, H., 77 Raisbeck, G., 19,21
Coyle, R. G., 87 Kelley, J. c., 77, 79 Reedman, J. H., 19
Cranstone, D. A., 57 Kelly, S. F., 309 Rennie, C. c., 162, 189
Cribb, J. L., 29 Kendall, M. G., 32, 77 Riddler, R. H., 236, 239
Culbert, R. R., 327 Kerr, P. F., 141, 145, 190 Ridge, J. D., 146, 147, 162,
Kinkel, A. R., 262 190
Davis, J. C., 25 Klichinov, V. A., 147 Roscoe, 77
De Geoffroy, J. G., 25, 77, Klovan, J. E., 327 Rosenberg, 21
80, 288, 326, 327, 337 Knopf, A., 190 Ross, J. R., 211
Dowsett, J. S., 215 Koch, G. S., 25 Roubens, M., 11,79
Drew, L. J., 33, 42, 49, 67, Koulornzine, T., 4
77 Krumbein, W. c., 25 Sangster, D. F., 148, 235,
Duckworth, E., 80 Kuzwart, M., 19 237,238
Savinskii, I. D., 32, 33, 43
Einaudi, M. T., 142, 175, Lee, Y. W., 21 Seigel, H. 0., 113,240,
176 Levinson, A. A., 19,326 309
361
362 AUTHOR INDEX
Shurygin, A. M., 33, 42, Smirnov, D. F., 141, 142, Wilde, D. J., 84
77 175 Wilson, H. D. B., 212
SiIlitoe, R. H., 110, 114 Smith, C. L., 285, 288 Wright, L. B., 306
Sinclair, A. J., 33 Solomon, H., 32
Singer, D. A., 32, 33, 42, Spector, A., 29 Young, G. A., 148
43
Slichter, L. B., v, 32, 34, White, L. G., 189 Zurbrigg, H. F., 212
77,78,80, 141 Wignall, T. K., 327 Zurflueh, E. G., 21
LISTED JOURNALS
Annals of Eugenics, 348

Annals of Mathematical Statistics, 346
Biometrics, 53
Biometrika. 13
Bulletin of the Australian Society of Exploration Geophysicists, 51
Bulletin of the Geological Society of America, 233
Bulletin of Mineral Industry Experimental Station of the University of Pennsylvania, 13
Canadian Institute of Mining and Metallurgy Bulletin, 12, 14, 15, 52, 74, 93, 139, 140, 210,
233,279,280,281,346,347
Canadian Journal of Earth Sciences, 279, 303, 348
Canadian Journal of Sciences, 233
Canadian Mining Journal, 12, 14, 74, 139, 303, 323, 348
Canadian Petroleum Geologists Bulletin, 303
Computers & Geosciences, 14, 53
Computer Journal, 93
Cost Engineering, 74
Earth and Planetary Sciences Letters, 139
Economic Geology, 14,53,74,93, 140,209,210,233,280,281,303,323,324,346-348
Geoexploration, 74
Geologisch Rundschau, 210
Geophysics, 52, 53, 93, 139, 140, 281, 303, 348
Geosciences of Canada, 303
Journal of the American Statistical Association, 12
Journal of Geochemical Exploration, 93, 281, 303, 346, 347
Journal of Geology, 281, 348
Journal of the Geological Society of the Philippines, 233
Journal of Geophysical Research, 348
Journal of Mathematical Geology, 13, 14,53,93,94,347,348
Journal of Research of the United States Geological Survey, 53
Management Sciences, 14
Marine Geology, 94
Mining Congress Journal, 74, 93
Mineralium Deposita, 13, 74, 140
Mining Engineering, 13, 93, 139
Mining Geology, 280
Mining Magazine, 74
Mining World, 93
Operations Research, 92
Operations Research Quarterly, 93
363
364 LISTED JOURNALS
Photogrammetric Engineering, 52
Quarterly of Colorado School of Mines, 14, 53
Science, 13
Transactions of Institute of Mining and Metallurgy, London, 13,93,347
Western Miner, 209, 346

ba4e3ddd028c72570ef868df39c9fe65

Uploaded by

Copyright:

Available Formats

ba4e3ddd028c72570ef868df39c9fe65

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ba4e3ddd028c72570ef868df39c9fe65

Uploaded by

Copyright:

Available Formats

DESIGNING OPTIMAL

STRATEG IES FOR

PLENUM PRESS • NEW YORK AND LONDON

© 1985 Plenum Press, New York

Three kinds of elements are required in order to carry out a pioneering

Chapter One Optimizing Mineral Exploration

Chapter Two Evaluation of the Probability of Detection of

Chapter Three Cost of Detection

Chapter Four Optimizing Ore Detection

4.4. Application of Optimization Theory to Ore Detection ............ 88

Chapter Five Application of the Optimization Methodology to

Chapter Six Designing Optimized Field Programs for the

Chapter Seven Optimized Search for Four Types of Contact

Types of Contact Metasomatic Deposits of the North

Chapter Eight Detection of Ni-Cu Ultramafic Deposits of the

Chapter Nine Optimized Airborne and Ground Search

9.12. Detection Probability and Optimization Tables for Cordillera

Chapter Ten Designing Optimized Field Programs for the

Chapter Eleven Detection of Vein-Gold Deposits of the North

Chapter Twelve Optimal Selection of Exploration Targets for

12.3. Optimal Selection of Exploration Targets without

Appendix: Measurement Conversion Table ................... 353

Subject Index .............................................. 355

Author Index ............................................... 361

Listed Journals ............................................. 363

OPTIMIZING MINERAL EXPLORATION

1.1. GENERAL STATEMENT

1.1.1. Purposes and Nature of Mineral Exploration

has no control. But, again, many other factors of a geographic, economic,

1.1.2. The Context of Mineral Exploration: Uncertainty and Risk

1.1.3. Quantification of Exploration Success in a Context of

1.1.3.2. Historical Success Ratio as an Estimate of the Probability

Probability of technical success

Probability of commercial success =

Probability of commercial success =

1.2. THE MITIGATION OF UNCERTAINTY AND RISK IN

1.2.2. Repetition of Trials

(Probability of success) = 1 - exp( -kN)

where k is an empirically determined constant and N is the number of trials. It

1.2.3. The Sequential Approach

targets. The final stage consists in the detailed three-dimensional investigation

1.2.3.3. Sequentiality in the Space Domain. Sequentiality in the space

1.2.4. The Optimizing Approach

1.2.4.2. Optimization of the Preliminary Planning Phase. In the

1.2.4.4. Optimization of the Evaluation Phase. The final phase of

1.3. CONCLUSION: SCOPE OF THE BOOK

REFERENCES AND SELECTED READINGS

Success in Mineral Exploration

Evaluation of the Probability of Ore Occurrence:

Evaluation of the Probability of Ore Occurrence:

Evaluation of the Probability of Ore Occurrence:

Evaluation of the Probability of Economic Worth:

EVALUATION OF THE PROBABILITY OF

2.1. THE DETECTION PROCESS

2.1.2. The Indirect Approach to the Detection

Detection Detection Type of Detection Procedures involved

surface visual recognition + physical

subsurface and physical sampling + laboratory

tunneling and underground physical sampling + laboratory

etc.) in reconnaissance surveys, or at control grid intersections, as is the case