science of targeting
TRANSCRIPT
Science of targeting: definition, strategies, targeting andperformance measurement
J. M. A. HRONSKY1* AND D. I. GROVES2
1BHP Billiton, Level 34 Central Park, 152 – 158 St Georges Terrace, Perth, WA 6000, Australia.2Centre for Exploration Targeting, School of Earth and Geographical Sciences, University of Western Australia,35 Stirling Highway, Crawley, WA 6009, Australia, and Redstone Resources Limited, Suite 3, 110 – 116 EastParade, East Perth, WA 6004, Australia.
Mineral exploration comprises three sequential steps: development of a business strategy, creation andapplication of a targeting model, and follow-up with direct detection in defined high-priority domains.The main geoscientific challenge is the conceptual targeting phase which can lower geological riskand ensure cost-effective direct-detection exploration. A fundamental tenet of conceptual targeting isthat ore deposits are part of much more extensive systems, and hence that targeting must be carriedout at global through province to district scales. The heterogeneous distribution of ore deposits andtheir power-law size frequency distribution in individual provinces leads to alternative ‘ElephantCountry’ and ‘First Mover’ strategies, both of which employ conceptual targeting, but at differentscales. The first stage of targeting science involves development of robust, multi-scale targeting modelsfor ore-deposit types, particularly larger examples. The targeting models can then be applied toidentify specific targets by interrogating databases compiled as layers of spatially referenced keythemes or parameters. At larger scales in immature terrains, a Hierarchical approach is commonly usedto progressively reduce terrains and identify targets, whereas a Venn-diagram approach, the basis ofmost GIS-based prospectivity analyses, is more commonly used in mature terrains where spatialdatabases are of higher, more homogenous quality. Target ranking is best achieved using amultiplicative probability approach in which it is required that all essential processes in a mineralsystem must have operated to form a significant ore deposit. In practice, one or more critical spatiallyreferenced parameters are used as proxies for the essential processes to develop a target score, whichis a semi-quantitative estimate of probability of the presence of a large ore deposit. Such target rankingcan be used in both proactive ground acquisition and reactive submittal-based project acquisition.Once targets have been defined and explored, it is important that there is critical feedback on therobustness of the targeting exercise such that new information is used to build superior databases and/or targeting models for future area-selection programs.
KEY WORDS: area selection, mineral exploration, ore deposit, targeting.
INTRODUCTION
Mineral exploration, comprising an initial targeting
stage (prediction) followed by a direct detection stage
(detection), does not enjoy steady-state success. Rather,
historically, there have been periods of accelerated
discovery of mineral resources followed by prolonged
periods where fewer major discoveries are made. This
has been due largely to the periodic development of
new techniques (e.g. ground geophysics, airborne
geophysics, satellite-based remote sensing methods,
low detection-limit geochemistry, new drilling meth-
odologies) that significantly improved direct detection,
with some, particularly the airborne techniques, also
indirectly aiding targeting via the generation of better
constrained spatial geological datasets.
The late 1990s and early 2000s represented a period
of significantly decreased discovery of world-class
deposits despite increased exploration expenditure
early in the period (Schodde 2004). This most likely
indicates that technologies currently deployed in ex-
ploration are reaching maturity and are increasingly
incapable of penetrating the deeper or more complex
cover below which the undiscovered deposits lie.
Clearly, there is a need to improve critical direct-
detection technologies (e.g. 3D integrated imaging, more
effective lower cost drilling, refined mobile-ion geo-
chemistry), but there is arguably a more pressing
need to develop more-sophisticated and predictive
conceptual-targeting methodologies to ensure that
the new improved technologies are applied in the
mineral districts with the greatest potential of hosting
*Corresponding author and present address: Western Mining Services (Australia) PL, Suite 26/17 Prowse Street, West Perth WA 6005,
Australia ([email protected]).
Australian Journal of Earth Sciences (2008) 55, (3 – 12)
ISSN 0812-0099 print/ISSN 1440-0952 online � 2008 Geological Society of Australia
DOI: 10.1080/08120090701581356
world-class ore deposits that meet modern metallurgical
and environmental standards.
This paper describes and discusses this targeting
process as a prelude to most of the following papers,
which illustrate varying approaches to conceptual
targeting, or methodologies that support it, for specific
metallic commodities or for particular mineralised
terrains.
SCIENCE OF EXPLORATION TARGETING
Hronsky (2004) argued that exploration targeting was
not simply the application of concepts, including both
genetic and deposit models, from economic geology
research, but was rather the integration of such
concepts with those derived from other fields, such as
geophysics, spatial analysis, mineral economics, deci-
sion science and probability theory, to deliver a busi-
ness outcome. It is thus a high-level scientific discipline
in its own right, being the critical conceptual phase in
the semi-quantitative prediction of the probability of ore
occurrence which leads to decisions on area selection. If
it is done poorly, it is irrelevant how efficient
and effective are the following detection stages of
exploration.
It is important to recognise that exploration targeting
lies at the pinnacle of scientific endeavour in applied
economic geology. This recognition provides a focus for
the type of outputs that are required from the dis-
ciplines, such as economic geology, which feed into the
science of exploration targeting. It also provides a
banner under which activities and tacit knowledge in
the field can be communicated and transferred to the
benefit of the mining and exploration industry. Recogni-
tion of conceptual targeting as a science discipline and
the training of high-quality personnel in its application
will be critical to the success of future, more scientific
and more cost- effective mineral exploration.
The following sections set out very briefly some of the
key elements and concepts in the generic discipline of
exploration targeting.
IMPORTANCE OF SCALE
The final stage of mineral exploration takes place within
a relatively small area that is effectively the deposit-
scale setting of the mineral deposit type sought by that
exploration. This relatively small area has been defined
by successive integrated phases of exploration targeting
from global through province to district scale (Figure 1),
progressively involving geodynamic, tectonic and litho-
structural concepts. It is a fundamental concept that the
relative efficiencies of both exploration targeting (pre-
diction) and detection are scale-dependent. Orebodies
tend to form as chaotic manifestations of much more
extensive systems that are inherently more predictable
at larger scales. As the scale increases, the number of
datasets that are required to make a robust scientific
decision on the prospectivity or potential endowment of
a terrain is also inherently lower. At the global scale,
the geodynamic and lithospheric setting within specific
time windows may be all that is required. For example,
orogenic-gold deposits (Groves et al. 1998) in Archean
greenstone belts would be selectively targeted in ca 2.7
Ga belts (Goldfarb et al. 2001) with short tectonic
histories and a thinned lithosphere (Bierlein et al.
2006). However, at the province scale, more specific
tectonic parameters and spatial datasets showing first-
order structures and lithostratigraphic packages will
be required. For example, the provinces most likely to
contain world-class orogenic-gold deposits should
have curvilinear crustal-scale shear zones, dominant
greenschist-facies domains and complex lithostrati-
graphic sequences, preferably with sedimentary
sequences overlying volcanic successions (Groves et al.
2003). From district to prospect scale, more specific
geological data that are spatially located with increasing
accuracy will be required if predictive concepts are to be
used. (Figure 1). Such data are seldom available at the
prospect scale prior to discovery. Thus, exploration
targeting (prediction) is progressively less effective with
decreasing scale and is replaced by detection at the
prospect scale (Figure 2a). As the targeted area is reduced
by decisions at a progressively smaller scale, the
flexibility decreases and costs increase (Figure 2b),
showing that the opportunity cost of poor initial target-
ing is high.
An important constraint on the quality of exploration
targeting decisions is the availability of high-quality
geoscientific data, particularly reliable regional-
geological map datasets supported by robust geochronol-
ogy at the global to province scale and high-resolution
geophysical and remotely sensed datasets at the district
to prospect scale. In developed countries, such as
Australia, Canada and USA, targeting is more effective
than in developing countries, particularly in Africa and
parts of Asia, where data availability is poor or data are
absent. However, datasets are commonly sufficiently
robust to define prospective exploration areas at the
global to province scales. One of the advantages of the
recognition of exploration targeting as a high-level
scientific discipline is that it allows definition of
the critical datasets required and hence can aid in
the formulation of strategic industry – government and
Figure 1 Hierarchy of scale-dependent targeting processes
and parameters to be considered in erection of targeting
models.
4 J. M. A. Hronsky and D. I. Groves
industry – academia alliances to progressively build
those datasets.
AREA SELECTION STRATEGIES
Area selection initially involves a number of business-
related constraints. These include political decisions on
which parts of the world are considered low to high risk
in terms of political and economic stability, which types
of ore deposits are considered suitable exploration
targets based on company philosophy, and the threshold
size that is acceptable for that ore deposit type given
that approximately two-thirds of all mineral wealth is
derived from a small number of world-class to giant
deposits (Schodde & Hronsky 2006). These decisions are
largely based on company size, and are commonly
company specific, so are not considered further. Only
the geological aspects of area selection are discussed
below.
A number of generic features of ore deposits have a
significant impact on the development of area selection
strategies that, in turn, impact on exploration targeting.
First, the spatial and temporal distribution of ore
deposits is extremely heterogeneous (Groves et al. 2005),
such that relatively few geological provinces are very
well endowed, whereas the majority contain only
sporadic mineral deposits. Second, the size-frequency
distribution of mineral deposits within any province
shows a power-law relationship. The orogenic-gold
deposits in the Kambalda – St Ives district are shown
as an example in Figure 3. Individual provinces also
show characteristic power-law size-frequency distribu-
tions. For example, the Abitibi gold province of Canada
and Eastern Goldfields gold province of Western Aus-
tralia have about an order of magnitude more deposits
at any given large size range than other Archean
greenstone belts globally (Goldfarb et al. 2001). Petro-
leum basins exhibit similar relationships, with the
Persian Gulf province also having about an order of
magnitude more producing wells of a given reservoir
size than in any North American basin. Third, the areal
extent of the geological halo, or ‘footprint,’ of an ore
deposit represents the size of the ore system and is
generally proportional to the size of the ore deposit
within it. The implication is that the larger deposits are
generally discovered first in any district or province if
all other factors are broadly equivalent. The history of
discovery of the komatiite-associated Ni – Cu sulfide
deposits of the Kambalda region, a newly recognised
class of deposit in 1966 (Woodall & Travis 1969), is shown
as an example in Figure 4. Similar patterns are
recognised throughout Australia, for example in the
historically important Broken Hill and Mt Isa base-
metal provinces, the Kalgoorlie gold province, and the
Hamersley iron province.
The two main conclusions that can be drawn from
the above are: (i) the best chances of discovery of a
world-class deposit are to explore where world-class
deposits are already known; and (ii) the largest deposits
are generally discovered in the early phases of mineral
exploration of a terrain. These conclusions provide the
basis for the two basic area-selection strategies that are
most successfully employed in mineral exploration: the
‘Elephant Country’ strategy and the ‘First Mover’ (or
‘Fast Follower’) strategy.
The Elephant Country strategy essentially exploits
the characteristic size-frequency distributions of metal-
logenic provinces. To be successful, several criteria
need to be met. First, the targeted ore-deposit type must
be one that actually clusters in provinces. Most deposits
do, but there are some important exceptions such as
Broken Hill-type base-metal mineralisation which tends
to occur in one, or at most three, extremely large,
isolated deposits in a terrain with otherwise small,
relatively insignificant deposits (Walters 1996). Second,
the known endowment characteristics of the province
must satisfy corporate objectives in that it is a high-risk
strategy to expect to discover a deposit that is much
larger than has already been discovered in the province.
Third, the targeted province must have some residual
potential such as extension of prospective environments
under inadequately tested transported cover. Finally,
the exploration group must be able to perceive some
competitive advantage in the province (e.g. improved
deposit and/or regolith model or more sensitive geo-
chemical or geophysical technique) as Elephant Coun-
try provinces are normally highly competitive in terms
of ground acquisition.
Figure 2 Contrast between global- to district-scale targeting
and project-scale exploration. (a) Relative effectiveness of
prediction based on targeting model and direct-detection
exploration. (b) Relative flexibility and cost of prediction
based on targeting model and project exploration illustrat-
ing the high cost of poor targeting.
Science of targeting 5
The First Mover (or Fast Follower) strategy seeks to
exploit the concept that the largest deposits will be
discovered in the earliest exploration phase, and there-
fore by the first exploration groups to move into the
province. To be successful, two criteria need to be
fulfilled. First, the exploration group must have the
scientific capability to recognise new exploration en-
vironments with high endowment potential together
with the ability to rapidly acquire ground within them.
Second, they must be assured that the targeted province
is genuinely immature in terms of its exploration
history, either because it has not been explored
previously, due to its geographic position or commer-
cial/political issues such as native title, or because the
introduction of new deposit models or exploration and/
or metallurgical technologies presents new opportu-
nities. For example, the province may be under deep and
conductive cover that could not be penetrated by pre-
existing geochemical or geophysical methods, or
known deposits could have been unresponsive to known
metal-extraction techniques. An example of how new
concepts and technology can revitalise a terrain is
shown for porphyry copper deposits in Chile in Figure 5.
Exploration companies must choose whether they
wish to employ one or both of these strategies. In some
cases, the choice of ore deposit type that is targeted will
dictate strategy. The Elephant Country approach re-
quires the company to have financial capability and
commercial skills to acquire ground in a competitive
environment and to have enhanced competence in
exploration technology applied to direct detection.
Conceptual targeting plays a role only in selection of
the best possible, yet affordable, ground at the district
scale in Elephant Country. In contrast, the key driver
for the First Mover (or Fast Follower) strategy is high-
calibre conceptual targeting at the global to province
scale (Figure 1) to recognise the geological opportunity
in an immature terrain, although this will commonly
be integrated with the political skills required to
recognise increasing stability, and have lower risk,
Figure 3 Endowment (current re-
serves plus cumulative produc-
tion) of orogenic-gold deposits in
the St Ives region of Western
Australia showing typical power-
law size frequency of deposits at
the district scale. Source: WMC
Resources 2000 company data.
Figure 4 Endowment (current re-
serves plus cumulative produc-
tion) of komatiite-associated
nickel – copper deposits in the
Kambalda region of Western Aus-
tralia against the year of discov-
ery, illustrating that the larger
deposits tend to be discovered
early in the exploration history
of a region. Source: WMC Re-
sources 2000 company data.
6 J. M. A. Hronsky and D. I. Groves
in a developing country or one emerging from political
or social instability. Thus, conceptual targeting plays a
vital role in both strategies but at different scales.
DEVELOPMENT OF TARGETING MODELS
Targeting models are commonly subdivided into con-
ceptual or process-based, and empirical models, but in
reality the two model types are interrelated. Conceptual
models are commonly based on data inducted from
empirical patterns, and most empirical models are
influenced to some degree by conceptual criteria. The
important general principles in developing targeting
models are set out below.
As developed further below, it is critical to recognise
that different targeting parameters are important at
different scales and to use them at the appropriate scale
in targeting models. If targeting models use data from
ore-formation process models (genetic models), it is
essential that a critical component be defined as a proxy
for the process that can be represented in commonly
available exploration datasets. In other words, the
critical component of the process must be able to be
defined in spatial datasets such as those used to produce
geological maps or aeromagnetic images. It is important
that the datasets be of as uniform quality as possible so
that the areas of superior data quality do not unduly
influence the targeting model, particularly at the
province to district scale. The targeting model should
also focus on those parameters that are essential to the
presence of mineralisation and, as far as possible,
discriminate them from other less-critical parameters,
that are allowable but not essential. Empirical mineral
occurrence data are clearly important to confirm or
deny the existence of the appropriate parameters to
produce the deposit type sought in exploration. How-
ever, it is important that minor mineral occurrences do
not unduly influence the development of the model.
A common weakness of many targeting models,
particularly those of more academic derivation, is that
they focus on prediction of the presence or absence of
mineralisation of a particular deposit type rather than
the probability of occurrence of large economically
important ore systems. However, it is vital for the
mineral industry that the models address factors that
discriminate large from small systems. Again, this
comes back to the scale of conceptual thinking, as
discussed further below. It is noticeable at the global to
province scale that the majority of significant mineral
deposits, particularly the world-class examples, have an
association with major, deep-seated structures that are
probably trans-lithospheric as well as trans-crustal in
scale, as many also lie on, or near, lithospheric
boundaries (Groves et al. 2005). This relationship
appears to be independent of whether the deposit is
syngenetic (either synmagmatic or synsedimentary) or
epigenetic, or the particular deposit type. It is also
apparent that a particular trans-lithospheric structure
may control the distribution of more than one deposit
type over multiple periods of mineralisation. An ex-
cellent example is provided by the so-called Carlin and
Battle Mountain trends of Nevada, USA, where Eocene
Carlin-type deposits are simply the most economic
deposit type in a province of synsedimentary Au – Zn –
Ba, porphyry Cu – Au, epithermal Au – Ag, disseminated
Au and hot-spring Au – Ag deposits that span the time
range from Devonian to Holocene due to reactivation
along major trans-lithospheric structures (Emsbo et al.
in press). A most important consideration is that these
controlling structures are normally not obvious at the
scale of the deposit where most detailed academic
research takes place.
It is clear also that all significant ore bodies are
associated with a tectonothermal system which is
orders of magnitude larger than the immediate deposit
environment. Such a system may be manifested by
major periods of crustal growth (e.g. orogenic-gold
Figure 5 Year of discovery of porphyry copper – gold deposits in Chile, illustrating that the new concepts or new technologies
can rejuvenate discovery in a mature Terrain. Note that the earliest discoveries tend to be some of the largest. Source: WMC
Resources 1999 company data.
Science of targeting 7
deposits: Goldfarb et al. 2001), widespread magmatism of
a specific age in a specific subduction environment (e.g.
porphyry Cu – Au deposits: Billa et al. 2004), or a
particular type of basin-filling package in a specific
structural setting (e.g. SEDEX deposits: Large et al.
2005). Although the genetic links may be somewhat
equivocal, it is nonetheless vital to recognise the
importance of the system concept (Wyborn et al. 1994)
in developing robust targeting models. An important
aspect is to recognise that the first-order parameters in
the targeting model may not be defined or even
discussed in research papers on the deposit type because
of the nature of most economic geology research.
TARGET IDENTIFICATION
Once a targeting model has been developed, it can then
be applied to identify specific targets within a province
or district. In order to achieve this, key targeting themes
or parameters must be compiled as layers of informa-
tion that represent key components of the targeting
model and can be depicted spatially. For most ore
deposit types that are targeted, these will comprise one
or more layers from geodynamic, tectonic, structural,
lithostratigraphic, and metamorphic and igneous pet-
rology datasets, together with available data on known
occurrences of the deposit type sought in the targeting
exercise.
There are two basic end-member approaches to the
process of target identification: the Hierarchical and
Venn-diagram systems (Figure 6). In practice, many
targeting exercises will represent a hybrid between the
two, because the hierarchical approach is commonly
used from the global to province scale, whereas the
Venn-diagram approach is more commonly used at the
district scale.
The hierarchical approach is based on the concept
that most targeting models comprise a series of critical
parameters that are manifested at different scales. Thus,
geodynamic parameters may be used to identify the
segment of a continent or country that is most
prospective (A in Figure 6); tectonic and crustal-scale
structural data may be used to define the most
prospective province (B in Figure 6); and a greater
variety of geological parameters (e.g. structural, basi-
nal, lithostratigraphic, igneous and metamorphic) may
be used to define the most prospective district (C in
Figure 6). The hierarchical approach is more generally
applied in a First Mover strategy when the initial area
of interest is initially extensive and/or the targeting
group has to compile most of the key targeting themes
themselves from primary data where there are poor
spatial datasets, for example in developing countries.
An important concept when targeting in such areas with
sparse data is that of permissible parameters in that if
there is no negative information on the presence of a key
factor, it should be considered neutral in any decision-
making process.
The Venn-Diagram approach is most commonly used
in areas of relatively high exploration maturity with
good quality and relatively uniform data, and is hence
normally linked to the Elephant Country strategy. It
essentially views target identification as the process of
locating areas where there is a conjunction of a number
of critical parameters in the targeting model: the area of
overlap of parameters D, E and F in Figure 6. This
philosophical approach underpins the methodology of
GIS prospectivity mapping (Agterberg 1974; Bonham-
Carter 1994; Brown et al. 2000), several examples of
which are presented in this thematic issue.
TARGET RANKING
The process of target ranking is essentially an extension
of the target identification process but is separated here
for clarity and to emphasise its potential importance.
Traditionally, target ranking has involved the assign-
ment of scores to each of a suite of parameters
considered essential to permissive, and then summing
these scores to provide a total target score. Normally,
the scores are weighted for the perceived relative
importance of each targeting parameters, and perhaps
for the degree of confidence in the assignment of the
score.
More recently, however, an alternative approach has
been proposed by workers such as Henley (1997)
and Lord et al. (2001), who advocate a multiplicative
probability-based approach similar to that used in the
petroleum industry. The probability-based approach to
target ranking is based on distilling the targeting model
down to parameters that are the proxies of critical
processes in the generation of the targeted deposit style.
This approach is described in detail by Kreuzer et al. (in
press). For example, epigenetic hydrothermal deposits
can be described in a minerals system (Wyborn et al.
1994) as involving a source, fluid conduit, trap and cap
or outflow system. In the probability-based approach,
each of these processes must have operated, and each
should be independent of the other. An example of such
a system is shown in Figure 7 for an orogenic-gold
system. In this case, the critical parameters that are
the proxies for the process must be recognised and
allocated to that part of the minerals system. An
example showing the critical proxies for the critical
Figure 6 Schematic diagram illustrating the two end-
member approaches to target definition: the hierarchical
and Venn-diagram methodologies. The former is commonly
used in immature terrains, whereas the latter is used in
mature terrains where it forms the basis for most GIS-based
prospectivity analyses.
8 J. M. A. Hronsky and D. I. Groves
parts of an orogenic-gold system are shown in Table 1.
This allocation of proxies or parameters to specific
processes within the minerals system allows formula-
tion of a ranking system that demands that critical
proxies for every process must be present for an area to
be prospective. The ranking system must give equal
weight to the presence of each process, rather than to
each parameter or proxy, and care must be taken that
the parameters are not measuring the same part of the
process and hence are overemphasised in the ranking
procedure. A probability that the critical process
operated in the minerals system is estimated, either
using 1.0 if it is proven to be present, 0 if it proves to be
absent, or 0.5 if there are no data to make a decision, or
using a fuzzy-logic system with ranking of favourability
where more than one critical parameter or proxy
represents one critical process in the system.
There are a number of advantages to this approach. It
clearly defines the critical processes in the targeting
model and the equivalent parameters that proxy for
those processes. It does not excessively penalise targets
where one or more parameters that represent a critical
process are permissive without definitive evidence that
it is present. If a critical process (e.g. source, fluid
Figure 7 Schematic diagram illus-
trating a minerals system model
for orogenic-gold deposits show-
ing source, fluid conduit, deposit
trap and cap or seal (adapted from
Groves et al. 1998). The key para-
meters that are proxies for these
critical processes are shown in
Table 1.
Table 1 Mineral systems model for orogenic-gold deposits.
Key parameter Proxies for parameter
Critical process—Thermal energy and ore-fluid source
Thinned lithosphere Short terrane history (Bierlein et al. 2006)
Major crust-forming event Geochronological data
Orogenic-gold mineralisation Known gold occurrences
Critical process—Plumbing fluid systems
Deep fluid conduits Crustal-scale shear zones preferably with lamprophyres
Focused fluid flow High-strain shear zones in low-strain belts
High-damage zone in lower order faults Jogs, thrust duplexes, fault intersections
Critical process—Trap (depositional) site
Structural trap Locked-up anticline, thrust-tip
Rheological trap Rheologically indexed rock types—contacts with high contrast
Chemical trap Reactive rocks with high Fe or C
Critical process—Fluid seal (cap)
Stratigraphic cap Impermeable rocks over structurally permeable rocks (e.g. sedimentary cap)
Structural cap Impermeable thrust stacks
Critical process—Outflow zone
Fluid dispersion Metal dispersion halos
The table shows the principle of critical processes and some of their key parameters and proxies which can be detected in geological
maps or from geological or geophysical databases normally available to a targeting team in an exploration company. It is illustrative
and is not intended to be a complete list of parameters and proxies at all scales. Adapted from Groves et al. (2000).
Science of targeting 9
conduit, trap or cap) is missing in the minerals system,
the probability of success of that target is zero because
one of the critical processes will rank 0 in the analysis.
It is thus more likely to produce a more realistic and
robust numerical separation between the scores for
potential world-class targets and weaker targets than
other methods. The methodology is also transparent and
has a high scientific integrity.
A weakness of the approach is that it is sensitive to
the nature of the minerals system model that is
developed and hence, in turn, sensitive to the choice
of critical processes and their proxies in terms of key
parameters for the targeting model. Minerals systems
are inherently more complex than petroleum systems,
where the three process steps of source – transport –
trap can be applied universally. The key is to define
those key parameters that are essential for the devel-
opment of a large ore deposit, while preserving some
parameters that are permissive in case there are data
gaps, particularly in immature terrains. In general,
this approach is recommended because it is an
effective and robust way of organising and interrogat-
ing data, it is an effective way of communicating the
targeting outcome, and it allows for a consistent and
transparent approach across multiple targeting exer-
cises in difficult terrains. However, it can never be
truly quantitative, because the number of critical
processes and related key parameters is ultimately a
somewhat arbitrary choice and hence affects the total
probability; few of the key parameters are likely to be
truly independent of one another, as it is necessary for
the methodology; and it is erroneous to assume that
lack of data equates to an equal probability of presence
or absence of a key parameter in spatially heteroge-
neous datasets: it is statistically more likely to be
absent.
As in all scientific endeavours, the quality of the
outcome will depend on the collective knowledge of the
targeting team and the robustness of the targeting
models they produce based on their understanding of
the critical components of minerals systems for a
variety of ore deposit types. The team should also
recognise the common pitfalls in ranking that are
independent of the quality of the targeting model
employed. First, it is important that the ranking
scheme is not biased toward mature exploration areas
of greater data density and therefore excessively
discriminates against areas with high potential endow-
ment but limited data. This typically happens when too
much emphasis is placed on known mineralisation
without adequately considering its context in the total
targeted terrain. Second, it is vital to separate model
parameters from the data that were used to derive
them. For example, the presence of magnetite-rich
bodies may be a critical parameter in several mineral
systems (e.g. hydrothermal iron ore, iron-oxide Cu – Au
or BIF-hosted orogenic-gold deposits), and may be
recognised by the presence of magnetite and/or a
magnetic anomaly and/or a gravity anomaly.
However, it should be treated as a single key para-
meter, not three independent parameters (magnetite,
magnetic anomaly, gravity anomaly) in the ranking
process.
PERFORMANCE MEASUREMENT AND FEEDBACKLOOPS
One of the major deficiencies of currently practiced area
selection is the general lack of follow-up to rigorously
estimate how effective it has been in improving the
probability of discovery of a large ore deposit. Such
quantification of the targeting exercise and team
performance is required to measure the degree of
success, and hence provide a quality control that leads
to improvement in the targeting process.
Rigorous application of the probability-based
approach, outlined above, is an important step towards
this achievement but requires robust calibration with
empirical case-history data. In reality, the exploration
industry seldom allocates resources to post-mortems of
targeting exercises, largely because of the transience of
staff, particularly during boom periods, the competing
lure of the next opportunity, and the relative difficulty
in accessing the relevant historical data. As a conse-
quence, the industry has very little information on
which are the most robust methodologies in exploration
targeting. Case histories in the public domain are
strongly biased towards those of successful exploration
programs, and almost invariably credit success to a
mixture of good teamwork, successful detection and
serendipity, with little emphasis on targeting models
and their value in the process (Australian Mineral
Foundation 1997, 2001).
However, if there are to be major improvements in
targeting capability, there is an urgent need to employ
better quantification of targeting performance. After
targets are generated, it is vital that information
obtained from testing the targets is fed back into the
targeting model and supporting databases. This is
particularly important in less-developed countries
where information obtained from field assessments of
province-scale target areas may have a substantial
impact on both the database and model. As exploration
moves more under cover, targeting models become an
essential tool to focus exploration, increase the geologi-
cal probability that any individual targeted prospect
will become an economic deposit, and hence reduce the
risk:reward ratio.
PROACTIVE CONCEPTUAL TARGETING VSREACTIVE TARGET EVALUATION
Pragmatically, there are two end-member approaches to
acquiring exploration properties. The first is essentially
the process described above, involving compilation of
databases over the area of interest, development of a
targeting model, generation of targets, and proactive
acquisition of ground over the highest ranked targets.
The second is to focus on evaluating properties sub-
mitted by other parties, commonly junior explorers,
followed by acquisition of the most prospective of those
submitted. Commonly, the targeting process is only
considered to be an integral part of the first approach.
However, this is somewhat of a misconception, since
there has to be a logical mechanism to prioritise
submittals. In order to robustly rank the submittals
10 J. M. A. Hronsky and D. I. Groves
and capture value for the company, a superior approach
to a submittal-based ground-acquisition program is to
have previously completed a targeting exercise for the
province or district of interest. In this way, any
submitted property can be evaluated more effectively
within the regional context, which largely controls the
potential of the mineral systems, and hence the potential
endowment of the submitted prospect, itself. Hence, all
acquisitions should be based on a framework provided
by the conceptual targeting process.
CONCLUSIONS
The targeting process discussed above is shown sche-
matically as a flow sheet in Figure 8. Following business
decisions that relate to the individuality of the mining
and/or exploration company, area selection is an
essential part of the mineral supply process. Although,
in the past, targeting success has been attributed
commonly to serendipity or favourable organisational
structure, world-class to giant ore deposits are predicted
to be more difficult to discover as districts with shallow
regolith became increasingly mature, and conceptual
targeting will assume increasing importance in the
future. This paper outlines the principles that should be
applied in area selection so that it becomes a better
understood, more effective, and more valued segment of
exploration business. Targeting needs to be recognised
in its own right as a scientific endeavour that integrates
information from the various geoscience disciplines,
translates critical processes and parameters from that
Figure 8 Summary of the targeting process from the business interface through developing the targeting model and spatial
databases to defining and testing targets, with a subsequent feedback loop to improve the targeting model and information in
the databases.
Science of targeting 11
information into spatially registered data layers or
themes, and interrogates those in terms of robust
targeting models to produce at least semi-quantitative
ranking of target areas at the global, province and
district scale. The science of targeting should underpin
all future exploration endeavours in order to reduce
geological risk and provide more cost-effective project
exploration using direct detection technologies and
methodologies.
ACKNOWLEDEGMENTS
JMAH acknowledges many former colleagues in WMC
Exploration Division and present colleagues at BHP-
Billiton, particularly Richard Schodde and Nicholas
Hayward, for contributions to the concepts presented in
this paper. DIG also acknowledges the benefits of
discussions with colleagues from Redstone Resources
and Newmont, particularly Stephen Gardoll and Musie
Gebre-Mariam, who have helped to bring targeting
concepts into better focus. The authors both acknowl-
edge helpful review comments by Mike Etheridge.
REFERENCES
AGTERBERG F. P. 1974. Automatic contouring of geological maps
to detect target areas for mineral exploration. Mathematical
Geology 6, 373 – 395.
AUSTRALIAN MINERAL FOUNDATION 1997. New Generation Gold Mines
‘97: Case Histories of Discovery, Conference Proceedings.
Australian Mineral Foundation, Glenside.
AUSTRALIAN MINERAL FOUNDATION 2001. New Gen Gold 2001, Con-
ference Proceedings. Australian Mineral Foundation, Glenside.
BIERLEIN F. P., GROVES D. I., GOLDFARB R. J. & DUBE B. 2006.
Lithospheric controls on the formation of provinces hosting
giant orogenic gold deposits. Mineralium Deposita 40, 874 – 886.
BILLA M., CASSARD D., LIPS A. L. W., BOUCHOT V., TOURLIERE B.,
STEIN G. & GUILLOU-FROTTIER L. 2004. Predicting gold-rich
epithermal and porphyry systems in the central Andes with
a continental-scale metallogenic GIS. Ore Geology Reviews 25,
39 – 47.
BONHAM-CARTER G. F. 1994. Geographic Information Systems for
Geoscientists—Modelling with GIS (Computer Methods in the
Geosciences 13). Pergamon Press, New York.
BROWN W. M., GEDEON T. D., GROVES D. I. & BARNES R. G. 2000.
Artificial neural networks: a new method for mineral pro-
spectivity mapping. Australian Journal of Earth Sciences 47,
757 – 770.
EMSBO P., GROVES D. I., HOFSTRA A. H. & BIERLEIN F. P. The giant
Carlin gold province: a protracted interplay of orogenic, basinal
and hydrothermal processes above a lithospheric boundary.
Mineralium Deposita (in press).
GOLDFARB R. J., GROVES D. I. & GARDOLL S. 2001. Orogenic gold and
geologic time: a global synthesis. Ore Geology Reviews 18, 1 – 75.
GROVES D. I., CONDIE K. C., GOLDFARB R. J., HRONSKY J. M. A. &
VIELREICHER R. M. 2005. Secular changes in global tectonic
processes and their influence on the temporal distribution of
gold-bearing mineral deposits. Economic Geology 100, 203 – 224.
GROVES D. I., GOLDFARB R. J., GEBRE-MARIAM M., HAGEMANN S. G. &
ROBERT F. 1998. Orogenic gold deposits: a proposed classification
in the context of their crustal distribution and relationship to
other gold deposit types. Ore Geology Reviews 13, 7 – 27.
GROVES D. I., GOLDFARB R. J., KNOX-ROBINSON C. M., OJALA J.,
GARDOLL S., YUN G. & HOLYLAND P. 2000. Late-kinematic timing
of orogenic gold deposits and significance for computer-based
exploration techniques with emphasis on the Yilgarn Block,
Western Australia. Ore Geology Reviews 17, 1 – 38.
GROVES D. I., GOLDFARB R. J., ROBERT F. & HART C. J. R. 2003. Gold
deposits in metamorphic belts: overview of current understand-
ing, outstanding problems, future research and exploration
significance. Economic Geology 98, 1 – 29.
HENLEY R. W. 1997. Risky business: the essential blending of
financial and scientific skills in the modern resources sector.
Window on New Zealand 1997: New Zealand Minerals and Mining
Conference Proceedings, pp. 29 – 33.
HRONSKY J. M. A. 2004. The science of exploration targeting. In:
Muhling J. ed. SEG 2004. Predictive Mineral Discovery Under
Cover, pp. 129 – 133. University of Western Australia, Centre for
Global Metallogeny, Publication 33.
KREUZER O. P., ETHERIDGE M. A., MCMAHON M. E. & HOLDEN D. J.
Probabilistic ore systems modeling: a new tool for quantitative
risk analysis and decision-making in exploration. Economic
Geology (in press).
LARGE R. R., BULL S. W., MCGOLDRICK P. J., WALTERS S., DERRICK G.
M. & CARR G. R. 2005. Stratiform and strata-bound Zn – Pb – Ag
deposits in Proterozoic sedimentary basins, Northern Australia.
Economic Geology 100th Anniversary Volume, pp. 931 – 964.
LORD D., ETHERIDGE M. A., WILLSON M., HALL G. & UTTLEY P. J. 2001.
Measuring exploration success: an alternative to the discovery-
cost-per-ounce method of quantifying exploration success.
Society of Economic Geologists Newsletter 45, 1, 10 – 16.
SCHODDE R. C. 2004. Discovery performances of the western world
gold industry over the period 1985 – 2003. In: PACRIM 2004
Proceedings, pp. 367 – 380. Australasian Institute of Mining and
Metallurgy, Melbourne.
SCHODDE R. C. & HRONSKY J. M. A. 2006. The role of world-class
mines in wealth creation. In: Dogget M. D. & Parry J. R. eds.
Wealth Creation in the Minerals Industry: Integrating Science,
Business, and Education, pp. 71 – 90. SEG Special Publication 12.
WALTERS S. J. 1996. An overview of Broken Hill-type deposits.
University of Tasmania, CODES Special Publication 1, 1 – 10.
WOODALL R. & TRAVIS G. A. 1969. The Kambalda nickel deposits,
Western Australia. 9th Commonwealth Mining & Metallurgy
Congress, 2, pp. 517 – 533.
WYBORN L. A. I., HEINRICH C. A. & JAQUES A. L. 1994. Australian
Proterozoic mineral systems: essential ingredients and map-
pable criteria. In: Australasian Institute of Mining and Metal-
lurgy Annual Conference Proceedings, pp. 109 – 115.
Received 25 August 2006; accepted 28 April 2007
12 J. M. A. Hronsky and D. I. Groves