SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 803
Probabilistic modelling of lithological
domains and its application to resource
evaluation
by X. Emery* and K.E. Gonza′lez†
Synopsis
The modelling of lithological domains is a critical step in mineral
resource and reserve evaluation. Deterministic models rely on an
interpretation of the available drill hole data and expert knowledge,
but do not account for the uncertainty in the spatial extent of the
lithological domains. Instead, probabilistic models based on
conditional simulation can be used to map the probabilities of
occurrence of the domains within the deposit, which reflect the
uncertainty in their presence or absence at unsampled locations.
The mineral grade model is then obtained by weighting the
predicted grades associated with each lithological domain by the
probability of occurrence of this domain. Such an approach
accounts for the continuity of the grades proper to each lithological
domain and for the uncertainty in the spatial extent of these
domains within the deposit. An application to a porphyry copper
deposit is presented.
Keywords: spatial uncertainty; plurigaussian simulation;
geological control; kriging; geostatistics.
Introduction
Lithological characteristics are often related to
the spatial distribution of grades in ore
deposits and must be taken into account for
mineral resource evaluation. The traditional
approach consists of delineating domains
defined by rock type, mineralization and/or
alteration, on the basis of the available drill
hole information and of the interpretation of
the orebody genesis. Evaluation of the mineral
resources in each domain then follows by
means of geostatistical tools.1–3 However, this
approach suffers from two limitations. First,
because the lithological domains are treated
independently, they ought to have hard
boundaries, i.e. with abrupt transitions in the
grade distribution when crossing a boundary.
Second, the mining geologist must delineate
the exact shape of each domain in the ore
deposit.3,4
This last limitation is critical and has
motivated the use of geostatistical techniques
to improve the lithological domain definition
and to reduce the uncertainty in the position of
their boundaries. One such technique is
The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107
T
r
a
n
s
a
c
t
i
o
indicator kriging that relies on a coding of the n
lithological information into indicator (binary)
variables. A variant is morphological kriging,
which accounts for the orientation of strata
defined by the interpreted geometry of the P
orebody.5 Both of these techniques (indicator a
and morphological kriging) allow determining p
which lithological domain is likely to prevail at
a specific location, leading to a more accurate e
model of the deposit. However, they are not r
fundamentally different from the traditional
approach, since no uncertainty in the domain
boundaries is assumed when predicting the
grades.
An alternative approach is to perform a co-
kriging of the grades (primary variable) and
the lithological information coded through
indicators as secondary variables.6,7 Unlike the
previous methodologies, here the prediction
relies on a single grade variogram and one
cannot impose particular conditions to each
lithological domain, such as the nugget effect
or anisotropy ratio of the variogram model.
Accordingly, the final grade model may not
reproduce the patterns of spatial continuity
proper to each domain.
Another option is local kriging, which is
based on a variogram model whose parameters
are chosen in each kriging neighbourhood by
cross-validation.3 However, a drawback of this
approach is the necessity to work with default
variograms models that depend on few
parameters, e.g. a model defined by a power
function of the separation distance. In
particular, it becomes difficult to handle
patterns such as local anisotropies, local
nugget effects or local ranges.
* Department of Mining Engineering, University of
Chile, Santiago, Chile.
† Golder Associates S.A., Santiago, Chile
© The Southern African Institute of Mining and
Metallurgy, 2007. SA ISSN 0038–223X/3.00 +
0.00. Paper received May. 2007; revised paper
received Oct. 2007.
▲
REFEREED PAPER DECEMBER 2007 803
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 804

Probabilistic modelling of lithological domains and its application

The objective of this work is to design and to implement
a methodology that improves the geological control for
predicting mineral grades, via the recourse to a probabilistic
modelling of the lithological domains. The methodology will
be presented through a case study corresponding to a
porphyry copper deposit.
Presentation of the case study
Radomiro Tomic is a porphyry copper deposit located in
northern Chile, at about 40 km from Calama city and at an
average elevation of approximately 3 000 metres above mean
sea level. The orebody is approximately 5 km long and 1 km
wide and is lodged along the West Fissure fault zone in the
Domeyko fault system. It can be split into five main
lithological domains, namely:8,9
➤ Gravel domain—sedimentary material situated in the
superficial layer and containing some exotic mineral-
ization in its base (chrysocolla, atacamite, copper wad
and gypsum)
➤ Leached domain—intensely leached-out material,
mainly containing iron oxides and hydroxides, clay
and quartz
➤ Oxide domain—mineralized material located below the
contact with gravel and containing copper oxides,
mainly of chrysocolla, malachite and atacamite
➤ Secondary sulphide domain—underlying the oxide
domain and consisting of chalcocite (upper part) and
covellite (lower part)
➤ Primary sulphide domain—consisting of bornite,
chalcopyrite or pyrite.
The available dataset comprises 140 555 samples from
exploration diamond drill holes, composited at 3 metres, with
information on the total copper grade and the lithology that
predominates in the sample.
Figure 1 presents a location map of the samples over a
vertical section crossing the area of interest. Statistics on the
assayed copper grades are given in Table I and their
histograms in Figure 2.
One observes that the direction of greatest geological
continuity dips about 7º with respect to the east-west
direction. To account for this dip in the geostatistical
analysis, the original coordinate axes (X,Y,Z) are rotated so
that the new ‘vertical’ coordinate axis (Z′) is orthogonal to
the direction of greatest continuity.
The spatial variability of copper grades is highly
dependent on the lithological domain under consideration
(Figure 3).
In the gravel domain, the grade variogram has a larger
range in the horizontal plane than along the vertical
direction, which can be explained because gravel is present as
a thin cover with a larger horizontal extension. In general,
the leached domain contains copper grades less than 0.2%
Cu; the grade variogram in this domain has a much smaller
sill than in the other lithological domains. Regarding the
oxide and secondary sulphide domains, copper grades display
a greater continuity along the vertical direction than in the
horizontal plane, with ranges below 100 metres in both
domains. In contrast, the grade variation of the primary
sulphide domain has a strong anisotropy, with a greater
variability along the vertical direction.
Probabilistic modelling of lithological domains
Several geostatistical techniques have been developed in the
past decades for modelling lithological domains. Most of them
are based on indicator kriging or on the conditional
simulation of categorical variables. In the following, the focus
is given to the simulation approach, through an application of
the so-called plurigaussian model, in which the lithological

Drill hole data Gravel

3000
Leached
Elevation (m)

2900
Oxide
2800
Primary
Sulphide
2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 Secondary
Easting (m) Sulphide

Figure 1—Location of the drill hole samples over a cross-section. Colour code indicates the lithological domain

Table I

Basic statistics on total copper grades

Number of data Minimum (% Cu) Maximum (% Cu) Mean (% Cu) Standard dev. (% Cu)

Gravel 7 252 0.00 7.35 0.20 0.58

Leached material 37 040 0.00 8.69 0.11 0.12
Oxide 48 967 0.03 15.04 0.63 0.56
Secondary sulphide 14 516 0.01 22.76 0.68 0.67
Primary sulphide 17 846 0.00 7.84 0.28 0.25
Total 125 621 0.00 22.76 0.41 0.51
Uninformed grades 14 934
▲

804 DECEMBER 2007 VOLUME 107 REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 805

Probabilistic modelling of lithological domains and its application

T
Entire deposit Gravel
Number of Data 125621 .700 Number of Data 7252
r
.200 mean .41 mean 0.20
std. dev.
coef. of var
.51
1.27
.600 std. dev.
coef. of var
.58
2.85
a
.500

Frequency
.150
n

Frequency
maximum 22.76 maximum 7.35
upper quartile .54 .400 upper quartile .10
.100 median .26 median .04
lower quartile
minimum
.10
.00
.300 lower quartile
minimum
.02
.00
s
.200
.050
.100 a
.000
.00 .50 1.00 1.50 2.00
.000
.00 .50 1.00 1.50 2.00 c
Total copper grade (%) Total copper grade (%)
t
i
Leached Oxide
.500 Number of Data
mean
37040
.11 .160 Number of Data
mean
48967
.63
o
std. dev. .56
.400
std. dev.
coef. of var
.12
1.09
.120
coef. of var .89 n
maximum 15.04
Frequency

Frequency
maximum 8.69
.300 upper quartile .14 upper quartile .76
median .09 .080 median .49
lower quartile .05 lower quartile .31
.200 minimum .00 minimum .03
.040
.100 P
.000
.00 .50 1.00 1.50 2.00
.000
.00 .50 1.00 1.50 2.00 a
Total copper grade (%) Total copper grade (%)
p
e
Primary sulphide Secondary sulphide
.250 Number of Data
mean
17846
.28
.160 Number of Data 14516 r
std. dev. .25 mean .68
.200 coef. of var .91 std. dev. .67
.120 coef. of var .99
Frequency
Frequency

maximum 7.84 maximum 22.76

.150 upper quartile .37
median .21 .080 upper quartile .84
lower quartile .11 median .52
.100 minimum .00 lower quartile .31
minimum .01
.040
.050

.000 .000
.00 .50 1.00 1.50 2.00 .00 .50 1.00 1.50 2.00
Total copper grade (%) Total copper grade (%)

Figure 2—Histograms of assayed total copper grades, for the entire deposit and for each lithological domain

Entire deposit Gravel

Copper grade variogram

Copper grade variogram

vertical
.300 horizontal .300
horizontal
.200 vertical .200

.100 .100

.000 .000
0. 100. 200. 300. 400. 500. 0. 100. 200. 300. 400. 500.
Distance (m) Distance (m)

Leached Oxide
Copper grade variogram

Copper grade variogram

.0160 vertical horizontal

.300
.0120 horizontal
vertical
.200
.3080

.0040 .100

.0000 .000
0. 100. 200. 300. 400. 500. 0. 100. 200. 300. 400. 500.
Distance (m) Distance (m)

Primary sulphide Secondary sulphide

Copper grade variogram

Copper grade variogram

.300 vertical horizontal

.400

.300 vertical
.200
.200
.100
horizontal .100

.000 .000
0. 100. 200. 300. 400. 500. 0. 100. 200. 300. 400. 500.
Distance (m) Distance (m)

Figure 3—Experimental (dotted lines) and modelled (solid lines) variograms of assayed total copper grades, for the entire deposit and for each
lithological domain
▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 REFEREED PAPER DECEMBER 2007 805
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 806

Probabilistic modelling of lithological domains and its application

domains are defined by truncating synthetic Gaussian ➤ location x within the gravel domain ⇐
⇒ Y1 (x) < y1
random fields.10–12 This model has been widely applied to the
➤ location x within the oxide domain ⇐
⇒ y1 ≤ Y1 (x) < y 1′
characterization of oil reservoirs and mineral deposits.13–16
Y2 (x) < y2
{
Its implementation consists of three main steps:
➤ location x within the leached domain ⇐
⇒ y1 ≤ Y1 (x) < y 1′
➤ Definition of a truncation rule that provides
y2 ≤ Y2 (x)
{
information about the allowed and forbidden contacts
⇒ y′1 ≤ Y1 (x)
➤ location x within the secondary ⇐
between lithological domains.
➤ Definition of the Gaussian field variograms, which sulphide domain Y3 (x) < y3
{
model the spatial continuity of the lithological domains. ⇒ y′1 ≤ Y1 (x)
➤ location x within the primary ⇐
➤ Simulation of the Gaussian fields, conditionally to the sulphide domain y3 ≤ Y3 (x)
{
available categorical data, and truncation to obtain the
simulated lithological domains. The thresholds {y1, y1′, y2, y3} (with y1′ > y1) are defined
on the basis of the proportions of the lithological domains
over the domain of interest.11
Modelling contacts between lithological domains
Modelling geological continuity
(truncation rule)
The variograms of the Gaussian random fields have to be
In the Radomiro Tomic copper deposit, all the lithological determined to complete the specification of the plurigaussian
domains are in contact with each other, except the gravel model. These variograms are related to the indicator
domain that cannot be in contact with the two sulphide variograms of the lithological domains, which can be inferred
mineral bearing domains (located below the water table). from the lithological information at the drill hole
This leads us to use three independent Gaussian random samples.11,17 In the following statements, we give the results
fields (denoted by Y1, Y2 and Y3) and to consider that the and present the fitting along the main directions of
domain prevailing at a given location x is defined in the anisotropy (horizontal plane and vertical direction in the
following fashion: rotated system of coordinates) (Figure 4).

Gravel Oxide
0.12
vertical
0.1 0.25
Indicator variogram

vertical
horizontal
Indicator variogram

0.08 0.2
horizontal
0.06 0.15

0.04 0.1

0.02 0.05

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Distance (m) Distance (m)

Leached Primary sulphide

0.25 0.07

0.06
0.2 vertical
Indicator variogram

Indicator variogram

horizontal
0.05 vertical horizontal
0.15 0.04

0.03
0.1
0.02
0.05
0.01

0 0
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000
Distance (m) Distance (m)

Secondary sulphide
0.08
0.07
vertical
horizontal
Indicator variogram

0.06

0.05
0.04
0.03

0.02

0.01

0
0 100 200 300 400 500 600 700 800 900 1000
Distance (m)

Figure 4—Experimental (dotted lines) and modelled (solid lines) indicator variograms along the horizontal and vertical directions
▲

806 DECEMBER 2007 VOLUME 107 REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 807

Probabilistic modelling of lithological domains and its application

T
➤ Variogram model for the first Gaussian field (Y1): Simulation of lithological domains r
γ1 = 0.04 cubic (100 m, 100 m) + 0.06 cubic (250 m,
The plurigaussian model is now used to create a set of 100 a
200 m) + 0.90 cubic (2 500 m, 200 m)
➤ Variogram model for the second Gaussian field (Y2): conditional simulations of the partition into lithological n
γ2 = 0.04 spherical (100 m, 100 m) + 0.30 spherical (2 domains over a grid with a 5 m × 5 m × 2.5 m mesh. As an s
000 m, 120 m) + 0.15 spherical (100 m, ∞) + 0.15 illustration, Figure 5 displays four simulations over the same
a
spherical (2000 m, ∞) cross-section as in Figure 1, conditioned to the information
observed at the nearby drill hole samples.
c
➤ Variogram model for the third Gaussian field (Y3):
γ3 = 0.40 cubic (200 m, 200 m) + 0.60 cubic (25 000 m, The simulations provide a set of plausible scenarios and t
10 000 m). allow assessing the uncertainty in the spatial distribution of i
In the previous equations, the distances given in brackets the lithological domains within the deposit. In particular, one o
can characterize these domains by mapping their probabilities
stand for the ranges along the horizontal and vertical n
directions respectively. The shape of the Gaussian variograms of occurrence at each location of the deposit (in practice, their
near the origin is linked to the regularity of the boundaries frequencies of occurrence over the simulations) (Figure 6).
between the lithological domains.17 In particular, the cubic The probability maps so obtained account for the drill
variogram model is smooth at the origin and is associated hole information, for the topological contacts between the P
with regular boundaries. In contrast, the spherical model lithological domains (via the truncation rule) and for their a
(used for the second Gaussian field to separate the oxide spatial continuity (indicator variograms). Near the drill holes
p
domain from the leached domain) is linear at the origin and (dipping straight lines that appear on the maps), one
entails a more irregular boundary between these two observes that the probabilities are close to zero or to one, e
domains. which means that there is little uncertainty. r

Simulation # 1 Simulation # 2
3000 3000
Elevation (m)
Elevation (m)

2900 2900

2800 2800

2700 2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m) Easting (m)
Simulation # 3 Simulation # 4
3000 3000
Elevation (m)
Elevation (m)

2900 2900

2800 2800

2700 2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m) Easting (m)

Gravel Leached Oxide Primary Secondary

sulphide sulphide

Figure 5—Four conditional simulations of the lithological domains. Conditioning data are superimposed

Gravel Leached
3000 3000
Elevation (m)
Elevation (m)

2900 2900
2800 2800
2700 2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m) Easting (m)
Oxide Primary sulphide
3000 3000
Elevation (m)
Elevation (m)

2900 2900

2800 2800

2700 2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m) Easting (m)
Secondary sulphide
3000
Elevation (m)

Scale for probabilities

2900

2800
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m)

Figure 6—Probabilities of occurrence of each lithological domain, obtained from a set of 100 conditional simulations
▲

The Journal of The Southern African Institute of Mining and Metallurgy VOLUME 107 REFEREED PAPER DECEMBER 2007 807
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM Page 808

Probabilistic modelling of lithological domains and its application

Application to the evaluation of mineral resources given location of the deposit, and that the predicted copper
grade at this location is 0.12% when using the gravel domain
To incorporate the lithological information in the grade data, 0.16% when using the leached domain data, and 0.55%
model, the simplest approach is to use a deterministic when using the oxide domain data. According to the previous
interpretation of the lithological domains, based on the drill equation, the expected grade is equal to:
hole data and expert geological knowledge. An example of
Predicted Grade = 0.15 × 0.12 + 0.45 × 0.16 +
such an interpretation is shown in Figure 7 for the same
0.40 × 0.55 = 0.31% Cu.
cross-section as in Figures 1 and 5.
The principle is then to predict (by kriging) the copper A visual inspection of the grade maps presented in
grade at any location from the data that belong to the same Figure 8 reveals strong differences between the two
domain as this location, by using the grade variogram approaches. The grade map based on the deterministic
corresponding to the domain under consideration (Figure 3). lithological model shows clear-cut discontinuities when
As mentioned in the introductory section, this approach crossing the assumed boundary between a high-graded
amounts to considering that the grades in a given lithological domain (oxide or secondary sulphide) and a low-graded
domain are independent of that in other domains, and that domain (leached or primary sulphide). As a consequence, the
the spatial extent of the domains exactly matches that of the grade distribution presents multimodality, as it can be seen in
interpreted lithological model. These assumptions lead to a the corresponding grade-tonnage curve (Figure 9).
grade model with hard boundaries when passing from one Although it relies on the available drill hole information,
domain to another one (Figure 8, left side). the deterministic lithological model inevitably depends on the
To account for the uncertainty in the lithological domain personal judgement of the mining geologist and is subject to
simplification. For instance, when examining the lithological
extent, an alternative approach is proposed, based on the
sequences along the drill holes (Figure 1), one notes that the
probability maps shown in Figure 6. It consists of weighting
oxide and leached domains are intermingled in a complex
the grades that are predicted in each lithological domain
way and that the actual boundaries between these domains
separately by the probability of occurrence of this domain.18
are likely to be much more irregular than they appear on the
The final grade model (Figure 8, right side) is therefore
interpreted model shown in Figure 7. As a result, errors in
defined as follows:
5 the lithological model affect the predicted grades and
Predicted Grade = ∑ Probability (kth domain) × Predicted tonnages, mainly because of misclassifications between high-
k=1 graded and low-graded domains, e.g. oxide zones wrongly
Grade (kth domain). assumed to be leached zones or vice versa.
For instance, assume that there is 15% chance of finding In contrast, the use of a probabilistic description of the
gravel, 45% chance of finding leached material, 40% chance lithological domains produces a grade model that accounts for
of finding oxide, and 0% chance of finding sulphide at a the uncertainty in the spatial extent of these domains. It

Interpreted lithological model Gravel

3000
Leached
Elevation (m)

2900
Oxide
2800
Primary
Sulphide
2700
3500 3500 3700 3800 3900 4000 4100 4200 4300 4400 4500
Secondary
Easting (m)
Sulphide

Figure 7—Deterministic modelling of the lithological domains. Conditioning data are superimposed

3000 3000
Elevation (m)

Elevation (m)

2900 2900
2800 2800
2700 2700
3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500 3500 3600 3700 3800 3900 4000 4100 4200 4300 4400 4500
Easting (m) Easting (m)

Scale for copper grade (%)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Figure 8—Copper grade models obtained by using a deterministic (left side) and a probabilistic (right side) modelling of the lithological domains
▲

808 DECEMBER 2007 VOLUME 107 REFEREED PAPER The Journal of The Southern African Institute of Mining and Metallurgy
SAIMM_Dec_47-53:Template Journal 12/6/07 8:00 AM
Probabilistic modelling of lithological domains and its application
1
Grade model based on
0.9 probabilistic lithological model
0.8 Grade model based on
deterministic lithological model
Fraction of total tonnage
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Cutoff grade (%)
Figure 9—Grade-tonnage curves associated with the copper grade
models of Figure 8. The calculations of tonnage account for the
following volumetric densities: 1.7 t/m3 for gravel material, 2.3 t/m3 for
leached material, 2.5 t/m3 for oxide material, 2.4 t/m3 for primary
sulphide material and 2.6 t/m3 for secondary sulphide material
results in a more homogeneous distribution of copper grades
(Figure 9), closer to that displayed in Figure 2 for the entire
deposit, and a smoother grade map. This smoothing effect is
inevitable if one wants to make the grade prediction more
accurate and to avoid misclassifications between lithological
domains. As a counterpart, because their exact location is
uncertain, the boundaries between the lithological domains
are no longer apparent in the grade model. These boundaries
can, however, be estimated from the set of plurigaussian
simulations and the related probability maps (Figures 5–6) or
by using the interpreted lithological model (Figure 7), leaving
aside the problem of grade prediction.
Conclusions
Geological modelling usually consists of a deterministic
interpretation and delineation of the main lithological
domains, using information from drill hole samples and
expert geological knowledge. This is undoubtedly a helpful
tool for understanding the genesis of the deposit, for defining
mineral resources and reserves, as well as for mine planning
and mineral processing.
However, when incorporating geological control in the
prediction of mineral resources, practitioners should be aware
that there always exists an uncertainty in the real extent of
the lithological domains, which may cause a loss of accuracy
in the grade models if it is not accounted for. To avoid this
difficulty, a probabilistic approach has been proposed, based
on the geostatistical simulation of the lithological domains
and on the calculation of their probabilities of occurrence
over the area of interest. These probabilities are then used for
weighting the grade prediction associated with each domain
so as to obtain the final grade model.
An application to a copper deposit has been presented, in
which the plurigaussian model has been used to simulate the
lithological domains. This model allows accounting for the
topological constraints (contacts) between the domains and
for their spatial continuity (indicator variograms).
The Journal of The Southern African Institute of Mining and Metallurgy
Page 809
T
Acknowledgements r
This research was funded by the National Fund for Science
a
and Technology of Chile (FONDECYT) and is part of project n
s
Nº1040690. The authors are grateful to Codelco-Chile for
providing the data-set used in this work. a
c
References t
1. DOWD, P.A. Geometrical and geological controls in geostatistical estimation i
and ore body modelling. Proceedings of the 19th APCOM Symposium.
Ramani, R.V. (ed.). Littleton, Colorado. Society of Mining Engineers,
o
1986. pp. 81–99. n
2. DUKE, J.H. AND HANNA, P.J. Geological interpretation for resource modelling
and estimation. Mineral Resource and Ore Reserve Estimation—The
1.8 2 AusIMM Guide to Good Practice. Edwards, A.C. (ed.). Melbourne. The
Australasian Institute of Mining and Metallurgy, 2001. pp. 147–156.
P
3. ALFARO, M.A. Control geológico en la estimación de reservas. Minerales,
1994. vol. 49, no. 49, pp. 25–32. a
4. ORTIZ, J.M. and EMERY, X. Geostatistical estimation of mineral resources p
with soft geological boundaries: a comparative study. Journal of the South
African Institute of Mining and Metallurgy, 2006. vol. 106, no. 8,
e
pp. 577–584. r
5. SOARES, A. Geostatistical estimation of orebody geometry: morphological
kriging. Mathematical Geology, 1990. vol. 22, no. 7, pp. 787–802.
6. ZHU, H. and JOURNEL, A.G. Indicator conditioned estimator. Transactions of
the Society for Mining, Metallurgy and Exploration, 1990. vol. 286,
pp. 1880–1886.
7. DOWD, P.A. Geological and structural control in kriging. Geostatistics Tróia
’92. Soares, A. (ed.). Dordrecht. Kluwer Academic, 1993. pp. 923–935.
8. CUADRA, P. and CAMUS, F. The Radomiro Tomic porphyry copper system,
northern Chile. Porphyry and Hydrothermal Copper and Gold Deposits: a
Global Perspective. Porter, T.M. (ed.). Adelaide. PGC Publishing, 1998.
pp. 99–109.
9. CUADRA, P. and ROJAS, G. Oxide mineralization at the Radomiro Tomic
copper deposit, Northern Chile. Economic Geology, 2001. vol. 96, no. 2,
pp. 387–400.
10. LE LOC’H, G. and GALLI, A. Truncated plurigaussian method: theoretical and
practical points of view. Geostatistics Wollongong ’96. Baafi, E.Y., and
Schofield, N.A. (eds.). Dordrecht. Kluwer Academic, 1997. pp. 211–222.
11. ARMSTRONG, M., GALLI, A., LE LOC’H, G., GEFFROY, F., and ESCHARD, R.
Plurigaussian Simulations in Geosciences. Berlin, Springer, 2003. pp. 160.
12. EMERY, X. Simulation of geological domains using the plurigaussian
model: new developments and computer programs. Computers &
Geosciences, 2007. vol. 33, no. 9, pp. 1189–1201.
13. BETZHOLD, J. and ROTH, C. Characterizing the lithological variability of a
Chilean copper deposit using plurigaussian simulations. Journal of the
South African Institute of Mining and Metallurgy, 2000. vol. 100, no. 2,
pp. 111–120.
14. SKVORTSOVA, T., ARMSTRONG, M., BEUCHER, H., FORKES, J., THWAITES, A., and
TURNER, R. Applying plurigaussian simulations to a granite-hosted
orebody. Geostats 2000 Cape Town. Kleingeld, W.J., and Krige, D.G.
(eds.). Johannesburg. Geostatistical Association of Southern Africa, 2001.
pp. 904–911.
15. SKVORTSOVA, T., BEUCHER, H., ARMSTRONG, M., FORKES, J., THWAITES, A., and
TURNER, R. Simulating the geometry of a granite-hosted uranium orebody.
Geostatistics Rio 2000. Armstrong, M., Bettini, C., Champigny, N., Galli,
A., and Remacre, A. (eds.). Dordrecht. Kluwer Academic, 2002.
pp. 85–99.
16. EMERY, X. and GONZÁLEZ, K.E. Incorporating the uncertainty in geological
boundaries into mineral resources evaluation. Journal of the Geological
Society of India, 2007. vol. 69, no. 1, pp. 29–38.
17. LANTUÉJOUL, C. Geostatistical Simulation: Models and Algorithms. Berlin,
Springer, 2002. pp. 256.
18. EMERY, X. and ORTIZ, J.M. Estimation of mineral resources using grade
domains: critical analysis and a suggested methodology. Journal of the
South African Institute of Mining and Metallurgy, 2005. vol. 105, no. 4,
pp. 247–255. ◆
▲
VOLUME 107 REFEREED PAPER DECEMBER 2007 809