Geoscience Reference
In-Depth Information
Table 14.13
Total resource by category, MIK Model, São Francisco
$X
*UDGH
JW
$X
*UDGH
JW
$X
*UDGH
JW
$X
*UDGH
JW
2XQFHV
$X
[
&XWRII
JW
7RQV
[
7RQV
[
7RQV
[
7RQV
[
and SAP). The Haz-Lo triangulation constrains extrapola-
tion towards the edges of the deposit. The block model was
built using sub-blocks to better adjust the model to the in-
terpreted wireframes. The edge blocks can be as small as
5 × 5 × 2.5 m. After grade estimation, re-blocking incorpo-
rates geologic contact dilution in the model, which is sig-
nificant when a Haz-Hi block is in contact with a Haz-Lo
block.
The Multiple Indicator Kriging method, described in
Chap. 9, estimates a distribution of possible values (a cumu-
lative conditional distribution function) based on the indica-
tor thresholds defined above. The point conditional distribu-
tion is obtained by kriging each indicator independently, and
is later post-processed to provide a block estimate. The block
estimate is the sum-product of the estimated probability for
each class multiplied by the declustered average grade for
each class. For the São Francisco block model, the average
grade was obtained for each discretization point within each
10 × 10 × 5 m block, and then averaged up to provide the es-
timated block grade. This is known as the “e-type” estimate,
which is obtained as follows:
• The minimum and maximum number of samples required
for estimation varied according to the pass, as this was
one parameter used to control smoothing.
• An octant search was used in all cases, since it aids in
declustering the estimated values.
• The anisotropic search ellipsoids used varied according
to the estimation domain. The search ellipsoids approxi-
mately follow the general orientation of each Domain.
The rotation convention used is a left hand (LH) rotation
around the
Z
-axis first, then a right hand (RH) rotation
around
X
′, and finally a RH rotation around
Y
′.
The resources of the São Francisco deposit were classified
on a block-by-block basis, using as basis the defined kriging
passes (Table
14.12
), which indicate the configuration and
quantity of information used to estimate each block. A flag
is stored in the block model indicating whether the block
was estimated on the first, second, or third pass, and this flag
thus corresponds to the measured, indicated, and inferred
categories. No further correction was required, since the ge-
ometry of the deposit and the geologic domains resulted in a
smoothly varying resource class.
Blocks estimated within a 20 m-search radius in the strike
and down-dip directions and 10 m in the across-strike direc-
tions were classified as measured. Blocks estimated with a
40 m search radius in the strike and down-dip directions and
20 m in the across-strike directions were classified as indi-
cated. Finally, all other estimated blocks beyond 40 m are
classified as inferred. Figure
14.35
shows a cross-sectional
view of the estimated grades in the model.
...
*
z
()
u
=
p
()*
u
cp
+
()*
u
c
++
p
()*
u
c
1
1
2
2
n
n
Here
z
*
(
u
) represents the block estimate,
p
i
(
u
) represents the
probability for each class defined, and
c
i
represents the mean
grade assigned to that class. Note that the MIK model con-
structed in this way does not have any explicit allowance for
the volume-variance effect. This is approximated using the
constrained kriging methodology described above.
14.2.8
MIK Resource Model: Grade-Tonnage
Curves
14.2.7
MIK Kriging Plans and Resource
Categorization
Table
14.13
shows the estimated resources by category and
for several cutoffs as of December 2001. At a 0.4 g/t Au
cutoff, the measured plus indicated diluted resource stands
at 32.5 million tons at a 1.23 g/t average grade, for about
1.288 million ounces of contained gold. There are an addition-
al 57.4 million tons in between the 0.13 g/t and the 0.40 g/t
Au cutoffs which is considered Run of Mine (ROM) material.
The kriging plan implemented for the MIK model has the
following characteristics:
• Three passes were implemented to estimate each of the
three domains. Table
14.12
shows the details for each
domain. The search radii was defined according to
resource classification criteria, see discussion below.
