Geoscience Reference
In-Depth Information
CuT, RC vs Pozos, Sulf. <=4m
• The models were discussed with the local geologists, who
reconciled them with their own experience and knowl-
edge of the deposit. The main directions of continuity
confirm current geologic knowledge. In addition, for
most Domains there is evidence of a short-scale anisot-
ropy that has a different direction than the longer-scale
continuity. This is interpreted as a mixture of geologic
controls (structures) at different scales, as confirmed by
local geologists.
Figure
14.20
shows the experimental correlograms and
models for Domain 1, for three directions that are close to
the main directions of anisotropy. As shown, two exponen-
tial structures have been modeled with an anisotropy ratio
of 5:1 in the N-S direction compared to the E-W direction,
and even slightly more anisotropic with respect to the verti-
cal direction. For this particular Domain, the first structure
showed slightly more continuity on a direction dipping ap-
proximately 20°, while the second structure did not show a
dip. Shallow-angle, cross cutting structures explained the
slightly different anisotropy direction of the first structure.
5.00
4.00
3.00
2.00
1.00
0.00
0.00
1.00
2.00
3.00
4.00
5.00
CuT, RC
Fig. 14.18
Q-Q plot, TCu, RC 10 m composites vs. blast holes, sulfide
(barren) cap with the oxidized mineralized zones. Within 1
or 2 m in the vertical direction, grades can change from bar-
ren (<0.1 % TCu) to 1 %. Soft contacts may exist between
high and low secondary enrichment units, where the grade
transition is much smoother, to the point that, based on grade
alone, it is difficult to define the contact itself.
Figure
14.19
shows an example of a soft contact. The
grade profile corresponds to two sulfide estimation domains,
and shows the average grade of all composites in each side of
the contact by class distance, in this case 20 m. Figure
14.19
shows that, despite the fact that the overall averages are dif-
ferent, locally and near the contact the grades are similar and
also somewhat depleted with respect to the global average.
Therefore, for these estimation units, it is reasonable to use
composites across the contact to estimate the grades of either
unit within a limited distance.
14.1.19
Change of Support to Estimate
Internal Dilution
The Cerro Colorado open pit mine works on a 10 m-bench
height and uses large equipment, such that the accepted se-
lective mining unit (SMU) is 20 × 20 × 10 m. This SMU size
defines, in theory, the target distribution of estimated grades
that should be achieved to accurately estimate recoverable
resources and reserves. Although the SMU is a convenient
concept, it is important to not lose sight of its limitations,
recall the discussion in Chap. 7.
The coefficient of variation (CV =
σ/m
) is a useful measure
of variability that can be used to characterize the variability
of the SMU distribution. This theoretical SMU distribution
can be used as a reference distribution and compared to the
estimated block model grades. A calibration of the estimated
grade distribution can be performed to better match the theo-
retical SMU distribution.
The dispersion variance is found for the assumed SMU
through either (i) the correlogram or variogram models de-
veloped for each Domain; or, (ii) as an “experimental” dis-
persion variance. For the second option, blast holes or other
production data should exist at a grid sufficiently smaller
than the SMU defined. If so, the available grades are simply
averaged into the SMU sizes.
The Discrete Gaussian (DG) method was used at Cerro
Colorado to derive a SMU distribution of TCu grades for
each Domain. The method was mentioned in Chap. 7, and
is described in detail in Journel and Huijbregts (
1978
). The
three elements that are needed for the DG model to pre-
dict a grade-tonnage curve for any SMU are (i) the cor-
relogram models for TCu and for each Domain; (ii) the
14.1.18
Correlogram Models
The correlogram function (Srivastava and Parker
1988
) was
chosen to characterize spatial continuity at Cerro Colorado
because experience has shown it to be more robust than the
traditional variogram with respect to outliers and trends,
being therefore generally easier to model. Correlograms
were calculated for TCu and SCu in each Domain using
10 m composites and blast hole concurrently. Down-the-
hole correlograms were used to define the nugget effect and
short scale continuity of the correlogram models. Experi-
mental correlograms were obtained in 37 directions using
SAGE2001 (Isaaks
1999
). Some observations from the cor-
relogram models were:
• Three rotation angles were used to define the directions
of anisotropy. In this particular case, there is sufficient
information (drill holes and blast holes) to warrant such
detailed correlogram models.
