Geoscience Reference
In-Depth Information
Fig. 14.67
Correlation between
the grain size variables. Minerals
are simulated independently
because of the small correlation
between minerals
ρρ ρ ρ
…
…
λ
ρρ ρ ρ
simulation to consider the non-stationary present throughout
the deposit. The mean for each PCA was determined using a
moving window average with a radius of 400 m in the hori-
zontal direction and 50 % anisotropy in the vertical direction.
1,1
2 ,1
n
,1
1
0 ,1
λ
=
1,2
2,2
n
,2
2
0,2
ρρ ρ ρ
…
λ
1,
n
2,
n
nn
,
n
0,
n
14.7.7
Modeling Nine Grain Size Variables
The right-hand side of this equation contains the correlation
between one of the grain size variables and the 23 input head
grade variables to be merged. The left hand side is the cor-
relation between all 23 PCA head grade variables; note that
the left hand side contains 1.0 on the diagonal and 0.0 for
all off diagonal terms because the PCA values are uncor-
related. This is done for the p
50
value for each mineral and
the same super secondary variable is used for modeling the
p
20
, p
50
and p
80
. This single super secondary variable allows
for the cosimulation of the three percentiles and only one ex-
haustive secondary variable. Without merging all secondary
variables into a super secondary, the grain size simulations
would have to consider 23 separate secondary variables
in order to use all the available information from the head
grade variables.
There are three Uranium minerals of interest: Brannerite,
Coffinite and Uraninite. The p
20
, p
50
and p
80
grain size for
each mineral has been measured at 497 locations. The cor-
relation between the percentiles of each grain size is repro-
duced by co-simulating the three percentiles, see Fig.
14.67
.
The densely sampled 23 head grade values is used to sup-
plement the lack of information for the grain size variables
by considering a super secondary variable which is the amal-
gamation of the 23 PCA head grade variables. This super
secondary variable is created differently for each mineral
because the correlations between the mineral grain sizes and
the PCA head grade variables differ. To generate this super
secondary variable, a linear combination of the PCA head
grades is determined from the following equations:
