Metallogenic Fingerprints of the Congo Shield with Predictions for Mineral Endowment Beneath the Congo Basin - Geology and Resource Potential of the Congo Basin - page 382

Table 19.1 Number of mineral deposits of selected element groups across (a) Central Africa (CA), (b) Central Africa Shield (CAS) and Congo
Shield (CS). Note that the number of deposits reflected in (a) and (b) exclude those of the Congo Shield
(a) CA Area in km 2 Au CrNiPgeTi CuZPbBa SnSb UThRee Total

Archean 168 34 27 28 7 264 5.04E+05

MPP 96 31 56 110 33 326 2.13E+06

Np 175 42 133 0 44 394 1.40E+06

NprA 51 8 18 9 8 94 7.25E+05

NprP 195 16 44 20 26 301 2.18E+06

Total 685 131 278 167 118 1379 6.94E+06

(b) CAS
Archean 66 1 5 0 5 77 3.96E+05

MPP 94 31 56 107 23 311 1.92E+06

Np 10 1 14 0 0 25 1.54E+05

NprA 51 8 18 9 8 94 7.18E+05

NprP 38 1 7 4 2 52 3.66E+05

Total 259 42 100 120 38 559 3.55E+06

(c) CS
Archean

101 9 43 3 14 170 2.16E+06

MPP 8 2 0 2 5 17 1.26E+05

Np 15 1 9 0 2 27 3.10E+05

NprA 9 1 1 0 1 12 1.03E+05

NprP 0 0 0 0 0 0 4.56E+03

Total 133 13 53 5 22 226 2.70E+06

Bonham-Carter 2001 ), r ij . The spatial coefficient ( r ij )isbased
on the

positive associations are greater than 0, and negative associ-
ations are less than 0; values close to zero indicate an
absence of spatial associations (i.e. reflecting a random
distribution). In a

' weight of evidence ' approach (Bonham-Carter 1996 ),

where

ND j

NT i \ D j =

' '

approach for evalu-
ating mineral prospectivity, it can be shown that ln r ij is an
approximation for the positive weight W i + , and the evalu-
ation is data-driven (Mihalasky and Bonham-Carter 2001 ,
their Appendix). Alternatively, in an expert system for
evaluating mineral prospects (e.g. PROSPECTOR,
Campbell et al. 1982 )W i + is referred to as the sufficiency
ratio (LS), and it is assigned by an expert (e.g. the evaluation
is knowledge driven by, say, a geologist).
The approximate standard error for ln r ij (Mihalasky and
Bonham-Carter ( 2001 , Appendix) and Bonham-Carter ( 1996 ,
Chap. 9 , this Topic) are given by

weight of evidence

r ij ¼

ATðÞ ; ATð =

in which A(T i ) is the area of the ith domain (T i ), and A(T • )is
the total area of all the domains in the study (A(T • )

A
(T i )). N(D j ) is the total number of deposits in the jth element
group (D j ). N(T i \

¼ ∑

D j ) represents the number of deposits of
group D j in domain T i .
The spatial coefficient, r ij represents the proportion of
deposits (for example gold—j) of all the jth deposits that
occur in the specified geologic domain (i) per unit area of all
similar geologic groupings. The spatial coefficient range
from 0 to infinity; it is equal to 1 if there is no spatial
association between a geological defined domain and an
element group (e.g. if the proportion of jth mineral is the
same as the proportion of area occupied by the ith). For
values of r ij > 1 (expected number of deposits is greater
than by chance), there is a positive association between
mineral j and geologic domain i; r ij <

s

1
NT i \

s ln r ij ¼

D j

19.4 Results

1 (fewer deposits
expected than by chance) indicates a negative association.
Because all negative associations are compressed in the
range from 0 to 1, and all positive associations fall in the
range of 1 to infinity, we use the natural log of r ij to eliminate
this skewness. ln r ij is now a symmetric value around 0:

The natural log of the spatial coefficients (fingerprint)
between the element group and each geological group (ln r ij )
and its approximate standard errors are given in Table 19.2
and these fingerprints are summarized graphically in
Figs. 19.4 (CA), 19.5 (CAS) and 19.6 (CS).