Page 230 - Geochemical Anomaly and Mineral Prospectivity Mapping in GIS

232 Chapter 7

opposed to complementary) spatial evidence (say, X 1 and X 2) in order to support the
proposition of mineral prospectivity. In mineral exploration, proximity to faults/fractures
and stream sediment geochemical anomalies can represent two sets of supplementary
spatial evidence of the presence of mineral deposits, because not all locations proximal
to faults/fractures contain mineral deposits and because not all stream sediment
geochemical anomalies necessarily mean the presence of mineral deposits. (After
application of equations (7.17)–(7.19), Pls X 1 X 2 is derived according to the
relationships of the EBFs explained above.)
According to Dempster’s rule of combination, only EBFs of two spatial evidence
maps can be combined each time. The EBFs of maps X 3,…,X n are combined with already
integrated EBFs one after another by re-applying either equations (7.14)–(7.16) or
equations (7.17)–(7.19) as deemed appropriate. The final integrated values of Bel are
considered indices of mineral prospectivity. Furthermore, because equation (7.14) is
multiplicative, whilst equation (7.17) is associative and commutative, the output
integrated values of Bel derived via the former are always less than the corresponding
output integrated values of Bel derived via the latter. This means that integrated values
of EBFs should not be interpreted in absolute terms but in relative terms (i.e., ordinal
scale) only and, therefore, in mineral prospectivity modeling integrated values of Bel
represent relative degrees of likelihood for mineral deposit occurrence.
As in Boolean logic modeling and in fuzzy logic modeling of mineral prospectivity,
an inference network is useful in combining logically EBFs of spatial evidence of
mineral prospectivity. The inference network used in the Boolean logic modeling (Fig.
7-4), which is more-or-less similar to the inference network used in the fuzzy logic
modeling (Fig. 7-15), is applied to logically integrate the EBFs of the spatial evidence
maps of epithermal Au prospectivity in the case study area. The geological reasoning
behind the integration of the EBFs of the spatial evidence maps of epithermal Au
prospectivity in the case study area is, thus, the same as in the earlier application of
Boolean logic modeling and similar to the earlier application of the fuzzy logic
modeling. The map of integrated Bel (Fig. 7-19A) shows a pattern of prospective areas
that is more similar to the pattern of prospective areas delineated via multi-class index
overlay modeling (Fig. 7-9A) and via fuzzy logic modeling (Figs. 7-16A and 7-17A)
than the pattern of prospective areas delineated via Boolean logic modeling (Fig. 7-5A)
and via binary index overlay modeling (Fig. 7-7A). However, unlike the earlier mineral
prospectivity maps, a prediction-rate curve (Fig. 7-19B) can be constructed for the
mineral prospectivity map in Fig. 7-19A with respect to the whole case study area
because, for the locations without stream sediment geochemical evidence, there are input
EBFs (i.e., Bel=0, Unc=1 and Dis=0) and thus output EBFs. Nevertheless, for proper
comparison of predictive performance with the earlier mineral prospectivity maps, a
prediction-rate curve with respect only to locations with stream sediment geochemical
evidence is also constructed (Fig. 7-19C). Using this curve, if 20% of the case study area
is considered prospective, the map of integrated Bel delineates correctly seven (or about
58%) of the cross-validation deposits (Fig. 7-19C). This predictive performance of the
evidential belief model (Fig. 7-19A) is the same as that of the fuzzy logic model (Fig. 7-