Evidential belief mapping of epithermal gold potential in the Deseado massif, Santa Cruz Province, Argentina

Ideally, at any exploration scale, evidential maps derived from multivariate and multi-source spatial datasets are integrated to create map of potentially mineralized zones that could be used guide further exploration of undiscovered deposits. Evidential maps represent recognition criteria, which are based on conceptual models and characteristics of known mineral deposits of interest. With every evidential map, however, there are associated uncertainties, which should be modelled properly so that users are aware of limitations of output mineral potential map.
Uncertainties in information depicted in maps could be either stochastic or systemic. Stochastic uncertainties could arise from incompleteness of spatial data pertaining to the target variable (i.e., locations of known mineral deposits of interest) or from efficiency/sufficiency of spatial data and/or map descriptions to provide precise information about recognition criteria. Systemic uncertainties arise from modeling procedures to assign weights to evidential information in terms of likelihood for mineral deposit occurrence.
On one hand, stochastic uncertainty can be dealt with by application, for example, of weights-of-evidence method to quantify variance of posterior probability of mineral occurrence given presence or absence of spatial evidence (Agterberg et al., 1989). On the other hand, systemic uncertainty could be dealt with by application of the theory of fuzzy sets (Zadeh, 1965) to assign fuzzy membership grades to classes of spatial data/information with respect to mineral potential proposition (e.g., Carranza and Hale, 2001).
Instead of handling stochastic and systemic uncertainties in mutual exclusion to generate mineral potential information, overall uncertainty in evidential maps can be managed by application of evidential belief functions or EBFs (An et al., 1994b). Previous applications of EBFs to mineral potential mapping are mostly knowledge-driven or based on expert opinion (e.g., Moon, 1990; An et al., 1994a, 1994b; Chung and Fabbri, 1993; Wright and Bonham-Carter, 1996). Knowledge-driven estimation of EBFs is suitable in cases where spatial data pertaining to the target variable (e.g., locations of known mineral deposits) are lacking or insufficient. However, for cases where a number of mineral deposits are known but expert opinion is wanting, a data-driven estimation of EBFs (i.e., based on maps of known deposits and evidential themes) could be followed (Carranza and Hale, 2003).
In this paper, we demonstrate an application of data-driven EBFs to generate and integrate evidential maps to predict potentially gold-mineralized zones in the Deseado Massif in southern Argentina.