A PREDICTIVE MODEL FOR BASEMETAL EXPLORATION IN A GIS ENVIRONMENT Alok Porwal* and Edmund Sides International Institute for Aerospace Survey and Earth Sciences (ITC), Delft, The Netherlands *Department of Mines and Geology, Govt. of Rajasthan, Udaipur, India porwal@itc.nl sides@itc.nl KEY WORDS: Fuzzy, Geology, Geophysics, GIS, Integration, Mathematical models. ABSTRACT A predictive model for mineral potential mapping based on fuzzy set theory is described. It is tested in the south-central part of the Aravalli province (western India), which hosts a number of conformable sediment-hosted basemetal deposits. Recognition criteria for basemetal mineralisation were identified on the basis of published work on metallogenesis in Aravalli province. A regional GIS was then established in ArcView GIS software using several public-domain geodata sets. These were reviewed, processed, reclassified and gridded to generate multi-class lithological, stratigraphic, structural, magnetic and lineament-density maps. Weights were assigned to each evidential map, and also to each class of the maps, on the basis of their significance as guides to the occurrence of basemetal mineralisation. These were used to calculate fuzzy membership values for all classes. The values thus determined were combined using fuzzy algebraic sum and fuzzy algebraic product operators to generate basemetal favourability maps for the province. It was observed that the fuzzy algebraic sum operator gives excessive areas of high favourability, while the fuzzy algebraic product operator tends to diminish favourability. The values obtained from these operations were therefore combined using fuzzy gamma operators to generate final favourability maps. Known basemetal occurrences were overlaid on the favourability maps to validate the procedure. It was found that most of the known mineral occurrences correlate with areas of high-predicted favourability, although there are several areas of high favourability that do not have any mineral occurrences. Work is continuing to check whether such areas genuinely represent areas warranting further exploration, or whether the modelling techniques used need further refinement. 1 INTRODUCTION Most statistical and probabilistic approaches to mineral potential modelling are based on the use of binary evidential maps, while real-world geodata is usually multi-class in nature. This necessitates reclassification of multi-class data into binary data, which may result in loss of valuable information. Moreover, the reclassification principles are normally based on available information, and may change, as more information becomes available. Models based on fuzzy set theory accept multi-class data and are sufficiently robust to assimilate the “informational fuzziness” (Zimmermann, 1985) that is inherent in most geodata. The input parameters can be selected either by using empirical methods based on statistical (or heuristic) evaluation of the spatial association of various geodata with mineral deposits or by using a genetic model. In this study the second approach was used, the fuzzy membership values of the model parameters being assigned subjectively by experts. For building and testing the model, the south central part of the Aravalli metallogenic province in Rajasthan, western India, was selected (see Fig. 1). The area has been relatively well explored and this work is documented in the literature. An area of about 37500 sq. km, falling between latitudes 23°30′ N and 26° N and longitudes 73°30′ E and 75° E is used. This area includes a number of major Zn-Pb-Cu and Pb-Zn deposits and many small and minor occurrences of basemetals as well as abandoned mining pits. 1.1 Geology and mineralisation of the test area Heron (1953) interpreted the geology of the Aravalli province in terms of three major Proterozoic orogenic cycles, represented by the Banded Gneissic Complex (BGC), the Aravalli Supergroup and the Delhi Supergroup. His scheme has remained the basic framework of reference for all subsequent revisions (e.g., Raja Rao, 1976; Roy, 1988; Sudgen et al., 1990; Gupta et al., 1995). The region is characterised by evidence of repeated tectonic deformation, metamorphism and magmatism (Roy et al., 1971; Naha and Halyburton, 1974; Roy, 1988; Sharma, 1988; Srivastava, 1988). Porwal, Alok International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1178