Gold prospectivity mapping and exploration targeting in Hutti-Maski schist belt, India: Synergistic application of Weights-of-Evidence (WOE), Fuzzy Logic (FL) and hybrid (WOE-FL) models

Abstract

Our study attempts to map gold prospectivity for deriving optimal exploration targets using data-driven, knowledge-driven and hybrid approaches. Prospectivity models viz. Weights of Evidence (WOE), Fuzzy Logic (FL) and a hybrid model that combines the data-driven and knowledge-driven components (WOE-FL) were applied to a part of the auriferous Hutti-Maski schist belt of 1352 km2 area with 20 known gold occurrences. With a pixel resolution of 500 m, 16 spatial evidential raster layers were created on a GIS platform encompassing viable predictive indicators, critical in gold exploration. Modelling inputs include essential ingredients and mappable criteria of the conceived orogenic gold mineral system in the study area. Multi-source geological data such as lithostratigraphic units, favourable litho-contacts, structural deformation sites, geochemical anomalies of selected gold pathfinder elements, and hydrothermal alteration zones derived from digital image processing of Landsat 8 OLI satellite imagery were integrated to generate prospectivity maps for delineation of future targets. A quantitative evaluation of the resulting three prospectivity maps was performed using concentration-area (C-A) fractal analysis, prediction–area (P–A) plot, fitting-rate curve (FRC) and area under curve (AUC). Comparative analysis indicates that the performance of the hybrid model (WOE-FL) stands out to be the most efficient, with a prediction rate of 87% and AUC of 91.40% compared to WOE and FL. A risk assessment was performed combining the outputs of prospectivity models that returned 10% of the study area as potential exploration targets out of which the low-risk exploration targets comprises merely 4.5% representing the optimal targets for gold exploration in the study area.