Geophysical exploration for concealed orogenic gold deposits is challenging because petrophysical contrasts between the gold lodes and their host rocks are normally subtle. Most greenstone-hosted lode‑gold deposits do not contain strong conductors as they normally contain <5 vol% of poorly conductive disseminated arsenopyrite and pyrite. Where there is no outcrop, reliance must be placed on lithological and structural interpretations from aeromagnetic data.Airborne Induced Polarization (AIP) Cole-Cole parameters from helicopter time-domain electromagnetic (HTEM) surveys are demonstrated to be effective in geophysical exploration for concealed gold deposits in the Kabinakagami Lake greenstone belt in Ontario, Canada. Interpretation reveals that most AIP positive responses are associated with clays and all known gold deposits in the study area are in resistive rock formations. Hence, resistivity-scaled chargeability (RSC) can be used to detect weakly chargeable zones within these resistive units.Deep Neural Network (DNN) predictive targeting analysis is utilized to integrate the AIP apparent resistivity, resistivity-scaled chargeability (RSC), EM induction time-constant, and magnetic data. The DNN results imply that the known gold occurrences coincide with high DNN probabilities, and hence exploration targets can be identified based on DNN analysis. From a geological perspective, gold prospects also occur at triple-point junctions between granite intrusions and at probable intersections between potentially hosting shear zones and oblique cross faults or dykes, both common structural geometries for orogenic gold deposits. Consequently, DNN predictive targeting results based both on HTEM and magnetic data, and consistent with expected structural geometries, are proposed as a vectoring tool in the generation of drill targets during exploration for concealed orogenic gold deposits in greenstone belts such as those of the Superior Province, Canada.
Read full abstract