AI-Powered predictiveGeological modeling of ICT pegmatite occurrences in Rwanda

Abstract

The abundant supply of lithium in the world can be used to mitigate global warming as it is one of the principal and essential components of Li-Fe batteries that can be used to power electrical cars reducing the amount of atmospheric emissions. The LCT pegmatites of Rwanda are of strategic importance as they are a source of Li, Cs, Ta, Nb, Sn, W mineralization. Efficiently locating pegmatite deposits represents a significant economic importance for Rwanda. This research uses the interaction of geophysics and artificial intelligence to discover the best approach to determine the most promising sites for pegmatite discoveries within the study area located to the north of the Gitarama granite intrusion. This study employs an airborne geophysical database including magnetic, gamma-spectrometry, resistivity data and geological database provided by the Rwanda Mines, Petroleum and Gas Board. These data were analyzed through a series of Neural Network algorithms using Oasis Montaj software. The results show that Fast Classification Neural Network with Eliot’s Input transfer function provides the optimal results in modeling LCT pegmatites prospectivity. We discovered four potential linear trends/zones as the most promising sites for further pegmatite exploration within the study area.