Transforming LCT Pegmatite Targeting Models into AI-Powered Predictive Maps of Lithium Potential for Western Australia and Ontario: Approach, Results and Implications
2025
Oliver Kreuzer | Bijan Roshanravan
Here, we present holistic targeting models for lithium–cesium–tantalum (LCT) pegmatites in Western Australia, the world’s largest supplier of hardrock lithium ores, and Ontario, an emerging hardrock lithium mining jurisdiction. In this study, the LCT pegmatite targeting models, informed by a review of this deposit type and framed in the context of a mineral system approach, served to identify a set of targeting criteria that are mappable in the publicly available exploration data for Western Australia and Ontario. This approach, which formed the basis for artificial intelligence (AI)-powered mineral potential modeling (MPM), using multiple, complimentary modeling techniques, not only delivered the first published regional-scale views of lithium potential across the Archean to Proterozoic terrains of Western Australia and Ontario, but it also delivered an effective framework for exploration and revealed hidden trends. For example, we identified a statistically verifiable proximity relationship between lithium, gold, and nickel occurrences and confirmed a significant size differential between LCT pegmatites in Western Australia and Ontario, with the former typically containing much larger resources than the latter. Overall, this regional-scale targeting study served to demonstrate the power of precompetitive, high-quality geoscience data, not only for regional-scale targeting but also for the development of camp-scale targets that have the resolution to be investigated using conventional prospecting techniques. Importantly, MPM does not generate ‘treasure maps’. Rather, MPM provides another tool in the ‘exploration toolbox’, and its output should be taken as the starting point for further investigations.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por Multidisciplinary Digital Publishing Institute