Mineral Reconnaissance Through Scientific Consensus: First National Prospectivity Maps for PGE–Ni–Cu–Cr and Witwatersrand-type Au Deposits in South Africa

This paper presents a pioneering approach to mineral prospectivity mapping (MPM) in South Africa by introducing a novel, consensus-based method for identifying potential deposits of Platinum Group Elements (PGE), Nickel (Ni), Copper (Cu), Chromium (Cr), and Witwatersrand-type gold (Au). The research outlines the development and validation of this method, which leverages scientific consensus through deep ensemble modeling. By simulating the decision-making process of multiple data scientists, the method captures uncertainties in workflow choices, producing more reliable and confident MPM outputs.

The validation of this method was conducted using well-known, resource-rich geological systems—the Bushveld Complex for PGE-Ni-Cu-Cr and the Witwatersrand Basin for Au. These regions serve as benchmark sites due to their extensive geological knowledge and the presence of mega-deposits, providing a robust test of the method’s effectiveness. The results demonstrate high agreement with existing geological and exploration data, identifying high-potential exploration targets in both known and new areas. For instance, new targets for PGE-Ni-Cu-Cr were identified northwest of the Bushveld Complex, and promising gold exploration areas were pinpointed west of the Witwatersrand Basin.

The broader impact of this research lies in its ability to increase trust in data-driven exploration tools within the mineral industry by offering a method that not only highlights prospective areas but also quantifies the confidence level in these predictions. This approach is particularly valuable as the industry increasingly relies on data-driven techniques for exploration, ensuring that decision-makers can base their strategies on scientifically validated and consensus-driven information.