3D Modeling, Stochastic Joint Gravity-Magnetic Inversion, and ML for Anomalous Zone Identification in a Mining Context.

This study proposes an integrated subsurface data methodology to identify zones with mineralization potential in mining contexts, focusing on detecting geological targets based on their geophysical properties. The case study encompasses an area of 400 km² around the Riotinto mine in the Iberian Pyrite Belt (southern Spain), a region internationally recognized for its significant accumulations of massive sulfides deposits. Our methodology integrates stochastic geological models derived from detailed mapping with a joint probabilistic inversion of gravity and magnetic data. Bouguer and magnetic anomaly digital maps are used to generate probabilistic density volumes of the target area. Additionally, petrophysical data from over a thousand rock samples were analyzed and used to construct predictive models of P-wave velocity and, total porosity using advanced Machine Learning (ML) techniques.

The generated 3D  models reveal the geometry of the main rock units. Geo-bodies can be differentiated within the multiparametric volume. These are characterized by high values for density and P-wave velocity, and low values for porosity. These rock units are key parameters for identifying mineralized structures. However, the available data on physical properties reveals an overlap between different lithologies and mineralized ore bodies which hinders the accurate discrimination of the latter. The models illustrate the presence of anomalous rock bodies, including mafic rocks located at shallow structural positions, and highly compacted slates at depths greater than 1250 m. These feature significant contrasts in their physical property values that could lead to false exploration targets. Considering this, we were able to establish a classification and prioritization system for zones based on their probability of containing mineralized bodies, identifying areas with greater potential of hosting ore structures in specific geological units. Finally, it is proposed to continue evaluating the applicability and effectiveness of this methodology in other geological and ore bearing settings, promoting its replicability and, aiding the development of more precise, efficient, and sustainable exploration techniques, aligned with the growing demand for strategic minerals necessary for a responsible energy transition.

*This work, funded under reference CPP2021 009072, has been supported by MCIN/AEI/10.13039/501100011033 (Ministry of Science, Innovation, and Universities/State Agency for Innovation) with funds from the European Union's Next Generation/PRTR (Recovery, Transformation, and Resilience Plan).