Machine Learning-Based Spatio-Temporal Prospectivity Modeling of Porphyry Systems in the New Guinea and Solomon Islands Region

Abstract

The discovery of new economic copper deposits is critical for the development of renewable energy infrastructure and zero-emissions transport. The majority of existing copper mines are located within current or extinct continental arc systems, but our understanding of the tectonic and geodynamic conditions favoring the formation of porphyry systems is still incomplete. Traditionally, exploration criteria are based on present-day geological and geophysical observations rather than the time-dependent evolution of subduction systems. Addressing this knowledge gap, our study connects the formation of porphyry systems, particularly enriched in copper, with subduction zone evolution, utilizing machine learning in a spatio-temporal mineral prospectivity framework. Incorporating Cenozoic intrusion-related copper-gold deposits in the New Guinea and Solomon Islands region, we develop a model that accurately predicts known mineral occurrences and identifies key features for potential porphyry mineralization in the study area. Key findings include the importance of the obliquity angle of subduction, which significantly affects strain partitioning, crustal fluid flow, and ore deposition, with angles between 10 and 50° favored for mineralization. Furthermore, rapid plate convergence and seafloor spreading half-rates ranging from 30 to 45 mm/yr potentially enhance mineralization prospects by promoting metasomatism and hydrous melting. This approach, integrating plate motion models with machine learning, provides new exploration criteria, enhancing our understanding of porphyry ore formation mechanisms and guiding future exploration in both active and abandoned subduction zones.