Identifying tectonic niche environments of South American porphyry magmatism through geological time: a spatio-temporal data mining approach

Abstract

Porphyry ore deposits are well known to be associated with arc magmatism related to subduction on the overriding plate. Furthermore, the regional mechanisms for magmatism and the resulting formations of porphyry deposits are well established. Specific parameters leading to these events have been inferred, but not formally tested. We aim to identify the specific set of tectono-magmatic parameters that result in a subducting slab producing particular types of magmatism on the overriding plate, and their link to the formation of ore deposits. We use a four-dimensional approach to reconstruct age-dated magmatism back through space and time to isolate the tectono-magmagic parameters leading to the formation of a metalliferous deposit during subduction. By utilising machine learning techniques we identify and quantify geodynamic parameters that are robust predictors of back-arc magmatism and porphyry formation. The 'random-forest' ensemble and 'support vector machines' learning classification methods are employed to prioritise parameters that are considered influential in the development of magmatism and the subsequent metallogenesis of porphyry ore deposits. We find that a combination of convergence rates and directions, seafloor age, subduction obliquity, and the distance to a trench edge help predict whether magmatism and related ore deposits occur.