Machine learning and tectonic reconstructions: Unlocking porphyry copper exploration in the western Tethyan region

Porphyry systems host most of the mineable copper reserves globally, a metal experiencing unprecedented demand due to global electrification and decarbonisation trends. While porphyry systems are known to form in magmatic arcs along subduction zones, the precise roles of factors within the subducting and overriding plates remain poorly constrained, complicating prospectivity mapping. In this study, we develop a machine learning-based mineral prospectivity model for porphyry mineralisation, trained on known occurrences and spatio-temporal features derived from a modified plate motion model for the western Tethyan region, incorporating reconstructions of ocean basins spanning 90 Ma to the present. This segment of the Tethyan convergence zones represents a complex tectonic environment shaped by the diachronous collision of the Arabian and Eurasian continents, and our plate motion model reconstructs the spatio-temporal evolution of subduction and collision processes in this region. The initial soft collision, where the thinned Arabian passive margin collided with southern Eurasia, began ~42 Ma along the eastern Bitlis suture zone, transmitting strain into eastern Anatolia, the Caucasus, and northwestern Iran. Collision propagated westward into central Anatolia and southeastward into the northwestern Zagros by the late Eocene (40–35 Ma), followed by central Zagros (35–25 Ma) and southeastern Zagros (25–15 Ma). We defined a segmented passive margin line representing collisional boundaries and timings to capture this diachronous process, integrating collision propagation, strain transmission, and crustal deformation across the western Tethyan region in our reconstruction model.
Our time-dependent mineral prospectivity model illustrates the temporal evolution of porphyry mineralisation across the western Tethyan Belt, highlighting several high-prospectivity zones that lack known deposits and thus represent promising exploration targets. Feature importance analysis reveals the complex mechanisms driving porphyry mineralisation, identifying key predictors: arc segment length, distance to the nearest trench edge, and the orthogonal component of the relative motion vector. The length and curvature of arcs emerge as critical features, with tightly curved arcs linked to enhanced compressional stress and fracturing, promoting magma ascent and porphyry formation. The median distance to the nearest trench edge for known deposits is about six degrees, which exceeds the typical arc distance from the plate boundary, suggesting a distinctive feature of porphyry processes in this region. The orthogonal convergence rate is also pivotal, with higher magnitudes correlated to mineralisation. This indicates rapid convergence enhances metasomatism and partial melting processes in the overriding plate, facilitating porphyry formation. Our results demonstrate the effectiveness of combining plate motion models and machine learning to advance mineral exploration along subduction zones in the western Tethyan Arabia-Eurasia convergence zone. This approach is adaptable to data-poor regions and other time periods globally, offering significant potential for identifying new porphyry targets.