Korzhinskii-Net: Physics-Informed Neural Network for Sub-Surface Mineral Prospectivity Modelling

2026-05-31

AI Summary

This study addresses a critical limitation in conventional mineral prospectivity modeling, which relies on surface proxy data while neglecting the key subsurface physical processes governing orebody localization. To overcome this, the authors propose the first two-dimensional radial physics-informed neural network (PINN) that couples Darcy flow, convective-diffusive heat transport, and softened saturation reaction kinetics. Notably, this framework embeds Korzhinskii’s metasomatic infiltration theory into a differentiable simulator and is trained under weak supervision using only surface and remote sensing observations. Evaluated across six mining districts and three mineral deposit types, the method substantially outperforms baseline approaches, achieving an average PR-AUC of 0.708 (compared to 0.235 for SVM) and an average fractional rank of 0.036 (versus 0.475 for baselines), thereby transcending the constraints of purely data-driven paradigms.

Abstract

Mineral prospectivity modelling (MPM) underpins exploration economics, yet most operational pipelines reduce to data-driven classifiers trained on shallow surface proxies. Such models are blind to the subsurface physics that actually localises ore: heat advection, fluid flow, and lithology-dependent precipitation. We present Korzhinskii-Net, a 2-D radial physics-informed neural network (PINN) that couples Darcy flow, advective-diffusive heat transport, and a softplus-saturated reaction rate into a single differentiable forward model, weakly supervised by surface and remote-sensing proxies. The network is named after Dmitri S. Korzhinskii (1899-1985), whose theory of infiltration metasomatism provides the physical scaffold. We evaluate Korzhinskii-Net on five ore provinces spanning four commodity classes -- Norilsk (Ni-Cu-PGE), Pechenga (Ni-Cu sulphide), Udokan (sandstone-hosted Cu), Sukhoi Log (orogenic Au), and Mirny (kimberlitic diamond) -- under a fair, leakage-controlled 5-fold cross-validation protocol with hard ring-shaped negatives. Korzhinskii-Net attains a mean PR-AUC of 0.885 versus 0.281 for the strongest classical baseline (gradient boosting), and a mean fractional rank of 0.019 versus 0.413. The improvement is consistent across all five provinces and four commodity systems, suggesting that physics-informed differentiable simulators, even when constrained only by global open-data proxies, can recover localisation patterns that pure feature-based learners systematically miss. We release the full pipeline and evaluation harness as open source.

Problem

• Mineral prospectivity modelling
• Subsurface physics
• Ore localisation
• Physics-informed neural network
• Geological proxies

Innovation

• Physics-Informed Neural Network
• Mineral Prospectivity Modelling
• Darcy Flow
• Advective-Diffusive Transport
• Differentiable Simulator