Thermodynamic-Inspired Explainable GeoAI: Uncovering Regime-Dependent Mechanisms in Heterogeneous Spatial Systems

2026-04-05

AI Summary

This study addresses the challenge that traditional geographically weighted regression and deep learning methods struggle to interpret the reversal of driver effects in heterogeneous spatial systems caused by state dependence. The authors propose the first thermodynamics-inspired, interpretable GeoAI framework, which integrates statistical mechanics with graph neural networks. By modeling spatial heterogeneity through a competition mechanism between system “burden” (E) and “capacity” (S), the framework constructs a phase-transition diagnostic model capable of distinguishing genuine shifts in physical mechanisms from mere statistical anomalies. Evaluated across six datasets—including housing markets, mental health, and wildfire PM2.5—the method successfully identifies driver role reversals missed by conventional models and accurately diagnoses the burden-dominated phase transition during the 2023 Canadian wildfire event, achieving both high predictive accuracy and mechanistic interpretability.

Abstract

Modeling spatial heterogeneity and associated critical transitions remains a fundamental challenge in geography and environmental science. While conventional Geographically Weighted Regression (GWR) and deep learning models have improved predictive skill, they often fail to elucidate state-dependent nonlinearities where the functional roles of drivers represent opposing effects across heterogeneous domains. We introduce a thermodynamics-inspired explainable geospatial AI framework that integrates statistical mechanics with graph neural networks. By conceptualizing spatial variability as a thermodynamic competition between system Burden (E) and Capacity (S), our model disentangles the latent mechanisms driving spatial processes. Using three simulation datasets and three real-word datasets across distinct domains (housing markets, mental health prevalence, and wildfire-induced PM2.5 anomalies), we show that the new framework successfully identifies regime-dependent role reversals of predictors that standard baselines miss. Notably, the framework explicitly diagnoses the phase transition into a Burden-dominated regime during the 2023 Canadian wildfire event, distinguishing physical mechanism shifts from statistical outliers. These findings demonstrate that thermodynamic constraints can improve the interpretability of GeoAI while preserving strong predictive performance in complex spatial systems.

Problem

• spatial heterogeneity
• regime-dependent mechanisms
• critical transitions
• nonlinearities
• phase transition

Innovation

• Thermodynamic-inspired GeoAI
• Spatial heterogeneity
• Regime-dependent mechanisms
• Graph neural networks
• Phase transition