Causal inference-driven framework for source verification: Integrating PMF, health risk assessment, and spatialized...

Soil heavy metal pollution in industrialized regions poses significant health risks and management challenges due to its complex source composition and nonlinear source-receptor relationships. This study establishes a novel framework with GCCM-driven causal inference as the core, constructs explicit causal links between high-risk sources' characteristic pollutants (from PMF-health risk assessment) and emission source spatiotemporal features (from spatialized inventories) to prioritize soil heavy metal pollution control in complex industrial regions. Focusing on a mining city in Yunnan, China, 460 soil samples were analyzed for As, Pb, Cd, Zn, and Ni. PMF identified 4 pollution sources, with health risk assessment highlighting As as the priority pollutant. High-resolution spatialized emission inventories reveal distinct spatiotemporal patterns among different pollution sources, which indicates a multi-source composite pollution pattern and the presence of complex pollution formation mechanisms in the study area. The GCCM analysis identified unidirectional causal relationships between smelting emissions and soil arsenic (As) concentrations (ρ=0.456, p < 0.01), surpassing the effectiveness of Pearson correlation and geographically weighted regression in capturing nonlinear source-receptor linkages. This study demonstrates causal emission-contamination links, enabling accurate, economically scalable soil pollution remediation for industrial regions.