Taking a typical mining area as an example, statistical methods and Positive Matrix Factorization（PMF）were integrated to qualitatively and quantitatively identify key regional pollution sources and their contributors. A spatial model was further constructed, considering the spatial heterogeneity of soil heavy metal pollution and its dominant environmental drivers, with the best environmental variables and spatial scale being selected. The results revealed that the sources of soil heavy metal pollution were natural sources, exhaust gas emission sources, slag emission sources, wastewater emission sources, and transportation sources, with contributions of 8.40%, 9.55%, 1.73%, 55.37%, and 24.99% of the total pollution, respectively. Notably, atmospheric deposition（q =0.113）and soil leaching（q=0.097）were identified as the primary input and output pathways for heavy metals. Among various spatial modeling strategies, the model that integrated both spatial pollution source characteristics and environmental variables demonstrated the highest predictive accuracy, outperforming the model based solely on dominant environmental factors or pollution source characteristics. The importance of incorporating spatial information to enhance model performance was highlighted by this finding. In particular, the Geographically Weighted Regression Kriging（GWRK）model was found to achieve superior predictive accuracy（mRadius=0.2916）when multiple data sources were integrated. Overall, a scientific foundation was provided for identifying high-risk soil pollution zones in mining regions, the understanding of ecological and environmental interactions between influencing factors and heavy metal contamination was enhanced, and valuable insights were offered for spatially targeted pollution control strategies.