Identification of multi-source pollution in peri-urban soil-water systems based on a self-organizing map

Peri-urban areas in developing regions face challenges in identifying multi-source heavy metal pollution due to dominant industrial emissions' masking effect. This study investigated heavy metal characteristics, sources, and transport from long-term lead smelting and coal industries in a peri-urban environment, revealing severe solid and aqueous pollution. Soils/sediments surrounding the lead smelter showed 90-100 % exceedance of background levels, with exceedance of risk screening values for Pb (50 %), As (60 %), Cd (95 %), and Zn (45 %), with Pb peaking at 116.3 times the limit. Surface waters had worse pollution, with dissolved As exceeding standards in most river samples and over 50 % showing elevated heavy metal concentrations. The pretreated SOM approach successfully resolved multiple pollution sources across environmental media. In soils/sediments, three patterns emerged: heavily contaminated surface layers dominated by smelting-derived Pb, As, Cd, and Zn (S-I); soils/sediments with distinct coal-related Cr and Ni enrichment (S-II); and deep soils maintaining near-background conditions (S-III). Waters segregated into five groups: river waters strongly affected by smelting emissions (W-IV); shallow groundwater exhibiting industrial-domestic mixing with elevated SO₄^2-, Cl^-, and Zn (W-III); agricultural-influenced groundwater with notable nitrate (W-II); reducing waters with Mn enrichment (W-V); and deep groundwater with minimal anthropogenic influence (W-I). Hydrus-1D simulations revealed that while the vadose zone attenuates metal migration, acid rain enhances it significantly (1.4-16.5 times). The Ca·Mg-HCO₃ to Ca·Mg-HCO₃·SO₄ hydrochemical evolution, alongside aquifer mixing via faults, underscores industrial impact on groundwater long-term. This study provides a robust methodological framework for accurate source identification in multi-source polluted systems and supports targeted remediation strategies.