A novel hybrid approach for quantitative source apportionment and source-oriented health risk of heavy metal(loid)s pollution in an old industrial city

Abstract

Heavy metal(loid)s (HMs) contamination in soil is a major environmental problem worldwide due to its threat to the environment and human health, particularly in old industrial cities. However, the accurate identification of sources, drivers and risks of HMs remains to be further studied. In this paper, a novel hybrid approach combining self-organizing map (SOM), Positive Matrix Factorization (PMF), Gradient Boosting Decision Tree (GBDT) and Monte Carlo simulation was applied to evaluate quantitative source identification, driving factors and source-oriented health risks of HMs in an old industrial city. The results showed that the concentrations of all eight HMs largely surpassed soil background values in the study area, indicating a high ecological risk. The results of SOM, PMF and GBDT models revealed that agricultural source (25.60%), mixed source (28.39%), natural source (13.81%), and industrial source (32.20%) were identified as the major sources of HMs. The river density (RIDE), road density (RODE), Fe₂O₃ content (Fe₂O₃), factory distance (FDI) were the key drivers of pollution sources. The health risk assessment suggested that the carcinogenic risk (CR) of all HMs exceeded the acceptable risk threshold of 1.00E-6, indicating a high cancer risk for adults and children, while the hazard quotient (HQ) of all HMs for all populations was negligible (HQ<1). As and Cr were the primary pollutants contributing to hazard index (HI), and the mixed source was the largest contributor to total carcinogenic risk (TCR), with a contribution of 50.18% in adults and 51.61% in children. This study provides novel insights into quantitative source apportionment and source-oriented health risks of HMs pollution, which can improve the contamination risk control and environmental management in old industrial cities.