ABSTRACT Due to the excavation of landfill sites, the safe disposal of high-content humus soil and the control of secondary heavy metal pollution have become urgent issues to be addressed. Exploring the forms of heavy metals in humus soil is crucial for a comprehensive and accurate assessment of the characteristics and toxic effects of heavy metal pollution. The present study analyzed the pollution characteristics and ecological risks of heavy metals in humus soils obtained from three typical municipal solid waste landfills in Zhejiang Province, China. The results indicated elevated concentrations of Cu, Zn, Cr, and Pb in the three landfills, with maximum concentrations reaching 5041, 6093, 2756, and 3576 mg/kg, respectively. The RAC (Risk Assessment Code) Risk Index and Potential Ecological Risk Index methods indicate that Cd in landfills has a high ecological risk and high bioavailability. Five machine learning models of Logistic Regression (LR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and Multi-Layer Perception (MLP) were employed to predict the concentrations of heavy metals. The RF model showed the best performance for predicting Cu with an average Area Under the Curve (AUC) of 0.99, while the MLP model performed the best for predicting the concentration of Hg. This study provides a scientific basis and practical guidance for the effective control and resource utilization of heavy metal pollution in landfill humus soil.
Read full abstract