Prediction of nitrate concentration and the impact of land use types on groundwater in the Nansi Lake Basin

J Hazard Mater. 2025 Jan 13:487:137185. doi: 10.1016/j.jhazmat.2025.137185. Online ahead of print.

Abstract

Groundwater faces a pervasive threat from anthropogenic nitrate contamination worldwide, particularly in regions characterized by intensive agricultural practices. This study examines groundwater quality in the Nansi Lake Basin (NSLB), emphasizing nitrate (NO3--N) contamination. Utilizing 422 groundwater samples, it investigates hydrochemical dynamics and the impact of land use on groundwater composition. Key methods include hydrogeochemical analysis, PCA, and the Duncan comparison method. The innovative aspect lies in using Multilayer Perceptron Artificial Neural Networks (MLP-ANNs) to predict NO3--N contamination. The results showed that NO3--N levels ranged from 0.004 to 177.72 mg/L, with approximately 43.6 % of the samples exceeding the safe drinking water limit of 10 mg/L (WHO 2022). Substantial spatial variability in the concentrations of major ions within aquifers, with NO3--N exhibiting the most significant fluctuations. The factors responsible for the hydrochemical composition of groundwater include recharge sources, water-rock interaction, prevailing groundwater environment, land use patterns, and related anthropogenic activities. Notably, land use types, primarily farmland and rural areas, exhibited a strong association with NO3--N. The MLP-ANNs achieved high prediction accuracy for NO3--N, with an AUC of 0.85. The MLP-ANN model identified heightened susceptibility to nitrate contamination in the central and southeastern regions, characterized by dense shallow wells (<60 m). Key factors include nitrogen-based fertilizer overuse, agricultural runoff, domestic wastewater discharge, and septic system leakage. The vulnerability is exacerbated by highly permeable loose rock pore water systems underlying intensively cultivated agricultural lands. This study elucidates the complex interrelation between natural processes and anthropogenic activities that influence groundwater quality, providing valuable perspectives that could guide the formulation of policies and practices aimed at promoting sustainable groundwater utilization and environmental conservation.

Keywords: Groundwater; Hydrogeochemistry; Land use; Machine Learning; Nitrate.