Identifying Dominant Factors and Combined Effects of Shallow Groundwater Nitrate Contamination in the Guanzhong Basin Using PSO-XGBoost-SHAP.

Authors

Kai Wang, Yahong Zhou, Jiayun Ji, Dan Wang, Tiantian Zhou

Published in

Water environment research : a research publication of the Water Environment Federation. Volume 98. Issue 9. Pages e70564.

Abstract

The Guanzhong Basin is a typical agricultural plain in China, where shallow groundwater nitrate contamination is jointly associated with multiple environmental factors. However, the dominant factors and their nonlinear response patterns remain unclear. In this study, shallow groundwater nitrate in the Guanzhong Basin was investigated using variables related to soil physicochemical properties, groundwater physicochemical parameters, hydrogeological conditions, climate, and land use/land cover. An XGBoost model optimized using particle swarm optimization (PSO) was combined with SHAP analysis to quantify the relative contributions and directions of association of environmental factors and to examine their condition-dependent combined patterns. The results showed that the PSO-XGBoost model effectively captured the nonlinear relationship between environmental factors and nitrate concentration, with satisfactory test-set performance (MAE = 1.3134, RMSE = 2.1502, R² = 0.9163). SHAP analysis indicated that precipitation, depth to groundwater, electrical conductivity, and vadose-zone thickness were the key predictors of shallow groundwater nitrate concentrations in the study area. Precipitation and electrical conductivity generally made positive contributions, depth to groundwater made a negative contribution, and vadose-zone thickness showed a nonmonotonic response pattern. SHAP dependence analysis further suggested that paired factors exhibited condition-dependent associations in model-predicted nitrate concentrations, providing evidence to support groundwater nitrate-risk management in the Guanzhong Basin.

PMID:
42723415
Bibliographic data and abstract were imported from PubMed on 11 Sep 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.