北运河水质变化特征及主导污染因子识别OA

As one of the major surface water systems in Beijing,the water environmental quality of the North Canal is significant for the ecological security of the capital and the green development of the city's sub-center.To systematically assess the temporal evolution pattern of its water quality and the variation characteristics of its dominant controlling factors,this study constructed a XGBoost regression model to predict water quality grades based on monitoring data from representative sections of the North Canal from 2021 to 2025,and employed the SHAP method to quantitatively analyze the influence of key indicators,including chemical oxygen demand(CODCr),ammonia nitrogen(NH3-N),and total phosphorus(TP),on water quality.The results showed that the water quality of the North Canal exhibited a staged evolutionary trend over the five-year study period,transitioning from significant fluctuations to gradual improvement and stabilization.The water quality grade gradually shifted from being predominantly Class Ⅳ in the early stage to mainly Class Ⅲ,although seasonal fluctuations and episodic pollution risks persisted in some months.The XGBoost model demonstrated high accuracy and stability in predicting water quality grades across different years,effectively capturing the temporal trend.The SHAP interpretation results further revealed an interannual succession pattern of dominant factors,such as NH3-N was the primary factor during the initial stage(2021-2022)and a structural shift driven by TP occurred in 2023,which followed by a stable regulation stage(2024-2025)characterized by the renewed dominance of NH3-N and relatively dispersed contributions from various pollution factors.While CODCr maintained a relatively stable level of importance throughout the entire period.Overall,the dynamic evolution of the dominant water quality factors reflects continuous changes in the basin's pollution load structure and the effectiveness of governance measures,providing a scientific basis for phased and refined water environmental management.