Identification and analysis of water pollution factors based on APCS-optimized input models

Rivers play a crucial role in urban ecosystems. However, urbanization has exacerbated water quality deterioration due to the discharge of untreated domestic sewage and industrial effluents. To accurately identify and quantify water pollution factors in the Nantong section of the Yangtze River, this study employed multi-model regression methods with inputs optimized by Absolute Principal Component Scores (APCS), including Multiple Linear Regression (MLR), Ridge Regression (RR), and Lasso Regression (LR). Results demonstrate that the APCS-RR model excels in handling highly correlated water quality datasets, enabling precise identification and quantification of seasonal contributions of pollution factors. Significant differences in water quality between dry and wet seasons were observed: inorganic salts and mineral pollution predominated during the dry season, whereas organic matter and nutrient pollution were dominant during the wet season. Strengthening industrial discharge control, optimizing agricultural practices, and implementing seasonal water quality management policies and measures can effectively address the challenges posed by seasonal water quality variations. The findings provide a data foundation for local water quality and environmental management strategies and offer a reference method for precise identification of water pollution factors in other regions.