Over the past few decades, rivers have become severely polluted as a result of receiving vast quantities of domestic and industrial wastewater. The identification of the major factors that influence water quality is crucial to understand the interactions of anthropogenic and natural factors and develop river restoration projects. In this study, the QUAL2Kw water quality model was used to quantitatively evaluate the most critical factors for water quality at two sites with different meteorological conditions and urban scales. The genetic algorithm (GA) was used to optimize the parameters in the model. The Monte Carlo simulation (MCS) method was used to assess the model uncertainty and sensitivity in all reaches for five water quality outputs (temperature, CBOD, DO, TP, and TN) in two seasons. The K-means clustering method associated with the sensitivity results was used to identify the major factors influencing the water quality in all reaches from the input data and the model parameters. The results showed that CBOD, TN, and TP were most sensitive to headwater and tributary quality. DO tended to be affected by more natural reactions than the other water quality indicators. In the cold and dry seasons and the more urbanized areas, river pollution was more severe, and the impact of natural reactions was reduced. The simulation results revealed the reliability of QUAL2Kw in modeling the quantity and quality of all river reaches. The method applied in this study is beneficial for the improvement and management of the water environment.
Keywords: Influencing factors; QUAL2Kw; Sensitivity analysis; Water quality.