Response of water quality to land use and sewage outfalls in different seasons

Abstract To better manage water environment in highly polluted rivers, the influence factors on water quality need to be investigated. With the effects of oxygen-demanding contaminants, it is difficult to resolve the complex interdependencies of the various factors using conventional methods. The Bayesian Networks (BNs), in which each variable only depends on its immediate parent variables, can solve this problem. In this study, the BNs were developed to assess the impacts of land use and sewage outfalls on Ammonia Nitrogen (AN) and Dissolved Oxygen (DO) concentrations in the Huaihe River Basin (HRB) for different seasons and spatial scales, where AN was a typical oxygen-demanding contaminant and the most serious contaminant in the area. The BNs gave the best explanations for variations in AN (NSE = 0.80) and DO (NSE = 0.72) concentrations by using land use and sewage outfalls data at the local scale (less than 20 km radii around monitor stations), suggesting that controlling water contaminant sources at local scales can improve water quality efficiently. AN negatively affected DO concentration, which was more significant in dry seasons. Wastewater from sewage outfalls was the largest contributor (26.2%) to AN pollution in dry seasons, which was weakened in wet seasons by an intensive dilution process. Farmland acted as a “sink” in dry seasons and as a “source” in wet seasons. The transition between two states was caused by large variations in surface runoff between dry and wet seasons. Urban land made a disproportionately large contribution to water pollution compared to other kinds of land use. These findings improve our understanding of influence factors on water quality and will contribute to effective river management.