Integrating spatiotemporal variability of non-point source pollution and best management practice efficiency to improve adaptive watershed management

As a major contributor to water quality degradation, non-point source (NPS) pollution exhibits pronounced spatiotemporal variability, posing significant challenges to watershed management under changing hydrological conditions. However, most existing control strategies rely on static assumptions, neglecting dynamic variations in pollutant loading and the performance of best management practices (BMPs) across time and space. This study proposed a new framework to assess the impact of spatiotemporal variability on watershed management, using the Liao River watershed in China as a case study. Copula and Bayesian models were coupled to provide a pollutant load scenario that integrates multiple hydrological conditions for BMPs configuration, capturing the nonlinear relationship between flow and water quality, and the associated uncertainties. Moreover, BMPs efficiency was evaluated using a self-organizing map combined with multiple regression, accounting for dynamic changes across watershed characteristics and operational durations. The results demonstrated that temporal variability in pollutant distribution led to substantial fluctuations in management costs, with a 38.91% increase in dry years and a 23.38% decrease in wet years. The spatial variability expanded the required control area by 25.99% and increased the control cost by 31.98%. The spatiotemporal variation in BMPs efficiency further affected management benefits. A five-year operational period for BMPs yielded the lowest management costs, which were 2.64% and 21.70% lower than those for one-year and ten-year periods. Additionally, spatial variability in BMPs efficiency increased management cost by 15.55%-28.97%. These findings provide quantitative insights to support adaptive BMPs layout under changing environmental conditions, thereby enhancing resilience in watershed management.