In recent years, non-point source pollution has become the main cause of water quality deterioration in some reservoirs in China. Taking the Panjiakou Reservoir as an example, the classical output risk model was improved by introducing a precipitation factor and terrain factor. Combined with high-resolution satellite precipitation products (GPM) and GF-6 satellite images, a high-resolution data-driven risk assessment method for non-point source pollution output was established to study the temporal and spatial distribution characteristics of non-point source pollution output risk in the Panjiakou Reservoir basin. The results showed that the non-point source pollution output risk was high in the study area in 2018. The areas with higher and highest risk of nitrogen pollution output accounted for approximately 70.6% of the total watershed area, whereas the higher risk of phosphorus pollution output accounted for approximately 21.9%. The temporal and spatial distribution characteristics of non-point source pollution output risk in the Panjiakou Reservoir basin were analyzed. It was found that the non-point source pollution output risk in the Panjiakou Reservoir basin increased first and then decreased from April to September. This was consistent with the spatial and temporal distribution of precipitation in the basin. Combined with the analysis of land use distribution characteristics, the upstream area of the basin was mainly cultivated land, whereas cities were concentrated in the downstream portion of the basin. Affected by agricultural production and human activities, the risk of non-point source pollution output was higher in these regions. In view of the temporal and spatial distribution characteristics of non-point source pollution output risk, it is necessary to formulate a reasonable agricultural fertilization method, plan the landscape layout of source-sinks, and construct vegetation buffer zones.
Keywords: Panjiakou Reservoir basin; high-resolution data-driven; non-point source pollution; output risk assessment; output risk model.