Abstract: Ever-increasing water environmental pollution, caused by non-point source (NPS) pollution, has seriously affected the safety of drinking water, which has drawn much more concern in recent years. Here, the secondary drinking water source protection of Chaihe Reservoir was selected. With a questionnaire survey, we proposed a new mechanistic method to assess the pollution sources of agricultural NPS pollution (e.g., agricultural planting, livestock farming and rural life) in this study. The dynamic equilibrium equation N_P=N_AP+(1-α)N_RF+(1-β)N_1F, was used to analyze dynamic change of nitrogen and phosphorus and estimate the total load of NPS pollution. The livestock farming and rural life pollution loads were assessed by export coefficient method. Additionally, the agricultural planting pollution load was mostly due to fertilizer application. Decision tree classification method was used to classify and extract the information of crop. The crop type was identified by land use, which was determined by remote sensing data and the theory of phenology. The amount of fertilizer application was calculated by cropyield evaluating model, which was based on the relationship between NDVI and measured cropyield. The results showed that crop type was sensitive to remote sensing data. Both nitrogen and phosphorus pollution from agricultural planting contributed the maximum loads, being 205.9 and 61.61 t·a^-1, respectively, which indicated that agricultural planting was the main source of NPS load in the study area. Our results can help the management of this drinking water source protection area, decreasing in NPS loads into downstream water bodies and consequently restoring the ecological environment.

Key words: energy balance function, heat island effect, spatiotemporal scale.