Abstract:Non-point source pollution has become an important source of pollution affecting the water environment of drinking water sources in recent years due to its wide distribution, complex production process and difficult prevention and control. This paper carried out remote sensing identification and evaluation analysis of non-point source pollution risk based on "source-sink" landscape in Chitian Reservoir. Firstly, high-resolution remote sensing image was used to extract different types of landscape distribution information of water source. Then, combined with the promotion/blocking effect of "source-sink" on pollution, considering the distance from the river and slope factor affecting the migration of pollutants, the non-point source pollution risk index was used to calculate the risk of non-point source pollution. And we analyzed the influence of spatial distribution of water source landscape on non-point source pollution based on Lorenz curve. The results showed that the risk of non-point source pollution was low, and the sub-watershed dominated by "sink" landscape accounted for 76.50% of the whole area. The risk of non-point source pollution in the basin was high in the east and low in the west. The extremely high risk areas mainly distributed in the southeast of the basin, which were dominated by "source" landscape types such as residential land and construction land, accounting for 1.28% of the overall regional area. And the western region was dominated by forest landscape with little pollution risk. The ratio of pollution load of "source" and "sink" landscape based on slope factor was greater than 1, indicating that the "source" landscape had a wide distribution range in the low-slope region. The landscape layout was more reasonable among slope elements. The pollution load ratio of "source" and "sink" landscape based on the factor of "distance from the river" was 2.30, indicating that the "source" landscape had a large distribution range near the river course and had a great impact on pollution. Therefore, the landscape pattern near the river should be further adjusted to reduce the impact of non-point source pollution. The method of "source-sink" landscape index calculation based on remote sensing is a rapid, objective and effective method for the identification and evaluation of non-point source pollution.