Land use change is one of the most important factors affecting the migration of non-point source (NPS) pollution. Watersheds in semi-urbanized areas are sensitive to human activities, especially the change of land use, which leads to significant changes in NPS pollution. In our research, by simulating the continuous dynamic migration path of NPS pollution in a typical small basin (Houxi River Basin) in a semi-urbanized area, a key indicator, namely the land use impact coefficient (LUIC), is proposed to explore the extent of the impact of land use on the export coefficient (EC) of NPS pollution. To reduce ECs, we set three scenarios for optimizing LUICs. The results show that: (1) Compared with the buffer zone in the mainstream of the river (1.16 km2), by optimizing land use/land cover types of areas with high LUIC (0.71 km2), the EC reduced by 0.58%, which is higher than that of reduced by 0.21%; (2) In terms of the sub-watersheds, the closer to the mainstream of the river, the more effective the reduction of EC by optimizing land use/land cover in areas with high LUIC; and (3) Compared with the ECs of areas with high LUIC under different buffer zone widths, the ECs of different LUIC values under a 90-meter buffer zone have a higher decline rate. Meanwhile, it is also necessary to consider the actual situation and reduce the costs of greening by selecting appropriate buffer zone areas. The significance of this research is that the proposed method is conducive to rapidly identifying the key areas of land use affecting the migration of NPS pollution. Through the rational planning of green space, it is possible to reduce ECs and lay a research foundation for proposing targeted land management, regional spatial planning, and strategies for the protection of basin ecological environments.
