[Objective] In order to clarify the spatiotemporal distribution characteristics of total nitrogen (TN) and total phosphorus (TP) loads from non-point source pollution in the Guanshan River watershed, as well as to identify their major pollution source contributions, and simultaneously to address the problem of insufficient estimation accuracy of the traditional export coefficient model when applied to watersheds with complex terrain conditions, this study introduced a rainfall factor (α) and a topography factor (β) to improve the traditional export coefficient model. Based on this improved model, the nitrogen and phosphorus loads from non-point source pollution in the Guanshan River watershed from 2021 to 2024 were estimated, and their temporal and spatial variation characteristics were systematically evaluated and analyzed. [Methods] In this study, the rainfall factor (α) was extracted by utilizing the daily runoff data recorded at the hydrological station located at the watershed outlet of the Guanshan River watershed during the period from 2021 to 2024, in combination with the observed rainfall data collected from the Dama Station, Xihe Station, Yuanjiahe Station, and Gushan Hydrological Station. Meanwhile, the topography factor (β) was extracted based on the Digital Elevation Model (DEM) data of the Guanshan River watershed. Subsequently, both the rainfall factor (α) and the topography factor (β) were incorporated into the framework of the traditional export coefficient model, thereby constructing an improved export coefficient model with enhanced applicability to the study area. [Results] (1) The improved export coefficient model effectively enhanced the simulation accuracy of nitrogen and phosphorus load estimation. Specifically, the NSE coefficients for total nitrogen and total phosphorus increased significantly from 0.18 and -2.65 to 0.98 and 0.93, respectively, while the relative error was reduced from a maximum value of 291.89% to a controlled range within 10%, demonstrating a substantial improvement in model performance.(2) During the period from 2021 to 2024, the total nitrogen and total phosphorus loads in the Guanshan River watershed decreased by 67.84% and 80.06%, respectively. These reductions further revealed the contribution rates of nitrogen and phosphorus loads from three major source categories, namely land use sources, rural domestic sources, and livestock sources. Among these categories, land use sources contributed the largest proportion, accounting for 81.32% to 94.94% of the total nitrogen and phosphorus loads. Within land use types, forest land made the highest contribution to both nitrogen and phosphorus loads, while the total nitrogen load intensity from cropland reached 2 778.63 kg/km², and the total phosphorus load intensity from construction land reached 709.30 kg/km². (3) On the temporal scale, both total nitrogen and total phosphorus loads exhibited an overall declining trend throughout the period from 2021 to 2024. On the spatial scale, Baihe Town and Guanshan Town were identified as the key priority areas for nitrogen and phosphorus pollution prevention and control within the watershed, primarily due to their diverse land use types and relatively higher intensity of human activities compared to other regions. [Conclusions] This study successfully constructed an improved export coefficient model by incorporating both the rainfall factor and the topography factor, which effectively enhanced the estimation accuracy of nitrogen and phosphorus loads from non-point source pollution in the Guanshan River watershed. The results demonstrate that the total nitrogen and total phosphorus loads in the Guanshan River watershed generally showed a decreasing trend from 2021 to 2024, thereby verifying the positive effects of the water source protection measures implemented in the region. However, it should be noted that the effectiveness of pollution control remains jointly influenced by multiple interacting factors, including interannual rainfall variability, changes in land use types, and livestock activities. Therefore, in the future, it is necessary to adopt precise zoned and categorized management strategies tailored to different sub-regions and pollution source types. This study provides a quantitative scientific basis for the effective prevention and control of non-point source pollution in the Guanshan River watershed, and in doing so, offers a valuable scientific reference for safeguarding the long-term water quality security of the South-to-North Water Diversion Project.