This study developed an improved export coefficient model by integrating rainfall and topographic correction factors to estimate non-point source nitrogen and phosphorus pollution loads and identify key pollution sources in the Ganjiang River Basin. The accuracy of the original and modified models was systematically compared, and correlation analysis was performed between nitrogen and phosphorus load intensity and monitored concentration data. The results demonstrated an increasing trend in both total pollution loads and load intensities from 2016 to 2020. Total nitrogen and total phosphorus exports increased by 15.99% and 16.37%, respectively, while corresponding load intensities rose by 15.89% and 16.85%. Spatially, the pollution distribution exhibited a characteristic north-high-south-low pattern with localized concentration, indicating higher contamination risks in downstream areas. Land use emerged as the primary source of nitrogen pollution, contributing 51.65% of total nitrogen exports, whereas livestock farming was identified as the dominant phosphorus source, accounting for 36.82% of total phosphorus outputs. The enhanced export coefficient model demonstrated significantly reduced relative errors compared to the original version. Statistical analysis revealed significant correlations (P<0.05) between annual average nitrogen or phosphorus concentrations and load intensities, confirming the improved model's superior accuracy. The refined model enables more precise assessment of watershed non-point source pollution, facilitates identification of major pollution sources, and supports targeted delineation of critical control zones, thereby providing valuable scientific support for non-point source pollution management and remediation strategies in river basins.