Spatial Assessment of Agricultural Non-Point Source Phosphorus Pollution in Six Contrasting Basins Using IECM and RUSLE Models

Agricultural nonpoint-source phosphorus (NPS-P) losses threaten receiving waters, but regional control is complicated by differences in source intensity, erosion sensitivity, and hydrologic connectivity. This study jointly applied an improved export coefficient model (IECM), the Revised Universal Soil Loss Equation (RUSLE), Global and Local Moran statistics, and Getis-Ord Gi* analysis to six contrasting basins. The outputs were cross-interpreted without a formal composite index. Average annual soil erosion ranged from 1.96 to 18.47 t ha−1 yr−1, with very slight and slight erosion dominating all basins. Annual NPS-P export ranged from 2169.55 to 12,028.32 t yr−1 (1.26–2.74 kg ha−1 yr−1), and cultivated land contributed 48.60–69.32% of modeled export. Under 999 random permutations, Global Moran’s I ranged from 0.615 to 0.834 (pseudo p = 0.001), and Gi* hot spots occupied 26.18–33.59% of valid cells. The basin-level perturbation analysis indicated greater ranking robustness for clearly high- and low-load basins than for intermediate basins, while the cultivated-land sensitivity analysis quantified the influence of the dominant coefficient. The framework supports regional screening, monitoring prioritization, and subsequent field verification rather than calibrated event-scale prediction.