The influence factor analysis for non-point source (NPS) pollution is very important to taking effective water pollution control measures. In this study, the self-organizing map (SOM) and linear model analysis were used to analyze the relationships between total phosphorus (TP) loads and influencing factors, both qualitatively and quantitatively. The land-use type, topography, and vegetation coverage were the main factors influencing the export of TP loads in Tangxun watershed. Slope and normalized difference vegetation index (NDVI) were chosen as characteristic indices of topography and vegetation coverage, respectively. For the whole watershed, the high TP loads were mainly distributed in areas with high slope and low vegetation coverage for a specific land-use type. For different land types, the slope significantly influenced the export of TP loads in waste/bare land and forest/green land while NDVI influenced the export of TP loads in forest/green land and farmland. In terms of multi-factor analysis, the comprehensive influence of slope and NDVI on TP loads showed as waste/bare land>forest/green land>farmland>rural/urban construction land.
Keywords: Complex non-point source; Influence factors; Phosphorus; Self-organizing map (SOM); Urban-rural fringe.