This study investigated the spatial distribution of BNP hot spots in the upper Shui-Li Creek watershed. Two spatial autocorrelation methods, Moran's I and G statistics, were selected for analyzing the spatial cluster of BNPs with different grid sizes. The results show that spatial autocorrelation analysis can indeed discover BNP hot spots in this watershed. The strength and significance of spatial dependence can be easily compared and monitored. For the global spatial autocorrelation analysis, both Moran's I and G statistics values are positive with whole different grid sizes and statistically significant (p ＜ 0.05). This means that the BNPs in study areas obviously have the characteristics of positive spatial autocorrelation. Furthermore, with the increase of the grid size, the value of Moran's I decreases, but the value of G statistics increases.