Abstract: Based on the graphs of soil erosion intensity at different spatial scopes in Yimeng Mountain area in 2004, and by using GIS technology and Fragstats software, 8 landscape pattern indices were selected to study the effects of the grain size of landscape pattern indices at different spatial scopes within the range of 20 to 65 m in Yimeng Mountain area, and the functional fitting analysis was applied to analyze the landscape pattern indices. With the increase of grain size, the landscape pattern indices showed different change trends, i.e., rising, declining, or unconspicuous change. As for the two study areas with different spatial scopes, the first scale domain of landscape pattern index was (20, 35), and the optimal spatial grain sizes were all 35 m. Through the fitting of the selected 8 landscape pattern indices by one-dimensional linear regression analysis, it was found that the NP, PD, LSI, PAFRAC, CONTAG, and AI could fit the change trends of the indices well, while the fitting results of LPI and SHDI were unsatisfactory.