Abstract:  On the basis of the second soil survey data of Jiangxi Province, this paper took K values of EPIC model as measuring indicators of soil anti-erodibility, compared the accuracy of different interpolation methods ， including IDW(inverse distance weighted) ， Spline and Ordinary Kriging, by the software of ArcGIS9.2 and GS+9.0 for the spatial interpolation of soil anti-erodibility in Jiangxi Province. The results show as follows: The spatial variability of soil anti-erodibility at the study area is moderate difference, with the range from 0.149 2 to 0.415 7, mean of 0.283 9, and variance coefficient of 18.06%. The nugget-to-sill ratio is 11.09%  , which indicates that the soil anti-erodibility has strongly spatial relativity in the study area. Taking account of cross-validation and interpolation results comprehensively, Ordinary Kriging interpolation is the best of three methods. Spatial distribution of soil anti  erodibility appears with the decreasing trend from the east, south, west to the central, north and northeast in the study area; in comparison with the mean of each soil sub-class anti-erodibility, both of them have the same law of spatial distribution.

Key words: soil anti-erodibility , spatial interpolation , spatial variation characteristics