Evaluation of interpolation models for rainfall erosivity on a large scale

Rainfall erosivity is an essential factor to describe the potential soil loss caused by rain, which can be expected to change in correspondence to climate changes. Spatial distribution pattern of rainfall erosivity is a guideline to erosion regional difference revelation and soil conservation regionalization. The focus of this article is to search an optimum spatial interpolation model for the mapping of rainfall erosivity on a large scale. In this research, average annual rainfall erosivity on China mainland is calculated through daily precipitation dataset from 711 weather stations over a period of 58 years (1951-2008). Precisions of 29 spatial interpolation models are compared including inverse distance weighting (IDW), radial basis function (RBF), kriging, cokriging (CK) and thin plate smoothing spline (TPS). Results indicate that three variables cubic TPS is the optimum model for spatial interpolation to rainfall erosivity on a large scale. Spatial characteristic of rainfall erosivity on mainland was analysed with the model employed. A increasing from northwest to southeast is identified and the highest annual rainfall erosivity occurs in the southwest area of Guangxi on the value of 25885 J. mm / (m(2) . h).