A Comparison On Soil Nutrient Interpolation Accuracy Of Mountainous Area And Hills

Based on MapGIS6.7 ArcGIS9.3, SPSS11.5 and other software, this paper use inverse distance weighted method, local polynomial interpolation, Kriging interpolation method for spatial interpolation of farmland soil nutrients in Yongshun County and Shaoyang County. In this paper, compare the accuracy of soil nutrient spatial interpolation results with root mean square error (RMSE) and mean prediction error (ME),then made soil nutrient distribution based on the comparison, after that, compare with the statistical characteristics of actual data to get the best method of spatial interpolation about mountains and hills. Hope that the result can cultivate for the farmer provides the science instruction, applies fertilizer rationally and scientic, reduces the environment pollution because of the nonessential investment, raises the production benefit, improve quality of life, lead to the resource conservation, the environment to be friendly truly. The main conclusions are as follows:(1) Whether the conditions of mountainous or hills, organic matter and nitrogen data are normally distributed and P and K are the log-normal distribution;(2) Whether the conditions of mountains or hills, the nutrient content in soil is relatively stable, soil nutrient what is a medium vary degree variation range are between 10%-100%. Among them,variation index of P that is much higher than other soil nutrients in Yongshun County was relatively large; variation of soil nutrients of Shaoyang County in a variety of little change are between 25.91%-54.04%;(3) Whether the conditions of mountains or hills, the soil nutrients, Whether to consider the case of soils,compare of inverse distance weighted method,local polynomial interpolation, and Kriging interpolation three methods, the RMSE of Kriging interpolation method and local polynomial interpolation are similar, interpolation results are better, inverse distance weighted interpolation are worst; (4) Whether the conditions of mountainous or hills, in the soil nutrients, if the data to conform to normal distribution,it must consider whether to distinct the soil classes to accuracy of interpolation;if not, do not considerthe distinction.