Study On Chestnut Soil Tier Thickness Model In The Typical Grassland

About 400 km2 areas of WuZhuMuQin typical grassland in Inner Mongolia was selected as the research subject which was against the phenomenon of the soil structural characteristics and wind erosion degradation. The surface characteristics of the typical grassland, three-dimensional coordinate data of the sampling points and chestnut soil tier thickness were obtained by means of GPS,GIS technologies combined with ground survey and sampling measurement methods,565 experimental data points were obtained. The basic data were dealt with by SuperMap software beforehand and managed by SQL Server database. The kriging method model and the BP neural network model based on Matlab was set up. The result of the model was analyzed.The result shows that the largest relative error for the kriging method forecasting in the flat grassland is 2.816%, the smallest one is 0.952%.The result of the BP neural network model for forecasting indicates that the largest relative error in the flat grassland is 2.747%, the smallest one is 0.195%;the largest relative error in the hillside of grassland is 6.366%, the smallest one is 1.897%;the largest relative error in low mountain hill of grassland is 13.172%, the smallest one is 0.693%. The forecast accuracy of the model meets the basic requirements. The forecast accuracy of the BP neural network and the kriging method can be as an effective way of forecasting the typical grassland chestnut soil thickness.Typical grassland degradation and desertification are essentially resulted from the wind erosion of chestnut soil. Space variation of soil and vegetation in typical grassland were researched by means of forecasting model of chestnut soil thickness.It may provides the basic data for the typical grassland desertification prevention and the scientific management.