Abstract:
Geostatistics combined with GIS were applied to analyze the spatial interpolation accuracy of soil nutrients such as organic matter,alkali-hydrolyzable nitrogen,available phosphate and available potassium in Lantian county. The accuracy of prediction obtained by Ordinary Kriging and Cokriging were compared by using cross-validation parameters and relative tolerance, and the result showed that there was a medium level of spatial variability in soil nutrients, and in the condition of significant correlation in soil nutrients, using Cokriging can narrow the range of errors, and reduce the root-mean-square prediction error (RMSE), and improve the accuracy level between actual value and predicted value. The conclusion is that Cokriging can improve the accuracy of prediction value of soil nutrients in medium scale region.
