Prediction of the spatial distribution of soil properties based on environmental correlation and geostatistics

As the most important determinants of soil quality, soil properties significantly influenced land use and ecological processes. On the landscape scale, a comprehensive understanding and consideration of soil spatial variation was essential for establishing an ecological and environmental process modeling. Spatial variation of soil properties (include bulk density (BD), soil organic matter (SOM) and total phosphorus (TP)) was analyzed and predicted according to environmental indicators based on data from 254 points of surface soil (0-20 cm) by digital terrainin and remote sensing image analysis technique in Hengshan county on the Loess Plateau. The relationship between soil properties , terrain attributes and remote sensing indices was analyzed. Finally, environment variables were used to predict spatial distribution of soil properties by multiple-linear regression analysis and geo-statistical. The results showed that BD and SOM were positively correlated with terrain attributes and remote sensing indices, but TP has little significant correlation with remote sensing indices. The multiple-linear stepwise regression model was relatively precise for the BD and SOM, but for TP, the predicted result was poor. The regression-Kriging method can effectively reduce residuals in prediction by eliminating smoothing effect, and its predicted values are quite close to the measured.