Mapping the Spatial Variability of Soil Properties: A Comparative Study of Spatial Interpolation Methods in Northeast China

We analyzed the variance characteristics of soil organic matter (SOM), total nitrogen (TN), extractable phosphorus (EP), and extractable potassium (EK), in Jiutai County, Northeast China, and compared different prediction methods for mapping of these four soil variables. The prediction methods used were geostatistical interpolation (ordinary kriging), inverse distance weight method, and the hybrid techniques (regression-kriging). A modified jackknifing method involving 40% partitions was used to examine the stability of validate the indices. Root mean square error (RMSE) was used as validation index, and mean RMSE was used to judge the prediction quality. The results showed that the hybrid interpolation regression-kriging can't be used in the region influenced by frequent and high-intensity human activity when the relationship between soil properties and environment factors were not obvious. The ordinary kriging was found to be the best method to fit the experimental semivariogram of SOM and EK. The inverse distance weight method fit well to predict the distribution of TN and EP. For SOM and EK, results showed that data values in the western part were higher than those in the eastern part. However, for TN and EP, there is no clear trend. Water and tillage erosion caused by human activity has weakened the structural influence and elevation and slope played key roles in the distribution of soil variables in the local area.