Digital mapping of apparent electrical conductivity using regression kriging and local variogram in Ardakan region

Converting point data to continuous one is the first step in order to use them in scorpan flowchart. For this purpose, different geostatistic methods are available which at present research regression kriging with local variogram was applied. For mapping apparent electrical conductivity at the area covering 72000ha located in Ardakan region, 700 readings in horizontal and vertical modes carried out by electromagnetic induction. Auxiliary data used in this study to represent predictive soil forming factors were terrain attributes, Landsat 7 ETM+ data and a geomorphologic surfaces map. At first, the relationship between independent variables (i.e. auxiliary data) and dependent variable (i.e. apparent electrical conductivity) was calculated by regression tree. Then, the residuals, derived from regression tree, were mapped by using kriging with local variogram. Finally, the residual and ECa maps were aggregated in order to produce the final maps. Results showed some auxiliary variables had more influence on predictive apparent electrical conductivity model which included: wetness index, geomorphology map and the first principal component analysis. Results also confirmed that regression kriging with local variogram had high performance; however, determination of coefficient, root mean square error and mean error calculated for model in vertical mode were 0.49, 37.74 and -1/07, respectively. These results are acceptable in digital soil mapping studies and hence, it is suggested using of regression kriging with local variogram for spatial prediction of soil properties in future studies.