Performance of soil electrical conductivity and different methods for mapping soil data from a small dataset

The performance of different methods for making soil maps from a small dataset was assessed. Soil samples obtained at five different depths were taken from 39 locations within an arable field in Sweden and analysed for clay content, soil organic matter content, pH (H2O), K-HCl, P-AL, K-AL and Mg-AL. Detailed and densely sampled soil electrical conductivity (EC) data with a distinct border between two different regions was used for dividing the field into two zones. Averages from these zones gave better prediction for most variables than interpolation without respect to the border. With the border taken into account, cokriging with EC as a covariable improved the prediction, whereas the improvement with ordinary kriging and inverse distance weighting (IDW) was insignificant. Direct interpretation of EC by a simple linear regression model gave reasonable predictions for clay content. In general, even the simplest interpolation method improved the prediction compared to field average.