A hybrid soil mapping approach using SOTER, SoLIM and Classification trees

The investigated mapping approach combines different (soil) mapping concepts including SOTER (FAO 2005), SoLIM (Zhu et al. 2001), expert knowledge and classification trees (Breiman et al. 1984). Northwestern Thailand was stratified into SOTER terrains using DEM data and geological information. Detailed soil mapping was tested in three pilot areas within different SOTER terrains in north-western Thailand. Therefore, classification trees were calibrated with soil data from transect-based sampling points. Maps of the pilot areas were created by implementing classification rules derived from Classification Tree (CART algorithm) and expert knowledge in SoLIM. Validation was performed using soil reference maps and independent sampling points. The reference soil maps contain information from transects, sampling points along local trails, sampling points for areas of low point density or high soil variability, reference profiles, LANDSAT and SPOT images, and topographic information. The mean sampling distance was around 200 m. The reference soil maps were manually created based on expert knowledge using ArcGIS 9.2 software. The hybrid mapping approach based soil maps showed high correspondence with the respective reference soil map and a very high degree of matching with independent sampling points. This hybrid mapping approach seems to be very useful for reconnaissance soil mapping.