Methods for digital soil mapping using a geographical information system

Soil maps are sources of important information for land planning and management, but are expensive to produce. This paper proposes testing and comparing single stage classification methods (Multiple Multinomial Logistic Regression and Bayes) and multiple stage classification methods (Classification and Regression Trees (CART), J48 and Logistic Model Trees (LMT)) using geographic information system and terrain parameters for producing soil maps with both original and simplified legend. The database was managed in ArcGis computer application in which the variables and the original map were related through training of the algorithms. The results from statistical software Weka were implemented in ArcGis environment to generate digital soil maps. The terrain parameters that best explained soil distribution were slope, profile and planar curvature, elevation, and topographic wetness index. The multiple stage classification methods showed small improvements in overall accuracies and large improvements in the Kappa index. Simplification of the original legend significantly increased the producer and user accuracies, however produced small improvements in overall accuracies and Kappa index.