Digital soil mapping of soil classes using decision trees in central Iran

In response to the demand for soil spatial information and in order to improve natural resource management outcomes through the development of soil suitability maps, the acquisition of digital auxiliary data and matching it to soil data is increasing. With the harmonization of these data sets, through computer based methods, so-called Digital soil Maps (DSM) are increasingly being found to be as reliable as traditional soil mapping but without the prohibitive costs. In this paper, we attempt to develop a decision tree model for spatial prediction of soil classes (sub-great group taxa) in an area located in central of Iran, covering 780 km(2). Using the conditioned Latin hypercube sampling method, 186 soil profiles were selected and then allocated in two taxonomic orders: Aridisols and Entisols. Environmental predictors used in this study to represent soil forming factors were terrain attributes, ETM+ image and some existing categorical maps. Results showed that some covariates had more significant effect on the prediction of soil classes including topographic wetness index, MRVBF, NDVI, geomorphology maps. Furthermore, the results indicated that decision trees can predict spatial prediction of soil map units with reasonable accuracy (65%).