Identifying sources of soil classes variations with digital soil mapping approaches in the Shahrekord plain, Iran

Mapping the spatial distribution of soil classes is useful for proper soil and land-use management. This study investigates the ability of different digital soil mapping (DSM) approaches to predict taxonomic classes up to the family level in the Shahrekord plain of Chaharmahal-Va-Bakhtiari province, Iran. A total of 120 pedons were dug at various map units of a semi-detailed soil map with 750-m intervals. After pedons description, soil samples were taken from different genetic horizons. Based on the pedon descriptions and soil analytical data, pedons were classified up to the family level. Different machine learning techniques such as artificial neural networks, boosted regression tree, random forest and multinomial logistic regression were used to test the predictive power for mapping the soil classes. Overall accuracy (OA), adjusted kappa index and brier scores (BS) were used to determine the accuracy of the prediction. The model with the highest OA (i.e., the highest adjusted kappa) and the lowest BS values was considered as the most accurate model for each soil taxonomic level. Results showed that the different models had the same ability for the prediction of the soil classes across all taxonomic levels while a considerable decreasing trend was observed for their accuracy at subgroup and family levels. The terrain attributes were the most important environmental covariates to predict the soil classes in all taxonomic levels, but they could not display the soil variation entirely. This shows that the unexplained variations are controlled by unobserved variations in environment, which can be due to management over the time. Results suggest that the DSM approaches have not enough prediction accuracy for the soil classes at lower taxonomic levels that focus on the soil properties affecting land use and management. Further studies may still be required to distinguish new environmental covariates and introduce new tools to capture the complex nature of soils.