Evaluating the Potential of Digital Soil Mapping Method to Map Soil Types in Bac Ninh Province

Abstract

     The multinomial logistic regression (MLR) model has been being widely used for soil mapping because it could help save time and costs for collecting and analyzing soil data points compared to conventional methods. This research aimed to assess the potential of mapping soil types in Bac Ninh by assessing the performance of Multinomial Logistic Regression (MLR) model in predicting soil types. Nine predictive variables were derived from the ancillary data including land use, altitude, slope, NDVI, PVI, RVI, Topographic Wetness Index and SAGA Wetness

Index. MLR model was constructed to predict soil classes at 2 levels: WRB-Reference Soil Group and intermediate level of Soil Group between Reference Soil Group and the full WRB soil name. The map quality was evaluated by the soil map purity estimated with an independent validation dataset. The diversity indices were calculated to assess the information content of the resultant maps. Assessment of the model’s prediction was based on the soil map purity, the Shannon’s entropy and a combined index. MLR yielded high map purity at the level of Reference Soil Group. When the taxonomic level changed from Reference Soil Group level to intermediate level, the map purity decreased while the value of the diversity indices increased. Therefore, soil mapping using MLR in predicting Reference Soil Group might be more efficient. However, at intermediate level, the model predicted higher diversity of soil map and thus the informative value estimated by the combined index was higher.