Document Type : Research Paper
Department of Reclamation of Arid and Mountainous Regions, Faculty of Natural Resource, University of Tehran, Iran
Kurdistan Agricultural and Natural Resources Research and Education Center, Kurdistan, Sanandaj, Iran
The University of California, Berkeley, USA
Land degradation is a global natural hazard that can be controlled by distinguishing susceptible areas. Although new approaches for determining areas prone to land degradation are necessary, spatial modeling of this hazard remains a challenge. This study aimed to investigate the efficiency of the weight of evidence (WOE) and evidential belief function (EBF) models for spatial modeling of land degradation in a semi-arid region in Iran. The trend of Net Primary Production (NPP) changes related to 2001-2020, obtained from MOD17A3, was taken into account to specify the inventory of land degradation in the study area. 120 random points were chosen as degraded points in areas with decreasing trend in NPP during 20 years. 70% of the dataset was randomly selected as a training set for the modeling step and 30% of them were selected as the testing set for the validation step. Fifteen geo-environmental factors including temperature, precipitation, slope, aspect, altitude, land use, normalized difference vegetation index, normalized difference salinity index, vegetation soil salinity index, normalized difference moisture index, visible and shortwave infrared drought index, electrical conductivity, and sodium adsorption ratio of groundwater, groundwater table, and annual depletion of groundwater resources were selected as influential factors or independent variables for modeling. The modeling process was done in ArcGIS software after calculating the values of EBF and WOE in excel. And finally, the efficiency of the models was analyzed using the area under the ROC curve. The findings illustrated that EBF with AUC = 0.72 had better performance for spatial modeling of land degradation in the Qazvin plain. Also according to the outputs of both models, north, northeast, northwest, west, southwest, and south of the Qazvin plain were susceptible to LD. The results of this research successfully suggested a new land degradation modeling method that can be used in different areas.