| چکیده انگلیسی مقاله |
Soil erosion by water is a significant problem in arid and semi-arid areas of large parts of Iran. Water erosion is one of the most effective phenomena that leads to decreasing soil productivity and pollution of water resources. Our case study is located in the Southern Isfahan province, in central of Iran. The area is severely affected by different type water erosion such as gullies, rills and badlands. The intention of this research is to predict the spatial distribution of the different water related erosion type susceptibilities using a probabilistic Maximum Entropy Model (MEM) approach based on the following environmental layers: lithology, soil textures, land use, precipitation, Normalized Difference Vegetation Index (NDVI) and topographic indices derived from a SRTM DEM with 30 m spatial resolution. In order to validate the stochastic modelling approach we divided the entire sample in a train (70%) and test (30%) dataset. We validated the model performance using the Area Under Curve (AUC) value. The model yield good (rill and gully erosion) to excellent (badland) results for both train and test data. The spatialized prediction of susceptibilities for rill, gully and badland erosion show that in total more than 40% of the study area affected by water erosion processes (4.8% rill erosion; 23.4% gully erosion and 17.9% badland erosion). The knowledge of susceptible areas is crucial for a proper land management and related soil conservation measures to guarantee a sustainable land use. |
