توصيفگر ها :
فرسايش خندقي , مدل آنتروپي , مدل دمپستر-شفر , مدل وزن شاهد , مدل هاي احتمالاتي
چكيده انگليسي :
Soil is one of the most important resources for the continuation of human life, which today has suffered a lot of destruction due to the increase in population and lack of proper management, especially gully erosion, which destroys a significant amount of land every year. Therefore, to prevent further destruction, it is necessary to manage and predict the sensitivity of the land to gully erosion. Different methods are used to predict the sensitivity to gully erosion, including bivariate and multivariate, probabilistic, and statistical methods. In this research, zoning was done with Weight of Evidence (WOE), Dempster-Shafer (DS) and Entropy models in Semiram watershed. At first, with reviewing aerial photographs, Landsat 8 images, Google Earth images and field visits, a distribution map of ditches was prepared. Then, the layers of factors affecting gully erosion were prepared in ArcGIS 10.8. These indices include two main categories of topographic indices and environmental indices. Topographic indices include: cross profile curvature indices, cross section curvature, convexity index, overall curvature, vertical distance index to the channel network, watershed area, waterway power index, The length factor - watercourse slope, topographic humidity index, height, slope, direction, sunlight analysis index and environmental indicators included: maps of soil type, soil depth, lithology, land use and the rain map. After preparing the layers, the modeling process was done using the mentioned models. In order to validate and determine the best model in terms of performance, the gully erosion samples were randomly divided into two groups of training or calibration (70%) and testing or validation (30%) based on location. After the modeling process, based on the samples of the training group, the accuracy of the model was determined by calibration data and receiver operating characteristic curve (ROC) and area under this curve (AUC) method. that the value of the area under the curve for Weight of Evidence, Dempster Shafer and entropy models was obtained as: 0.886, 0.875, 0.882 respectively. According to the results obtained from the entropy model, which had the best performance among other models, the factors of soil type, altitude, vertical distance to the waterway, lithology, rainfall, land use, and vegetation have the greatest effect on the sensitivity of gully erosion, respectively in the study area.
