چكيده لاتين
Soil is one of the main elements of the life of living organisms on the planet. Undoubtedly, soil is the most important source of food production for humans on a global scale, however, many factors threaten the quantity and quality of soil, Water erosion has affected about 99 billion hectares of land in the world and is one of the most widespread factors threatening soil in terms of quantity and quality. Among the types of water erosion, gully erosion is its advanced and acute form, which causes a large amount of soil to be wasted, As a result, the amount of sediment produced in this type of erosion has a significant volume. Therefore, the study, identification and zoning of areas prone to ditch erosion has become very necessary. The purpose of this research is to identify the factors affecting the occurrence of gully erosion, forecasting and zoning the probability of gully erosion in the Alamarvdasht watershed in Fars province using Maxent and Random Forests models. The position of the formed trenches was registered using satellite images, Google Earth software and Global Positioning System (GPS). The influential variables in the erosion of the ditches of the Alamarvdasht watershed include topography-index indicators (height, slope, Aspect, SPI, TWI ),
ArcGIS and SAGAGIS software. Normalized Diference Vegetation index (NDVI) and land use have been prepared using data from Sentinel 2 and Landsat in R and ArcGIS software environment. distance from the roads and the distance from the Stream using the Euclidean distance function, geological map (extracted from the geological map of Iran), the amount of precipitation for the thirty-year period from 1351 to 1381, (statistics of the National Meteorological Organization) Soil type and soil erodibility were also prepared in ArcGIS software using soil samples collected from the study area. After assigning the values related to the independent indicators to the points taken from the ditches, modeling for zoning or predicting areas prone to ditch erosion in the study area, in the GIS & R software environment using Maxent and Random Forests statistical models were done. The mentioned models have considered 70% of the data as training data and 30% of the data as test data in the modeling process. Then, the accuracy of the implemented models was checked based on the RMSE, MAD, MSE and R_Square accuracy evaluation indices, and as a result, 2 ditch erosion zoning maps of the Alamarvdasht watershed extracted from the 2 mentioned models were prepared in the ArcGIS software environment. and have been examined and compared. In the meantime, the Maxent model has a higher accuracy than the Random Forests models by registering a value of 0.997 for the R_Square index.
