Determining prone areas to gully erosion and the impact of land use change on it by using multiple-criteria decision-making algorithm in arid and semi-arid regions

Gully erosion is one of the types of water erosion which causes the loss of fertile soil, depletion of soil moisture, and so on. Due to the importance of the identification of the areas that are prone to this type of erosion in the management and appropriate control of this phenomenon, the present study has tried to determine the prone areas to gully erosion in the southern regions of Fars province using Multiple-criteria decision-making (MCDM). For this purpose, First, fuzzy maps are prepared for each parameter with membership functions, and then the analytic hierarchy process (AHP) method was implemented to determine the weight coefficients of each parameter by pairwise comparison based on their importance in determining areas prone to gully erosion. The map of the gully-prone areas was prepared by multiplying the weight coefficients in the fuzzy layers. Since there are too many influential parameters in gully erosion (19 parameters), the feature selection algorithm is carried out to select the most prominent parameters. It is also decided to apply best subsets regression to determine the strength of the relationship between various parameters as well as their effects on gully erosion. Finally, land use changes (i.e., urbanization) and their effect on gully erosion are investigated. The results showed that the areas located in the central regions of the study area (about 15%) are more susceptible to erosion. The results of the feature J48 algorithm (i.e., Best-First and Greedy-Stepwise method) showed that the most important effective parameters in gully erosion are normalized difference vegetation index (NDVI), slope, topographic wetness index (TWI), altitude, terrain ruggedness index (TRI), lithology and land use. Then, the gully erosion map obtained based on these data. The comparison of the map prepared by selection data and the map prepared by all data show that the gully erosion map with selection data has high accuracy (AUC = 0.80%). The best subsets regression showed that the NDVI, TWI, altitude, TRI, land use variables with R-2 = 0.79, AdjR(2) = 0.76 and C-p = 1.6 are the most important variables in the study of gully erosion. In the end, it is concluded that land use change (i.e., urbanization) and non-compliance with soil protection principles have increased the erosion rates in the region. These results can help the researchers to identify areas prone to erosion and slow down the soil erosion process in the area by taking soil conservation practices.