Spatial modeling of groundwater capacity using a combination of maximum entropy method and random forest method in GIS environmentCase study: Ardabil Ghorichay catchment

Groundwater is one of the most important natural resources in arid and semi-arid regions. The purpose of this study is to identify areas that have groundwater capacity and to prioritize the factors affecting it. In this study, 11 indicators affecting groundwater capacity including Slope, Elevation, Aspect, Distance from River, Drainage Density, Distance from Fault, Topographic Wetness Index, and Topographic Position Index, lithology, Land use and Relative Slope Position were used. 30% of the totals of 230 wells were randomly placed in the validation data group and 70% in the training data. To prioritize the effective factors and zoning of groundwater potential in Ghorichay watershed, the random forest method was used using ArcGIS and to evaluate the model of relative performance curve (ROC) and Area Under the curve surface (AUC). The results showed that the groundwater capacity of about 8% of the watershed is higher at the outlet of the watershed. According to the VIP diagram, the TWI layer with a value of 0.329 and the distance from the river layer with a value of 0.175 was the most and the least influential factors on groundwater capacity, respectively. The area below the AUC curve showed an accuracy of 87% in the training phase to identify areas with groundwater potential. The result of this study can be used in groundwater management in the Ghorichay watershed.