GIS-based Groundwater Spring Potential Modelling and Assessment Mapping in the the Omarak Watershed

Nowadays, in most countires, water supply in order to achieve the objectives of sustainable development is one of the most important challenges. Because of this, one of the important tools in the protection, management and exploitation of water resources, is to determin groundwater areas. Therefore, the purpose of this study, is to prepare the potential map of groundwater springs, using a well-known machine-learning model (i.e. random forest) and a statistical model (i.e. frequency ratio model) and comparing the efficiency of these methods in the Omarak watershed, Tehran Province. First, 18 factors influencing the emergence of springs including: lithological formations, the distance from the fault, fault density, elevation classes, slope percentage, slope direction, slope length factor, curvature maps, distance from the stream, stream density, maximum height, wetness index, relative slope position, soil texture, terrain roughness index, flow convergence index and land use cover were selected and their maps were prepared in the ArcGIS10.5 and SAGA systems. After the Multicollinearity test and classification of the effective layers, using the natural fracture method, then, the percentage of groundwater potential frequency in each layer obtained using the overlap of the distribution map of the springs with each of the layers. The relative operating characteristic (ROC) curve was used to evaluate the performance of the mentioned models and the area under the curve (AUC) of the random forest models and the frequency ratio were 88 and 72%, respectively. The results indicated that both methods are suitable to estimators to prepare the groundwater source potential map in the studied area. However, the random forest model with a higher area under the curve was introduced as a better method to identify and zoning the potential of groundwater springs.