bioclim algorithm

GGroundwater is considered as one of the most remarkable sources of fresh water. The aim of the current research is to apply four machine learning models of Random Forest (RF), Support Vector Machine (SVM), Bioclim, and Domain to groundwater potential mapping in Kahurestan watershed, Hormozgan province. The innovation of the research is to employ Bioclim and Domain algorithms to groundwater potential simulation, to compare them with the two techniques of RF and SVM and to combine these four models by an innovative and new equation. For this purpose, 11 criteria including slope percent, slope aspect, plan curvature, profile curvature, soil adjusted vegetation index (SAVI), modified normalized difference water index (MNDWI), slope length and steepness factor (LS), stream power index (SPI), topographic wetness index (TWI), land use, and distance to streams were considered. Also, the data of 113 high-discharge wells were used for simulation (70%) and validation (30%)processes. The collinearity test was performed prior to modeling which indicated that there was no relationship between the variables. Evaluation of the modeling performance with the ROC curve showed that all four methods used had very good accuracies and AUC values higher than 90% for prediction. The survey on the weight of the criteria based on the RF method demonstrated that the land use/cover and distance to streams criteria has the highest weight. The final map revealed that 21.4% of the area under study has good groundwater potential.