Mapping groundwater potential zones using relative frequency ratio, analytic hierarchy process and their hybrid models: case of Nzhelele-Makhado area in South Africa

Abstract

Identification of groundwater potential zones (GWPZ) is as equally important as determining the most effective groundwater mapping models for efficient production of reliable GWPZ maps. In this study, relative frequency ratio (RF) with predictor value and analytic hierarchy process (AHP) models, including two RF-AHP hybrid models (HM-A and HM-B) were applied and compared to evaluate their effectiveness and accuracy in groundwater potential mapping. For HM-A model, RF was used to rank the subclasses of groundwater controlling parameter and AHP provided the weight of that parameter. For HM-B, it was vice versa. Eight parameters (slope, geomorphology, lithology, lineament density, drainage density, soil, rainfall and LULC) were considered for this assessment. The RF, AHP, HM-A and HM-B show the AUC value of 72.47%, 60.55%, 71.96% and 49.85%, respectively. The models based on RF-derived ranks show better accuracies (>70%), thus suggesting that ranks have more influence in the performance of models than weights.