Novelty and International Appeal Statement
Through understanding on spatio-temporal variability of groundwater level in tropical savanna climatic region with heavily stressed aquifers and its hydrological, geological, and climatic (HGC) controls, has immense role in long-term sustainable management of the resource. This study attempts to develop prediction model for GWL taking the dominant HGC control as explanatory variables using ML algorithm. To achieve the aforesaid objective, firstly the spatio-temporal variability of GWL in the study region was comprehended and the dominant HGC controls for observed spatio-temporal variability were identified using entropy approach. To the best of our knowledge this study is the first of its kind in groundwater variability prospective. Further, the grid-wise relationship among the groundwater level and the features in the dominant HGC control was established using Random Forest learning algorithm to make short- and long-term predictions of GWL with reasonable accuracy in the study region with limited data. The proposed framework is a novel attempt to characterize spatio-temporal variability in GWL in tropical regions with highly heterogeneous aquifers and predict the future trends in GWL at regional level.