Prediction Of Groundwater Level Using Advance Machine Learning Techniques

Climate change is a highly debated topic today, with various organizations conducting initiatives to understand its impact. The Intergovernmental Panel on Climate Change (IPCC) is recognized for publishing authoritative reports on climate change and its effects. In Chhattisgarh, a state where groundwater is the primary source of irrigation, the declining groundwater level is a significant concern. This study employs advanced prediction methods using Machine learning (ML) algorithms to address this issue. Rainfall and temperature data are utilized for training various models, such as Random Forest (RF), K Nearest Neighbor (KNN), Artificial Neural Network (ANN), and Support Vector Machine (SVR). We utilized the Regional Climatic Models (RCMs) to determine the potential impact of climate change. We conducted a study using 38 rain gauge stations, obtaining observed data from government organizations and simulated data from RCMs. Two bias correction methods were performed: power transformation (PT) and quantile mapping (QM). The most suitable circulation model was selected based on regression and Nash Sutcliff efficiency coefficients. Time series analysis was then performed to determine trends. Results showed that power transformation performed better than quantile mapping after bias correction and was therefore used for further research.Among all the ML models, the Random forest model gave the best results with minimum root mean square error, so the random forest was selected for predicting the future trajectory of groundwater.