Ground Water Level Estimator With Back Propagation Neural Network Classification Using Machine Learning Approach

The groundwater is among the most priceless resources. The continual depletion of water is among the major factors impacting the development of the economy and society. The utilization of the water will depend on how easily it can be reached. The problem of water scarcity is a result of population growth and climate change. This would have increased the need for groundwater. However, there are differences in the groundwater's distribution. Three different types of models periodic, polynomial, and rainfall models that take more time and have less accuracy are used in a single pass learning process to predict groundwater. The data from the more than 100 observation wells in each of these districts were reviewed by these models, which subsequently produced seasonal models to illustrate the behavior of the groundwater. We state that the annual groundwater availability for use in the future is achieved utilizing a rainfall level predictor in our proposed method. The system created a precise classification in order to determine the estimated groundwater level with various variability. We estimated groundwater level in our suggested system at different groundwater well locations in India using a BPNN model. Additionally, we conducted sensitivity research using several feature combination scenarios, comparing the accuracy and reliability of the predictions that were generated. In this work, the usage and comparison of three data-driven models for predicting short-term groundwater levels is the main topic. The objective is to develop a new data-based method for forecasting groundwater levels that is extremely accurate and that can help water managers, engineers, and other stakeholders manage groundwater levels in a more effective and sustainable way