Review of Groundwater Potential Storage and Recharge Zone Map Delineation Using Statistics based Hydrological and Machine Learning based Artificial Intelligent Models

Abstract

Abstract - Groundwater accounts for 63% of agriculture irrigation and 80% of household water supplies. In Many parts of the country, the water table is going down by 1 to 2 meters per year due to over utilization. It may result in up to a 20 percent decrease in food production. Given the huge impact this invisible resource has on the economy, environment, and society, we need to improve the scientific understanding, estimation, and governance of groundwater. The creation of groundwater’s possible storage area (GWPSZ) and regional recharge (GWRZ) zone maps can be helpful in this regard. They depend on statistics-based, machine learning (ML) based, and hybrid models. This paper reviews the work done by multiple researchers who have used geospatial techniques using satellite imagery sensing analytics in GIS, followed by AHP or Multi Influencing Factors (MIF) pairwise comparison to characterize, forecast GW levels, and generate GWPSZ and recharge GWRZ maps. We also reviewed the research on historical aquifer data using ML-based regression analysis, random forest (RF), supervised algorithms like support vector machine, nonparametric ML algorithm decision tree model, and ensemble hybrid multi-wavelet ANN models for the prediction of the GWL variability and storage loss/deceleration. Though some papers focused on the use cases like irrigation scheduling and predicting geothermal well locations or designing community cooling hubs, a comprehensive approach for village-level community water demand and supply assessment and decisionmaking is missing.