An Empirical Evaluation of Machine Learning Algorithms for Groundwater Quality Classification

Abstract

Groundwater is an effective monitoring system is essential, one of the most vulnerable resources. The use of spatial data to measure spatial changes in groundwater one of the most key things of soil monitoring. As a result, the most important water constituents based on groundwater characteristics is critical for an effective soil monitoring programmed the development of an efficient reference system that estimates. We evaluated the performance of neural network (NN)-based algorithms and event prediction models (EPM)) to estimate the severity of SS in some Indian regions throughout this study. Using 16 years of We developed a regional and local model remote sensing dataset to estimate the SS of the entire Indian basin and each catchment in the study area. Based on EPM and NN regional models had accuracy and SS of 88%, 96%, 88%, and 87%, The estimation and SS outperformed both the regional and spatial NN by 50-84% and 71-84%, whereas the local model was the empirically derived model, respectively. Consequently, according to the findings, machine learning methods should be used to accurately and continuously monitor groundwater quality parameters. In complex topography of India and other similar land classifications.