Authors : Taqveem Ali Khan

Volume/Issue : Volume 8 - 2023, Issue 11 - November

Google Scholar : https://tinyurl.com/2cs5s7pt

Scribd : https://tinyurl.com/4e8ec6xn

DOI : https://doi.org/10.5281/zenodo.10250605

Abstract : This study explores the application of regression models and artificial neural networks (ANNs) in predicting Total Dissolved Solids (TDS) in groundwater within two distinct regions of Aligarh city - the Northern Area Samples (NAS) and the Southern Area Samples (SAS). It aims to identify the key predictors of TDS in both areas and to compare the effectiveness of the two modelling approaches. In the NAS, sulphate, bicarbonate, sodium, and chloride were found to be the major TDS predictors, with the strongest being sulphate. In contrast, the SAS showed sodium, magnesium, potassium, and chloride as the main predictors, with sodium as the most influential. The ANN models displayed strong validity with high R square values between observed and predicted neurons. The study concluded that the ANN predictive models for TDS produced more accurate results than multilayer regression models, thereby demonstrating their broader applicability in groundwater quality characterisation and predictive modelling. The findings of this study can contribute to more effective water resource management strategies, especially in areas heavily reliant on groundwater.

Keywords : Predictive Modeling, Water Resource Management, Urban Sprawl,