Assessing Multi-Criteria Decision Analysis Models for Predicting Groundwater Quality in a River Basin of South India

Abstract

India is the largest consumer of groundwater in the world, and it suffers from a groundwater crisis due to the overexploitation of groundwater and the deterioration of its quality at an alarming rate. Rapid urbanization, a growing population, and mismanagement are major driving forces behind these groundwater issues. Thus, increasing problems of water scarcity and water-quality deterioration threaten the sustainability of the water supply. This necessitates the development of novel approaches to assess prevailing groundwater quality scenarios at a large scale, which can help protect this vital freshwater resource from contamination. In this study, for the first time, the effectiveness of three Geographical Information System (GIS)-based Multi-Criteria Decision Analysis (MCDA) models (i.e., ‘Unit Weight’, ‘Rank Sum’, and ‘Analytic Hierarchy Process’) was explored for predicting groundwater quality in a river basin of Southern India. The seasonal concentrations of groundwater quality parameters, viz., Cl−, TDS, TH, F−, NO3−-N, Na+, Mg2+, Ca2+, K+, and SO42−, were considered for generating their thematic layers. Each thematic layer was classified into suitable feature classes based on the WHO guidelines for drinking water. The thematic layers and the feature classes of individual groundwater quality parameters were assigned relative weights according to the theories of the three MCDA models mentioned above. These thematic layers were then aggregated in GIS to develop Groundwater Quality Index (GQI) maps of the study area for pre-monsoon and post-monsoon seasons. Furthermore, the accuracy of the developed GQI maps was validated using relative operating characteristic curves. The results of the validation indicated that the GIS-based Analytic Hierarchy Process (AHP) model outperformed with prediction accuracies of 71.4% in the pre-monsoon season and about 85% in the post-monsoon season. However, the performances of the Unit Weight and Rank Sum models were found to be average with prediction accuracies varying from 68% to 63% and 64% to 68%, respectively. Thus, the GIS-based AHP model can serve as a reliable scientific tool for predicting seasonal groundwater quality at a river basin scale. It can be very helpful to the policymakers for devising viable management strategies for groundwater protection as well as for ensuring a sustainable water supply.