Investigation and Comparative Assessment of Surface Water Quality for Drinking Purposes by Using Relief Algorithm, GIS, and Machine Learning: A Case Study of Mahanadi River Basin, Odisha (India)

Abstract

Surface water is the best source of drinking water available. However, climate change, over-pumping, and a variety of contaminants have all led to the depletion of this valuable resource. Conducting surface water quality assessments for home usage, especially drinking water, is essential to safeguarding human health and effectively managing resources. In this study, this work has highlighted an evaluation of surface water quality of river Mahanadi, Odisha, for drinking purposes using Relief Algorithm (RA) based WQI (RA-WQI), with reliability-based MLs (Machine Learning) such as Weight of Evidence (WOE) have been employed. For this, water samples from 19 locations were taken for a period of 2018-2023, to test 20 physicochemical parameters in the selected sampling sites. The findings indicated that although pH changes, the water is alkaline and its value spanned from 7.73 to 7.9. The concentration of coliform and TKN is found to be higher at all locations. The highest levels of Cl- and SO42- are located close to the downstream area. Based on the results, anions and cations are observing a shift in the trend, i.e., Fe2+ > B+ and Cl- > SO42- > NO3- > F- respectively, throughout the occupied duration. Further, the calculated RAWQI revealed that 63.16% belong to poor water quality while 31.57% of sites come under the zone of excellent water. However, 5.26% of samples indicated an unsuitable water class. The analysis primarily revealed that at 8 samples, the main cause could be deterioration of domestic water, illegally dumped municipal solid waste, and agricultural runoff were the leading sources causing adulteration of the river’s water quality. As a result, a renowned ML models, such as WOE, were adopted and it suggests location SP-(9) was the most polluted in comparison with other locations, followed by SP-(8), (19), and (2) respectively. Following this, the analytic findings also suggests from the highest RA-WQI values that consists of 488, 243, 277 and 285 at this location. However, it was relevant that the degree of pollution at these stations was more closely linked to a wide range of expanding human activities, such as excessive water use, fertilizer effects, agricultural runoff, and industrial activity in and around the river corridor. According to the drinking water quality indices, the surface water in the area under investigation is classified as suitable for human consumption. Thus, the results illuminate the preservation and distribution of drinkable and irrigable surface water supplies, and provide decision-makers with a valuable resource for implementing successful surface water protection strategies in the area under study.