Journal Article

Indexical and artificial neural network modeling of the quality, corrosiveness, and encrustation potential of groundwater in industrialized metropolises, Southeast Nigeria

Egbueri, Johnbosco C, Agbasi, Johnson C, Nwazelibe, Vincent E, Unigwe, Chinanu O

2023

Overview

Adequate evaluation, monitoring, and prediction of groundwater resources are essential because humans heavily rely on groundwater for drinking, domestic, and industrial needs. The current study aimed at evaluating the quality of groundwater for drinking and industrial purposes in Awka and Nnewi urban metropolises (southeastern Nigeria) using indexical and artificial neural network (ANN) methods. The temperature of the studied groundwaters was found to range from 23 to 28 °C. The pH values revealed that the waters are acidic, though the groundwaters in Awka are acidic more than those in Nnewi. The majority of the analyzed physicochemical parameters (conductivity, total dissolved solids, Cl, SO4, HCO3, and Ca) examined were found to be below acceptable standard limits. In the metropolises, integrated water quality index (IWQI) classified over 75% of the groundwaters as unfit for drinking. Except for the Revelle index (RI), which classified 70% of the water samples within the Awka metropolis as slightly affected by salinization and 90% as strongly affected by salinization in the Nnewi metropolis, all other corrosivity and encrustation potential indices (Larson–Skold index (LSI), chloride–sulfate mass ration (CSMR), Langelier index (LI), aggressive index (AI), Ryznar stability index (RSI), and Puckorius (PSI)) utilized classified all the groundwater as having a high corrosivity. This demonstrates that the groundwaters in both metropolises have higher corrosion potential than encrustation potential. Additionally, the eight ANN models produced in this study function admirably. The ANN models performed well in the order IWQI > LSI > RSI > CSMR > LI > AI > PSI > RI according to R2 values. High performance of the models was validated by the R2, residual error, relative error, and sum of square error values. The findings of this paper would offer valuable insights for sustainable and strategic management of the groundwater resources.

• Publisher: Springer Nature B.V
• Subject: Artificial neural networks / Beverage industry / Corrosion potential / Deixis / Dissolved solids / Drinking / Drinking behavior / Drinking water / Encrustation / Errors / Groundwater / Groundwater quality / Indexes / Neural networks / Physicochemical properties / Rationing / Salinization / Strategic management / Total dissolved solids / Values / Water analysis / Water quality / Water quality standards / Water resources / Water sampling / Water supply