Enhancing water quality prediction: a machine learning approach across diverse water environments

ABSTRACT

Rapid urbanization and industrialization have significantly contributed to the pollution and degradation of water bodies through urban runoff, industrial discharge, and inadequate wastewater treatment infrastructure. This study explores the role of machine learning (ML) as a decision-support tool for addressing water quality challenges across diverse environments, including surface water, groundwater, seawater, and wastewater. We evaluated the performance of seven ML models for predicting water quality parameters using data from six years (2014–2019) of the Melbourne Eastern wastewater treatment plant, encompassing energy consumption, climate variables, and wastewater characteristics. Among the models tested, the gradient boosting model demonstrated the highest predictive accuracy, achieving a statistical measure coefficient of determination R2 score of 0.75. Key findings highlight the importance of integrating climate and water quality data to improve prediction accuracy and identify critical water parameters for enhancing future models. This review provides insights into the applicability of ML techniques in water quality management and identifies potential avenues for further research in predictive modeling.