Predicting Water Quality with Artificial Intelligence: A Review of Methods and Applications

• Main Authors: Irwan D., Ali M., Ahmed A.N., Jacky G., Nurhakim A., Ping Han M.C., AlDahoul N., El-Shafie A.
• Other Authors: 55937632900
• Format: Review
• Published: Springer Science and Business Media B.V. 2024
• Subjects: Deep learning; Forecasting; Generative adversarial networks; Learning systems; Numerical methods; Potable water; Agricultural activities; Daily activity; Existing problems; Learning models; Modeling process; Performance; Prediction methods; Quality of water; Quality parameters; Water quality predictions; Water quality

The water is the main pivotal sources of irrigation in agricultural activities and affects human daily activities such as drinking. The water quality has a significant impact on various aspects and thus this review aims to addresses existing problems related to water quality prediction methods that have been found in the literature. We explore numerous quality parameters incorporated in the modelling process to measure the quality of water. Furthermore, we review the commonly adopted artificial intelligence-based models which have been utilized to forecast the water quality. 83 studies published from 2009 to 2023 were selected and reviewed based on their success in modelling and forecasting the water quality in multiple regions. We compared these articles in terms of parameters, modelling algorithms, time scale scenarios, and performance measurement indicators. This paper is beneficial to researchers that have interests to conduct future studies related to water quality forecasting. Additionally, we discuss a variety of modelling methods such as deep learning (DL) that have proven to boost the efficiency compared to traditional machine learning (ML) models. As a result, the hybrid-DL models were found to outperform other models such as standalone ML, standalone DL, and hybrid-ML. This study shows a significant limitation of the data-hungry DL models which require a big data size for modelling. Hence, at the end of this review study, we discuss the potential of some methods such as generative adversarial networks (GANs) and attention-based transformer to open the door for water quality prediction improvement. GAN has shown promising performance in other domains for synthetic data generation. The potential usage of GAN for water quality domain can overcome the limitations of lack of data and enhance the performance of the predictive models reviewed in this study. Similarly, transformer was found to be state of the art model for time series prediction and thus it can be good candidate to predict water quality. � 2023, The Author(s) under exclusive licence to International Center for Numerical Methods in Engineering (CIMNE).