Applications of deep learning in water quality management: A state-of-the-art review

Abstract

• Water quality (WQ) management using deep learning (DL) approaches reviewed critically. • DL-based forecasting of WQ parameters reviewed for different water bodies. • Publications on hybrid DL models for WQ management assessed comprehensively. • Importance of the IoT and cloud computing towards DL-based WQ management outlined. Excellent water quality (WQ) is an indispensable element in ensuring sustainable water resource development. It is highly associated with the 3rd (good health and well-being), the 6th (clean water and sanitation), and the 14th (life below water) listed items of the United Nations’ Sustainable Development Goals. Thus, policymakers have always been seeking strategies to manage WQ efficiently. Recent advancements in computational technologies have created enthusiasm for using artificial intelligence, particularly deep learning (DL), in WQ management. This review provides a comprehensive overview of the application of DL in WQ management, covering developments from 2011 to 2022, in maintaining the temporal relevance of this review. In this paper, a brief description of different variants of DL models, including the recurrent neural network (RNN), long short-term memory network (LSTM), convolutional neural network (CNN), etc, are presented. The distinct approaches in the optimization, hybridization and relevant data pre-processing techniques suitable for the DL models, are also discussed. This is the first review paper that extensively discusses the application of DL models for forecasting WQ parameters in various water bodies, such as rivers, lakes, coastal areas, etc. The emergence of the Internet of Things (IoT) and cloud computing that revolutionized DL approaches in WQ management are also presented. This review paper serves as a complete easy guideline for the researchers in the field of DL-based WQ management. The findings of this review paper may help policymakers to enhance their decision-making process with the hope that regional environmental welfare can drastically be improved.