Explainable AI for Water Quality Monitoring: A Systematic Review of Transparency, Interpretability, and Trust

Water quality monitoring is essential for protecting public health, sustaining ecosystems, and achieving sustainable development goal 6 (clean water and sanitation). While recent advances in artificial intelligence (AI), particularly machine learning, have improved the accuracy and responsiveness of water quality assessment, the opaque nature of many AI models limits transparency, stakeholder trust, and regulatory compliance. Explainable AI (XAI) offers a viable solution by enabling human-understandable insights into model behavior. This article presents a preferred reporting items for systematic reviews and meta-analyses-guided systematic review of 60 peer-reviewed articles (2011–2025), sourced from Scopus, to evaluate the evolution, application, and effectiveness of XAI in water quality monitoring. Key techniques such as Shapley additive explanations (SHAP), local interpretable model-agnostic explanations, and counterfactual reasoning have been applied across random forest, XGBoost, and long short-term memory models. Results indicate a surge in XAI adoption post-2022, with dominant use cases in groundwater prediction, surface water quality forecasting, and real-time monitoring in Internet-of-Things-enabled smart cities. While SHAP remains the most widely used method, multimodal and hybrid frameworks are emerging to address challenges such as data heterogeneity and model complexity. The review identifies persistent barriers including computational scalability, lack of standardized evaluation metrics, and limited deployment in low-resource settings. It proposes future research directions for integrating XAI with digital twins, causal inference, and edge computing to achieve robust, transparent, and equitable water management systems.