Artificial Intelligence Applications in Environmental Management: Integrating Stormwater and Wastewater Systems

Abstract

The accelerating complexity of urban water management demands a paradigm shift from conventional reactive approaches towards proactive, data-driven governance. This article presents a comprehensive review and synthesis of artificial intelligence (AI) applications at the nexus of stormwater and wastewater management, examining how machine learning, deep learning, Internet of Things (IoT)-enabled sensing networks, digital twin frameworks, and AI-driven decision support systems are reshaping environmental management practice. Drawing on 94 peer-reviewed studies published between 2018 and 2025, we document performance benchmarks across four core application domains: flood forecasting and early warning, wastewater treatment process optimisation, combined sewer overflow prediction, and integrated catchment-scale water quality modelling. Our meta-analysis reveals that ensemble deep learning models consistently outperform standalone methods, reducing root mean square error (RMSE) for short-term runoff prediction by 28–47% relative to physics-based benchmarks. Transformer-based architectures demonstrate exceptional capacity for multi-variate, long-horizon forecasting where temporal dependencies span multiple rainfall seasons. We further identify critical barriers to operational deployment, including sensor data heterogeneity, model interpretability deficits, regulatory inertia, and infrastructure funding asymmetries between high- and low-income municipalities. A proposed Integrated AI Environmental Management (IAEM) framework is presented that harmonises real-time sensing, adaptive control, stakeholder engagement, and regulatory compliance pathways. The article concludes with a research agenda prioritising federated learning for privacy-preserving multi-utility data sharing and physics-informed neural networks for improved generalisation under climate-change-altered hydrological regimes.