Environmental Resilience, Climate and Sustainability

ABSTRACT

Urban flooding can occur when short-duration rainfall produces runoff faster than local drainage and storage systems can safely accommodate it. SINK (Smart Intelligent Network for Rainwater Capture and Flood Mitigation) is proposed as a closed-loop infrastructure framework coupling short-horizon rainfall/runoff-risk prediction with controllable local storage. Forecast-informed real-time rainwater-harvesting control is already established; therefore, SINK is not presented as a first-of-its-kind smart-tank concept. Its proposed research contribution is the explicit integration of local drainage-state feedback, uncertainty-aware emergency-capacity allocation, distributed storage control, a hard hydraulic safety supervisor, and causal benchmarking that separates passive storage, automated control and AI-assisted optimization. The intended prediction horizon is 60 min with a 5-min operational control interval, subject to catchment travel time and forecast skill. A synthetic numerical proof-of-concept uses a 1-ha catchment, 200 m³ storage capacity and a 45-mm five-step rainfall event. Under the stated assumptions, peak discharge is 0.200 m³/s for the no-storage baseline, 0.133 m³/s for passive storage, 0.120 m³/s for predictive rule control, and 0.100 m³/s for an illustrative AI/MPC case. These values demonstrate internal feasibility only. Calibrated SWMM simulation and field measurements are required before any empirical effectiveness claim

Keywords: Urban Flooding; Rainwater Harvesting; Artificial Intelligence; Real-Time Control; Model Predictive Control; Iot; Detention Storage; SWMM; Uncertainty; Climate Resilience