CAREER: Advancing a Participatory Framework for Decentralized Stormwater Management

Many stormwater systems in the United States were designed for past rainfall and water level conditions and are now strained by heavier storms, tidal backwater, rising groundwater, and aging infrastructure. These interacting stresses can overwhelm drainage networks, causing flooding, sewer overflows, and disruptions to homes, roads, businesses, and public services. Decentralized gray and green measures, such as rain gardens, permeable surfaces, and small detention facilities, can reduce these risks, but they are often planned separately without fully considering compound hazards, system interactions, costs, or service reliability across neighborhoods. This Faculty Early Career Development Program (CAREER) project develops an AI/ML-accelerated, stakeholder-informed socio-hydrologic decision framework for evaluating and optimizing decentralized stormwater portfolios under changing environmental conditions. By helping engineers, planners, and residents compare stormwater options based on performance, cost, recovery, and service reliability, the project serves the national interest by strengthening infrastructure resilience, reducing flood-related losses, and supporting more transparent decision-making. Educational activities will train students and practitioners in hydrologic modeling, data science, responsible use of AI/ML tools, digital twins, and community-informed infrastructure planning. The research has three integrated aims. First, it develops and validates metrics for physical resilience, including system capacity, adaptability, and recovery, and for service reliability, including risk reduction and recovery support across neighborhoods, under combinations of rainfall, tides, sea level rise, and elevated groundwater. Second, it combines stakeholder surveys, behavioral choice experiments, agent-based modeling, and optimization to determine when and why community priorities change the siting, sizing, and selection of gray and green infrastructure portfolios compared with hydraulic only designs. Third, it trains AI/ML-accelerated surrogate models using detailed compound hazard simulations and integrates them with the socio-hydrologic decision methods in a participatory digital twin testbed. This testbed will enable rapid exploration of tradeoffs among cost, resilience, and service reliability. The project will produce transferable methods, open-source software, reusable data sets, and teaching materials for resilient stormwater planning in coastal and urban communities. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.