AI-assisted design of nature-based solutions for integrated flood and pollution management

Flood risk and water pollution are escalating challenges under climate change and rapid urbanisation. Conventional grey infrastructure provides limited flexibility and sustainability, while nature-based solutions (NbS)—such as wetlands, riparian buffers, and urban green corridors—offer multiple co-benefits by reducing flood hazards, improving water quality, and supporting biodiversity. Yet, designing NbS remains challenging due to diverse local conditions, competing objectives, and deep uncertainties.

This project will develop an AI-assisted framework for NbS design and evaluation, combining advanced predictive modelling with multi-objective optimisation. Methodological approaches include:
•    Data integration of remote sensing, hydrological, water quality, and land-use datasets.
•    AI-based modelling, leveraging LSTMs, Transformers, and graph neural networks to predict flood and pollution dynamics, coupled with process-based models such as SWAT and HEC-HMS.
•    Multi-objective optimisation using evolutionary algorithms and reinforcement learning to balance flood risk reduction, pollution mitigation, and ecological co-benefits.
•    Uncertainty quantification through Bayesian inference and factorial analysis to ensure robustness under future climate scenarios.

Potential research directions include scaling NbS design from local to basin levels, evaluating socio-economic trade-offs, and exploring adaptive pathways under climate change. The project will deliver a transferable decision-support tool for policymakers and planners, advancing AI applications in sustainability science and contributing to resilient, cost-effective solutions for water security.