Adaptive flood risk management: A decision support system integrating deep learning, digital twins, and economic risk assessment

Floods are among the most destructive climate-related disasters, with their frequency and severity increasing due to climate change and urban expansion. In response to rising claims and insufficient adaptation measures, insurers are progressively withdrawing from high-risk areas, thereby shifting the responsibility for risk management to businesses and municipalities, who must either implement their own solutions or resort to self-insurance. Effective flood risk management requires accurate forecasting, robust financial impact assessments, and decision support systems (DSS) to inform adaptation strategies. Within the framework of the European Union (EU) Floods Directive, this study develops an integrated, AI-powered DSS that combines deep learning-based flood forecasting (ConvLSTM models), economic vulnerability modelling (Joint Research Centre methodology), digital twin simulations, and predictive analytics to support data-driven adaptation planning. The framework was initially applied to assess pluvial, fluvial, and coastal flood risks in the coastal city of Nice, France, and subsequently extended to over 100 public and private sites across three urban municipalities. The findings demonstrate that this methodology improves the accuracy of risk assessments and provides a structured basis for capital allocation, insurability evaluation, and the optimization of adaptation investments. The multi-site deployment revealed significant governance, legal, and behavioural constraints, with public authorities and family-owned businesses responding differently despite comparable risk information. This research shows that integrating AI and digital twin technologies advances the EU Floods Directive’s objectives by enhancing risk mapping, preparedness, and transparency, while supporting public–private partnerships and extending protection to vulnerable populations at risk of losing insurance coverage.