M.S. Thesis Defense Seminar of Jack Baer

Design Storm Underestimate Flood Hazard and Risk Derived from Stochastic Storm Transposition

Flood hazard and risk data are essential to flood management and resilience efforts, including development plans, construction requirements, and insurance mandates. However, many current datasets rely on design storms, which use simplified representations of extreme rainfall and assume a linear relationship between storm and flood frequency. We investigate whether these simplifications could lead to underestimates of flood hazard and risk, leveraging stochastic storm transposition (SST) to generate 5,000 years of synthetic storms with realistic space-time structures for a 2,800 km² watershed in eastern North Carolina. We advance the existing SST framework by transposing tropical cyclones (TCs) non-uniformly in order to maintain their landfall timing and to preserve realistic rainfall fields. We then simulate the watershed’s response to both TC and non-TC events using a computationally efficient hydrodynamic model, generating probabilistic hazard maps and risk estimates from the model output. Finally, we compare our probabilistic hazard and risk datasets against those produced using traditional design storms. We find that relative to SST, the design storm approach underestimates the return periods of floods by an average of 19 to 36 years, and expected annual damage by approximately 65 percent. We attribute these discrepancies to nonlinearities in the relationships of storms, hazards, exposure, and risk. Our findings demonstrate that probabilistic methods and realistic rainfall fields are crucial to flood risk assessment, suggesting that current approaches may not adequately prepare communities for flooding and emphasizing the critical need for FEMA’s Future of Flood Risk Data initiative.