Variability of Best-Estimate Flood Depth Return Periods in Coastal Louisiana

Abstract

Estimates of surge-based flood depth exceedance curves are useful to inform flood risk management strategies. Estimated return periods associated with flood depth exceedances naturally vary over time, even under assumptions of stationarity, due to the irreducible randomness associated with storm events as new observations accrue with each passing year. We empirically examine the degree to which best-estimates of coastal Louisiana floodplains have changed over time and consider implications for risk management policies. We generate variation in estimated 100-year flood depths by truncating a historical data set of observed tropical cyclones to end in years ranging from 1980 to 2016, adopting three procedures for updating various inputs to an existing flood risk model using the truncated data set to identify which factors are most important in driving variation in risk estimates over time. The landscape used for modeling hydrodynamics is kept constant, allowing us to isolate the impacts of randomness in storm occurrence from other factors. Our findings indicate that the 100-year floodplain extent has substantially expanded in populated areas since 1980 due to these effects. Due to the low frequency at which flood maps are updated, it is possible that thousands of coastal residents are misclassified as being outside of the 100-year floodplain relevant to flood insurance rates and other regulations.