Quantification of the Impact of Uncertainties in Flood Risk Projections Across the Delaware River Basin

The Delaware and Raritan River Basins in the U.S. Northeast have experienced significant flooding over recent decades. Understanding flood risk is crucial for improving resilience and community planning, both under current and future conditions. In this study, we assess projected flood hazards by downscaling climate model data to a spatial resolution suitable for hydrologic modeling. Using eight bias-correction and statistical downscaling techniques, we process 100 members of the Community Earth System Model version 2 Large Ensemble (LENS2). We then input the downscaled data into the Hillslope Link Model (HLM) TETIS hydrologic model to generate continuous streamflow simulations from 1981 to 2100. These simulations are used to create inundation maps with the Height Above Nearest Drainage (HAND) method. We overlay the inundation maps with building data (2020) to estimate the number of impacted buildings and the associated socioeconomic impacts. Our analysis reveals that, in the near future, flood peak changes are generally small, with some areas showing decreases of 10%–40%, and most areas showing increases of around 5%. In the far future, on the other hand, more extreme changes are expected, with upstream areas seeing increases of 10%–20%, while areas close to the outlet seeing decreases of 20%–50%. We also find significant uncertainty in flood hazard projections, with variability depending on the downscaling method used. Despite small projected changes in inundated areas in the near future, impacts on buildings and populations are expected to increase, with population impacts ranging from 5% to 18% for extreme annual exceedance probabilities.