H214-01
Combined modelling of US fluvial, pluvial, and coastal flood hazard under current and future climates
Wednesday, 16 December 2020: 19:00
Virtual
Oliver Wing1, Niall Quinn1, Christopher Sampson1, Andrew Smith Dr1, Jeison Sosa1, Mike Amodeo2, Edward Joseph Kearns2, Sharai Lewis-Gruss2, Jeremy R Porter2, Paul D Bates3, Jeffrey C Neal3 and Guy Schumann4, (1)Fathom, Bristol, United Kingdom, (2)First Street Foundation, New York City, United States, (3)University of Bristol, School of Geographical Sciences, Bristol, United Kingdom, (4)Remote Sensing Solutions, Pasadena, United States
Abstract:
A consortium of scientists across academia and industry have come together to produce the first consistent and complete picture of flood hazard across the contiguous US. We simulate all major flood perils under both present and future climate conditions based on a suite of state-of-the-art physical models. Flood inundation models are executed at ~30m resolution (downscaled to ~3m, depending on local data availability) with a 2D shallow water model built using openly available data. Inland flood modelling has been significantly improved using recent advances in channel schematization, extreme flow estimation, and defence representation. Storm surges at the coast are climate-adjusted to represent sea level rise up to 2020 and changing (sub)tropical cyclone behaviours using hurricane models and the CMIP5 climate model ensemble. Fluvial and pluvial boundary conditions are similarly updated to 2020 with these atmospheric models, employing calibrated HBV runoff models for the former. Statistical techniques are employed to sample plausible flood events at coasts where both inland and coastal flood processes can interact. The climate models are also used to project US flood hazards out to 2035 and 2050 due to climate change under the RCP4.5 scenario.
We validate the inundation maps against local-scale models from FEMA and the Iowa Flood Center (IFC), as well as against available claims data from the NFIP. The 100-year flood maps obtained similarity scores (accounting for under and over prediction) of 78% and 87% against FEMA and IFC models, respectively. The 5-year flood map was 69% similar to equivalent data from the IFC. Further, the 5-year flood extent reproduced the building flood exposure evident from the rate of NFIP claims. It is likely that the US-wide flood model can therefore represent flood hazard within the error bounds of local-scale modelling skill.
We find 70% more US buildings are at risk of a 100-year flood than stipulated by FEMA, with climate change inflating this exposure by a further 8% within the next 30 years. Newly identified risk is found across the nation, but particularly in inland western and southeastern states. Climate change impacts are most widely felt in the Gulf and East coasts. The model is openly available for anyone to gain a detailed understanding of US flood risk via https://floodfactor.com/.