Forecasting Hurricane Impacts on U.S. Coasts

Extreme weather events like hurricanes cause loss of life, property damage, harm to the environment and disruption of local economies. In the US alone, extreme weather accounts for 1.8 trillion dollars in damages since 1980 ( NCEI, 2021 ). With the ongoing coastal development and climate change, costs will increase. It is therefore important to understand and predict the impact of these events so that planning and emergency measures can be taken.  

The NOPP Hurricane Coastal Impacts ( NHCI ) project's goal is to improve our understanding of the physical processes of flood and erosion hazards and their impacts. The project team consists of 10 working groups each dedicated to collect and aggregate topography and land use data, make meteorological forecasts, make in-situ or remote-sensed observations and to forecast hydrodynamic and morphodynamic impacts.  

The  Deltares -led Group (with  IHE Delft)  is one of three hazard and impact modelling teams. The group has made coastal impact forecasts during the hurricane seasons of 2022-2024. These are “research-grade” forecasts (not to be confused with National Hurricane Center operational forecasts) and will facilitate innovation in our ability to better prepare coastal communities for extreme weather events. 

In this storymap we outline the approach and results for three hurricanes: Ian (2022), and Idalia and Lee (2023)

The approach is to forecast the hazards of flooding and erosion, and impacts on buildings and roads with a suite of innovative models and workflows, developed at Deltares in collaboration with partners (see Figure on the right).

Our models are built using input from the other teams such as information on topography, bathymetry and vegetation, and is driven by forecasts of winds and pressure from COAMPS-TC ( Doyle et al., 2015)  for hurricane conditions.

The output consists of waves, water levels, flood depths and extents, as well as building damage and morphological change.

The workflow is managed by CoSMoS which was first successfully applied for risk assessments on the US West Coast ( Barnard et al., 2014 ), but is now used to produce real-time forecasts. 

The CoSMoS system consists of three numerical models:  

• HurryWave for the wave heights, periods and directions on the ocean scale.
•  SFINCS  for the total water levels (including tide and surge) and for the overland flooding depth and extent. 
•  XBeach  for the coastal erosion and breaching.  

Hurrywave (Van Ormondt et al., in prep) is based on the wave action equations and using a simplified numerical scheme, it computes wave heights and wave periods on the ocean basin scale about 50 times faster than conventional wave models. The model includes wave propagation and refraction, wave breaking, wave-wave interactions and wind wave growth and is driven by wind fields forecasted by COAMPS-TC. 

In this project we use SFINCS ( Leijnse et al. 2021 ) in a novel way to compute the surge in coastal waters starting at the 500 meter depth contour. The model is driven by forecasted winds and pressure from COAMPS-TC and tidal constituents from TOPEX/Poseidon. The water levels drive overland flooding which is computed in 16 individual domains stretching from the Mexican to the Canadian border. In each model the so-called sub-grid approach is applied so that we can compute the water levels on a rather coarse (200 meter) grid but output the flood depth at the high (5 meter) resolution provided by the topographic input data source.

XBeach - Dune Erosion and Breaching Model XBeach ( Roelvink et al., 2009 ) is used to compute the morphological change of erosion, accretion and breaching of coastal barrier islands. In this project we developed 700+ XBeach domains all along the sandy coastal areas in the Gulf of Mexico and the Eastern Seaboard. Because XBeach is quite computationally intensive we reduced the extent and resolution of the model as much as possible. 

Each of the model schematizations needs at least elevation and roughness fields as inputs. Within this project, the Coastal National Elevation Database ( CoNED , by the USGS) and the Continuously Updated Digital Elevation Model (CUDEM, by NOAA) are used as the main sources for the elevations. Besides that, landuse maps from the National Landcover Database ( NLCD , by the USGS) and vegetation maps by USGS are used to estimate the roughness.

For the real time predictions of total water levels, wave heights and morphological impacts, the models are forced by tidal water levels based on TOPEX/Poseidon and spatially varying wind, air pressure and precipitation fields provided by the Navy COAMPS-TC model ( Doyle et al., 2015 ). Since COAMPS-TC products only are available during during hurricane conditions, NOAA's GFS products are used in non-cyclonic conditions.  

Driven by these meteo forecasts the system was run continuously for the entire hurricane seasons (1 June – 1 November) of 2021 until 2024.  

In this storymap we will focus on the flooding, erosion and impacts of the most impactful hurricanes, Ian in 2022, and Idalia and Lee in 2023. 

Including Uncertainties

Forecasted hurricane tracks are inherently uncertain. They can take a different course, move slower or faster and change in strength. To obtain insight in the effect of these uncertainties it is useful to consider an ensemble of hurricane variations, and the resulting flood impact.

We generated an ensemble of N (=50) members, extracted the best track from the COAMPS-TC meteo model, applied DeMaria (2009) for cross-track and along-track position and intensity, and created parametric wind fields using Holland (2010). We ran the simulations on Amazon AWS EC2 using Argo workflows, and produced output for every ensemble member. These are aggregated to produce a 90% exceedance flood map.

Here you see the ensemble of tracks for three different days: 27, 28 and 29 August. You can see that the spreading narrows over time, and also that the uncertainty band of the water level forecast at Cedar Key narrows and improves relative to the observation as the hurricane is closer to land fall.

The forecasts generated by the NHCI project support proactive measures by communities and emergency management agencies. The detailed impact predictions enable targeted evacuation plans and infrastructure protection strategies.

Realtime and historic forecasts for selected hurricanes can be viewed in the webveier. Select a storm scenario in the top left and choose your output parameter using the radio buttons.

Moving forward, the project will focus on refining its predictive models, addressing uncertainties through ensemble forecasting, and enhancing the granularity of urban area analyses. These efforts aim to improve the accuracy and usability of hurricane impact forecasts.

The NHCI project represents a significant advancement in our ability to forecast the impacts of hurricanes on coastal areas. Through ongoing research and model development, the project contributes to the broader goal of enhancing coastal resilience against the increasing threat of extreme weather events.

DOI: 10.13140/RG.2.2.28728.76808