Evaluating the numerical modeling of storm surges induced by hurricanes

Coastal zones are increasingly threatened by extreme sea level (ESL) events. Storm surges (i.e. sea level variations due to meteorological drivers) are one of the most hazardous components of ESLs, especially in regions prone to tropical cyclones. This study aims to explore factors affecting the performance of numerical modeling in simulating storm surges induced by hurricanes. The focus is on the tropical Atlantic region, covering the Caribbean Sea and the Gulf of Mexico. Several historical hurricanes causing severe coastal impacts are simulated. The skill of the simulations to reproduce the storm surge contribution to ESLs is evaluated against recorded values from tide-gauge stations. The modeled peak surge maxima and the hourly time series are analyzed during these extreme events. The factors explored in this study encompass the numerical model, oceanic and atmospheric forcings, physical parameterizations, spatial resolution, and baroclinic/barotropic modes. Two ocean models (ADCIRC and NEMO) are intercompared using a similar configuration: domain, spatial resolution (~9 km), bathymetry and barotropic mode. The sensitivity of the atmospheric forcings is assessed by comparing storm surges induced by ERA5 reanalysis data and parametric wind models usually applied for hurricanes (e.g. Dynamic Holland Model, Generalized Asymmetrical Holland Model). The effect on storm surge due to non-linear interactions with the astronomical tide and variations in mean sea level is also investigated, as well as the sensitivity to different wind stress schemes. In addition, the baroclinic contribution to ESLs is studied using a configuration with 75 vertical levels. Finally, the role of the spatial resolution on the modeled storm surges is evaluated with a high-resolution domain of about 500 m in coastal areas. The analysis of the numerical experiments reveals some interesting insights. ADCIRC and NEMO can simulate storm surges due to tropical cyclones in a similar way compared to tide gauges. In general, the ERA5 forcing outperforms the various parametric wind models for storm surge modeling, in terms of maximum values, correlation, and duration of extreme events. Non-linear interactions of tides and mean sea level with storm surges have minimal contribution in the storm surges induced by hurricane events. However, the baroclinic response significantly improves the storm surge estimations in some coastal areas (e.g. along the southeastern Florida peninsula).  All the authors would like to thank the Government of Cantabria through the FENIX Project GFLOOD.