Collier County, southwest Florida, experienced catastrophic flooding and building losses during Hurricane Ian (Cat-4, 2022) and Hurricane Irma (Cat-3, 2017). This study evaluates how sea level rise (SLR) and hurricane characteristics influenced these events to inform future preparedness. A 3D vegetation-resolving surge-wave model coupled with a parametric hurricane wind and precipitation model was used to simulate flooding for both storms. Estimated structural losses were $126.3 M (Irma) and $1.95B (Ian), with NFIP-insured losses within 7-12% of FEMA claims. Ian's inundation corresponded to a 100-10,000-year flood, while Irma's was mostly ~100 years, based on comparison with ~300 synthetic storm simulations. Losses were comparable to those expected from a 100-year flood (Irma) and 3500-year flood (Ian). Sensitivity analyses revealed that SLR, higher intensity, larger size, slower speed, more perpendicular track, northern landfall, and high tide amplified Ian's impact. Climate change-driven SLR and intensity increases substantially heightened flood risk and damages.

Hurricane Irma in 2017 and Hurricane Ian in 2022 caused significant coastal flooding and flood-induced building loss in Florida, particularly in southwest Florida (Fig. 1) where the hurricanes made landfall. Both hurricanes underwent Rapid Intensification (RI, i.e., increase of wind speed by 15 m/s within 24 hours) a few times before their Florida landfall. Forecast tracks for both hurricanes, particularly Ian, changed significantly prior to their landfall, due to interaction between the hurricanes and large-scale weather systems. Irma and Ian caused catastrophic damage in Florida, and rank as the second-costliest and the costliest hurricanes in Florida history. Ian ranks as the third-costliest hurricane in the U.S. history, next to Katrina in 2006 and Harvey in 2017.

Hurricane Irma made landfall as a Saffir-Simpson Hurricane Wind Scale (SSHWS). Category-3 hurricane (with sustained wind of 178-208 km/h) on the south of Marco Island at 19:30 UTC on September 10, 2017, as primarily a windstorm which caused about $320 M of business and residential building damage in Collier County. According to the FEMA (Federal Emergency Management Agency) National Flood Insurance Program (NFIP), Collier County received approximately $52.4 M in flood insurance claims. At 19:05 UTC, September 10, 2022, Hurricane Ian made landfall at Cayo Costa Island, 30 miles north of Collier County, as a peak Category-4 (sustained wind of 209-251 km/h) hurricane, caused catastrophic building damage of approximately $2.2 B to 3515 buildings, with $1.7 B primarily flood damage to residential buildings in Collier County. However, the latest FEMA flood insurance payout data for Collier County during Ian shows the flood loss is ~$777.4 M, ~ 46% of the total estimated damage, comparable to the penetration ratio (percent of NFIP-insured households) of Collier County. During Irma, wind and flood damages on structures in Collier County were comparable but flood damage during Ian was dominant over wind damage, and much of the Irma damage was due to non-coastal damage due to wind and rainfall outside Collier County. Although both wind and flood hazards play significant roles in hurricane-induced building damage and that their combined effects should be considered for accurate damage assessments and disaster response planning, this study focuses on flood-induced building loss due to the lack of wind-induced building loss data which are proprietary property of private insurance companies.

Using a high-fidelity three-dimensional vegetation-resolving surge-wave modeling system CH3D-SWAN and its surrogate version, the Rapid Forecasting and Modeling System (RFMS), and extensive available data, this study simulates the coastal flooding and flood-induced building losses in Collier County during Irma and Ian and examines the causes for their dramatic differences. We first conduct simulations of the surge, wave, and flooding during Irma and Ian and compare them to extensive observed data at tidal gauges and High-Water Mark (HWM) locations. With the simulated maximum flood elevation during Irma and Ian, damage functions provided by the FEMA-HAZUS were utilized to estimate the flood-induced building loss during Irma and Ian and results compared to the FEMA flood insurance loss payout data. Validation of county-level flood loss estimation during hurricanes can aid the improvement of FEMA damage functions and provide accurate estimation of flood loss during flood events of various return periods. However, due to uncertainty of the data (first floor elevation, buildings, and details of insurance claims and payouts), parcel-level analysis will require substantial effort and will not be performed in this study.

To gain insight on future hurricane response and planning, we assess the sensitivity of coastal flooding, in terms of Total Inundation Area (TIA) and Total Inundation Volume (TIV), to various factors that may affect the storm surge and flooding, including such hurricane parameters as intensity (central pressure deficit), Radius of Maximum Wind (RMW), forward speed (V), azimuth (angle of approach), and landfall location, as well as sea level rise and tide during Hurricane Ian. In combination with future hurricane trends predicted by climate models, extensive analysis of historical hurricane data can be used for projecting potential coastal flooding and losses during future hurricanes.

The best tracks of Irma and Ian are shown in Fig. 2. Figure 3 shows the shifting official forecast tracks (OFCL) for Irma and Ian at 48, 24, 12, and 0 hours prior to landfall. Hurricane Irma and Ian made landfall on Marco Island and Cao Costa, respectively, after underwent rapid intensifications prior to landfall. Irma's RIs occurred on August 30-31 and September 4-5, followed by weaking with two less significant intensifications prior to its landfall as a Cat-3 hurricane at Marco Island on September 10. Ian had RIs on September 25 and then September 28, just seven hours prior to its landfall at Cayo Costa as a high-end Cat-4 hurricane. Rapid intensification (RI) greatly elevates coastal flood hazards by increasing surge and flood impacts, particularly when it occurs within 24 hours of landfall, leaving minimal time for evacuation. Moreover, RIs made the forecasting of hurricane tracks a major challenge during Irma and Ian, as shown by the shifting parallel forecast tracks in Fig. 2. Consequently, any slight error in the cross-track direction would result in a significant error in landfall location.

For example, 12 hours before Irma's landfall at Marco Island, it was forecasted to make landfall at Sarasota, 210 km to the north. 48 hours prior to Ian's landfall at Cayo Costa, it was projected to landfall at Tampa, 225 km to the north. Based on the forecasted landfall location of Tampa, several Florida counties in the Tampa Bay region issued evacuation orders on September 26, two days prior to Ian landfall. Evacuees from Tampa to Orlando were hit with rainfall-induced flooding. Counties near the eventual landfall location did not issue evacuation orders until September 27, which resulted in more losses. Therefore, to improve hurricane preparation of coastal communities, it is imperative to learn the impact of hurricane forecast error, which has been found to increase due to rapid intensification (RI) of recent hurricanes like Irma and Ian, on coastal flooding and to improve the accuracy and efficiency of forecast models. Even if hurricane forecast becomes more accurate at 48 hours prior to landfall, nevertheless, it is essential to improve the efficiency of the coastal flood forecasting system, which usually takes hours to days to run, to facilitate timely evacuations.

Observations from Hurricanes Michael (2018) and Ian (2022) show that RI can drive storm surges exceeding 4 meters due to sharp increases in wind speed and pressure drops near landfall. Coupled atmosphere-ocean models simulations confirm that RI enhances wind stress and wave setup, amplifying surge height and inundation extent, especially in low-slope coastal areas. Climate projections indicate RI will become more frequent and intense with warming, compounding flood risks. One study, using downscaled synthetic cyclones and surge-rainfall models, showed that RI disproportionately increases 100-year flood exceedance probabilities in the North Atlantic. Another analysis found anomalously high subsurface temperatures (2-3°C above average) on the West Florida Shelf during Hurricane Ian, fueling its rapid transition from Category 3 to 5 over ~12 hours. A related simulation study attributed this intensification to strong latent heat fluxes and eyewall dynamics, despite vertical wind shear. Additional research highlighted the role of warm shelf waters and limited upwelling in delaying storm-induced cooling and sustaining intensification. Furthermore, the frequency of RI events within 400 km of coastlines has tripled since 1980, underscoring the growing threat to coastal communities. Together, these findings show that RI significantly amplifies storm surge and inundation, demanding improved forecasting and adaptation strategies.

In the following, we present the results of our simulations on Irma and Ian flooding and flood losses, with verification based on available data. Followed by an assessment of how the flooding and flood losses during Irma and Ian compare with those of various return periods, e.g., the 1% Annual Exceedance Probability (AEP) flood, aka as Base Flood Elevation (BFE) or the so-called 100-year flood, to help us understand the adequacy of using the FEMA flood map as a guidance for community coastal flood preparedness. Special attention is given to the roles of various hurricane parameters in affecting coastal flooding and flood losses. Capability of existing surge-wave models for simulating the effect of RI on coastal flooding is discussed.