*2.1. Site Description and Data Overview*

The study site is Corsica Island, located in the Mediterranean Sea (Figure 1). Since 2013, the approach implemented for the whole French Mediterranean coastline consists in mapping coastal sectors whose altitude is below 2 m. These sectors are considered potentially exposed to coastal flooding. Indeed, in past studies, the 100-year return period event was computed using the Sète tide gauge located at least 400 km from the coasts studied (red symbol in Figure 1). Moreover, the analysis did not take into account the combination of wind, waves, and water level as a joint return period. For coastal management, it is necessary to assess if this 2 m value is valid in our study area.

**Figure 1.** Location of Corsica Island and Sète (red symbol). (**a**) The map is adapted from Google Earth. (**b**) Wind provenance and crossed distance. The map is adapted from ROL Corse (Coastal Observatory Network of Corsica).

> Figure 1b highlights that the island is under the influence of different wind regimes coming from far as the Libeccio (can cross more than 1000 km) or coming from closer as the Mezzogiorno (~10 km). The waves induced by the wind can be swell when they cross long distance or wind waves if they stay close to their location of generation. Because of the distance from other coasts, we can assume that the west coast of Corsica can be impacted by swells, whereas the south and east coast mostly by wind waves.

> In this area, the tidal range is from 0.2 to 0.4 m as it is a microtidal regime [25]. For this reason, we consider a direct approach using the SWL instead of storm surges. Indeed, Haigh et al. [26] showed that for long return periods, when the ratio of the tidal to non-tidal component is lower than 2, direct and non-direct approaches present good agreement. Kergadallan [27] showed that this condition (<2) is observed for the French Mediterranean coasts.

> In order to manage our statistical analysis, validate data, and run hydrodynamic simulations, we used different products listed at Table 1.


**Table 1.** Data used in the study, spatial resolution, temporal resolution, and sources.

The National Oceanic and Atmospheric Administration (NOAA) provides the NWW3 MED hindcast [28,29] consisting of sea states, simulated using the third generation wave model Wavewatch 3 (WW3) with the parametrization proposed by [30]. Offshore wind conditions are also provided.

The SWL hindcast was generated by the BRGM using the MARS-2DH model [31]. This dataset is available from 1979 to 2010 and gives an assessment of the regional hydrodynamics based on tidal components and meteorological reanalysis CFSR [32].

The bathymetry and topography used after in the nearshore numerical modeling chain were provided by the Shom [33] and IGN.

We used in-situ data from Candhis and ISPRA (Table 1) to correct the Hs in the NWW3\_MED hindcast using a linear correction (see Appendix A). This correction reduces the bias and is relevant for both mean and Hs > 4 m. For extreme Hs, the bias reduction is ~1 m. In addition, we used the in-situ datasets provided by the Shom and CNES to validate the MARS\_MED\_ BRGM hindcast (Appendix A), and in-situ data coming from tide gauges provided by ISPRA and Météo-France in order to validate wind speed and direction.
