*3.2. Model Setup*

The model mesh covers the entire Gulf of Mexico, horizontally from 80.7◦ W to 97.9◦ W, and zonally from 18.1◦ N to 30.7◦ N (Figure 1). There are two open boundaries. One is located in the Caribbean Sea, connecting the east border of the Mexico and the west edge of Cuba. The other open boundary is located at the North Atlantic Ocean with the north point in the edge of Florida and the south point at the border of Cuba. The mesh contains 119,566 nodes and 214,297 elements. The finest resolution is about 50 m. There are 40 sigma layers. The model is three-dimensional barotropic, so salinity and temperature are not simulated nor discussed. The model is cold started with a time step of 1 s. The output time interval is set to be 30 min. The time periods for cold fronts and Hurricane Barry are from 20 December 2013 to 30 January 2014 and 20 June to 30 July 2019, respectively. Open boundary is only forced by tides, a combination of 10 tidal constituents (M2, S2, N2, K2, K1, O1, P1, Q1, MF, and MM). It is predicted by a tide model called TMD [54]. Wind stress and air pressure at mean sea level forcing at surface are obtained from the global Climate Forecast System Reanalysis (CFSR: https://climatedataguide.ucar.edu/climate-data/climate-forecast-system-reanalysis-cfsr) data with horizontal resolution of 0.5 degree by 0.5 degree.

### *3.3. Model Validation*

The skill scores of FVCOM in simulating water elevation and along-channel velocity are 0.7 and 0.67 (Figure 2), which shows "excellent" performance based on Wu et al. [55]. The low-pass filtered water elevation and velocity are also in line with the observed data, which are categorized as "very good" with the skill scores of 0.51 and 0.60 (Figure 2). Discrepancies between the modeled and observed low-pass filtered water levels and along channel velocities may be caused by the uncertainty of bathymetry of these highly active tidal inlets. As mentioned earlier, tidal passes of Barataria bay are severely eroded. The model bathymetry may therefore have errors, leading to larger uncertainties in model results. Since our focus is on the weather-induced hydrodynamics, variables to be examined below are all low-pass filtered with a cut off frequency of 0.6 cycles per day.

**Figure 2.** Validation of water level and along-channel velocity using observation in January 2014 at Caminada Pass, (**<sup>a</sup>**,**<sup>c</sup>**) are validations of water elevation and surface flow at Caminada Pass, (**b**,**d**) are validations of low-pass filtered water elevation and along channel velocity at Caminada Pass. Skill score for water level and along channel simulation are 0.7 and 0.67, which are categorized as "excellent".
