This talk will highlight the atmospheric and storm surge modeling efforts for Hurricane Gloria (26-28 September 1985) in the New York Metropolitan Region. There have been no realistic atmospheric and storm surge model simulations of historic landfalling hurricanes in this urban-coastal environment. This event provides a good test case for these numerical models. Gloria resulted in the highest recorded storm surge of ~7 feet (2.1 m) at the Battery, New York City in the last 60 years. Luckily, Gloria made landfall during a low tide, but the overall damage was still estimated at around $100 million.

The tropical storm bogus scheme within the Penn-State National Center of Atmosphere Research (PSU-NCAR) MM5 was used in concert with the North American Regional Reanalysis (NARR) to initialize Gloria at 0000 UTC 26 September 1985 down to 4-km grid spacing. The data was ingested into the Weather Research and Forecasting (WRF) model in order to utilize its vortex-following moving nest algorithm. The WRF winds and pressures at 12- and 4-km grid spacing were one-way coupled with the Advanced Circulation Model for Coastal Ocean Hydrodynamics (ADCIRC) model at grid spacings down to ~30 m around the New York City (NYC) area. A few different planetary boundary layer (PBL) schemes were also tested. For example, using the YSU scheme, the simulated storm at 1800 UTC 27 September (hour 41) made landfall over central Long Island ~15 km to the east of the observed and was ~2 h slow. In contrast, using the GFS scheme, the simulated storm made landfall at 1700 UTC about 30 km west of the observed. The WRF realistically simulated the pressure evolution of Gloria along the coast, with the WRF pressures of Gloria within ~5 mb of observed over Long Island. The peak simulated winds at 30-m were 35-40 m s-1 over coastal Long Island, which is within 3 m s-1 of observed, although the 10-m winds were 5-7 m s-1 too weak and the horizontal extent of the tropical storm force wind field was less than observed.

Using 10-m winds and pressures from WRF into ADCIRC produced a low bias of 0.75 – 1.0 m at the Battery, NY and other stations. Using 30-m winds instead of 10-m resulted in a negative surge error of 0.25-0.5 m, improving peak surge by ~0.5 meter. Lastly, WRF model winds were also used within the SWAN (Simulating Waves Nearshore) model along the East Coast to in order to ingest wave radiation stresses into ADCIRC. The inclusion of waves reduced the peak surge negative error to less than 0.2 m. Finally, a comparison run using FVCOM (Final Volume Community Model) ocean model will be compared using the same WRF forcing and ADCIRC grid.