Hurricane Sandy inflicted heavy damage in New York City and the New Jersey coast as the second costliest storm in history. A large-scale, unstructured grid storm tide model, Semi-implicit Eulerian Lagrangian Finite Element (SELFE), was used to hindcast water level variation during Hurricane Sandy in the mid-Atlantic portion of the U.S. East Coast. The model was forced by eight tidal constituents at the model’s open boundary, 1500 km away from the coast, and the wind and pressure fields from atmospheric model Regional Atmospheric Modeling System (RAMS) provided by Weatherflow Inc. The comparisons of the modeled storm tide with the NOAA gauge stations from Montauk, NY, Long Island Sound, encompassing New York Harbor, Atlantic City, NJ, to Duck, NC, were in good agreement, with an overall root mean square error and relative error in the order of 15–20 cm and 5%–7%, respectively. Furthermore, using large-scale model outputs as the boundary conditions, a separate sub-grid model that incorporates LIDAR data for the major portion of the New York City was also set up to investigate the detailed inundation process. The model results compared favorably with USGS’ Hurricane Sandy Mapper database in terms of its timing, local inundation area, and the depth of the flooding water. The street-level inundation with water bypassing the city building was created and the maximum extent of horizontal inundation was calculated, which was within 30 m of the data-derived estimate by USGS.

Washington, DC, the capital of the U.S., is located along the Upper Tidal Potomac River, where a reliable operational model is needed for making predictions of storm surge and river-induced flooding. We set up a finite volume model using a semi-implicit, Eulerian-Lagrangian scheme on a base grid (200 m) and a special feature of sub-grids (10 m), sourced with high-resolution LiDAR data and bathymetry surveys. The model domain starts at the fall line and extends 120 km downstream to Colonial Beach, VA. The model was used to simulate storm tides during the 2003 Hurricane Isabel. The water level measuring 3.1 m reached the upper tidal river in the vicinity of Washington during the peak of the storm, followed by second and third flood peaks two and four days later, resulting from river flooding coming downstream after heavy precipitation in the watershed. The modeled water level and timing were accurate in matching with the verified peak observations within 9 cm and 3 cm, and with R2 equal to 0.93 and 0.98 at the Wisconsin Avenue and Washington gauges, respectively. A simulation was also conducted for reconstructing the historical 1936 Potomac River Great Flood that inundated downtown. It was identified that the flood water, with a velocity exceeding 2.7 m/s in the downstream of Roosevelt Island, pinched through the bank northwest of East Potomac Park near DC. The modeled maximum inundation extents revealed a crescent-shaped flooding area, which was consistent with the historical surveyed flood map of the event.