Development of Large-Scale Unstructured Grid Storm Surge and Sub-Grid Inundation Models for Coastal Applications
Doctor of Philosophy (Ph.D.)
Virginia Institute of Marine Science
Harry V. Wang
Yinglong J. Zhang
Carlton H. Hershner
John M. Klinck
Storm surge and inundation induced by hurricanes and nor'easters pose a profound threat to coastal communities and ecosystems. These storm events with powerful winds, heavy precipitation, and strong wind waves can lead to major flooding for cities along U.S. Coasts. Recent examples of Hurricane Irene (2011) in North Carolina and Virginia and Hurricane Sandy (2012) in New York City not only demonstrated the immense destructive power by the storms, but also revealed the obvious, crucial need for improved forecasting of storm tide and inundation. in part I, a large-scale unstructured-grid 3-D barotropic storm tide model SCHISM (Semi-implicit Cross-scale Hydroscience Integrated System Model) is developed with open ocean boundary aligning along the 60-degree West longitude to catch most Atlantic hurricanes that may make landfall along U.S. East and Gulf Coasts. The model, driven by high-resolution NAM (North America Mesoscale) and ECMWF (European Centre for Medium-Range Weather Forecasts) atmospheric fields, was coupled with Wind Wave Model (WWMIII) to account for wave effects, and used to simulate storm surge in 3-D barotropic mode rather than the traditional 2-D vertical average mode. For Hurricane Sandy, the fully coupled wave-current interaction 3-D model using ECMWF atmospheric forcing performs the best. The storm tide results match well with observation at all nine NOAA tidal gauges along the East Coast. The maximum total water level in New York City, is accurately simulated with absolute error of amplitude less than 8 cm, and timing difference within 10 minutes. The scenarios of "2-D" versus "3-D" and "with" versus "without" wind wave model were compared and discussed in details. Overall, the wave contribution amounts to 5-10% of surge elevation during the event. Also, the large-scale model with similar setup is applied to hindcasting storm tide during Hurricane Irene and the results are excellent when compared with observed water level along Southeast Coast and inside Chesapeake Bay. in part II, a high-resolution sub-grid inundation model ELCIRC-sub (Eulerian-Lagrangian CIRCulation) was developed from the original finite-volume-based ELCIRC model. It utilized the sub-grid method for imbedding high-resolution topography/bathymetry data into the traditional model grid and delivering the inundation simulation on the street level scale. The ELCIRC-sub contains an efficient non-linear solver to increase the accuracy and was executed in the MPI (Message Passing Interface) parallel computing platform to vastly enlarge the water shed coverage, and to expand the numbers of sub-grids allowed. The ELCIRC-sub is first validated with a wetting/drying analytic solution and then applied in New York City for Hurricane Sandy (2012). Temporal comparisons with NOAA and USGS water level gauges showed excellent performance with an average error on the order of 10 cm. It accurately captured the highest surge (during Hurricane Sandy) at Kings Point on both maximum surge height and the explosive surge profile. Spatial comparisons of the modeled peak water level at 80 locations around New York City showed an average error less than 13 cm. The modeled maximum modeled inundation extent also matched well with 80% of the FEMA flooding map. in terms of robustness and efficiency for practical application, ELCIRC-sub surpasses the prototype model UnTRIM2.
© The Author
Liu, Zhuo, "Development of Large-Scale Unstructured Grid Storm Surge and Sub-Grid Inundation Models for Coastal Applications" (2018). Dissertations, Theses, and Masters Projects. William & Mary. Paper 1550153651.