USF FVCOM Tropical Cyclone Inundation Testbed Progress by Robert H. Weisberg, Lianyuan Zheng and Yong Huang College of Marine Science University of South Florida St. Petersburg, FL March 7, 2010 in UNC-Chapel Hill, NC
Models used 1. FVCOM: tidal and storm surge simulations; version unSWAN: wave simulation; public accessible version which is different from unSWAN coupling with ADCIRC.
Model setup 1. Grid: prepared by UND, converted to FVCOM input format, and shared with the LSU group. Node #: Cell #: Victoria and Corpus Christi
2. Tide open boundary condition: 8 principal tidal constitutes’ harmonic constants were provided by UND, and we generated the open boundary sea level data for the period July 1 through October 31, 2008, and shared with LSU. 3. Ike wind and pressure fields were provided by UND on a 2km structured grid. We interpolated these to the FVCOM unstructured grid. Wind and pressure data were then shared with LSU. Model setup
Is the interpolation of the OWI structured wind field onto the FVCOM unstructured grid accurate? Location: (-90,26) 00:00 on 9/12/2008
Tidal Simulations Case1: 2D tidal simulation with spatially varying Manning coefficients (provided by UND); Case2: 3D tidal simulation with constant bottom roughness (z 0 =0.01m) and BL scaling; Case3: 3D tidal simulation with spatially varying bottom roughness based on the formulation provided by UND. The 3D simulations used 11 vertical sigma layers (uniformly distributed). The models were run for 4 months. The first 20 days were for tide spin-up; the next 100 days were used for harmonic analysis (using matlab t_tide package).
Model-Data Comparisons
Reduced Set Model vs Observed All Set
IKE Storm Surge Simulations We ran 3 cases paralleling the tidal simulations by adding the Ike wind and pressure forcings.
Model vs Observed
IKE Wave Simulations For the Hurricane IKE wave simulations, the direction resolution is 4º (i.e. 90 directions) and the frequency resolution is 0.04~1.00 Hz (i.e. 35 frequencies). Time step is 10 min.
Model vs Observed NDBC 42001: Central GOMNDBC 42040: East of New Orleans
Model vs Observed NDBC 42035: Near Galveston Bay NDBC 42055: SW of GOM
What have we learned? Tidal simulations are only slightly sensitive to 3D versus 2D, or constant versus spatially varying bottom friction. Surges, in contrast, are sensitive to these factors because the currents are much stronger for hurricane events when compared with tides over much of the domain. Semi-diurnal tides are well simulated, but diurnal tides are not. Both diurnal and semi-diurnal tides are poorly simulated in Victoria and Corpus Christi region.
What have we learned? Based on the bottom friction parameters provided by UND, the 3D simulations overestimate the surges in the Galveston Bay region. Is this due to Manning coefficient calibration? SWHs are sensitive to direction resolution. Our experiments suggest that a resolution of at most 6 degrees is necessary near the coast. We used 4 degrees.
Acknowledgements We thank the FVCOM and SWAN groups for their support in sharing model developments. We thank UND for providing the initial model setup (grid, bathymetry) and forcing (tide, wind and pressure) information. Rick’s project leadership has been excellent. We also appreciate SURA’s (Gary and Linda) role in coordinating computer resources, without which we would not be able to perform. We appreciate the Testbed resources supplied through NOAA IOOS and the collaborations with all of our Testbed colleagues in addressing the important matters before us.