The ESPreSSO hydrodynamic model (<http://www.myroms.org/espresso/>) has been operated by Rutgers University since October 2009 as a data-assimilative nowcast/forecast system for the Mid-Atlantic Regional Association Coastal Ocean Observing System (MARACOOS, <http://maracoos.org/>), part of the U.S. Integrated Ocean Observing System (<http://www.ioos.noaa.gov>). The underlying circulation model is the Regional Ocean Modeling System (ROMS; <http://www.myroms.org>), a finite-difference, hydrostatic, primitive equation ocean model that solves for the free surface elevation and three dimensional flow patterns, temperature, and salinity.
The ESPreSSO configuration of ROMS has 5 km horizontal resolution and 36 layers in vertical terrain-following coordinates. Bathymetry and land-sea masking is from the National Geophysical Data Center (NGDC) Coastal Relief Model. The vertical turbulence closure is the k-kl option of the Generalized Length Scale (GLS) formulation. Air-sea fluxes of momentum and heat are computed using bulk formulae applied to ROMS ocean surface conditions and meteorological conditions (wind velocity, rain, downward long- and short-wave radiation, and marine boundary layer temperature, pressure, and relative humidity) from the NOAA NAM atmospheric forecast. Ocean open boundary values are from a global forecast that uses the HYbrid Coordinate Ocean Model (HyCOM) with assimilation of satellite and in situ data with the Navy Coupled Ocean Data Assimilation (NCODA) system. ESPreSSO river inflows are from daily U.S. Geological Survey (USGS) stream gauge data, and tidal harmonic boundary variability is determined from a regional tidal model.
Output from the ESPreSSO forecast system is saved every 2 hours. Access to model results is made practical through the use of Thematic Real-time Environmental Distributed Data Services (THREDDS) technology, which allows subsets of large data sets to be accessed directly via Open-source Project for a Network Data Access Protocol (OPeNDAP) over the Internet from remote locations without transferring the entire multi-gigabyte model output.
The ESPreSSO data assimilation methodology consists of using the incremental strong constraint 4-dimensional variational (IS4DVar) approach to optimally adjust the model state within a 3-day duration analysis interval that precedes each 72-hour forecast. The assimilation cycle is repeated daily (hence the 3-day analysis windows overlap) and the first 24 hours of each forecast is retained as the "best estimate" of the ocean state for that day. It was these "best estimate" data that were analyzed here.
The data assimilated include surface currents from the MARACOOS HF-Radar (CODAR) network, sea surface temperature from satellite infrared (AVHRR) and microwave (AMSR-E) radiometers, and sea surface height anomalies from the Jason-2 altimeter satellite. In addition, a regional high-resolution climatology based on a 4-dimensional weighted least squares mapping of historical hydrographic data is assimilated to constrain biases in temperature and salinity introduced by the boundary conditions and/or internal model drift.
Significant wave height, dominant wave period, and wave direction were prescribed as SWAN TPAR format files on the model grid boundary with a spatial resolution of a boundary point every 25 grid cells using results from the NOAA Wavewatch III global multi-grid model, updated every 3 hours. A JONSWAP (JOint NOrth Sea WAve Project) spectral shape was assumed at these boundary points. Wind forcing was provided at 3-hour resolution from the NOAA North American Mesoscale (NAM) model (12 km resolution) over its domain, with forcing at the most offshore portions of the grid (outside the NAM grid) provided by the NOAA Global Forecasting System (GFS) model at 0.5 degree resolution. The SWAN directional resolution was 6 degrees (60 bins), determined via sensitivity analysis as the coarsest (and hence least computationally expensive) resolution that does not result in the "Garden-Sprinkler Effect" (GSE), wherein swell traveling over large distances inaccurately disintegrates into non-continuous wave fields as a result of frequency and directional discretization. The minimum frequency bin should be set to a value less than 0.7 times the lowest expected peak frequency and the maximum frequency bin should be set at least 2.5-3 times the highest expected peak frequency expected. In order to determine appropriate values, the peak periods from 43 NDBC buoys throughout the wave model domain were analyzed (when available) over the one year period of the study, yielding 297,533 hourly observations. The 99th and 1st percentiles of peak period were 15 s and 3 s, corresponding to frequencies of 0.07 Hz and 0.33 Hz, noting that these values may be biased by buoy limits of detection at high and low frequencies. The frequency range was therefore specified as 0.04-1 Hz. SWAN was allowed to internally determine the frequency resolution as one tenth of each frequency bin for best performance of the discrete interaction approximation (DIA) method of nonlinear 4-wave interactions, resulting in 34 frequency bins. Bottom friction calculations used the Madsen formulation with a uniform roughness length scale of 0.05 m. This value was selected for the best comparison of model output and buoy observations within the domain, and does not correspond to physical roughness values or the bottom roughness used in stress calculations. Wind generation and whitecapping parameterizations follow the modified Komen approach prescribed by Rogers et al. (2003), which reduces inaccurate attenuation of swell energy by whitecapping. Wave model outputs of bottom orbital velocity, bottom representative period, and bottom wave direction were output hourly and interpolated onto the ESPreSSO model grid.
NDBC observations were used for model validation; ten full-spectra wave buoys are within the domain. Due to maintenance problems observations from all buoys were not available over the entire time period and data from Texas Tower Station 44066 for January 2011, when the buoy was adrift but still reporting, were manually removed. Using the GM method and the same bottom roughness of 0.005 m used to process model output, surface and bottom parameters were calculated from buoy spectra and compared to model output at the same locations. The same person that conducted this processing step conducted each subsequent processing step.
References:
Rogers, W.E., Hwang, P.A., Wang, D.W., 2003. Investigation of Wave Growth and Decay in the SWAN Model: Three Regional-Scale Applications. J. Phys. Oceanogr. 33, 366-389.
Warner, J.C., Armstrong, B., He, R., Zambon, J.B., 2010. Development of a Coupled Ocean-Atmosphere-Wave-Sediment Transport (COAWST) Modeling System. Ocean Modelling 35, 230-244.
Wave direction, bottom orbital velocities, and bottom periods are calculated internally by the wave model. Near-bed current magnitude and direction are taken from the hydrodynamic model, with the reference height taken as the distance from the cell vertical midpoint to the seabed. GM requires that the current velocity be taken above the wave boundary layer (WBL) but within the log-profile current velocity layer. If the thickness of the WBL calculated using GM exceeds of one or more of the deepest grid cells, the current estimate and associated reference height are used from the deepest grid cell at each location where the reference height exceeds the width of the WBL. An estimate must be used for the maximum reference height where the log-profile velocity layer assumption is valid. As discussed in Grant and Madsen (1986), the thickness of the log-profile layer based on laboratory experiments is approximately 10% of the current boundary layer thickness (Clauser, 1956). Because tidal currents, storm currents, and mean flow have a boundary layer thickness on the order of magnitude 10's of meters (Goud, 1987), a maximum value for reference height is set as 5 m. The GM bottom boundary layer model also requires a value for bottom roughness; a uniform value of 0.005 m is used throughout the domain.
References:
Madsen, O.S., 1994. Spectral wave-current bottom boundary layer flows, Proceedings 24th Conf. Coastal Eng., pp. 384-398.
Glenn, S.M., 1983. A Continental Shelf Bottom Boundary Layer Model: The Effects of Waves, Currents, and a Moveable Bed. Dissertation, Massachusetts Institute of Technology and Woods Hole Oceanographic Institution, Cambridge, MA, 237 pp.
Glenn, S.M., Grant, W.D., 1987. A suspended sediment stratification correction for combined wave and current flows. J. Geophys. Res. 92, 8244-8264.
Goud, M.R., 1987. Prediction of Continental Shelf Sediment Transport Using a Theoretical Model of the Wave-Current Boundary Layer. Dissertation, Massachusetts Institute of Technology and Woods Hole Oceanographic Institution, Cambridge, MA, 211 pp.
Grant, W.D., Madsen, O.S., 1986. The continental-shelf bottom boundary-layer. Annu. Rev. Fluid Mech. 18, 265-305.
Grant, W.D., Madsen, O.S., 1982. Movable bed roughness in unsteady oscillatory flow. J. Geophys. Res. 87, 469-481.
Grant, W.D., Madsen, O.S., 1979. Combined wave and current interaction with a rough bottom J. Geophys. Res. 84, 1797-1808.
Madsen, O.S., 1994. Spectral wave-current bottom boundary layer flows, Proceedings 24th Conf. Coastal Eng., pp. 384-398.
Madsen, O.S., Poon, Y., Graber, H.C., 1988. Spectral wave attenuation by bottom friction: theory, Proceedings 21st Int. Conf. Coast. Eng., pp. 492-504.
Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.