Marine Ecosystem Dynamics Modeling Laboratory

The 1995-2006 Simulation/Assimilation

1a. First generation of FVCOM Results-GOM

This includes the monthly, daily, 3-hourly distributions of the subtidal (40-hour low-pass filtered) currents, surface water temperature and salinity in the Gulf of Maine with inclusion of Georges Bank.

1b. First generation of FVCOM Results-GB

This includes the monthly, daily, 3-hourly distributions of the subtidal (40-hour low-pass filtered) currents, surface water temperature and salinity in the Gulf of Maine with inclusion of Georges Bank.

2). Second generation of FVCOM-GoM/GB Results

This includes examples of the comparison between the first and second generation model results for 1998, 1999 and 2000. For other years, please contact Dr. Chen at UMASSD.

3). Third generation of FVCOM-GoM/GB Results

This includes examples of the comparison between the first and third generation model results for 1999 and 2000. For other years, please contact Dr. Chen at UMASSD. This version of FVCOM will be placed into an operation for the 5-day forecast soon.

 

FVCOM is the first unstructured grid model that is capable to run prognostically for a long-term simulation or assimilation in the GoM/GB region. To validate the capability of the integrated GoM/GB model system built by a joint effort of SMAST/UMASSD and WHOI, we have applied this system to simulate and assimilate the 1995-2006 seasonal variability of stratification and circulation in the GoM/GB region. Driven by realistic tidal forcing, MM5-predicted wind stress and heat flux, freshwater discharge from rivers and upstream inflow condition on the Scotian Shelf, FVCOM has successfully provided the 3-dimensional distribution of water stratification and circulation with a time interval of 2 minutes from 1995 to present. Detailed model-data validation results are being written into several papers. 1995-2006 assimilation results are output with an hourly interval and stored in MEDM hard disks. To whom it may be interested, please contact Dr. Chen at UMASSD or Dr. Beardsley at WHOI.

The 1995-2006 simulation and assimilation model runs were carried out by the following steps.

Firstly, start the model run under homogenous condition on 1st December, 1994 with the only realistic tidal forcing and continue to integrate until 15 December, 1994. The initial condition of the flow field at each grid point is specified using the Foreman tidal forecast program built on the FVCOM tidal model output. The purpose of running the model under this homogenous tidal condition is to spin up the model for tidal condition.

Secondly, re-start the model run on 15 December 1994 with the “hot-start” initial condition of T and S specified using December climatologic hydrographic field. Keep numerical integration until 31 December to spin up the baroclinic circulation driven by stratified tidal rectification and buoyancy forcing.

Thirdly, add the real-time wind stress, heat flux, river discharges, and upstream inflow condition starting on 1st January, 1995 and continue the numerical integration through the data assimilation approach until 2006. The nudging assimilation method is used to merge the model-predicted 5 day averaged surface water temperature to the daily objective-mapped satellite-derived SST (with a 25 years of satistical coherence spatial and temporal scales, the current to the hourly-sampled observed currents at moorings and bottom-mounted ADCP, and 3-D water temperature and salinity to hydrographic survey data.

The “upstream” boundary of the GoM/GB FVCOM domain cuts across the Scotian Shelf and upper slope just west of Banquereau Bank (Fig 10).  For realistic long-term (> 1yr) coupled physical/ecosystem hindcasts, the inflow of water, heat, salt, nutrients and biota must be specified accurately as a function of time and space along this boundary.   We use here the following statistical/theory-based approach coupled with local data.  The initial GLOBEC 1995-1995 hindcasts were conducted using a statistical model developed by J. Loder (BIO) to compute the along-shelf transport of Scotian Shelf water due to wind and buoyancy forcing. This transport had a simple cross-shelf structure over the inner shelf based on historical data, and the along-slope transport of slope water was set to zero. These simple upstream boundary conditions produced realistic flow and stratification conditions in the GoM, suggesting that realistic multiyear hindcasts could be achieved with a more comprehensive boundary condition that included an along-slope buoyancy-driven current over the upper slope.    

Recently we have improved our existing boundary condition to include a more realistic Scotian Shelf current (with mean and seasonal buoyancy forcing and water property variation) and a slope current with time-dependent transport and water properties (T/S, nutrients, and other ecosystem components).  For the 1998-2006 hindcast, existing data collected over the Scotian Shelf (e.g., the Halifax transect data), the RIVSUM index of St. Lawrence estuary freshwater discharge, and other information have been used to construct the shelf boundary condition time series. For the slope current, a climatological annual and seasonal mean current structure will be determined using existing hydrographic and moored data. This will be used in the process-oriented studies.  For the final 1998-2006 hindcast, the slope water transport and/or water properties will be varied to reproduce as accurately as possible the observed change in water properties in the NEC, and especially the appearance of LSW during 1998.  We envisage a sequence of physical-only model runs to determine the best set of upstream cross-margin boundary conditions for the process experiments and the final high-resolution continuous hindcast.

The monthly averaged in-situ temperature and salinity were nudged at the upstream open boundary to resolve the seasonal-interannual variability of stratification and circulation due to the inflow over the Scotian shelf and slope. The model results have been validated by comparing directly with field data for T/S and currents. The model tides are in good agreement with available surface elevation and current data, with overall uncertainties for the dominant M2 component of less than 3 cm in amplitude, 5° in phase, and 3 cm/s in the tidal current major axis (Chen et al., 2006e).  The model subtidal currents and stratification also compare well with existing in-situ measurements, capturing the seasonal cycle in vertical stratification and increased around-bank circulation on GB.  By adding the upstream T/S conditions on the Scotian shelf, the model captures the interannual variability of salinity in the GoM. In particular, the model reproduced the significant low salinity water inflow over the Scotian shelf during spring of 1996 and 1998.

«Previous Posted on January 17, 2014