PROJECT TITLE:
Circulation modeling of the southeastern Bering Sea

PRINCIPAL INVESTIGATORS:
A. Hermann, P. Stabeno, D. Haidvogel, D. Musgrave

PROGRESS:

The overall objective of this project is to develop an an eddy-resolving circulation model of the southeastern bering Sea which includes tides and tidal mixing; this circulation model is intended for use as input to a suite of biological models focusing on walleye pollock and higher trophic levels. In the first year of this project a rigid-lid, primitive equation model with 20 kilometer resolution and no tidal dynamics was expanded to include seasonal variability. Much of the research of the second year has focused on the implementation of a free-surface, primitive equation, eddy-resolving regional Bering Sea model at 4 kilometer resolution with both tidal and subtidal forcing. The development of a suitable open boundary condition for such a model is unfortunately not trivial, required much experimentation, and has slowed down the overall effort. A suitable technique has been developed, however, which involves a nudging technique combined with a telescoped horizontal grid (Figure 1).

Concurrently with the regional model development, a global spectral element ocean model (SEOM) has been used to generate tides for use by the free-surface regional model. A tidal run with the five major tidal components of the Bering Sea (M2, S2, N2, K1 and O1), on a grid with finest resolution in the Bering Sea (Figure 2) has been converted to equivalent amplitudes and phases, for use in boundary conditions on the regional model (Figure 3).

Results of the new global and regional simulations have been very encouraging.  In particular we have reproduced the observed tidal residual circulation around the Pribilof Islands, while replicating the Aleutian North Slope Flow, the Bering slope current, and the shallow inflow through Unimak Pass (Figure 4). The regional model is now being driven with realistic winds (from ECMWF) and climatological heat flux (from Oberhuber's analysis of COADS data), as well as tidal forcing from global model. Having performed well in barotropic (tidal) simulations, SEOM is now being implemented as a multi-layer (and ultimately continuously stratified) primitive equation model of the world ocean. A new global grid has been developed with greater emphasis on the full equatorial and coastal waveguides, which should help capture more of the interannual variability of the North Pacific (including the Alaskan Stream and Alaska Coastal Current, which influence the Bering Sea) due to El Nino dynamics.
 

Figure 1. General layout, grid and boundary technique for the regional model of the southeastern Bering Sea. Model bathymetry is contoured, and land (Alaska) is colored green. Units on the axes are in meters. The interior box is resolved with ~4km spacing. Representative gridpoint locations (shown as crosses) illustrate telescoping of the grid beyond this interior box. The full domain is surrounded by closed walls. At red gridpoints we nudge the free surface height towards its proper tidal value, obtained from the global model (SEOM). At blue gridpoints we nudge the depth-integrated velocity towards values appropriate to the Aleutian North Slope Current and the Alaskan Stream/Alaska Coastal Current.
 
 

Figure 2. The grid of the Spectral Element Ocean Model (SEOM) used to generate tidal information for the regional model. Structure internal to each quadrilateral element is represented by a polynomial basis set. Minimum grid spacing is approximately 20 km.
 
 

Figure 3. Amplitudes (shaded; in m) and relative phase (contoured; in degrees) of the M2 tidal component, derived from the global simulation.
 
 
 

Figure 4. Output from the finely-resolved portion of the regional, free surface model (SCRUM), driven by ECMWF winds, climatological heat flux, tidal results from the global model (SEOM), and subtidal forcing based on current meter and hydrographic data. Depth-integrated velocities (in meters s-1) are shown superimposed on free-surface height (shaded; in meters). Model bathymetry is contoured; land is in black. Distances on axes are in meters. Unimak Pass appears at the bottom of the domain; the Pribilof Islands appear as black dots near the top of the domain.

SCIENTIFIC ACCOMPLISHMENTS:

1.  Tides and tidal residuals

A significant tidal residual signal has been observed in current meter records from mornings near the Pribilof Islands. Our regional model replicates this signal when forced with either the M2 tides alone or a mixture of five tidal components (Figure 5). It is likely this residual circulation, in conjunction with quasigeostrophic currents captured by the model, plays an important role in advecting young pollock in the vicinity of Pribilof Islands.
 

Figure 5. Close-up of subtidal circulation and free-surface height (with bathymetry contoured) near the Pribilof Islands.
 
 
 

2.  Nutrient mixing across isobaths

One of the core issues of the SEBSCC program is how nutrients are supplied to the shelf area, to feed the annual production there. As an investigation of cross-shelf nutrient flux, we initialized a three-dimensional scalar field with the value of the local the bathymetric depth, i.e.

g(x,y,z,0) = h(x,y)

where g is the tracer and h is the model bathymetry. We then tracked the evolution of this tracer field over a 60 day run of the model with M2 tidal forcing and subtidal inflows corresponding to the Aleutian North Slope Current and the Alaskan Stream/Alaska Coastal Current. As the simulation proceeds, cross-isobath mixing and advection are indicated by the difference between the value of the tracer g(x,y,z,t) and the value of the local bathymetry (gxs = g - h). A positive difference indicates either flow of initially deeper (and typically more nutrient-laden) water parcels into shoal areas, or sub-grid-scale mixing across isobaths. Results at day 60 are shown in Figure 6. Mixing and/or advection of deeper water across the shelf break is clearly implied, and a large tongue of deeper waters flows into the Southeast Bering Sea through Unimak Pass. A tongue of initially deep water penetrates up Pribilof Canyon, though its surface signal is far weaker than the Unimak tongue.  An intriguing aspect of the surface pattern is its strong resemblance to the spatial pattern of primary production suggested by Springer at al (1986).
 

Figure 6. Results of the tracer experiment, with corresponding velocities and free-surface height.
 
 

3.  Vertical mixing

A related issue to horizontal transport of nutrients is their vertical transport. We presently use Mellor-Yamada level 2.0 mixing dynamics to calculate effective vertical viscosity and diffusivity in the model. A recent run of the model with five tidal constituents, ECMWF wind forcing and Oberhuber heat flux forcing suggests that the most intense vertical mixing occurs not in the shallow region of Bristol Bay, but rather at the shelf break (Figure 7). This issue needs to be addressed with more realistic bathymetry, however; the present implementation of the model limits model bathymetry to the greater than 50 meters throughout the domain.
 

Figure 7. Logarithm (base 10) of the vertical viscosity (m^2/s) calculated by the Mellor-Yamada level 2.0 mixing algorithm of the regional model, along a vertical section near the Pribilof Islands. The location of model gridpoints is also shown in this figure.
 

APPLICATIONS:

Presentations:

Hermann, Albert J. 1998.  An open boundary technique for the simultaneous modeling of tidal and subtidal dynamics in the coastal gulf of Alaska and the Bering Sea.  International conference on coastal ocean and semi-enclosed seas: circulation and ecology modeling and monitoring.  Sept. 8-12, 1998, Moscow, Russia.

World Wide Web:

 A Bering Sea model results home page (http://www.pmel.noaa.gov/~hermann/sebscc.html) has been set up with sample output (including animations) from model runs.

STEPS TO COMPLETION:

Interannual comparisons of circulation fields must be delivered to fulfill the initial contract to Phase I of SEBSCC; these need to be made available to other researchers through the World Wide Web.  We will approach this goal first using subtidal boundary conditions based on current meter and hydrographic data for the ANSC and the AS/ACC. Since computer resources are limited, we will presently focus on the years 1995 and 1997.  Hydrographic and current meter measurements are available for each of these years, and they were remarkably different in character. In particular, 1997 was much warmer, and produced a coccolithophore bloom over broad areas of the shelf. Subtidal boundary conditions appropriate to the two different years, as well as ECMWF winds specific to those years, will be employed for this purpose. Heat flux measurements specific to these two years are unfortunately not available; climatological values will be used instead.  Results from a layered version of SEOM will be used as of boundary condition on the regional model; this will be replaced by continuously stratified results as they become available.
 

Revised time line

10/98 Interannual runs of regional model available (tides from SEOM; subtidal information from data)

12/98 Coupling of physical model results with biological models

3/99 Three-dimensional coupling of global and regional physical models

Projected publication: "An interannual comparison of circulation in the southeastern Bering Sea" (submit in spring 1999).  The intended funding to complete this work is from Phase II of SEBSCC.