Introduction

The purpose of this overview is to review physical processes in the Southern Ocean that affect the ecosystem. Given the surface intensified nature of oceanic ecosystem, the bulk of the overview will consider physics near the ocean surface. However, the Southern Ocean is weakly stratified so some physical processes influence the whole water column. Additionally, dynamics will be different in the shallow coastal areas as compared to the deep ocean so it is necessary to discuss these two general regions separately. It is quite likely that the nature of the ecosystem over the continental shelves is different from that in the open ocean so this division of the subject seems natural.

The most obvious surface physical structure is the mixed layer that is driven by wind mixing and density convection. These layers are thicker in the winter and, in the Southern Ocean, can be 100 to 600 m thick. During the warmer months, surface warming and freshwater from melting ice create one or more stratified seasonal mixed layers of 10 to 50 m thick (which is more typical of mixed layers in non-polar regions of the World Ocean). The deepest mixed layers seem to occur in the Weddell sea, while open ocean and coastal areas have winter mixed layers of 100 to 200 m or so. The reasons for these regional differences have to do with the water stratification and the atmospheric forcing.

A less obvious general feature of the Southern Ocean is that thermodynamic processes are much more important than at lower latitudes. This statement may be obvious, but much of our modeling experience is with non-polar oceanic systems and we may be misunderstanding or misrepresenting processes in polar environments. One particular feature of polar environments is that water density, at temperatures near freezing, is controlled strongly by salinity and is only weakly influenced by temperature. It is possible, then, for water at depth to be warmer than that above it. For example, winter mixing creates near-freezing water down to 100 m or so which is capped in the summer by warmer water. Organisms can then experience relatively warm surface water (2°C), cold subsurface water (less than -1.5°C) and deeper water as warm as 1°C. Given the large Q10 of biological processes in the polar regions, this may represent the extreme range for the temperature tolerance of some organisms and it can be experienced during daily vertical migration. The temperature structure may also create ecologically desirable depth zones for some organisms. The focus of these comments is on temperature with the assumption that salinity variations do not have a similar effect on ecological processes. The truth of this assumption is not known (to me, in any case).

Sea ice is a clearly perceived feature of the Antarctic environment and it is well known to have a strong influence on thermodynamic exchange across the ocean surface. It also provides a mechanism to create dense water by releasing salt during freezing or to create stratified surface layers through melting. Ice freezing in one place and melting in another produces a salt flux which may be much larger than the surface precipitation- evaporation imbalance. A less obvious, perhaps, effect of ice is to create a habitat or a transport mechanism for some organisms (for example, krill).

Observations of marine ecosystems in the Southern Ocean are very difficult because of the large size of the system and its remoteness. Satellite based observations have been very helpful in other parts of the ocean; they are less so in the Southern Ocean because of the persistent cloud cover and the high latitude (some satellites are limited in the latitude coverage). One can obtain good information on ice concentration, surface elevation and wind stress from sensors that are not strongly affected by clouds. There is considerable difficulty in using either infrared or visible band instruments that are completely blocked by clouds. The presence of ice may have an effect on altimeter measurements because of the change in reflection characteristics. The scatterometer (which estimates wind stress) will be rendered useless by ice cover since the instrument depends on capillary waves on the ocean surface. The effect of partial ice cover on the performance of either instrument is not known to me.

All physical models need information about fluxes (momentum, heat, freshwater) at the ocean surface. The Southern Ocean is deficient in ship traffic so climatologies based on ship reports are poor compared to similar estimates for other parts of the ocean. Some attempts have been made to use atmospheric circulation models from global weather centers (for example, the European Centre for Medium Range Weather Forecasting) to estimate ocean-atmosphere exchange for the Southern Ocean (D. Olbers, personal comm. referring to unpublished work by A. Stossel). These estimates of forcing have the benefit of being available 2 to 4 times per day, but the quality of the estimates have not been verified extensively.

In general, there is a mismatch in scales between biological and physical processes and this difference manifests itself in the different character of models for these systems. Many biological organisms in the ocean occur as small patches with vertical extents of a few meters and horizontal extents of a few tens of meters. These scales are much smaller than can be represented by circulation models at the present time. There will need to be some sort of "sub-grid scale" parameterization to represent the biological processes in the current family of ocean circulation models. On the other hand, the length scales in oceanic flow are known to be 1 km or larger (except for internal waves, vertical convection and turbulence). Therefore, it may be possible to limit the resolution of the circulation model and put computational resources into a fine grid to consider patch level biological processes. Careful thought is required when combining physical and biological models.

Large Scale Dynamical Processes

The general dynamical balance in the Southern Ocean is that the flow is forced by the atmospheric westerlies that drive eastward flow (the ACC). Since the flow is driven at the surface, vertical shear develops which is supported by lateral density gradients (thermal wind balance). When the shear is sufficiently large, the flow becomes dynamically unstable (baroclinic instability). The eddies that are created deform the interior density surfaces (makes them lumpy) creating internal form drag that transmits downward the momentum transferred from the surface winds. This instability releases potential energy (reduces the slope of the density surfaces) and in the process transmits dense water towards the equator (poleward heat flux). When the momentum reaches the bottom, form drag (pressure difference across a bottom feature) on the bottom topography removes the momentum to slow the flow. A balance occurs when the bottom form drag balances the surface wind stress (Johnson and Bryden, 1989; Marshall et al. 1993; Olbers, 1994). The transient eddies in this process are a critical part of the downward transmittal of momentum and so must be represented properly (either explicitly in an eddy resolving model or parametrically as some sort of vertical viscous process).

It is important to note that flow paths that circle Antarctica and do not contact continental boundaries or bottom topography (generally the layer from 500 m to 2000 m depth) are required to have transient eddies to transfer momentum downward (dynamics do not allow a geostrophically balanced, zonally averaged, net meridional transport). These eddies produce a poleward heat flux, but they should also transport other water properties (nutrients, plankton, etc) across the ACC. These fluxes have not been measured to my knowledge.

One important feature of the Southern Ocean is that the water is weakly stratified, compared to stratification at subtropical latitudes. Weak stratification gives rise to a short dynamical scale, the first internal radius of deformation, which ranges from 20 km to 8 km, decreasing to the south (Inoue, 1985). This scale defines the width of jets and the size of mesoscale eddies and has a considerable impact on sampling and numerical modeling. In general, both model grids and sampling are at scales that are larger than this dynamical scale which produces aliasing in observations and bad behavior in numerical models. The weak stratification has the further effect that flow disturbance at the bottom of the ocean penetrate to the surface so the ACC is much more strongly affected by bottom topography than flow in mid-latitudes. Additionally, the small dynamical scale means that smaller features in the bottom topography can have large-scale dynamical effects and must be included in circulation models.

One final comment on the eddies is that they can create a convergence of momentum that produces narrow, meandering jets (Eliason-Palm flux). Treguier and Panetta (1994) use a model to show that two jets can be created for a broad region of wind forcing like the Southern Ocean, but the two jets they model are separated by 50 internal radii instead of 5 internal radii which is observed in the ACC. Altimetry can be used to verify the location and size of the ACC jet cores without having to use a geoid (Gille, 1994). The newly released geoid from the US Navy should make these analyses of altimetric observations easier.

Regions of strong eddies seem to be confined to limited geographical regions and to the area between the Subantarctic Front and the Polar Front (Nerem et al. 1994). Polar Frontal zone should be packed with eddies and the effect of this strong stirring on the ecosystem is not clear. Eddies may not have a strong influence on the ecology of the area since they provide stirring but not sources and sinks (transport into or out of the region. Recall that the effect of the eddies is to cause a convergence of momentum and it should be accompanied by convergence of other properties.). Is this true? This region of the ACC is clearly biologically active as evidenced by extensive diatomaceous (siliceous) sediments observed under the mean position of the ACC.

Models of the large scale circulation have been generally successful, but the major test has been the total transport through Drake Passage which was measured for 14 months (1979-80). Both FRAM and Semtner-Chervin models calculate transports for the ACC (190-200 Sv) that are larger than measured values (130 Sv) in spite of realistic forcing. Why?

Large scale circulation models do not represent well processes near the continental slopes. Small scale (1-5 km) bottom features also are shown to be important (small gaps in mid-ocean ridges are important for guiding flow or providing flow path for abyssal water). These small features of topography can have dynamical and thermodynamical effects but are not well represented in these models because of the grid box nature of the model and the (relatively) large values of grid spacing.

Dynamical Processes Near Continents

Most of the work on Antarctic shelves has either been general surveys or has focused on the processes that create dense bottom water. There are relatively few studies of general circulation along any section of the shelf that makes this topic an open area of study. This lack of observations means that we do not know the relative importance of wind and thermodynamic forcing on these shelves. Furthermore, we do not know the magnitude or character of tidal currents nor do we know about the general circulation directions. The weak stratification and deep mixed layers mean that geostrophic calculations of flow are of limited use. It is very likely that there are strong seasonal differences in physical processes because of ice. We can be guided by a number of mature studies of the Arctic Ocean, but the two systems are rather different so one can not just import ideas without careful consideration.

Continental shelf circulation is generally forced both locally and remotely. Local forcing occurs through wind (mainly) and surface thermohaline (less important) forcing. Additional local forcing occurs at the coast due to fresh water runoff from the continent or at the shelf break if there is a strong coastal current (say, an oceanic boundary current) that is driven by large scale processes. Remote forcing occurs because waves propagate along continental shelf and slope which bring information, after some time delay, about flow variations far removed from the region of interest. Since these waves propagate with the coast to the left (looking in the propagation direction in the Southern Hemisphere), the direction of influence for a region is easily determined.

Surface fluxes over Antarctic shelves are poorly known because of the limited number of coastal stations and the small number of ships that transit the area. Information from coastal weather stations must be used with care due to the tall mountains that are common around Antarctica. Some effort is being made to place automated weather stations away from the coast on low islands. These stations will be a great help, but it will take a while for a useful network to be put in place. Another option for estimating surface fluxes to the ocean is through global atmospheric circulation models. However, the quality of these estimates over the continental shelf has not been checked.

A second interesting feature of Antarctic shelves is that, except near the Weddell Sea, the southern part of the ACC flows along the shelf break. The bulk of the water in this flow is Circumpolar Deep Water (CDW, basically North Atlantic Deep Water which has been absorbed into the deep levels of the ACC) which is relatively warm (around 2°C), oxygen poor and nutrient rich. Because of the tilt of density surfaces in the ACC, this water mass occurs at about the depth of the shelf break at many places around the Antarctic continent. This water is observed to flood the continental shelf, by some dynamical process that has not been identified, bringing with it heat and nutrients. The heat provides a balancing effect for the strong winter cooling and may explain why wide areas of the Antarctic Continent thaw during the summer. For example, this may explain why the central areas of the Antarctic Peninsula are relatively ice free compared to areas in the Weddell Sea at the same latitude (separated by the relatively narrow Peninsula mountain range). This import of warm, salty water onto the shelves creates a relatively persistent temperature structure with cold (-1.5°C) water at the bottom of the mixed layer (100 m) and water as warm as 1.5°C below. This layer of deep, warm water may be very important for many biological processes.

A further effect of this import of CDW is that nutrients are brought onto the shelf and this may have some bearing on the observations of relatively high nutrient concentrations, even in the summer (although other processes may explain this excess, such as light limitation, micro-nutrient limitation, top-down control of primary production, etc.).

Ice is a critical element in the coastal environment because of its influence on exchanges with the atmosphere and because of the role that it plays in the thermodynamics of the coastal environment. Ice coverage is different (different thicknesses and different timing for freezing and thawing) in different regions around the continent due to heating from offshore (import of CDW across the shelf break), because of different heating during the summer (different stratification) or because of ice transport by the circulation. Ice has also been proposed to have an effect on the ecosystem by providing a habitat for some organisms. All of these processes are likely to have different relative importance in different parts of the Antarctic coastal environment.

Timely Questions and Observations

Can products calculated from atmospheric forecast models be used to estimate fluxes between the atmosphere and ocean (wind stress, heat and freshwater fluxes) with sufficient precision to be useful in the next generation of models (whether they are mixed layer [z,t], coastal or oceanic)? Do these models work well near the coast or does some sort of regional atmospheric model representing the local land topography and driven by the large scale atmospheric conditions (or some other correction) need to be used to specify correctly the atmospheric fluxes over the coastal areas?

FRAM and Semtner-Chervin models (both of which are based on the model code developed by Kirk Bryan at NOAA-GFDL) have large transports (190-200 Sv) relative to measured values in spite of realistic topography and surface forcing. Why? Possible difficulties could be 1) incorrect surface forcing, 2) need for daily varying surface forcing rather than averaged monthly fluxes, 3) poorly represented bottom topography and bottom slope, 4) poorly represented coastal areas, 5) problems with bottom form drag calculation in areas of steep topography, 6) excess viscosity required to remove small scale numerical errors, or 7) numerical problems using a B-grid model where the internal radius of deformation is resolved by the grid.

Problems focusing on the ACC, especially away from the "choke points" could benefit from a cluster of moorings in the open ocean away from mid-ocean ridges and sea-mounts. The study would look at eddy effects in the free ACC (combined with altimetry and scatterometer observations). Moored biological sensors could address effects of eddies on transport of phytoplankton or nutrients.

References

Inoue, M., 1985. Modal decomposition of the low-frequency currents and baroclinic instability at Drake Passage. J. Phys. Oceanogr., 15, 1157-1181.

Johnson, G. C. and H. L. Bryden, 1989. On the size of the Antarctic Circumpolar Current. Deep-Sea Res., 36, 39-53.

Klinck, J. M., 1992. Effects of wind, density and bathymetry on a one-layer Southern Ocean model. J. Geophys. Res., 97, 20,179-20,189.

Marshall, J., D. Olbers, H. Ross and D. Wolf-Gladrow, 1993. Potential Vorticity Constraints on the dynamics and hydrography of the Southern Ocean. J. Phys. Oceanogr., 23, 465-487.

Nerem, R. S., E. J. Schrama, C. J. Koblinsky and B. D. Beckley, 1994. A preliminary evaluation of ocean topography from the TOPEX/POSEIDON mission. J. Geophys. Res., 99, 24,565-24,583.

Olbers, D., 1994. Is there a direct relation between the zonal wind stress and the poleward heat transport in the Southern Ocean? unpublished manuscript, 12pp.

Stossel, A., P. Lemke and W. B. Owens, 1990. Coupled sea ice-mixed layer simulations for the Southern Ocean. J. Geophys. Res., 95, 9539-9555.

Treguier, A. M. and R. L. Panetta. 1994. Multiple zonal jets in a quasi-geostrophic model to the Antarctic Circumpolar Current. J. Phys. Oceanogr., 24, 2263-2277.