Modelling of the ACC and polar gyres

This page is part of the topic The Southern Ocean in the instrumental period

Observations are still not as frequent and spatially well distributed as needed to establish a well-proven view which would allow the unambiguous determination of forcing mechanisms that might be subject to change. To fill this gap, the dynamics are evaluated using results from models. However, the models themselves have shortcomings, so their results have to be considered with as much care as the ones derived from observations alone. Indeed, some models throw up contradictions that cannot be solved until new generations of models or observations become available.

A number of modelling studies, including both general circulation models (GCM) and simplified theoretical models, have been carried out in an effort to improve our understanding of the basic circulation of the ACC. These studies examined the controls on its transport strength and the mechanisms for import and export of various water masses. The GCM studies involved either the Southern Ocean only or fully global domains, and in most cases included the polar gyres, especially the cyclonic Ross Sea and the Weddell Sea Gyres. These gyres are intimately linked to the ACC, sharing their northern boundaries with the southernmost front of the ACC. Excluding the geographical areas covered by these two gyres, the ACC approaches close to the Antarctic continent (Orsi et al, 1995) and brings its relatively warm Circumpolar Deep Water (CDW) masses close to the edge of the Antarctic Ice Sheet.

A focus of most modelling studies has been to identify the forces that constitute the dynamical balance for the ACC. At the latitude of the Drake Passage, where the ACC is unbounded, the current receives a zonally-continuous momentum input from the dominantly westerly wind. Using the output from various OGCMs, efforts have been made to identify the leading dynamical process that can remove zonal momentum from the ACC at the same rate as the wind input. Given the unbounded nature of the current, it is perhaps not surprising that a classical Sverdrup balance is not able to balance the wind input (Gille et al., 2001), and that instead bottom form stress counteracts the input of momentum from the wind (Grezio et al., 2005). The amount of bottom stress is related to the representation in models of the bottom topography; models with a smoothed representation of topography produce less form stress and thus higher transport values for the ACC than those with rougher topography. Increasing the wind stress within the Southern Ocean increases the ACC transport (Gnanadesikan and Hallberg, 2000).

Eddies that result from hydrodynamic instability of the mean flow of the ACC also play a role in the momentum balance. At high latitudes the length scale at which oceanic flows are affected by the Earth’s rotation is rather small - of the order of several kilometres - which means that energetic eddies are relatively small. Indeed they may be smaller than the typical grid cell in an OGCM. Most modelling studies have been carried out at a relatively coarse resolution, in which case they would not simulate eddies well. Some others, of higher resolution, do provide eddy-permitting simulations (e.g. Maltrud and McClean, 2005). The eddy-kinetic energy in OGCMs varies as a function of the model grid resolution, and this in turn has a significant influence on the simulated transport of the ACC. In some regions of the ACC, eddies cause an upgradient transfer of kinetic energy into the mean flow, while for the major part of the ACC the transfer is downgradient (Best et al, 1999). Because coarse-resolution models represent eddy processes and topography inadequately, they produce an unrealistic simulation of the ACC transport, and of the overall Southern Ocean circulation.

Under some scenarios of modelled climate change, there are significant changes in the wind field that drives the ACC. The response of the ACC to a change in the wind field occurs in the remarkably short period of two days (Webb and de Cuevas, 2006). The response is largely barotropic (induced by sea surface elevation) and controlled by the topography, with the changed wind stress quickly transferred by the barotropic flow into the bottom topography as form stress. This is an important finding in the context of climate change, as it suggests that changes in atmospheric circulation can be quickly transmitted into changes in ocean circulation. The ability of the ACC to respond quickly to the wind may explain the observed poleward shift of the ACC over recent decades. An analysis of an OGCM in which the observed poleward shift of the ACC was simulated lends support to the idea that human-induced climate change is currently influencing the ACC and will continue to do so over the coming century (Fyfe and Saenko, 2005).

Although numerical modelling of the ACC and adjacent polar gyres sheds some light on the behaviour of the ACC and its interaction with the gyres, at least two pressing questions remain poorly addressed. First, the transport volume of the ACC remains poorly constrained in different OGCMs, particularly so in models typically used in IPCC simulations, which show a wide discrepancy in transport values even where the external forcing is similar (Ivchenko et al., 2004). Analyzing the ACC transport in 18 coupled atmosphere-ocean models Russell et al (2006b) found that compared to the observed transport estimate of 135 Sv, the coupled models produced a spread ranging from a low of -6 Sv to a high of 336 Sv. They concluded that it is difficult, at present, to get the Southern Ocean “right” in coupled atmosphere-ocean models. This shortcoming reflects the lack of high resolution in many model simulations, and should be overcome as Southern Ocean eddy resolving ocean models become increasingly prevalent. Secondly it is currently difficult to model the interaction of the ACC and polar gyres with the edge of the Antarctic Ice sheet. To properly tackle this problem requires an OGCM to be coupled interactively to an ice sheet model.