Simulated changes in the extratropical Southern Hemisphere winds and currents

Authors


Abstract

[1] The results from 12 global climate models show a remarkably consistent strengthening and poleward shifting of the zonal wind stress through the 20th and 21st centuries at extratropical Southern Hemisphere latitudes. Changes in the zonal circulation of the ocean in the region are broadly consistent with the changes in zonal wind stress. In particular, the climate models simulate a strengthening and a poleward shift of the Antarctic Circumpolar Current. The strengthening of the zonal wind stress also results in intensifying northward Ekman transport across the Antarctic Circumpolar Current which, in the unblocked latitudes of Drake Passage, implies increasing southward geostrophic transport in the ocean below about 2000 m. Zonal wind stress changes such as these may be expected to enhance the mesoscale eddy activity in the Southern Ocean.

1. Introduction

[2] With a volume transport of 130–140 Sv (1 Sv = 106 m3/s), the Antarctic Circumpolar Current (ACC) is the most powerful current in the world ocean. It connects all the major ocean basins, effectively exchanging heat, salt, and other tracers between them, and has an important effect on the Earth's climate. The absence of land at the latitudes of the Drake Passage means that the net northward ageostrophic mass transport can only be balanced by geostrophic flows in the deep ocean, which in turn connects the Southern Ocean to the other ocean basins around the world [Gnanadesikan, 1999].

[3] In two recent studies [Fyfe and Saenko, 2005; Saenko et al., 2005] we have examined the changes in the Southern Ocean circulation simulated in response to increasing concentrations of atmospheric greenhouse gases (GHGs) using the Canadian Centre for Climate Modelling and Analyses (CCCma) global climate model (GCM). Our major conclusion was that as part of the simulated warming mid-latitude zonal winds in the Southern Hemisphere (SH) strengthen and shift poleward. This translates into changes in the strength and position of the ACC, and in the mass exchange across the ACC. Here we confirm and expand on these results for an ensemble of GCMs.

2. Data and Methodology

[4] We analyze decadal mean observed and simulated SH zonal wind stress τx(λ, ϕ) and oceanic barotropic stream function Ψ (λ, ϕ) (λ is longitude and ϕ is latitude). We consider data from NCEP2 and ERA40 reanalyses, as well as output from 12 GCMs. The model output was downloaded from an archive hosted by the Program for Climate Model Diagnosis and Intercomparison (PCMDI). The models used are: CCCMA-CGCM3-1 (1), GFDL-CM2-0 (2), GFDL-CM2-1 (3), INMCM3-0 (4), MIROC3-2-MEDRES (5), MRI-CGCM2-3-2A (6), CNRM-CM3 (7), IPSL-CM4 (8), MIUB-ECHO-G (9), MPI-ECHAM5 (10), NCAR-CCSM3-0 (11) and GISS-MODEL-E-R (12). Documentation for the models is available on the PCMDI website located at http://www-pcmdi.llnl.gov. We consider pre-industrial control runs, 20th century runs (with historical GHG, aerosol, and in some cases, volcanic, and solar forcing) and 21st century runs following the Intergovernmental Panel on Climate Change (IPCC) A2 emissions scenario.

[5] The treatment of Ψ(λ, ϕ) is as follows. At each longitude in the SH Ψ(λ, ϕ) is fitted to an error function [Marquardt, 1963]. This leads to an approximation for the vertically integrated zonal current U = ∂Ψ/∂(Reϕ):

equation image

where Re is the earth's radius. In this way U(λ, ϕ) can usefully be described in terms of it's strength ηU(λ), position ΦU(λ) and width σU(λ). Similarly, after fitting equation imageτxdϕ to an error function we obtain τx(λ, ϕ) ∼ ητ(λ) exp(−(ϕ − Φτ(λ))2τ2(λ)). Figure 1 shows the ERA40 climatological zonal wind stress before fitting (left) and after fitting (right). The fitted field is a good approximation: capturing the strength and position of the maxima in the Indian Ocean (near 90°E), south of New Zealand (near 180°E) and off the tip of South America (near 90°W).

Figure 1.

Climatological (1991–2000) ERA40 zonal wind stress τx (left) before fitting and (right) after fitting. The outermost contour is 0.05 Pa and the contour interval in 0.025 Pa.

[6] We first consider the extent to which these climate models reproduce the present-day zonal wind stress climatology. Figure 2 shows the NCEP2, ERA40 and GCM-mean climatological wind stress strength ητ(λ) (top) and position Φτ(λ) (bottom). The GCM mean profile captures the maxima in the Indian Ocean and off the tip of South America but misses the observed maximum south of New Zealand. (The dark shading shows the 2.5% to 97.5% percentile intervals assuming normally distributed GCM parameter values.) The most notable model error is in the Pacific sector where the wind stress strength and position are significantly underestimated by most of the models. The GCM mean (denoted with curly brackets) zonal mean (denoted with square brackets) strength, position and width are {[ητ]} ∼ 0.18 ± 0.02 Pa, {[Φτ]} ∼ −48.6 ± 1.6° and {[στ]} ∼ 11.5 ± 0.6° respectively. (Here, and henceforth the interval will indicate the 2.5% to 97.5% percentile intervals for the parameter values.) Clearly, the simulated climatological wind stress is systematically too weak (by about 10%), too equatorward (by about 4°) and too narrow (by about 1°) relative to the reanalyses. We shall keep these model biases in mind as we continue forward with the analysis.

Figure 2.

Climatological (1991–2000) (top) zonal wind stress strength (ητ(λ)) and (bottom) position (Φτ(λ)). The dark shading shows the 2.5% to 97.5% percentile intervals assuming normally distributed GCM parameter values.

3. Results

3.1. Zonal Wind Stress Changes

[7] From the pre-industrial period to the end of the 20th century the simulated changes in the zonal wind stress parameters are Δ{[ητ]} ∼ 0.009 ± 0.004 Pa, Δ{[Φτ]} ∼ −0.87 ± 0.27° and Δ{[στ]} ∼ −0.17 ± 0.20°, respectively. In words, the GCMs indicate a modest, yet statistically significant, wind stress strengthening and poleward shifting over the 20th century. Toward the end of the 21st century, as the concentration of atmospheric GHGs increases, these changes are projected to become much more pronounced. The simulated 20th through 21st century change in the zonal wind stress parameters are Δ{[ητ]} ∼ 0.038 ± 0.008 Pa, Δ{[Φτ]} ∼ −2.8 ± 0.9° and Δ{[στ]} ∼ −0.7 ± 0.5°, respectively. In percentage terms, the strength of the simulated zonal wind stress increases by about 25%. Despite minor local differences in the response, all of the models predict a circumpolar strengthening and poleward shifting of the mid-latitude zonal wind stress over the Southern Ocean, as shown in Figure 3 for a subset of the GCMs. The subset is for those GCMs for which control, 20th century and A2 scenario output is available for Ψ(λ, ϕ). Locally, the zonal wind stress in the Pacific sector displays particularly large changes: strengthening by about 40%, shifting poleward by 3.5° and narrowing by 1.5°.

Figure 3.

Simulated pre-industrial (1851–1860) and end-of-the 21st century (2091–2100) (left) profiles of zonal wind stress strength (ητ(λ)) and (right) position (Φτ(λ)) for a subset of the 12 GCMs. The subset is for those GCMs for which control, 20th century and A2 scenario output is available for Ψ(λ, ϕ). Figure 3 (left) shows that τx strengthens so the pre-industrial profiles are below the 21st century profiles, with red indicating strengthening. Figure 3 (right) shows that τx shifts southward so the pre-industrial profiles are above the 21st century profiles, with blue indicating poleward shifting. The encircled numbers on the left correspond to the GCMs listed in the Introduction.

[8] In summary, the GCMs simulate a consistent strengthening and poleward shifting of the zonal wind stress over the Southern Ocean through the 20th and 21st centuries. We now consider to what degree these simulated changes in the zonal wind stress translate into changes in the Southern Ocean circulation.

3.2. Southern Ocean Zonal Circulation Changes

[9] A complete theory capable of predicting how the ACC responds to changes in zonal wind stress is not available. However, a simple theoretical argument based on residual-mean theory [Rintoul et al., 2001; Karsten et al., 2002; Marshall and Radko, 2003] can be made to suggest that strengthening and poleward shifting of the zonal wind stress over the Southern Ocean should result in strengthening and poleward shifting of the ACC [Fyfe and Saenko, 2005]. However, the simple theory is complicated by the fact that in reality the pathway of the ACC is constrained by ocean bottom topography. In addition, global warming involves other changes to the surface climate (e.g., sea-ice melt) which may also affect the Southern Ocean circulation. The theory also assumes steady state conditions while the simulated climate system is evolving in time.

[10] Despite its limitations the residual-mean theory helps explain the simulated response of the ACC to global warming. In particular, as the theory suggests the GCM mean vertically-integrated zonal current strengthens and shifts poleward. Specifically, the 20th through 21st century change in the GCM mean and zonal mean current strength, position, and width are Δ{[ηU]} ∼ 23.6 ± 8.5 m2/s, Δ{[ΦU]} ∼ −1.0 ± 0.9° and Δ{[σU]} ∼ −1.8 ± 0.9°, respectively. We note the relatively small shift but pronounced narrowing. Figure 4 shows the pre-industrial and the end-of-the 21st century profiles of zonal current strength ηU (left) and position ΦU (right) for the GCMs for which Ψ output is available. A comparison of Figure 4 with Figure 3 indicates that the changes in the depth-integrated zonal current are broadly consistent with the changes in zonal wind stress.

Figure 4.

As in Figure 3 but for (left) the depth integrated zonal current strength (ηU(λ)) and (right) position (ΦU(λ)).

3.3. Southern Ocean Meridional Circulation Changes

[11] Simple dynamical considerations indicate that the simulated changes in zonal wind stress must translate into changes in the meridional circulation in the Southern Ocean. We consider the time mean, zonally and vertically integrated approximate momentum balance in the Southern Ocean between meridional Ekman transport and geostrophic transport [Munk and Palmen, 1951; Gille, 1997; Rintoul et al., 2001]:

equation image
equation image

where p is pressure, vg is meridional geostrophic velocity (i.e., the portion of the meridional velocity explained by the zonal pressure gradient), ρo is the reference density of sea water, f is the Coriolis parameter, and H is the depth of the ocean. This balance links the Southern Ocean with the other ocean basins, and is consistent with findings from high-resolution ocean models over a broad range of southern latitudes [Gille, 1997]. The GCM mean 20th through 21st century change in the meridional Ekman transport (i.e., the left-hand-side of (2)) along the path of maximum −τxof (located at Θ = Φτ − στ2 cot Θ) is 4.3 ± 1.6 Sv. This increase would be more than twice as large (i.e., to match the 25% increase in the zonal wind stress strength) if not for the poleward shift of τx. To see this consider that the Ekman transport M along a constant latitude path of length L is −([τx]/ρof)L, from which it follows that ΔM/M ∼ Δ[τx]/[τx] − Δf/f + ΔL/L. For a poleward shift Δf/f > 0 and ΔL/L < 0.

[12] From the point of view of wind-driven changes in the deep ocean circulation, it is of particular interest to consider the unblocked latitudes of Drake Passage (between −55° and −62°). In this region, the zonally integrated pressure gradient everywhere above the shallowest ocean bottom topography is zero. This implies, according to (2) and (3), that a strengthened northward Ekman transport must eventually result in strengthened southward geostrophic currents in the deep ocean (i.e., below about 2000 m). From Figure 5, showing the GCM mean profiles of −([τx]/ρof)L for pre-industrial and end-of-21st century times, we conclude that the deep southward geostrophic transport must increase in the darkly shaded (i.e., unblocked) latitudes by as much as 5–10 Sv. This increasing southward transport in the deep Southern Ocean is much more robust than the decreasing southward transport simulated in the deep western boundary current of the North Atlantic [e.g., Cubasch et al., 2001]. This, of course, does not mean that there is any contradiction between these two results. The water for the intensified deep southward transport in the Southern Ocean need not be supplied from the North Atlantic alone.

Figure 5.

GCM mean profiles of Ekman transport −([τx]/ρof)L for pre-industrial times (1851–1860: blue), the end-of-the 21st century times (2091–2100: red) and their difference (black). The dark-shading indicates the unblocked latitudes of Drake Passage. Note that this calculation is for the original unfitted profiles of τx.

[13] High resolution ocean models (R. Hallberg and A. Gnanadesikan, The role of eddies in determining the structure and response of the wind-driven Southern Hemisphere overturning: Results from the Modeling Eddies in the Southern Ocean project, submitted to Journal of Physical Oceanography, 2005) indicate that zonal wind stress changes such as these may affect the oceanic mesoscale eddy activity. Indeed, changes in the eddy-induced circulation could potentially negate the direct response of the residual overturning circulation on isopycnal surfaces. However, a satisfactory estimate of these meso-scale eddy processes is unobtainable in the GCMs considered here, all of which parameterize the effects of the mesoscale eddies on the large-scale circulation.

4. Summary and Discussion

[14] An ensemble of 12 global climate models simulate a consistent strengthening, poleward shift and narrowing of the zonal wind stress over the Southern Ocean through the 20th and 21st centuries. However, as a group the GCMs produce a present-day zonal wind stress climatological distribution which is too weak, too equatorward and too narrow relative to reanalysis. Simulated changes in the Southern Ocean zonal circulation which are associated with the changes in zonal wind stress include a strengthening, poleward shift and narrowing of the Antarctic Circumpolar Current. We infer, based on balance considerations, that intensifying northward Ekman transport across the Antarctic Circumpolar Current is balanced by increasing southward transport in the deep ocean below about 2000 m. Recent high-resolution ocean model results suggests that zonal wind stress changes such as reported here enhance the mesoscale eddy activity in the Southern Ocean. Finally, we note the importance that changing Southern Hemisphere winds and eddies may have on the oceanic uptake of anthropogenic carbon [Mignone et al., 2006].

Acknowledgments

[15] We acknowledge the modeling groups for providing their data, the PCMDI for collecting and archiving the data, the JSC/CLIVAR Working Group on Coupled Modelling (WGCM) and their Coupled Model Intercomparison Project (CMIP) and Climate Simulation Panel for organizing the analysis activity, and the IPCC WG1 TSU for technical support. The IPCC Data Archive at Lawrence Livermore National Laboratory is supported by the Office of Science, U.S. Department of Energy. We also thank George Boer and Ken Denman for their insightful comments.

Ancillary