Does the position of the Southern Ocean westerly winds represent a negative feedback on anthropogenic carbon dioxide?

Increasing stratification associated with global warming has been posited to serve as a positive feedback on global warming, reducing the uptake of anthropogenic carbon dioxide. We suggest that the poleward shift of westerly winds, may represent a negative feedback on anthropogenic carbon dioxide. It is well known that the ocean represents a major sink for anthropogenic carbon dioxide, taking up roughly 1/3 of the total emissions each year (Sabine et al., 2004). The effects of climate change on this uptake have been the subject of much speculation for many years. Sarmiento et al. (1998) showed that both the warming associated with anthropogenic CO2 increase and the increased hydrological cycle resulted in a decrease in the uptake of anthropogenic carbon. However, these results were based on models that did not predict a strong impact of increased radiative forcing from carbon dioxide on the pattern of surface winds.

Particularly since the mid-1970's, both the Northern and Southern Annular Modes have trended toward the high-index state (i.e. high-index years have become more frequent) (Graf, et al., 1995; Hurrell, 1995; Thompson, et al., 2000). In the Southern Hemisphere, this means that the zone of strongest westerlies has shifted poleward into the circumpolar channel by as much as 5° of latitude over the last 40 years. Fyfe et al. (1999) and Shindell et al. (1999) have linked these trends in the annular modes to tropospheric warming. Palynologists, paleoceanographers and glaciologists have identified a larger shift in the same direction during the period of warming at the end of the last ice age (Heusser, 1989; Lamy et al., 1998; Moreno et al., 1999; McCulloch et al., 2000; Stuut and Lamy, 2004).

 

Figure 3: Zonal wind stress (N/m²) around Antarctica for the CM2.0 and CM2.1 Model Runs at the start of the run (Year 1861, left panels) and the end of the run (Year 2060, middle panels). The difference due to a doubling of the atmospheric carbon dioxide concentration is shown in the right panels.

 

Two coupled climate models recently developed at the Geophysical Fluid Dynamics Lab (GFDL) offer an unprecedented opportunity to investigate the impact of wind stress changes on ocean circulation in a coupled system. The models are constructed using two atmospheric codes that solve the equations of motion in slightly different ways but contain virtually identical parameterizations for physical processes such as radiation, convection, and atmospheric boundary layer mixing. The air-sea heat flux and water fluxes are almost similar in the two models: the winds, however, are not. The mid-latitude storm tracks in both hemispheres are shifted poleward in CM2.1 relative to CM2.0, with the larger shift (order 3°-4°) in the Southern Hemisphere (Fig. 1 left panels).  Moreover, in both models changes in radiative forcing result in an intensification of the westerlies over the Southern Ocean Fig 1, right panels) . The magnitude of these differences is much larger than those seen in previous coupled models.  As discussed in the companion talk given by Gnanadesikan, the differences in wind stress in the CM2.0 and CM2.1 control runs result in significant changes in the Southern Ocean circulation. Can we say anything about how both the shift in the control runs and the changes in the scenario runs might affect the carbon cycle?

We demonstrate that a great deal of insight into this question can be gained by examining an ideal age tracer that was run in these models. The ideal age is set to 0 in the surface layer and increases 1yr/yr below this layer. It therefore measures the time since a parcel was last at the surface. If we assume that water at the surface was in equilibrium with the atmosphere we can use the ideal age of a water parcel to calculate an "inferred CO2 burden" associated with this parcel. Strictly speaking, this ideal age will not be correct, as the surface waters are not in complete equilibrium with the atmosphere, but it will give us a rough scaling for how much anthropogenic carbon could be taken up by the circulation. Comparison of the distribution of inferred anthropogenic carbon with the observed distribution of anthropogenic carbon shows that our technique captures all of the major features of the observed distribution. The technique predicts that CM2.1, with its more realistic westerly wind positions, would have a larger anthropogenic uptake than CM2.0 for the present-day control simulation

The differences between the control simulations support our hypothesis that the position of the winds could represent a negative feedback on atmospheric carbon dioxide. When we examine the climate change scenarios we find even more support for the hypothesis. In CM2.1, the uptake over the Southern Ocean becomes more and more important over time, despite the fact that the ventilation of deep waters decreases. Thus the position of Southern Hemisphere winds represents an important feedback on atmospheric carbon dioxide concentration that has been neglected in most climate discussions to date.