The Brewer-Dobson circulation : interannual variability and climate change

A quasi-geostrophic theoretical model for the frequency-dependent response of the zonalmean flow to planetary-wave forcing at Northern Hemisphere (NH) midlatitudes was applied to 4D-Var ECMWF data for six extended winter seasons. Linear regression analyses yielded height-dependent estimates for the thermal damping time, and for a scaling parameter which includes the aspect ratio of the meridional to the vertical length scale of the response. The estimated thermal damping time is ~2 days in the troposphere, 7-10 days in the stratosphere, and 2-4 days in the lower mesosphere. The results indicated that the theoretical model is applicable to midlatitude wintertime conditions. In the low-frequency limit, the response to the wave driving is given by the Brewer-Dobson circulation (BDC), which has been the focus of our further research. ERA-40 reanalysis data for the period 1979-2002 were used to examine several factors that significantly affect the interannual variability of the wave driving. The individual zonal wave-1 and wave-2 contributions to the wave driving at 100 hPa exhibit a significant coupling with the troposphere, predominantly their stationary components. The stationary wave-1 contribution to the total wave driving significantly depends on the latitude of the stationary wave-1 source in the troposphere. The results suggest that this dependence is associated with the varying ability of stationary wave-1 activity to enter the tropospheric waveguide at mid-latitudes. The wave driving anomalies were separated into three parts: one part due to anomalies in the zonal correlation coefficient between the eddy temperature an eddy meridional wind, another part due to anomalies in the zonal eddy temperature amplitude, and a third part due to anomalies in the zonal eddy meridional wind amplitude. It was found that year-to-year variability in the zonal correlation coefficient between the eddy temperature and the eddy meridional wind is the dominant factor in explaining the year-to-year variability of the poleward eddy heat flux. Using the ECHAM middle-atmosphere climate model, it was found that the midwinter NH wave driving exhibits a highly significant increase (12%) if CO2 concentrations are doubled. The magnitude and large statistical significance of the increase due to stationary waves only was found to be comparable to that of the total increase. However, dividing the response into the different wavenumber components yielded a more subtle picture, with a decrease in transient wave-1 to less than 50% of the 1×CO2 value and an increase in transient wave-5 to almost the double value. Although transient wave-5 is usually thought not to contribute substantially to the wave driving of the stratosphere, its increase constitutes about 1/6 of the total increase of the wave driving in a 2×CO2 climate.