Previous: Trends in temperatures Next: Conclusions Up: Ext. Abst.

 

Brewer Dobson circulation

The Brewer-Dobson circulation is a global-scale cell in the stratosphere in which air rises in the tropics and then moves polewards and downwards, mostly in the winter hemisphere Because it describes Lagrangian-mean. transport it can not be diagnosed directly from the UM integrations. Instead we have calculated the transformed Eulerian mean residual velocities ( $\overline{v^{*}}$, $\overline{w^{*}}$) (Andrews & McIntyre 1976; 1978) which approximate the mean meridional mass transport for seasonally averaged conditions (e.g., Holton 1990). Averaged over the 60 years the residual velocities indicate the model is able to correctly reproduce the classical Brewer-Dobson picture of a meridional overturning circulation (see Fig. 4). In agreement with observations the maximum upwelling occurs in the summer hemisphere.

\begin{figure} \vspace{11.5 cm} \special{psfile=/home/butchart/umfigs/sparcf04.ps hscale=100 vscale=100 hoffset=-90 voffset=-260 angle=0 } \end{figure}

Fig. 4. Sixty year mean residual velocities ( $ \overline{v}^{*}, \overline{w}^{*}$) for each season of run B (run A results are very similar). The contours show the strength, in mms$^{-1}$, of the vertical component, $\overline{w}^{*}$, with dashed contours indicating descent and blue shading ascent.

In each model year the smallest upward mass flux entering the lower stratosphere was in June-August (JJA) and the largest in December-February (DJF)(Fig. 5), presumably because of the stronger extra-tropical wave-driving in the northern winter. In the ``1990s'' the modeled mass fluxes are in broad agreement with mass fluxes derived from observations (Rosenlof 1995). At the same time, the mean vertical velocity from 12$^{\circ}$N to 12$^{\circ}$S (not shown), varied from $\sim0.16$ mms$^{-1}$ in JJA to $\sim 0.3$ mms$^{-1}$ in DJF, slightly less than the 0.2-0.4 mms$^{-1}$ inferred from the so-called ``tape recorder'' signal seen in water vapor measurements (Mote et al. 1995). The tape recorder signal is a manifestation of the annual cycle in the tropical lower-stratospheric temperatures resulting from the annual cycle in the upwelling (Yulaeva et al. 1994). Our simulated annual cycle in tropical lower stratospheric temperatures (see Fig. 6, for run B) had the correct amplitude and, notably, the amplitude did not change significantly over the 60 years, despite the predicted cooling of the lower stratosphere. Consistently, the trends in the tropical upwelling were roughly the same in every season (Fig. 5) with the combined effect of a 18% (19% for run A) increase in the annual mean upwelling by 2051.

\begin{figure} \vspace{7.0 cm} \special{psfile=/home/butchart/umfigs/sparcf05a.p... ...sparcf05b.ps hscale=40 vscale=40 hoffset=200 voffset=-50 angle=0 } \end{figure}

Fig. 5. Seasonal and annually averaged upward mass fluxes at 68 hPa. The curves are 11 year running means and the straight lines are least squares fits to the annual mean results.

Fig. 6. The trend in the mass flux is fairly uniform throughout the year as indicated by the absence of any change in the decadal mean amplitude of the annual cycle in the mean temperatures from 20$^{\circ}$N-20$^{\circ}$S at 68 hPa.

The main reason for this increase in upwelling was increased extra-tropical planetary wave driving. For steady seasonal mean conditions the Cambridge ``downward control'' principle (Haynes et al. 1991) gives the vertical mass flux poleward of latitude $\phi_{0}$ at height $z$, in terms of the vertical integral of the zonal forces, $\cal F$, above; or more specifically

\begin{displaymath}2\pi a \int_{z}^{\infty}\left\{ \frac{\rho_{0} a^{2} {\cal F}... ...\overline{m}/\partial\phi}\right\} _{\overline{m}=const.} dz', \end{displaymath}


where $a$ is the earth's radius and $\rho_{0}$ a basic state density. The integration is up a line of constant zonal mean absolute angular momentum $\overline{m}=\overline{m(z,\phi_{0})}$ (Haynes et al. 1991). At the latitudes used in our calculations these surfaces were nearly vertical, hence we used, instead, constant latitude. For the model, $\cal F$ is the sum of the resolved planetary wave forcing given by the Eliassen-Palm (EP) flux divergence, numerical dissipation and unresolved forcing represented by the parameterized orographic gravity wave drag up to 20 hPa and Rayleigh friction above that. The latter was, however, negligible below 1 hPa and, because of the density weighting of the integrand, had only a small effect on the mass fluxes in the lower stratosphere.

Fig. 7. Seasonal and annually averaged extra-tropical downward mass fluxes at 68 hPa derived from the planetary wave driving by the Cambridge downward control principle. The curves are 11 year running means and the straight lines are least squares fits to the annual mean results.

Quantitatively the mass fluxes derived from the EP flux divergence agreed well in every season with those obtained directly from the residual vertical velocities (cf. Figs. 5 and 7). The downward control mass fluxes were slightly larger, partly because of the absence of a small negative contribution from the Rayleigh friction--in the mesosphere zonal mean zonal winds at the turnaround latitudes (i.e. where the vertical velocities change direction) were generally easterly--and also probably because of the missing contribution from orographic gravity wave drag which was not archived from our runs. However, the dominant contribution to the extra-tropical downward flux at 68 hPa came from the EP flux divergence and, in every season, increased over the 60 years (Fig. 7). Furthermore, the 60 year increase of $6.4\times10^{8}$ Kgs$^{-1}$ in the wave-driven annual mean downward mass flux in run B, for example, balances most of the $8.7\times10^{8}$ Kgs$^{-1}$ increase in the tropical upwelling, confirming that the strengthening of the Brewer-Dobson circulation was predominantly wave driven in the model.

The increased wave driving strengthening the Brewer-Dobson circulation is most likely a direct consequence of more wave activity emanating from the trosposphere in the region of the trunaround latitudes where the residual vertical velocities change direction (Fig. 3). In contrast to the analysis of ```doubled CO$_{2}$'' (Rind et al. 1998), contributions to increased wave-driving from changes in the focussing of the waves was small because of the absence of any significan change in the direction of the EP flux vector (again see Fig.  3)

Previous: Trends in temperatures Next: Conclusions Up: Ext. Abst.