The traditional view in climate science has been that the stratosphere, which represents only about 10-20% of the atmosphere in terms of mass, can play only a limited role in climate change. However, there has been increasing evidence in recent years that the stratosphere is a sensitive component of the climate system, which can affect the troposphere through various coupling mechanisms. This article reviews the current state of knowledge, and discusses possible mechanisms for this coupling.
The temperature structure of the stratosphere represents a balance between radiative and dynamical heating. To a very large extent, radiative transfer in the stratosphere is a clear-sky process and is well understood (Andrews et al. 1987); the only significant exception occurs in the aftermath of a large volcanic eruption. The radiative equilibrium state is dynamically stable nearly everywhere and there are no rapid small-scale adjustments to be considered as there are in the troposphere. The one place where stability is not guaranteed is in the tropics, where the seasonal migration of the solar heating maximum about the equator leads to a process of quasi-horizontal inertial adjustment (Dunkerton 1989). This process has not received much attention, presumably because it is resolved to some extent in climate models, but it is still not well understood in detail. A point of concern in this respect is that the "traditional" form of the hydrostatic primitive equations, widely used in climate models, becomes invalid at the equator (de Verdière et al. 1994).
The dynamical heating is associated with vertical (diabatic) motion which is partly caused by radiative heating, but is primarily caused by the breaking and dissipation of Rossby waves and gravity waves emanating from the troposphere. The extent to which the wave-induced forcing is the dominant factor increases as one considers longer time scales, reaching a wave-induced "downward control" regime in the steady-state limit (Haynes et al. 1991). Thus, on climatological time scales one can regard the stratospheric temperature field as being pulled away from radiative (-inertial) equilibrium by wave-induced forces. In the extratropical stratosphere, the wave-induced forces are almost exclusively westward and act to drive air poleward, thus producing a meridional mass circulation that is generally rising in low latitudes and sinking in high latitudes (Holton et al. 1995). The resulting dynamical heating is negative in low latitudes and positive in high latitudes.
The dominant wave-induced forcing in the stratosphere is believed to come from tropospherically generated planetary-scale Rossby waves, and this forcing maximises in winter because that is the season of greatest tropospheric baroclinicity and greatest land-sea thermal contrasts. This is reflected in the fact that the meridional mass circulation, although two-celled, is dominantly directed towards the winter pole, and it leads to a significant warming (and weakening) of the polar night vortex relative to its radiatively determined state. This points to a mechanism whereby differences in wave-induced forcing lead directly to differences in stratospheric polar night temperatures. Such differences are most apparent in the asymmetry between the Northern and Southern Hemispheres: the Southern Hemisphere, with less zonal inhomogeneity, produces less planetary-wave forcing and therefore less dynamical warming of the vortex; in the Northern Hemisphere, the forcing is sometimes strong enough to completely destroy the vortex in a "sudden stratospheric warming". By the same token, interannual variability and long-term trends in wave-induced forcing can lead directly by this mechanism to corresponding changes in stratospheric temperatures. It is important to note that the temperature effect of wave-induced forcing is to a first approximation zero in the global mean, at any pressure level, because rising motion in one place must be compensated by sinking motion elsewhere; thus the signature of such changes is compensating temperature changes in the tropics and extratropics. This compensation has been noted in the annual cycle, with the coldest tropical temperatures occurring in Northern Hemisphere winter when the wave-induced forcing is greatest (Yulaeva et al. 1994).
There nevertheless remain significant uncertainties in both qualitative and quantitative aspects of the wave-induced forcing. A qualitative uncertainty is that while the wave-induced forcing is greatest in the winter hemisphere, the maximum radiative imbalance and maximum upwelling in tracer fields are found on the summer side of the equator. Various reasons have been advanced for this discrepancy, but they all require knowledge of the tropical angular momentum budget and thus will be difficult to test for two reasons: first, there is great sensitivity to small errors in this budget, and second, the quality of stratospheric wind analyses in the tropics is very poor.
The principal quantitative uncertainties in wave-induced forcing have to do with the role of gravity waves, which are essentially undetected in stratospheric analyses. Inferences regarding the role of gravity waves have therefore had to come from systematic errors in climate models, most notably the well-documented "cold pole" problem. Based on the arguments presented above, the cold pole problem is most plausibly attributed to insufficient wave drag. On the presumption that the planetary-wave forcing in GCMs is reasonable (since planetary waves are well resolved even by relatively coarse models), attention naturally focuses on gravity waves. Although some gravity waves will be represented in GCMs, a GRIPS study has shown that their horizontal wavenumber spectra of kinetic energy are shallow (Koshyk et al. 1999). Thus for most models, most of the drag will be unrepresented; typical horizontal length scales are believed to be in the range 10-200km.
It is widely accepted that the drag exerted by breaking and dissipating gravity waves is responsible for the pole-to-pole solstitial circulation in the mesosphere (Andrews et al. 1987). The principle of downward control shows how mesospheric wave drag can affect temperatures in the stratosphere, particularly in the polar night where there is strong sensitivity of temperatures to dynamical heating (because of the long radiative time scale). Several middle atmosphere GCMs have eliminated most of the cold pole problem by using a strong Rayleigh friction in the upper stratosphere or mesosphere. Although such "tuning" with Rayleigh drag may be an effective way of obtaining reasonable mean states, it can lead to erroneous meridional circulations in response to climate perturbations because Rayleigh drag is not momentum-conserving (Shepherd et al. 1996). More physically based gravity-wave drag parameterisation schemes have been developed, but there is little consensus on the most appropriate framework. More seriously, in all these schemes the amplitude and shape of the gravity-wave spectrum must be determined empirically, but current measurements provide little information on either of these quantities.
As the resolution of climate models increases, more of the gravity-wave spectrum is explicitly resolved and the cold pole bias tends to reduce as more wave drag is explicitly represented (Hamilton et al. 1995; Jones et al. 1997). However, even calculations performed with the GFDL SKYHI GCM at 0.6 degree resolution (which can resolve gravity waves longer than about 100km) reveal a cold bias of up to 20K in the upper stratosphere. Thus, given the extreme computational cost of ultra-high-resolution simulations, this is not a feasible approach at the present time. Moreover, there is no guarantee that the gravity-wave spectrum will be realistic, since the principal forcing mechanisms will be physical parameterisations such as convective adjustment in the troposphere, acting near the grid scale of the model.
Although most attention with regard to gravity-wave drag has focused on the Southern Hemisphere, it is interesting that gravity-wave drag has been found to be crucial for obtaining realistic simulations of the Northern Hemisphere winter in some modest-resolution models (Boville 1995; Beagley et al. 1997). A major scientific issue in climate change involves possible changes in the Arctic wintertime vortex and the impact of those changes on ozone depletion (WMO 1999, Chapter 12; see further discussion below). The fact that the realism of present-day simulations of the Arctic vortex can depend critically on a parameterisation unconstrained by measurements is therefore a cause for concern.
Until recently, a consistent failing of middle atmosphere GCMs has been their inability to simulate the quasi-biennial oscillation (QBO). Now the situation is changing rapidly. First, QBO-like oscillations were found in highly simplified GCMs with better than 1km vertical resolution and very little horizontal dissipation (Takahashi 1996; Horinouchi and Yoden 1997). Subsequently they were found in very high resolution versions of some GCMs (although these simulations are typically quite short). Yet more recently, several groups have found such oscillations in moderate resolution GCMs using a parameterised gravity-wave drag that is sensitive to critical lines. It should be noted that a "QBO-like oscillation" simply means a downward propagating oscillation between equatorial easterlies and westerlies with a quasi-regular period. Whether the period is close to biennial is not really a key feature, since the period is expected to depend on the amplitude of the tropical wave-drag forcing (either resolved or parameterised) which is unlikely to be correct. It remains unclear exactly what is required for a model to produce a QBO-like oscillation, and whether the oscillations that have been seen are occurring for the right reasons. Some models exhibit curious parameter dependences, and some show a tendency for an intriguing seasonal synchronisation.
The distribution of radiatively active trace gases in the stratosphere is an important component of climate forcing, including the shape of the tropopause itself. This distribution is controlled by dynamics in two ways. First, through the temperature distribution which affects chemical reaction rates; and second (generally more importantly), through transport and mixing by winds. The same meridional mass circulation that produces dynamical heating and cooling also transports chemical species poleward in the stratosphere; this is the so-called Brewer-Dobson circulation (Andrews et al. 1987). The balance between this overturning circulation (which on its own has a time scale of about five years) and quasi-horizontal mixing due to breaking planetary waves produces a characteristic picture for long-lived tracers (such as CH4 and N2O) with tropospheric sources and stratospheric sinks of quasi-horizontal tracer distributions having an upward bulge in the tropics. The control of stratospheric chemical distributions by transport and mixing induced by breaking planetary waves points to a possible feedback mechanism between tropospheric dynamics and radiative forcing, via the stratosphere. Thus, it is essential to understand what controls the Brewer-Dobson circulation. In the last several years, there have been significant advances in the understanding and quantification of this process (see WMO 1999, Chapter 7).
A key quantity determining the stratospheric distribution of long-lived trace gases is the length of time it takes to get from the tropical tropopause (the entry point to the stratosphere) to the location in question. Because of mixing, there is no single pathway, but rather a distribution of pathways; an air parcel at a given location is composed of many components with different transport histories and, therefore, different transit times. The distribution of such transit times is known as the "age spectrum", and the mean of the distribution is the "mean age". Although the age spectrum cannot be measured directly, the mean age can be inferred from measurements of long-lived tracers with linear trends, such as SF6 (e.g. Volk et al. 1997); this provides a valuable constraint on models. Generally speaking, model ages are too "young", suggesting that the models mix too rapidly (Waugh et al. 1997). This suggests a potential climate drift problem for simulations with full chemical-radiative interactions, and it can be expected to compromise predictions of stratospheric residence times (e.g. of halogens). These problems may be expected to become more evident in the coming years as groups begin to perform long simulations using full chemical modules.
The picture of a meridional mass circulation modified by quasi-horizontal mixing has been refined in several respects in the last few years. Transport "barriers" have been identified at the edge of the polar vortices as well as in the subtropics. The isolation of tropical upwelling, together with the annual cycle of temperature at the tropical tropopause mentioned earlier, leads to a remarkable phenomenon dubbed the "tropical tape recorder" (Mote et al. 1996): since the tropical tropopause is a local temperature minimum, it controls the amount of water vapour entering the stratosphere through "freeze drying". The annual cycle in temperature is therefore imprinted on the water vapour distribution, which gets carried upward with remarkably little attenuation. From this signal one can infer a vertical transport time scale of about 0.2-0.4 mm/s (greatest in NH winter) --- corresponding to only 20-40m per day --- which is consistent with ascent rates derived from diabatic circulation calculations.
Because the tropopause is higher in the tropics than in the extratropics, there is a dynamical distinction to be made between isentropic surfaces lying entirely within the stratosphere and those which intersect the tropopause (lying in the stratosphere only in the extratropics). The former are well isolated from the troposphere, while on the latter, known as the "lowermost stratosphere" (Holton et al. 1995), rapid quasi-horizontal isentropic mixing with the troposphere is possible. In this region, evidence of intrusions of tropospheric air is evident from transport studies as well as direct dynamical-chemical signatures. Any air mixed into this region cannot drift upwards into the deep stratosphere and will eventually have to find its way back into the troposphere. Nevertheless, the degree of inmixing is a key factor in radiative forcing of climate through its control of the water vapour and ozone distribution in the lowermost stratosphere, which is the part of the stratosphere most crucial for the radiative balance at the surface.
A key factor affecting tropospheric climate is the height of the tropopause. It is rather remarkable that this question, so basic to climate, remains poorly understood. Considerations of radiative-convective adjustment suggest that the tropopause be identified with the top of the region of moist convection. There are, however, important feedbacks from the stratosphere itself. The dynamical heating associated with the Brewer-Dobson circulation acts to raise the tropopause in the tropics and to lower it in the extratropics. In addition, the radiative equilibrium temperature is determined by trace gas distributions in the stratosphere, particularly ozone and water vapour, and these distributions are controlled by the transport and mixing processes described above. In particular, the poleward and downward transport of ozone in the Brewer-Dobson circulation acts, through radiation, to lower the tropopause in the extratropics. Thus, the latitudinal variation in the height of the tropopause is controlled to a considerable extent by stratospheric processes.
There are three principal mechanisms by which the stratosphere can affect tropospheric climate. The first is through radiative transfer, either by changes in the amount of solar radiation that reaches the surface (e.g. after a volcanic eruption), or by changes in the amount of downwelling longwave radiation emitted by the stratosphere (e.g. because of stratospheric ozone depletion). The impact depends very sensitively on the vertical, latitudinal, and seasonal structure of the changes in the radiatively active substances, particularly in the vicinity of the tropopause (Forster et al. 1997; Hansen et al. 1997).
The fact that the distribution of radiatively active substances is controlled by the Brewer-Dobson circulation together with quasi-horizontal inmixing into the lowermost stratosphere emphasises that climate models need to represent these processes with sufficient fidelity in order to capture this sensitivity.
The second and third mechanisms by which the stratosphere can affect tropospheric climate take account of the basic dynamical fact that tropospherically forced waves propagate up, while zonal-mean anomalies propagate down. Thus, the second mechanism is that the stratosphere can affect the "upper boundary condition" of the troposphere by affecting the propagation characteristics of tropospheric waves. This possibility goes back to the classic work of Charney and Drazin, but there has been remarkably little investigation of this issue in recent years. The possibility of wave reflection at the tropopause has obvious implications for regional climate perturbations.
The third mechanism is then the downward propagation of zonal-mean anomalies, whose mechanism is "downward control". Such downward influence has been seen in model studies (Kodera et al. 1996). Since the zonal-mean anomalies are themselves caused by wave-induced forces whose ultimate origin is the troposphere, this provides a purely dynamical troposphere-stratosphere feedback loop, which may account for the well-documented troposphere-stratosphere anomaly correlations seen in observations (Baldwin et al. 1994). Indeed, there has been much recent interest in the possibility that the North Atlantic Oscillation could be coupled with the strength of the wintertime Arctic vortex (Thompson and Wallace 1998), thereby accounting for the unusual strength (and coldness) of the Arctic vortex through most of the 1990s and the resulting halogen-induced severe ozone depletion (WMO 1999, Chapter 7). Whether the stratospheric vortex is an essential component in a feedback loop involving the NAO, or simply responds to the NAO, remains unclear: in long-term climate simulations with coupled models, results from different groups differ in this respect. (However, even a model without a real stratosphere will still have some capacity for this mechanism in its uppermost levels.)
Andrews, D.G., J.R. Holton, and C.B. Leovy, Middle Atmosphere Dynamics, 489 pp, Academic Press, 1987.
Baldwin, M.P., X. Cheng and T.J. Dunkerton, Observed correlations between winter-mean tropospheric and stratospheric circulation anomalies, Geophys. Res. Lett., 21, 1141-1144, 1994.
Beagley, S.R., J. Grandpre, J.N. Koshyk, N.M. McFarlane, and T.G. Shepherd, Radiative-dynamical climatology of the first generation Canadian middle atmosphere model, Atmos-Ocean, 35, 293-331, 1997.
Boville, B.A., Middle atmosphere version of CCM2 (MACCM2): Annual cycle and interannual variability, J. Geophys., Res., 100, 9017-9039, 1995.
de Verdière, A. Colin and R. Schopp, Flows in a rotating spherical shell: the equatorial case, J. Fluid Mech., 276, 233-260, 1994.
Dunkerton, T.J., Non linear Hadley circulation driven by asymmetric differential heating. J. Atmos. Sci., 46, 956-974, 1989.
Forster, P.M. de F., R.S. Freckleton and K.P. Shine, On aspects of the concept of radiative forcing, Clim. Dyn., 13, 547-560, 1997.
Hamilton, K., R.J. Wilson, J.D. Mahlman, and L.J. Umscheid, Climatology of the SKYHI troposphere-stratosphere-mesosphere general circulation model, J. Atmos. Sci., 52, 5-43, 1995.
Hansen, J., M. Sato and R. Ruedy, Radiative forcing and climate response, J. Geophys. Res., 102, 6831-6864, 1997.
Haynes, P.H., C.J. Marks, M.E. McIntyre, T.G. Shepherd, and K.P. Shine, On the "downward control" of extratropical diabatic circulations by eddy-induced mean forces, J. Atmos. Sci., 48, 651-678, 1991.
Holton, J.R., P.H. Haynes, M.E. McIntyre, A.R. Douglass, R.B. Rood, and L. Pfister, Stratosphere-troposphere exchange, Revs. Geophys., 33, 403-439, 1995.
Horinouchi, T., and S. Yoden, Wave-mean flow interaction associated with a QBO like oscillation in a simplified GCM, J. Atmos. Sci., 55, 502-526, 1998.
Jones, P.W., K. Hamilton, and R.J. Wilson, A very high resolution general circulation model simulation of the global circulation in Austral winter, J. Atmos. Sci., 54, 1107-1116, 1997.
Kodera, K., M. Chiba, H. Koide, A. Kitoh, and Y. Nikaidou, Interannual variability of the winter stratosphere and troposphere in the Northern Hemisphere, J. Meteor. Soc. Japan, 74, 365-382, 1996.
Koshyk, J.N., B.A. Boville, K. Hamilton, E. Manzini, and K. Shibata, The kinetic energy spectrum of horizontal motions in middle-atmosphere models , J. Geophys. Res., to appear, 1999.
Mote, P.W., K.H. Rosenlof, M.E. McIntyre, E.S. Carr, J.C. Gille, J.R. Holton, J.S. Kinnersley, H.C. Pumphrey, J.M. Russell III, and J.W. Waters, An atmospheric tape recorder: The imprint of tropical tropopause temperatures on stratospheric water vapor, J. Geophys. Res., 101, 3989-4006, 1996.
Shepherd, T.G., K. Semeniuk, and J.N. Koshyk, Sponge layer feedbacks in middle atmosphere models, J. Geophys. Res., 101, 23447-23464, 1996.
Takahashi, M., Simulation of the stratospheric quasi-biennial oscillation using a general circulation model, Geophys. Res. Lett., 23, 661-664, 1996.
Thompson, D.W.J., and J.M. Wallace, The Arctic Oscillation signature in the wintertime geopotential height and temperature fields., Geophys. Res. Lett., 25, 1297-1300, 1998.
Volk, C.M., J.W. Elkins, D.W. Fahey, G.S. Dutton, J.M. Gilligan, M. Loewenstein, J.R. Podolske, and K.R. Chan, On the evaluation of source gas lifetimes from stratospheric observations, J. Geophys. Res. 102, 25543-25564, 1997.
Waugh, D.W., T.M. Hall, W.J. Randel, K.A. Boering, S.C. Wofsy, B.C. Daube, J.W. Elkins, D.W. Fahey, G.S. Dutton, C.M. Volk, and P. Vohralik, Three-dimensional simulations of long-lived tracers using winds from MACCM2, J. Geophys. Res., 102, 21493-21513, 1997.
WMO, Scientific Assessment of Ozone Depletion: 1998, World Meteorological Organization (WMO), Geneva, 1999.
Yulaeva, E., J.R. Holton, and J.M. Wallace, On the cause of the annual cycle in tropical lower stratospheric temperatures, J. Atmos. Sci., 51, 169-174, 1994.