Portal de Periódicos da CAPES

Impacts of increases in greenhouse gases and ozone recovery on lower stratospheric circulation and the age of air: Chemistry-climate model simulations up to 2100

2011; American Geophysical Union; Volume: 116; Issue: D7 Linguagem: Inglês

10.1029/2010jd015024

2156-2202

Makoto Deushi, Kiyotaka Shibata,

Climate variability and models

Journal of Geophysical Research: AtmospheresVolume 116, Issue D7 Climate and DynamicsFree Access Impacts of increases in greenhouse gases and ozone recovery on lower stratospheric circulation and the age of air: Chemistry-climate model simulations up to 2100 Makoto Deushi, Makoto Deushi [email protected] Atmospheric Environment and Applied Meteorology Research Department, Meteorological Research Institute, Tsukuba, JapanSearch for more papers by this authorKiyotaka Shibata, Kiyotaka Shibata Atmospheric Environment and Applied Meteorology Research Department, Meteorological Research Institute, Tsukuba, JapanSearch for more papers by this author Makoto Deushi, Makoto Deushi [email protected] Atmospheric Environment and Applied Meteorology Research Department, Meteorological Research Institute, Tsukuba, JapanSearch for more papers by this authorKiyotaka Shibata, Kiyotaka Shibata Atmospheric Environment and Applied Meteorology Research Department, Meteorological Research Institute, Tsukuba, JapanSearch for more papers by this author First published: 13 April 2011 https://doi.org/10.1029/2010JD015024Citations: 7AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Abstract [1] Long-term changes in lower stratospheric wave forcing and the distribution of mean age of air were examined using multidecadal simulations carried out with a chemistry-climate model in which changes in ozone concentration and the climate of the middle atmosphere were projected through the twenty-first century. Changes in wave forcing between future (2085–2099) and past (1985–1999) periods show clear seasonal variation in the simulations. In both summer hemispheres, subtropical wave forcing significantly strengthens and extends to the extratropics in the lower stratosphere, with the exception of the Antarctic region where resolved wave forcing is decreased in spring and summer as a result of an earlier breakdown of the polar vortex in the future period. This summer strengthening and extending in wave forcing is likely related to the westerly wind shift of the lower stratospheric easterly wind. In the future period, mean age of air is decreased at all locations and seasons at 50 hPa. However, the decrease is nonuniform in each season and region. A large decrease is simulated over the northern extratropics in summer, in accordance with a strengthened residual circulation. Over the Antarctic, a local maximum decrease appears in December, related to the earlier breakdown of the polar vortex. Key Points Chemistry-climate model simulations through the twenty-first century Future changes in wave forcing and mean age in the stratosphere 1. Introduction [2] Many comprehensive climate models have been used to project future changes in stratospheric circulation through the twenty-first century. Recently, these climate models have projected increases in the tropical upward mass flux in the lower stratosphere in response to higher concentrations of greenhouse gases (GHGs) [e.g., Butchart and Scaife, 2001; Sigmond et al., 2004; Butchart et al., 2006, 2010; Garcia and Randel, 2008]. [3] The tropical upward mass flux is generally calculated from residual vertical velocities by using transformed Eulerian mean (TEM) equations [Andrews et al., 1987]. It is used as a measure of the strength of stratospheric mean meridional circulation: the Brewer-Dobson circulation (BDC) inferred from the distributions of water vapor and ozone [Brewer, 1949; Dobson, 1956]. The BDC is a global two-cell overturning circulation in the middle atmosphere. It is driven primarily by extratropical wave forcing via the "downward control" principle [Haynes et al., 1991], in which, on an average annual basis, a tropospheric air mass primarily enters the stratosphere in the tropics and returns to the troposphere in the extratropics [e.g., Holton, 1995]. [4] The mean age of stratospheric air [Hall and Plumb, 1994] can be used as a measure of the strength of the BDC. The mean age of air is defined as the mean transport time since a stratospheric air mass previously made contact with the troposphere [e.g., Waugh and Hall, 2002]. It has been suggested that an acceleration of the BDC may lead to a decrease in the mean age of air [Austin and Li, 2006; Garcia et al., 2007; Garcia and Randel, 2008]. Garcia and Randel [2008] attributed an increase in tropical upwelling, accompanied by a decrease in mean age as GHG concentrations rise, mainly to an increase in wave forcing in the subtropical lower stratosphere. [5] Some recent studies used chemistry-climate models (CCMs) to show that long-term changes in ozone have significant effects on the BDC [Austin and Li, 2006; Li et al., 2008; McLandress and Shepherd, 2009; Oman et al., 2009]. CCMs incorporate fully coupled stratospheric chemistry processes and can simulate interactions between radiatively active gases and dynamical fields. Oman et al. [2009] found in their simulations that ozone depletion over the past decades is the main cause of the decrease in mean age and that subsequent ozone recovery through the first half of the twenty-first century contributed to an increase in mean age. However, the net impact of ozone recovery with increasing sea surface temperatures from increasing GHGs still results in a decrease in mean age in their simulations. [6] Although all recent stratospheric models simulate decreases in stratospheric mean age over the past 30 years as a result of changes in GHG and/or stratospheric ozone abundances [Austin and Li, 2006; Garcia and Randel, 2008; Oman et al., 2009; Butchart et al., 2010], measurements of carbon dioxide (CO2) and sulfur hexafluoride (SF6) from balloon flights during the same period show a weak upward trend in mean age (0.24 ± 0.22 years per decade) in the middle stratosphere from 30 to 5 hPa in the northern midlatitudes [Engel et al., 2009]. As Waugh [2009] pointed out, the models may misrepresent the trend, although the large uncertainties in this observational trend should also be taken into account. [7] The main purpose of this study was to investigate the seasonal behavior of long-term changes in wave forcing, residual circulation, and mean age by using CCM simulations from 1960 to 2100, which account for long-term changes in tropospheric and stratospheric circulation induced by trends in ozone and GHG concentrations. Because long-term trends in mean age have been examined mainly by using averaged annual states, less attention has been paid to seasonal variation. Li et al. [2008] showed that BDC trends exhibit different seasonality between past (1960–2004) and future (2005–2100) simulations, in which seasonality is closely related to Antarctic ozone depletion and its projected recovery. However, they did not investigate seasonal behavior in mean age trends. [8] Here, we use transient climate simulations from the CCM developed at the Meteorological Research Institute (MRI-CCM) [Shibata et al., 2005; Shibata and Deushi, 2008a, 2008b]. The simulations were conducted for the period 1960–2100 under the forcing prescribed according to the CCM Validation Activity (CCMVal-2) for Stratospheric Processes and their Role in Climate (SPARC) reference simulation 2 (REF2) scenario [Eyring et al., 2010]. The MRI-CCM simulates the decreasing trend in mean age above 30 hPa over the past 30 years, similarly to other CCMs in CCMVal [Butchart et al., 2010]. Therefore, we focus on the behavior of mean age in the lower stratosphere, where available observational data are too scarce to reliably detect mean age trends. 2. Model and Simulations [9] The dynamics module of the MRI-CCM is the MJ98 general circulation model (GCM) [Shibata et al., 1999]. It employs a spectral-transform method, using triangular truncation at the maximum wave number 42 (T42) with a horizontal resolution of about 2.8° by 2.8° in longitude–latitude space. It has 68 layers (L68) extending from the surface to the mesopause (0.01 hPa ≈ 80 km), with a layer thickness of 500 m between 100 to 10 hPa. The scheme of Hines [1997] is employed as a nonorographic gravity wave (GW) parameterization, and the scheme of Iwasaki et al. [1989] is used for orographic GW forcing, with longer vertical wavelength forcing being switched off. Because shorter vertical wavelength waves do not propagate into the stratosphere but travel in the troposphere, only the nonorographic GW forcing affects stratospheric circulation. Horizontal diffusion is of the form Δ2, with an ordinary e-folding time (18 h) in the troposphere but a much larger e-folding time (150 h) in the middle atmosphere [Shibata and Deushi, 2005]. [10] The MRI-CCM chemistry module incorporates full chemistry processes in the stratosphere with heterogeneous reactions on two types (I and II) of polar stratospheric clouds and sulfate aerosols [Shibata et al., 2005]. Abundances of radiatively active gases, such as ozone (O3), methane (CH4), and nitrous oxide (N2O), prognosticated in the chemistry module are fed back to the radiation scheme in the dynamics module. The MRI-CCM uses an ideal age tracer for the computation of mean age. As a boundary condition, the concentration of the age tracer linearly increases in time at the surface of the Earth. The current MRI-CCM version adopts a new hybrid semi-Lagrangian transport scheme for long-lived chemical species, including the age tracer, greatly decreasing the positive bias in mean age [Eyring et al., 2010] compared with the previous version [Eyring et al., 2006]. Detailed descriptions of the new transport scheme and model performance are given by Shibata and Deushi [2008a, 2008b]. [11] An ensemble of two transient runs for 1960–2100 were conducted under the CCMVal REF2 scenario [Eyring et al., 2007, 2010] with a spin-up run of about 10 years by the MRI-CCM. The two transient runs used the same forcings and boundary conditions but slightly different initial conditions. In this paper, the ensemble mean results of the two REF2 runs are presented. Surface boundary data came from the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emissions Scenarios (SRES) GHG scenario A1B (medium) [Intergovernmental Panel on Climate Change (IPCC), 2000] and the Ab halogens scenario of World Meteorological Organization/United Nations Environment Programme (WMO/UNEP) [2003]. Boundary data for sea surface temperature and sea ice distributions were taken from an IPCC AR4 simulation of an MRI coupled atmosphere–ocean model (MRI-CGCM2.3) [Yukimoto et al., 2005] under the A1B (medium) GHG scenario. Natural forcings resulting from volcanic aerosols and the 11 year solar cycle are not included. The CCMVal-2 intercomparison activities indicate that the MRI-CCM has a larger ozone loss trend than observations and other models in the late 20th century, if we take 1960 as a reference year (i.e., 1960 baseline-adjusted time series) [Eyring et al., 2010]. However, the 1980 baseline-adjusted trend of the MRI-CCM is in much better agreement with other model trends in the extratropics and almost within the 95% confidence interval of the multimodel ensemble, although the trend in the tropics is still larger compared to observations and other models. Considering this ozone trend in the MRI-CCM simulation, the ensemble simulation is only analyzed for the period of 1985–2099 in this paper. [12] The distribution in mean age in three-dimensional (3-D) space was computed in units of days from an interpolation of the simulated monthly mean concentrations of the age tracer. The zonal mean equatorial concentration of the age tracer at 100 hPa is used as a reference for the computation of mean age. To investigate relationships between the wave forcing and distribution of mean age, the Eliassen-Palm (EP) flux and residual circulation were calculated by using a mass-weighted isentropic zonal mean (MIM) analysis [Iwasaki, 1989; Tanaka et al., 2004]. [13] The MIM analysis is a full extension of the conventional TEM analysis and is equivalent to the TEM analysis under quasi-geostrophic assumptions. The MIM analysis has the advantage of expressing finite wave amplitudes, and it treats lower boundary conditions more accurately than the conventional TEM analysis [Tanaka et al., 2004]. Miyazaki and Iwasaki [2008] showed, for example, that in the MIM analysis, the dynamically derived vertical velocity is in good agreement with the diabatically derived one around the Antarctic polar vortex, whereas in the TEM analysis, the dynamically derived vertical velocity tends to be overestimated (underestimated) outside (inside) the Antarctic polar vortex. In this study, the MIM derivation was applied for the daily mean dynamical fields in the REF2 simulations. 3. Results Zonal Winds [14] Climatological means were computed for the past period (1985–1999) and the future period (2085–2099). Figure 1 shows differences between the two periods (the future period minus the past period) for the zonal mean zonal wind (U), zonal wind acceleration from the EP flux divergence (EPFD) of resolved waves and from the parameterized nonorographic GW dissipation (NOGWD), and residual mass stream function for June, July, and August (JJA). In Figure 2, values for December, January, and February (DJF) are shown. We computed 95% significance levels for the differences using the two-tailed paired Student's t test, on the assumption that each year and season are independent. Figure 1Open in figure viewerPowerPoint Latitude–altitude cross sections averaged over June–July–August (JJA) for (a) U (in m s−1) in the past period (1985–1999) and (b) difference (in m s−1) between U in the future period (2085–2099) and the past period (the future period minus the past period); zero-wind contours for the past period (black line) and the future period (green line) are shown. (c and d) Same as Figures 1a and 1b but for EPF (arrows; in kg s−2) and EPFD (shading; in m s−1 d−1), (e and f) same as Figures 1a and 1b but for NOGWD (in m s−1 d−1), and (g and h) same as Figures 1a and 1b but for residual velocities (arrows; in m s−1) and mass stream function (shading; in 109 kg s−1). In Figures 1b, 1d, 1f, and 1h, hatched regions indicate statistically significant areas at the 95% level for the shaded differences. In Figure 1d, arrows show statistically significant differences at the 95% level for which vertical components have absolute values over 400 (kg s−2). In Figure 1h, arrows show statistically significant differences at the 95% level. Figure 2Open in figure viewerPowerPoint Same as Figure 1 but for December–January–February (DJF). [15] In the future period, simulated zonal winds during the summer and winter increase in the midlatitude upper troposphere and stratosphere. The increase is generally larger in the winter hemisphere than in the summer hemisphere (Figures 1b and 2b). Through the thermal wind balance, changes in the zonal winds correspond to the enhanced meridional temperature gradients between the tropical upper troposphere and the lowermost extratropical stratosphere (not shown), which were also simulated in double CO2 experiments using the MJ98 GCM [Kodama et al., 2007]. The midlatitude zero-wind contour in the lower stratosphere in summer shifts upward in the future period as compared to the past period as a result of increases in westerly zonal winds (Figures 1b and 2b). At 50 hPa, the increase in midlatitude zonal winds is simulated not only in summer and winter but also throughout the year (Figure 3a). Figure 3Open in figure viewerPowerPoint Latitude-time cross sections of monthly averaged (a) U (in m s−1) at 50 hPa, (b) EPFZ (in 10−4 kg s−2) at 100 hPa, (c) EPF (in m s−1 d−1), (d) NOGWD (in m s−1 d−1), and (e) meridional and (f) vertical components of residual velocities (in 10−2 m s−1) at 50 hPa for the past period (contour) and differences (shading) between the future and past periods. Hatched areas show statistically significant differences at the 95% level. (g) Same as Figure 3a but for zonally averaged mean age (in years) for the past period. Contour interval is 0.3 years. (h) Same as Figure 3g but for differences between the future and past periods. Contour interval is 0.05 years; light (dark) shading indicates the difference below −0.5 (−0.7) years; the difference is statistically significant at the 95% level in all seasons and areas. [16] Over the Antarctic, the decrease in zonal winds during summer and winter is projected in the future period to occur in the lower to middle stratosphere (Figures 1b and 2b), resulting in the lowering of the zero-wind contour in summer (Figure 2b). At 50 hPa, the Antarctic zonal wind decreases from April to December, and the decrease is seasonally enhanced from October to December, indicating an earlier breakdown of the polar vortex (Figure 3a). The Antarctic zonal wind decrease in the future period is consistent with meridional temperature gradient changes through the thermal wind balance. The meridional temperature gradient in the past period is steep as a result of Antarctic cooling through a radiative response to severe ozone depletion, whereas the gradient in the future period becomes less steep as a result of the projected ozone recovery [e.g., Li et al., 2008]. Observational and reanalysis data clearly indicate that in the past decades, significant cooling with large depletions of ozone has occurred in the austral spring in the lower stratosphere of the Antarctic [Randel and Wu, 1999]. Wave Forcing [17] Over the Antarctic, the vertical components of the EPF (EPFZ) at 100 hPa predominantly decrease from November to January (Figure 3b), accompanying a prominent decrease in EP flux convergence throughout the southern high-latitude stratosphere (Figure 2d), as manifested by an intense divergence change at 50 hPa in Figure 3c. This is primarily a result of changes in traveling conditions for planetary waves: an easterly wind shift in the stratospheric westerly wind (Figure 2b) hinders planetary wave propagation into the stratosphere over the Antarctic, leading to subsequent weaker wave forcing in the southern high-latitude stratosphere [Li et al., 2008]. [18] In the future period, EP flux convergence in JJA increases prominently around the zero-wind contour in the northern midlatitude lower stratosphere, accompanied by enhanced EPFZ (Figure 1d). The area of increased EP flux convergence extends from the subtropics to the high latitudes and tilts slightly upward. Although a similar increase is also found in the southern lower stratosphere, the area is confined to the subtropics with a shallower vertical depth and a weaker increase in EPFZ than in the Northern Hemisphere (NH). In DJF, the southern increase extends poleward, except for the Antarctic, with enhanced EPFZ, whereas the northern increase is confined in the subtropics with a weaker EPFZ increase than in the Southern Hemisphere (SH; Figure 2d). [19] In the northern lower stratosphere, the area and magnitude of the increases in EP flux convergence and EPFZ are significantly broader and larger in summer than in winter (Figures 1d and 2d). Similarly, the area of the southern increase in summer is broader than that in winter, although its magnitude is not markedly different between winter and summer. [20] Figure 3c clearly illustrates the summer enhancement (poleward extension and/or intensification) of EP flux convergence in the seasonal cycle at 50 hPa. Subtropical EP flux convergences are also enhanced in both hemispheres throughout the year, as shown in Figure 3c. However, extratropical EP flux convergences in the future period are enhanced or reduced in both hemispheres, depending on the season. In the NH, EP flux convergence generally decreases from autumn to spring. It increases in summer (May to August), and at high latitudes, it also increases in December. In the SH, EP flux convergence increases in summer (February to April), following a prominent decrease in spring (November to January). In addition, it increases in winter (June to August) over the Antarctic. [21] Coincident with the summer enhancement of EP flux convergence in the future period, EPFZ at 100 hPa intensifies in the midlatitudes in summer (Figure 3b), as was also indicated in the multidecadal simulations of the Canadian Middle Atmosphere Model [McLandress and Shepherd, 2009]. The summer enhancement of EPFZ at 100 hPa partly accounts for the summer increase of EP flux convergence at 50 hPa, although the increases in EPFZ and the EP flux convergence are not always concurrent (Figures 3b and 3c). The upward shift of the zero wind contour (i.e., a westerly wind shift in the stratospheric easterly wind) in the summer midlatitude lower stratosphere is likely related to the increase of EP flux convergence around the zero-wind contour, because the Charney-Drazin theorem [Charney and Drazin, 1961] suggests that its upward shift allows large-scale waves, which are typically dominated by stationary components, to propagate deeper into the lower stratosphere, where they break. [22] In the NH, the EP flux convergence in the lower and middle stratosphere is generally decreased over the midlatitudes and high latitudes during winter, with the decreased EP fluxes in the high latitudes (Figure 2d). However, the EP flux convergence is increased in the subtropical stratosphere as well as the extratropical upper stratosphere. The EP fluxes in the midlatitude lower stratosphere are concurrently increased with more equatorward direction. These changes in the northern EP fluxes and their convergence during winter were similarly indicated in the simulation of double CO2 experiments using the MJ98 GCM [Kodama et al., 2007]. As mentioned above, the increase of westerly zonal winds during winter (Figure 2b), leading to the stronger Arctic vortex, were also simulated in the double CO2 experiments. [23] The stronger Arctic vortex might partly contribute to the future changes of resolved wave activities in the northern winter stratosphere. This is because the stronger westerly zonal winds produce more favorable condition for vertical propagation of planetary waves, increasing the vertical group velocity [Kodama et al., 2007]. As a result, EP flux convergence of planetary waves in the extratropical upper stratosphere increases concurrent with the decrease in EP flux convergence in the lower and middle stratosphere, as shown in Figure 2d. The stronger Arctic vortex might also make propagation of planetary waves more equatorward [Kodama et al., 2007]. Figure 2d shows the increase in equatorward EP fluxes in the midlatitude lower stratosphere, partly accounting for the increase of EP flux convergence in the subtropics. The future changes in wave sources in the lower stratosphere and the troposphere can also affect the increase (decrease) of EP fluxes in the northern midlatitude (high latitude) lower stratosphere during winter (Figure 2d), which is also indicated in the future changes of EPFZ at 100 hPa (Figure 3b). [24] Figures 4a and 4b show vertical profiles of midlatitude EPFD and U in the summer hemispheres. At midlatitudes in summer in the future period, the increase of U in the upper troposphere and lower stratosphere (UTLS) appears to shift the EPFD vertical profiles upward in the UTLS compared with those in the past period. As a result, westward zonal wind acceleration resulting from EP flux convergences becomes stronger around the height of the zero-wind contour at 60 hPa, canceling the zonal wind increase as a secondary effect. Below that, the westward acceleration in the upper troposphere becomes weaker, enforcing the zonal wind increase as a secondary effect (see red dashed lines in Figures 4a and 4b). Figure 4Open in figure viewerPowerPoint Vertical profiles of U (solid lines; in m s−1) and EPFD (dashed lines; in 10−1 m s−1 d−1) averaged over (a) 40°N–50°N in June and (b) 25°S–35°S in January for the future period (green) and the past period (black). The red line indicates the EPFD difference between the future and past periods. (c) Same as Figure 4a but for U (solid lines; in m s−1) and NOGWD (dashed lines; in 10−2 m s−1 d−1) averaged over 30°S–40°S in December. (d) Same as Figure 4c but for vertical fluxes of eastward (solid lines) and westward (dashed lines) momentum (in 10−4 Pa) computed using nonorographic GW parameterization. [25] The parameterized GW forcing substantially affects the lower stratosphere in the simulations of both the past and future periods [Sigmond et al., 2004; Butchart et al., 2006, 2010; Li et al., 2008]. Vertical profiles of the nonorographic GW forcing (NOGWD) and U in the southern midlatitudes in summer are indicated in Figure 4c for both the past and future periods. Those of the vertical fluxes of the eastward and westward GW momentum are also shown in Figure 4d. In the nonorographic GW parameterization scheme [Hines, 1997] used in this ensemble simulation, the GW source spectrum is isotropically launched to eight equally spaced azimuths at the surface, and the GW source is latitudinally symmetric and fixed to be constant, independent of surface atmospheric conditions [Shibata and Deushi, 2005]. Therefore, differences in GW forcing between the future and past periods are generated not through changes in the GW source but through changes in propagation conditions, i.e., the background dynamical fields. [26] In both the future and past periods, eastward GW momentum (solid lines in Figure 4d) is mainly deposited below the middle troposphere due to the lower tropospheric westerly zonal winds. As a result, its vertical flux into the stratosphere is very small, producing no significant GW forcing in the stratosphere. However, considerable numbers of westward GWs propagate into the stratosphere in both the future and past periods, indicating that tropospheric filtering is significantly weaker for them than for eastward GWs. [27] Compared to the past period, vertical flux of westward GW momentum (dashed lines in Figure 4d) into the free troposphere and the lower stratosphere is larger in the future period, primarily owing to the increase in the westerly zonal wind there (solid lines in Figure 4c). Because the westward GWs propagating into the lower stratosphere are almost dissipated below about 30 hPa as a result of background easterly winds in the summer stratosphere, total westward GW momentum deposition in the lower stratosphere becomes larger in the future than the past. As a result, the maximum westward zonal wind acceleration becomes about 1.3 times larger in the future (−0.6 m s−1 d−1) than the past (−0.45 m s−1 d−1) (Figure 4c). Note that westward GW momentum deposition largely accounts for NOGWD in the lower stratosphere. [28] The vertical profiles of NOGWD also show an upward shift of the breaking level of GWs in the future period. As a result, the westward zonal wind acceleration from NOGWD becomes stronger above, and weaker below, the zero wind level (∼50 hPa), as shown by the red line in Figure 4c. This phenomenon is similar to EPFD changes in the future period. Meridional structures in GW forcing between the past and the future periods (Figures 1f and 2f) show that in the lower stratosphere in both summer hemispheres, GWs exert stronger forcing from low to middle latitudes above about 70 hPa in the future period, accompanied by weaker forcing below this level. Stronger GW forcing is also simulated in the winter hemisphere, but it is generally confined to the subtropics in the lower stratosphere. [29] Figure 3d shows the seasonal cycle of the NOGWD difference at 50 hPa between the future and the past periods. Seasonal variation in low and middle latitudes is similar to that of the EPFD differences shown in Figure 3c. The NOGWD in the future period increases in the subtropical lower stratosphere throughout the year. The increased NOGWD is seasonally enhanced in summer, causing it to extend to the midlatitudes. This summer enhancement is more obvious in the SH than in the NH, in contrast to the EPFD increase. It is likely that differences in NOGWD are also related to upward shifts of summer zero-wind lines in the future period. Although the parameterized GW forcing trend in the lower stratosphere has been discussed [e.g., Li et al., 2008; Butchart et al., 2010], an increase in GW forcing, or in EP flux convergence, in the summer lower stratosphere has seldom been addressed. Figure 3d also exhibits significant decrease in NOGWD in the southern midlatitude lower stratosphere from April to August, as a result of the decreased deposition of the westward GW momentum. This is primarily because the large increase of the westerly zonal wind in the southern midlatitude lower stratosphere (Figure 1b) produces more favorable condition for upward propagation of the westward GWs, leading to less deposition of their momentum there. Residual Velocities and Mass Stream Function [30] The residual mass stream function, which consists of a two-celled structure with upwelling in the tropics and downwelling in the extratropics, is stronger in the winter hemisphere than in the summer hemisphere, as shown in Figures 1g and 2g. In JJA in the future period, residual circulation strengthens over the whole northern extratropical stratosphere, especially in the subtropical lower stratosphere (Figure 1h). This strengthening can be inferred from the significant increase (westward acceleration change) of resolved and parameterized wave forcing (Figures 1d and 1f) through downward control. The positive relationship between residua