• Baldwin, M. P., , and T. J. Dunkerton, 1999: Downward propagation of the Arctic Oscillation from the stratosphere to the troposphere. J. Geophys. Res., 104 , 3093730946.

    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., , and T. J. Dunkerton, 2001: Stratospheric harbingers of anomalous weather regimes. Science, 294 , 581584.

  • Baldwin, M. P., , and D. W. J. Thompson, 2009: A critical comparison of stratosphere–troposphere coupling indices. Quart. J. Roy. Meteor. Soc., 135 , 16611672.

    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., , D. B. Stevenson, , D. D. W. Thompson, , T. J. Dunkerton, , A. J. Charlton, , and A. O’Neill, 2003: Stratospheric memory and skill of extended-range weather forecasts. Science, 301 , 636640.

    • Search Google Scholar
    • Export Citation
  • Charney, J. G., , and P. G. Drazin, 1961: Propagation of planetary disturbances from the lower into the upper atmosphere. J. Geophys. Res., 66 , 83109.

    • Search Google Scholar
    • Export Citation
  • Christiansen, B., 2001: Downward propagation of zonal mean zonal wind anomalies from the stratosphere to the troposphere: Model and reanalysis. J. Geophys. Res., 106 , 2730727322.

    • Search Google Scholar
    • Export Citation
  • Christiansen, B., 2003: Evidence for nonlinear climate change: Two stratospheric regimes and a regime shift. J. Climate, 16 , 36813689.

    • Search Google Scholar
    • Export Citation
  • Gerber, E. P., , L. Polvani, , and D. Ancukiewicz, 2008: Annular mode time scales in the Intergovernmental Panel on Climate Change Fourth Assessment Report models. Geophys. Res. Lett., 35 , L22707. doi:10.1029/2008GL035712.

    • Search Google Scholar
    • Export Citation
  • Gerber, E. P., and Coauthors, 2010: Stratosphere–troposphere coupling and annular mode variability in chemistry–climate models. J. Geophys. Res., 115 , D00M06. doi:10.1029/2009JD013770.

    • Search Google Scholar
    • Export Citation
  • Gillett, N. P., , and D. W. J. Thompson, 2003: Simulation of recent Southern Hemisphere climate change. Science, 302 , 273275.

  • Harnik, N., 2009: Observed stratospheric downward reflection and its relation to upward pulses of wave activity. J. Geophys. Res., 114 , D08120. doi:10.1029/2008JD010493.

    • Search Google Scholar
    • Export Citation
  • Harnik, N., , and R. S. Lindzen, 2001: The effect of reflecting surfaces on the vertical structure and variability of stratospheric planetary waves. J. Atmos. Sci., 58 , 28722894.

    • Search Google Scholar
    • Export Citation
  • Karpetchko, A., , E. Kyrö, , and B. M. Knudsen, 2005: Arctic and Antarctic polar vortices 1957–2002 as seen from the ERA-40 reanalyses. J. Geophys. Res., 110 , D21109. doi:10.1029/2005JD006113.

    • Search Google Scholar
    • Export Citation
  • Lin, P., , Q. Fu, , S. Solomon, , and J. M. Wallace, 2009: Temperature trend patterns in the Southern Hemisphere high latitudes: Novel indicators of stratospheric change. J. Climate, 22 , 63256341.

    • Search Google Scholar
    • Export Citation
  • Neff, W., 1999: Decadal time scale trends and variability in the tropospheric circulation over the South Pole. J. Geophys. Res., 104 , 2721727251.

    • Search Google Scholar
    • Export Citation
  • Neff, W., , J. Perlwitz, , and M. Hoerling, 2008: Observational evidence for asymmetric changes in tropospheric heights over Antarctica on decadal time scales. Geophys. Res. Lett., 35 , L18703. doi:10.1029/2008GL035074.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , and N. Harnik, 2003: Observational evidence of a stratospheric influence on the troposphere by planetary wave reflection. J. Climate, 16 , 30113026.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , and N. Harnik, 2004: Downward coupling between the stratosphere and troposphere: The relative roles of wave and zonal mean processes. J. Climate, 17 , 49024909.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , S. Pawson, , R. Fogt, , J. E. Nielsen, , and W. Neff, 2008: The impact of stratospheric ozone hole recovery on Antarctic climate. Geophys. Res. Lett., 35 , L08714. doi:10.1029/2008GL033317.

    • Search Google Scholar
    • Export Citation
  • Randel, W. J., 1987: A study of planetary waves in the southern winter troposphere and stratosphere. Part I: Wave structure and vertical propagation. J. Atmos. Sci., 44 , 917935.

    • Search Google Scholar
    • Export Citation
  • Randel, W. J., , D. E. Stevens, , and J. L. Stanford, 1987: A study of planetary waves in the southern winter troposphere and stratosphere. Part II: Life cycles. J. Atmos. Sci., 44 , 936949.

    • Search Google Scholar
    • Export Citation
  • Shaw, T. A., , and T. G. Shepherd, 2008: Raising the roof. Nat. Geosci., 1 , 1213.

  • Thompson, D. W. J., , and S. Solomon, 2002: Interpretation of recent Southern Hemisphere climate change. Science, 296 , 895899.

  • Thompson, D. W. J., , M. Baldwin, , and S. Solomon, 2005: Stratosphere–troposphere coupling in the southern hemisphere. J. Climate, 62 , 708715.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131 , 29613012.

  • Waugh, D. W., , W. J. Randel, , S. Pawson, , P. A. Newman, , and E. R. Nash, 1999: Persistence of the lower-stratospheric polar vortices. J. Geophys. Res., 104 , 2719127201.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Vertical–time lag section of the correlation coherence for wave 1 averaged from 45° to 80°S at 500 hPa with vertical levels between 1000 and 1 hPa for 2-month overlapping periods from April–May to November–December and for time lags between −10 and 10 days. Only correlations that are statistically significant at the 99% levels are contoured, and the contour interval is 0.1.

  • View in gallery

    The wave-1 correlation phase in degrees associated with the correlation coherence in Fig. 1 for −4- (solid line) and +4 (dashed line)-day time lags during (left) May–June and (right) October–November.

  • View in gallery

    (top) The climatological seasonal cycle of the vertical wavenumber averaged between 55° and 75°S and (bottom) zonal-mean cross sections of the climatological vertical wavenumber in (left) May–June and (right) October–November in the Southern Hemisphere. The contours (units 10−5 m−1) are shown at 0.01 (thick line); 2, 4 (dashed line); and 6–30 in jumps of three (solid). Shading indicates regions of wave evanescence (m2 < 0).

  • View in gallery

    (top) The climatological seasonal cycle of the meridional wavenumber averaged between 100 and 30 hPa and (bottom) zonal-mean cross sections of the climatological meridional wavenumber in (left) May–June and (right) October–November in the Southern Hemisphere. The contour interval is 1 rad−1, and the thick line represents the 0.01 contour. Dark shading indicates regions of wave evanescence (l2 < 0). Overlying the climatological seasonal cycle of the meridional wavenumber is the seasonal cycle of the vertical wavenumber averaged between 10 and 1 hPa. The light shading indicates regions of wave evanescence (m2 < 0). The dashed line shows the m = 0.01 m−1 contour.

  • View in gallery

    Ratio of the climatological meridional component of the Eliassen–Palm flux at 35°S integrated between 100 and 10 hPa to the climatological vertical component at 100 hPa integrated between 35° and 90°S (solid line) and the ratio of the vertical component at 10 hPa integrated between 35° and 90°S to the vertical component at 100 hPa integrated between 35° and 90°S (dashed line) from June to December. The integrated flux components were smoothed with a 10-day running average.

  • View in gallery

    As in Fig. 1, but for 45°–80°N October–November to April–May.

  • View in gallery

    The wave-1 correlation phase in degrees associated with the correlation coherence in Fig. 6 for −6- (solid line) and +6 (dashed line)-day time lags during (left) November–December and (right) February–March.

  • View in gallery

    As in Fig. 3, but averaged between 55° and 75°N and (bottom) in (left) November–December and (right) February–March in the Northern Hemisphere.

  • View in gallery

    As in Fig. 4, but for (top) The climatological seasonal cycle of the meridional wavenumber averaged between 100 and 30 hPa; (bottom) (left) November–December and (right) February–March in the Northern Hemisphere. Overlying the climatological seasonal cycle of the meridional wavenumber is the seasonal cycle of the vertical wavenumber averaged between 3 and 1 hPa.

  • View in gallery

    As in Fig. 5, but at 35°N and integrated between 35° and 90°N from October to April.

  • View in gallery

    (left) The climatological seasonal cycle for 2-month overlapping periods of the time-lagged correlation (from −30 to 30 days) of the SAM index between 20 and 1000 hPa (shaded), and the wave-1 cross-coherence correlations between 20 and 500 hPa for wave-1 averaged from 45° to 80°S (solid lines). (right) The time-lagged correlation of the SAM index (shaded) and wave-1 cross coherence (isolines) at the 20-hPa levels with all levels from 1000 to 1 hPa for the September–December period.

  • View in gallery

    As in Fig. 11, but for the NAM index (shaded) and the wave-1 cross-coherence correlations for wave-1 averaged from 45° to 80°N (solid lines), and (right) the NAM index and cross-coherence correlations for the December–March period.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 56 56 8
PDF Downloads 43 43 2

Downward Wave Coupling between the Stratosphere and Troposphere: The Importance of Meridional Wave Guiding and Comparison with Zonal-Mean Coupling

View More View Less
  • 1 Center for Atmosphere Ocean Science, Courant Institute of Mathematical Sciences, New York University, New York, New York
  • 2 Cooperative Institute for Research in Environmental Sciences, University of Colorado, and NOAA/Earth System Research Laboratory/Physical Sciences Division, Boulder, Colorado
  • 3 Department of Geophysics and Planetary Sciences, Tel Aviv University, Tel Aviv, Israel
© Get Permissions
Full access

Abstract

The nature of downward wave coupling between the stratosphere and troposphere in both hemispheres is analyzed using the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) dataset. Downward wave coupling occurs when planetary waves reflected in the stratosphere impact the troposphere, and it is distinct from zonal-mean coupling, which results from wave dissipation and its subsequent impact on the zonal-mean flow. Cross-spectral correlation analysis and wave geometry diagnostics reveal that downward wave-1 coupling occurs in the presence of both a vertical reflecting surface in the mid-to-upper stratosphere and a high-latitude meridional waveguide in the lower stratosphere. In the Southern Hemisphere, downward wave coupling occurs from September to December, whereas in the Northern Hemisphere it occurs from January to March. A vertical reflecting surface is also present in the stratosphere during early winter in both hemispheres; however, it forms at the poleward edge of the meridional waveguide, which is not confined to high latitudes. The absence of a high-latitude waveguide allows meridional wave propagation into the subtropics and decreases the likelihood of downward wave coupling. The results highlight the importance of distinguishing between wave reflection in general, which requires a vertical reflecting surface, and downward wave coupling between the stratosphere and troposphere, which requires both a vertical reflecting surface and a high-latitude meridional waveguide.

The relative roles of downward wave and zonal-mean coupling in the Southern and Northern Hemispheres are subsequently compared. In the Southern Hemisphere, downward wave-1 coupling dominates, whereas in the Northern Hemisphere downward wave-1 coupling and zonal-mean coupling are found to be equally important from winter to early spring. The results suggest that an accurate representation of the seasonal cycle of the wave geometry is necessary for the proper representation of downward wave coupling between the stratosphere and troposphere.

* Current affiliation: Lamont-Doherty Earth Observatory and Department of Applied Physics and Applied Mathematics, Columbia University, Palisades, New York

Corresponding author address: Dr. Tiffany A. Shaw, Center for Atmosphere Ocean Science, Courant Institute of Mathematical Sciences, New York University, 251 Mercer St., New York, NY 10012. Email: tshaw@cims.nyu.edu

Abstract

The nature of downward wave coupling between the stratosphere and troposphere in both hemispheres is analyzed using the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) dataset. Downward wave coupling occurs when planetary waves reflected in the stratosphere impact the troposphere, and it is distinct from zonal-mean coupling, which results from wave dissipation and its subsequent impact on the zonal-mean flow. Cross-spectral correlation analysis and wave geometry diagnostics reveal that downward wave-1 coupling occurs in the presence of both a vertical reflecting surface in the mid-to-upper stratosphere and a high-latitude meridional waveguide in the lower stratosphere. In the Southern Hemisphere, downward wave coupling occurs from September to December, whereas in the Northern Hemisphere it occurs from January to March. A vertical reflecting surface is also present in the stratosphere during early winter in both hemispheres; however, it forms at the poleward edge of the meridional waveguide, which is not confined to high latitudes. The absence of a high-latitude waveguide allows meridional wave propagation into the subtropics and decreases the likelihood of downward wave coupling. The results highlight the importance of distinguishing between wave reflection in general, which requires a vertical reflecting surface, and downward wave coupling between the stratosphere and troposphere, which requires both a vertical reflecting surface and a high-latitude meridional waveguide.

The relative roles of downward wave and zonal-mean coupling in the Southern and Northern Hemispheres are subsequently compared. In the Southern Hemisphere, downward wave-1 coupling dominates, whereas in the Northern Hemisphere downward wave-1 coupling and zonal-mean coupling are found to be equally important from winter to early spring. The results suggest that an accurate representation of the seasonal cycle of the wave geometry is necessary for the proper representation of downward wave coupling between the stratosphere and troposphere.

* Current affiliation: Lamont-Doherty Earth Observatory and Department of Applied Physics and Applied Mathematics, Columbia University, Palisades, New York

Corresponding author address: Dr. Tiffany A. Shaw, Center for Atmosphere Ocean Science, Courant Institute of Mathematical Sciences, New York University, 251 Mercer St., New York, NY 10012. Email: tshaw@cims.nyu.edu

1. Introduction

Planetary waves represent the most important source of dynamical coupling between the stratosphere and troposphere. They are generated in the troposphere by orography and continent–ocean heating asymmetries and propagate upward into the stratosphere where they either dissipate and initiate a downward-propagating zonal-mean response or they are reflected downward toward the troposphere. The focus of many recent studies has been on zonal-mean stratosphere–troposphere coupling (e.g., Baldwin and Dunkerton 2001; Thompson et al. 2005). This type of coupling has been shown to be important for seasonal prediction in the troposphere during Northern Hemisphere winter (Baldwin et al. 2003). Wave reflection in the stratosphere, which impacts the troposphere (called downward wave coupling), is another important source of dynamical coupling; however, its intraseasonal impact has not been extensively studied.

In the Northern Hemisphere, Perlwitz and Harnik (2003) showed that significant downward wave-1 coupling between the stratosphere and troposphere occurs during January, February, and March (JFM). They used a time-lagged singular value decomposition analysis to isolate downward-propagating wave-1 signals and found statistically significant correlations for time lags when wave-1 patterns in the stratosphere lead those in the troposphere. The wave reflection signal coincided with a vertical reflecting surface (a region of negative vertical wavenumber) in the stratosphere above 10 hPa. The vertical reflecting surface was diagnosed using the wave geometry diagnostic of Harnik and Lindzen (2001). In Perlwitz and Harnik (2003), the JFM period was contrasted with September, October, and November (SON) in the Northern Hemisphere, which did not show significant wave-1 correlations between stratospheric and tropospheric levels for time lags when the stratosphere leads the troposphere. The lack of downward wave coupling in SON was consistent with the wave geometry diagnostic, which showed no reflecting surface during that time.

In the Southern Hemisphere, Randel (1987) noted wave reflection in his study of planetary wave propagation for two 120-day winter seasons (1 June–28 September 1983 and 1984). He used a cross-spectral correlation analysis to study the vertical propagation of planetary waves. The upward propagation time scales from this analysis method agree with those from the time-lagged singular value decomposition analysis used by Perlwitz and Harnik (2003). In the winter of 1983, Randel (1987) found significant correlations for time lags when the stratosphere leads the troposphere. These correlations coincided with downward-pointing Eliassen–Palm (EP) vectors in the lower stratosphere (Randel et al. 1987). However, the cause of the wave reflection was not studied in detail because the focus of the study was on upward-propagating waves, which represent the dominant signal. Harnik and Lindzen (2001) used their wave geometry diagnostic to show that a vertical reflecting surface forms toward late winter as part of the climatological seasonal cycle in the Southern Hemisphere; however, the impact on the troposphere was not discussed. Overall, much less is known about downward wave coupling in the Southern Hemisphere.

These previous studies provide observational evidence for downward wave coupling between the stratosphere and troposphere in both hemispheres during the winter season. They highlight the importance of a vertical reflecting surface as a condition required for wave reflection. Here, we explore the seasonal cycle of downward wave-1 coupling between the stratosphere and troposphere. We show that downward wave coupling only occurs during those times of the year with a particular wave geometry, which involves both a vertical reflecting surface and a high-latitude meridional waveguide. This leads to a distinction between wave reflection in general, which requires a vertical reflecting surface in the stratosphere, and downward wave coupling between the stratosphere and troposphere, which requires both a vertical reflecting surface and a high-latitude meridional waveguide.

We subsequently investigate the relative roles of zonal-mean and downward wave coupling between the stratosphere and troposphere in the seasonal cycle of both hemispheres. Recent studies have pointed to nonzonal decadal variability and trends in the Southern Hemisphere troposphere and stratosphere (Neff et al. 2008; Lin et al. 2009). Perlwitz and Harnik (2004) showed that winters (i.e., JFM) in the Northern Hemisphere are either reflective and characterized by downward wave-1 coupling or nonreflective and characterized by zonal-mean coupling. Understanding the roles of zonal-mean and downward wave coupling in the seasonal cycle is important for understanding atmospheric variability and its response to climate change. A better characterization of dynamical coupling between the stratosphere and troposphere on intraseasonal time scales is also important for model validation. Only models with a proper representation of both zonal-mean and wave coupling can be used to assess the impact of the stratosphere on tropospheric climate and climate change, which is an important topic of current research (Shaw and Shepherd 2008).

Section 2 describes the reanalysis dataset and analysis methods used in this study. In section 3, we investigate downward wave coupling in the Southern and Northern Hemispheres. In section 4, we assess the relative roles of zonal-mean and downward wave coupling during the seasonal cycle in the Southern and Northern Hemispheres. The paper concludes with a summary and discussion in section 5.

2. Data and analysis approach

The data used in this study are the daily three-dimensional geopotential, monthly-mean zonal-mean zonal wind, and temperature fields from the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) dataset covering the period from January 1979 to August 2002 (Uppala et al. 2005). As noted by Harnik and Lindzen (2001) and Perlwitz and Harnik (2003), vertical reflecting surfaces in both the Southern and Northern Hemispheres typically form above 10 hPa. The ERA-40 dataset was chosen because it extends from the surface to 1 hPa and has a good representation of the stratosphere. In the analysis that follows, all of the ERA-40 results were found to be in good agreement with the results from the Modern Era Restrospective-Analysis for Research and Application (MERRA) and the Japanese 25-year Reanalysis Project (JRA-25) datasets and are therefore robust among the various reanalysis products.

To isolate the upward- and downward-propagating planetary wave signals we use the cross-spectral correlation diagnostic of Randel (1987). The diagnostic considers two geopotential height Fourier coefficients of wavenumber k at two different latitudes and heights:
i1520-0442-23-23-6365-e1
i1520-0442-23-23-6365-e2
and it determines their in-phase and out-of-phase correlations
i1520-0442-23-23-6365-e3
where
i1520-0442-23-23-6365-e4
is the linear correlation at time lag τ between S1k(t) and S2k(t), and σZ1k(σZ2k) is the square root of the variance of Z1k(x, t)[Z2k(x, t)]. The in-phase and out-of-phase correlations are then used to compute the correlation coherence and correlation phase for the two wave coefficients:
i1520-0442-23-23-6365-e5
The correlation coherence and phase are used to understand the correlation between a wave at a given reference level (in the troposphere or stratosphere) with other levels in the vertical as a function of time lag. Here, we focus on zonal wavenumber 1 and use a reference latitude band from 45° to 80° and a reference level at 500 hPa in both hemispheres.1 We choose to average wave 1 over a reference latitude band, rather than consider a specific latitude, to account for the change in position of the maximum heat flux throughout the seasonal cycle. Note that the correlation coherence and phase do not provide any information about the zonal propagation of the wave. However, correlations at sufficiently small time lags likely reflect the stationary component. We high-pass filter the wave signals to remove variability greater than a year so as to focus on intraseasonal variability. The upward-propagating wave-1 signal is identified by statistically significant correlations for time lags when the reference wave-1 field at 500 hPa leads the wave field at levels in the stratosphere. Here, we are interested in the downward-propagating wave-1 signals (associated with wave reflection), which are identified by significant correlations for time lags when the reference wave field lags the wave field at levels in the stratosphere. Upward or downward wave propagation is confirmed using the correlation phase: upward (downward)-propagating waves have a westward (eastward) phase tilt with height. Further details regarding the cross-spectral correlation diagnostic can be found in section 2b of Randel (1987).
The second analysis technique used is the wave geometry diagnostic of Harnik and Lindzen (2001). The diagnostic separates the more commonly used index of refraction (c.f. Charney and Drazin 1961) into vertical (m) and meridional (l) wavenumber contributions. The wavenumbers are diagnosed from the solution to the wave equation associated with the conservation of potential vorticity in spherical coordinates linearized about a zonal-mean basic state (presented here for illustrative purposes in Cartesian coordinates):
i1520-0442-23-23-6365-e6
where ψ is the weighted wave geopotential height, z and y are the vertical and meridional coordinates, N2 and f are the zonal-mean basic-state Brunt–Väisälä frequency and Coriolis parameter, and
i1520-0442-23-23-6365-e7
is the squared index of refraction with U, qy is the basic-state zonal-mean zonal wind and meridional potential vorticity gradient, c is the wave phase speed, and h is the scale height. The last equality in (7) is the corresponding dispersion relation. Here, we focus on zonal wavenumber 1 and set c = 0, that is, we consider stationary wave 1. The coefficients of the wave Eq. (6) are calculated using monthly-mean zonal-mean zonal wind and temperature data. The vertical and meridional wavenumbers are subsequently diagnosed from the solution to the wave equation:
i1520-0442-23-23-6365-e8
where Re denotes the real part. The separation into vertical and meridional wavenumbers distinguishes between regions that represent barriers for vertical and meridional propagation. As for the squared index of refraction, waves propagate in the vertical direction when m2 > 0, are evanescent when m2 < 0, and are reflected when m2 = 0. Similarly, waves propagate in the meridional direction when l2 > 0, are evanescent when l2 < 0, and are reflected when l2 = 0. The wave geometry diagnostic is used to determine whether the basic state for a given month in the seasonal cycle is reflective in the vertical and meridional directions. Further details can be found in Harnik and Lindzen (2001).

3. Downward wave coupling and meridional wave guiding

a. Southern Hemisphere

We examine downward wave coupling in the Southern Hemisphere by considering wave-1 propagation in 2-month overlapping periods from April to December. Figure 1 shows the wave-1 cross correlations with a reference latitude band from 45° to 80°S at 500 hPa with vertical levels from 1000 to 1 hPa for time lags between −10 and 10 days. Correlations that exceed 0.18 are statistically significant at the 99% level and only significant correlations are contoured in Fig. 1. Since the reference level is in the troposphere, positive lags have the troposphere leading the stratosphere, whereas negative lags have the stratosphere leading the troposphere. The correlations only reflect intraseasonal variability because we have removed variability greater than 1 yr as discussed in section 2. According to the cross correlations there is significant upward propagation (significant correlations for positive lags) from April to December. The upward propagation time from 500 to 5 hPa varies throughout the seasonal cycle from 3 to 4 days in agreement with Randel (1987). The correlations are very weak during November–December, which is expected because of the transition to westward zonal winds and the formation of a critical surface for stationary wave 1—that is, where the zonal wind equals zero (Charney and Drazin 1961). Figure 2 shows the vertical phase structure in May–June (left panel) and October–November (right panel) for +4-day lag (dashed line), which corresponds to the time lag when the correlations maximize in the stratosphere. The upward propagation for positive lags is confirmed by the vertical phase structure from representative 2-month periods. The phase tilt is westward with height above 500 hPa in both May–June and October–November, which indicates upward wave propagation.

While there is significant upward propagation during most of the year, there are only significant correlations for downward propagation (i.e., for negative lags) from September to December. The correlations for negative lags maximize in October–November. The downward propagation time from 5 to 500 hPa is approximately 4 days. Figure 2 shows that the phase tilt for the −4-day-lag correlation (solid line) is eastward with height above 500 hPa in both May–June and October–November, which confirms that the negative time lags are associated with downward wave propagation. In October–November, there is a very large eastward phase tilt between 200 and 100 hPa and the tilt in the stratosphere is very small. A similarly large westward phase tilt between 200 and 100 hPa is found for the upward-propagating wave pattern (Fig. 2, dashed line, right panel). The combined phase structure of the upward- and downward-propagating waves is suggestive of a vertical standing wave and is indicative of a wave in the presence of a permanent vertical reflecting surface. The phase tilt with height in May–June is much smoother and is not suggestive of a standing wave pattern.

To better understand the nature of downward wave coupling throughout the seasonal cycle, we consider the evolution of the wave geometry. Figure 3 shows the climatological seasonal cycle of the zonal wave-1 vertical wavenumber (m) averaged between 55° and 75°S (top panel) along with zonal-mean cross sections of the vertical wavenumber during May–June (bottom-left panel) and October–November (bottom-right panel). The shading indicates regions of wave evanescence, which are bounded by the m = 0 surface, called a vertical reflecting surface. The solid line indicates the m = 0.01 m−1 surface. The dashed lines indicate regions where m = 2, 4 m−1, that is, where m is small. It is clear from the seasonal cycle of the vertical wavenumber that there are two times during the year when there is a vertical reflecting surface in the stratosphere: early winter (May–July) and late spring (September–November). Evidently, the evanescent (shaded) region below 100 hPa from March to September is too narrow (much smaller than a vertical wavelength) to block upward wave propagation because the cross correlations for positive lags are large during that time (see Fig. 1). The reflecting surface during October–November extends farther toward the equator than the vertical reflecting surface in May–June, which is confined toward the pole. Note that the lowering of the evanescent region from October to January is consistent with the formation of a critical surface for planetary waves discussed previously.

The climatological seasonal cycle of the vertical wavenumber suggests that there should be wave reflection in both May–June and October–November. However, the cross correlations only show statistically significant downward wave coupling during the latter period. The two diagnostics suggest that a vertical reflecting surface is not sufficient for downward wave coupling between the stratosphere and troposphere. We thus examine the second wave geometry diagnostic, which is the meridional wavenumber. Figure 4 shows the seasonal cycle of the zonal wave-1 meridional wavenumber (l) averaged between 100 and 10 hPa (top panel) along with zonal-mean cross sections of the meridional wavenumber during May–June (bottom-left panel) and October–November (bottom-right panel). The dark-shaded regions represent regions of wave evanescence bounded by the l = 0 surface, which represent a boundary to meridional propagation. The solid line indicates the l = 0.01 rad−1 surface. We see that meridional propagation is confined to a specific latitudinal region, which we refer to as the meridional waveguide. Overlying the seasonal cycle of l is the seasonal cycle of m averaged between 10 and 1 hPa. The lighter-shaded regions represent regions of vertical wave evanescence, and the dashed line represents the m = 0.01 m−1 surface. In the bottom panels of Fig. 4, we also include the shading representing vertical wave evanescence and the m = 0.01 m−1 surface, which is identical to the solid lines in the bottom panels of Fig. 3. Figure 4 clearly indicates that throughout most of the seasonal cycle (January–August) there is a very wide meridional waveguide (the north and south evanescent regions are separated by 30°–40° of latitude). This wide meridional waveguide allows waves to propagate equatorward into the subtropical region where they would encounter a subtropical critical surface, which extends to the equatorward boundary of the waveguide. From September to December, however, there is a high-latitude meridional waveguide from approximately 40° to 70°S. Thus, unlike the case for the vertical wavenumber, there is a clear difference in the meridional wavenumber between the May–June and October–November periods.

The combined l and m seasonal cycles in Fig. 4 reveal that from October to December the high-latitude meridional waveguide is bounded above by a vertical reflecting surface. This configuration forms a bounded wave geometry (cf. the October–November zonal-mean cross section, Fig. 4, bottom-right panel), and it coincides with the maximum downward wave coupling in Fig. 1. In contrast, the wave geometry during May–June is not bounded (Fig. 4, bottom-left panel). In particular, there is no high-latitude meridional waveguide, and thus the waves can propagate meridionally into the subtropics instead of propagating vertically. Even if the waves were to propagate vertically the vertical reflecting surface does not extend far enough equatorward to induce wave reflection. As a result, there is no significant downward wave coupling during May–June.

The hypothesis is that the absence of a high-latitude meridional waveguide in May–June results in more meridional wave propagation instead of strict vertical propagation. A high-latitude waveguide acts to guide vertically propagating waves upward toward the vertical reflecting surface, which increases the likelihood of downward wave coupling. To confirm that meridional wave guiding is important for guiding vertically propagating planetary waves we consider the seasonal cycle of wave-1 EP fluxes. Figure 5 shows the climatological seasonal cycle of the ratio of the meridional component of the wave-1 EP flux at 35°S averaged between 100 and 10 hPa to the vertical flux at 100 hPa averaged between 35° and 90°S (solid line) from June to December. The ratio quantifies how much of the wave-1 activity flux entering the stratosphere at 100 hPa between 35° and 90°S propagates equatorward. It decreases by a factor of 3 between May–June and October–November. This confirms that there is significant meridional propagation in June–August, which is consistent with the lack of a high-latitude meridional waveguide during that time. The meridional propagation is significantly decreased from September–November, which is consistent with the presence of a high-latitude meridional waveguide. Figure 5 also shows the ratio between the vertical flux at 10 hPa and the vertical flux at 100 hPa both averaged between 35° and 90°S (dashed line). This ratio quantifies how much of the wave-1 activity flux entering the stratosphere at 100 hPa between 35° and 90°S propagates vertically through the stratosphere and reaches 10 hPa. The ratio peaks during midwinter (August–September) at the time when there is no reflecting surface. This also agrees with the seasonal evolution of the wave geometry. Note that the sum of the two EP ratios exceeds one during certain times, which suggests the stratosphere is a source of wave 1 during these times of the year. Randel et al. (1987) noted a wave source in the upper troposphere (between 300 and 100 hPa). These EP flux ratio diagnostics support the hypothesis that a bounded wave geometry, particularly in the meridional direction, is important for preventing meridional propagation toward the equator and for guiding the waves upward toward the vertical reflecting surface and subsequently back downward toward the troposphere.

b. Northern Hemisphere

Given the importance of meridional wave guiding in the Southern Hemisphere, we also examine its role in the Northern Hemisphere. As discussed in the introduction, Perlwitz and Harnik (2003) investigated wave-1 reflection during JFM and SON (3-month-average periods) in the Northern Hemisphere from 1979 to 2001. They argued that an index based on a vertical reflecting surface was useful for isolating years with downward wave coupling. In the previous section, it was shown that in the Southern Hemisphere the presence of a vertical reflecting surface is not a sufficient criterion for downward wave coupling (cf. winter in the Southern Hemisphere). Since Perlwitz and Harnik (2003, 2004) only investigated wave reflection during fall and mid-to-late winter, we investigate the full seasonal cycle of downward wave coupling and wave geometry. We repeat all of the diagnostics from the previous subsection.

Figure 6 shows the seasonal evolution based on 2-month overlapping periods from September to May of the wave-1 cross correlations, with a reference latitude band from 45° to 80°N at 500 hPa with vertical levels from 1000 to 1 hPa for time lags between −10 and 10 days. There is significant upward planetary wave propagation (significant correlations for positive lags) from September to March. In general, the correlations for positive lags in the Northern Hemisphere are statistically significant over a wider range of time lags than those in the Southern Hemisphere. Figure 7 shows the vertical phase structure in November–December (left panel) and February–March (right panel) for +6-day lag (dashed line). The upward propagation for positive lags is confirmed by the vertical phase structure from representative 2-month periods. The phase tilt is westward with height above 500 hPa in both November–December and February–March, which indicates upward wave propagation.

In the Northern Hemisphere, the seasonal cycle of the cross correlations associated with downward wave propagation (correlations for negative time lags in Fig. 6) have a similar evolution as those in the Southern Hemisphere except that they occur 2 months earlier in the seasonal cycle. The shift in the seasonal evolution of the correlations between the two hemispheres is consistent with the fact that planetary waves are stronger in the Northern Hemisphere. Note that the downward wave correlations during April–May are not accompanied by significant upward correlations. Thus, we assume that during April–May the coupling will be dominated by different dynamics related to the final vortex breakdown, in particular, nonlinear wave dynamics and reflection from a critical surface rather than simple linear reflection from a vertical reflecting surface. Figure 7 shows that the phase tilt for the −6-day-lag correlation (solid line) is eastward with height above 500 hPa in both November–December and February–March, which confirms that the negative time lags are associated with downward wave propagation. The phase tilt with height in both November–December and February–March is very smooth and not indicative of a standing wave pattern as found in the Southern Hemisphere during October–November.

Figure 8 (top panel) shows the climatological seasonal cycle of the zonal wave-1 vertical wavenumber averaged between 55° and 75°N along with zonal-mean cross sections of the meridional wavenumber during November–December (bottom-left panel) and February–March (bottom-right panel). As in the Southern Hemisphere, there are two times during the year when there is a vertical reflecting surface in the stratosphere: early winter (November–December) and late winter (February–March). The vertical reflecting surface is very high in the stratosphere in November–December and very low from March onward. In November–December and February–March, the vertical reflecting surface extends from 1 to 3 hPa; however, the vertical reflecting surface in November–December extends farther south than that in February–March (Fig. 8, bottom panels). The seasonal cycle of the vertical wavenumber agrees with the results of Perlwitz and Harnik (2003): the 3-month JFM average clearly has a vertical reflecting surface, whereas the 3-month SON average does not. We know from Perlwitz and Harnik (2003) that the vertical reflecting surface during JFM is not a permanent reflective surface, it is associated with the interannual variability of the stratospheric jet, which results from major stratospheric sudden warming events.

The lack of downward wave coupling during November–December despite the presence of a vertical reflecting surface can once again be understood using the meridional wavenumber. Figure 9 (top panel) shows the climatological seasonal cycle of the zonal wave-1 meridional wavenumber averaged between 100 and 30 hPa along with zonal-mean cross sections of the meridional wavenumber during November–December (bottom-left panel) and February–March (bottom-right panel). Overlying the seasonal cycle of l is the seasonal cycle of m averaged between 3 and 1 hPa. As in the Southern Hemisphere, there is a very wide meridional waveguide throughout most of the year (April–December). This wide meridional waveguide allows for meridional propagation into the subtropics where the waves would encounter a subtropical critical surface. From January to March, however, there is a well-defined high-latitude meridional waveguide from approximately 45° to 75°N. In February–March the climatological wave geometry is not completely bounded as it is during October–November in the Southern Hemisphere (cf. the bottom-right panels in Figs. 4 and 9). However, there is still a high-latitude meridional waveguide, which is important for focusing wave energy in the vertical. Perlwitz and Harnik (2003) showed that downward wave coupling during individual JFM months is associated with a bounded wave geometry (see their Figs. 12c and d).2

Finally, we confirm that meridional wave guiding is important for focusing planetary wave propagation in the vertical direction by considering the seasonal cycle of wave-1 EP fluxes. Figure 10 shows the climatological seasonal cycle of the ratio of the wave-1 meridional component of the EP flux at 35°N averaged between 100 and 10 hPa to the vertical flux at 100 hPa averaged between 35° and 90°N (solid line) from October to April. This ratio decreases from early to late winter (November–April), though the decrease is weaker than that in the Southern Hemisphere. However, there is still an indication that for a given amount of vertical flux through 100 hPa there is more meridional flux through 35°N during November–December than during February–March, consistent with the meridional wavenumber evolution. Figure 10 also shows the ratio of the vertical flux at 10 hPa and the vertical flux at 100 hPa, both averaged between 35° and 90°N (dashed line). This ratio is nearly constant (with wave pulses that likely reflect the large variability and insufficient data to properly capture the climatology), which is consistent with the existence of downward wave reflection during November–March. However, the relative amount of wave reflection is overall much smaller in the Northern Hemisphere than in the Southern Hemisphere (cf. Figs. 5 and 10). The decrease in the ratio in April–May is consistent with the breakup of the polar vortex and a downward progression of the critical surface for stationary waves, which prevents the waves from reaching 10 hPa. Note that the sum of the two EP ratios is never greater than one, which indicates that the stratosphere is not a source of wave 1 as it was in the Southern Hemisphere. Overall, the EP flux ratio diagnostics support the hypothesis that a high-latitude meridional waveguide plays an important role in downward wave coupling in both hemispheres.

4. The relative role of wave versus zonal-mean coupling in the seasonal cycle

The previous section showed that the seasonal evolution of downward wave coupling and wave geometry for wave 1 in the Southern and Northern Hemispheres are similar in many respects. However, there are also important differences. First, the phase structure of the downward-propagating wave is different in the two hemispheres. In the Southern Hemisphere, the cross correlations and phase for negative lags were suggestive of a standing wave pattern and indicative of a wave in the presence of a permanent vertical reflecting surface. In the Northern Hemisphere, however, the phase tilt with height was not suggestive of a standing wave. This is most likely due to the interannual variability in downward wave coupling in the Northern Hemisphere. Second, the cross correlations for upward-propagating waves in the Northern Hemisphere are statistically significant over a broader range of time lags than those in the Southern Hemisphere (cf. the correlations for positive time lags in Figs. 1 and 6). This suggests a longer time scale of the upward planetary wave activity pulses in the Northern Hemisphere, which can be seen in the EP flux ratios (cf. Fig. 5 to Fig. 10). Harnik (2009) found that the time duration of the upward planetary wave activity pulse entering the stratosphere from the troposphere is an important factor determining whether planetary waves are reflected or absorbed. In particular, short (long) pulses were preferentially reflected (absorbed). Wave absorption is analogous to wave dissipation and induces a downward-propagating zonal-mean response and zonal-mean stratosphere–troposphere coupling. Overall, these differences between the hemispheres suggest that the relative importance of intraseasonal downward wave and zonal-mean stratosphere–troposphere coupling is different between the hemispheres.

A common diagnostic for quantifying the zonal-mean coupling between the stratosphere and troposphere is time-lagged correlations of the northern and southern annular mode (NAM and SAM) time series at 10 hPa with all vertical levels. Baldwin and Dunkerton (1999) and Perlwitz and Harnik (2004) computed time-lagged correlations of the NAM index and Thompson et al. (2005) computed similar correlations for the SAM index. Christiansen (2001) used the zonal-mean zonal wind at 60°N and 60°S to calculate downward zonal-mean correlations in the extratropical Northern and Southern Hemispheres. The zonal-mean coupling is identified by enhanced correlations for positive time lags exceeding a few days. Here, we use the SAM- and NAM-index correlations to assess intraseasonal zonal-mean coupling and compare it to the downward wave coupling between the stratosphere and troposphere. The SAM and NAM indices at each pressure level were determined by applying an empirical orthogonal function analysis on zonal-mean geopotential height data south of 20°S and north of 20°N, respectively (Baldwin and Thompson 2009).

Figure 11 (left panel) shows the seasonal cycle based on 2-month overlapping periods of the time-lagged correlations of the SAM index between 20 and 1000 hPa (shaded contours) and the wave-1 cross correlations between 20 and 500 hPa averaged from 45° to 80°S (solid contours) for time lags between −30 and +30 days. Here, we use the 20-hPa level as the reference level. The 20–1000- and 20–500- hPa-level combinations were selected because the zonal-mean and wave-1 correlations maximize at those levels, respectively. The SAM data were high-pass filtered to remove all variability greater than one year so as to remove the interannual variability associated with trends in the timing of polar vortex breakdown in the stratosphere, which could dominate the zonal-mean correlations between the stratosphere and troposphere in spring. The wave-1 cross correlations from April to December in Fig. 11 are identical to the correlations at 20 hPa in Fig. 1 except that the time lags are reversed. During summer, the wave-1 cross correlations maximize between 0- and −1-day lag and are not associated with upward wave propagation but rather with hydrostatic balance in the column. During winter and spring the wave-1 cross correlations for negative time lags shift to −3- to −4-day lag and are associated with upward wave propagation. Downward wave propagation (solid contours for positive lags) occurs from September to December as discussed in section 3. The seasonal cycle of the SAM-index correlations (shaded contours) suggest that the zonal-mean coupling is weak throughout the seasonal cycle. The largest SAM correlations occur near zero time lag, which is more indicative of instantaneous coupling than downward-migrating wave–mean flow interactions. The clear separation between the wave-1 cross correlations and SAM-index correlations in the seasonal cycle suggests that zonal-mean and downward wave coupling are distinct and dominate during different times of the year. Figure 11 (right panel) also shows the SAM-index correlations (shaded contours) and the wave-1 cross correlations (solid contours) for a 20-hPa reference level as a function of height averaged from September to December (right panel). From September to December the SAM-index correlations are large above 200 hPa but do not extend below 200 hPa. The wave-1 cross correlations, however, maximize around 500 hPa. It is clear that downward wave coupling dominates the downward coupling between the stratosphere and troposphere in the Southern Hemisphere during spring. Zonal-mean coupling during this time is very weak. This is consistent with the shorter time scale of the upward planetary wave activity pulses (Harnik 2009).

The corresponding seasonal evolution of time-lagged correlations of the NAM index is shown in Fig. 12 (left panel). The wave-1 cross correlations from September to May in Fig. 12 are identical to the correlations at 20 hPa in Fig. 6, except that the time lags are reversed because we use the 20-hPa level as a reference level here. During summer, the wave-1 cross correlations maximize between −1- and −2-day lag and are not associated with upward wave propagation. During winter and spring, the wave-1 cross correlations show a clear transition from instantaneous coupling to upward propagation as the correlations for negative time lags in winter maximize between −5- to −6-day lag. Downward wave propagation (solid contours for positive lags) occurs from December to May as discussed in section 3b. The seasonal cycle of the NAM-index correlations (shaded contours) shows significant zonal-mean coupling during the same time period. The time scales of coupling extend beyond 15 days. The zonal-mean coupling in the Northern Hemisphere is likely due to stratospheric sudden warmings, which represent strong variations in the NAM on intraseasonal time scales. Figure 12 (right panel) shows the NAM-index correlations (shaded contours) and the wave-1 cross correlations (solid contours) for a 20-hPa reference level as a function of height averaged from December to March. During the active period, the NAM-index correlations extend below 200 hPa and are largest in the troposphere right at the surface for time lags up to 20 days, which is consistent with previous studies (e.g., Baldwin and Dunkerton 1999). The NAM correlations in the midtroposphere are weak. The downward wave coupling indicated by significant cross correlations at negative time lags maximizes around 500 hPa. Clearly, both zonal-mean and wave processes contribute to downward coupling during Northern Hemisphere winter, consistent with the results of Perlwitz and Harnik (2003). Perlwitz and Harnik (2004) suggested that these processes are more exclusive in their downward impact when investigating individual years. The significant contribution of zonal-mean processes is consistent with the longer time scale of the upward planetary wave activity pulses (Harnik 2009).

5. Summary and discussion

We have examined downward wave-1 coupling between the stratosphere and troposphere in the Southern and Northern Hemispheres using the ERA-40 reanalysis dataset. Wave coupling was identified using the cross-spectral correlation analysis of Randel (1987), which isolates upward- and downward-propagating wave signals. In particular, statistically significant cross correlations for time lags when the stratosphere leads the troposphere indicate downward wave coupling between the two layers. The seasonal cycle of the vertical and meridional wavenumbers was calculated using the diagnostic of Harnik and Lindzen (2001). This diagnostic reveals vertical reflective surfaces and meridional wave guides in the basic-state flow, which are used to understand the wave propagation. Finally, SAM and NAM time-lagged correlations between different vertical levels were compared with wave-1 cross correlations to understand the relative roles of zonal-mean and downward wave coupling between the stratosphere and troposphere throughout the seasonal cycle.

In both the Southern and Northern Hemispheres, downward wave-1 coupling occurs when there is a preferred configuration of the stratospheric jet, which is associated with a high-latitude meridional waveguide and a vertical reflecting surface. In the Southern Hemisphere, this wave geometry forms during September–December as part of the seasonal cycle and is bounded (i.e., the high-latitude meridional waveguide in the lower stratosphere is bounded above by the vertical reflecting surface). In the Northern Hemisphere, there is a high-latitude meridional waveguide and a vertical reflecting surface, and hence significant downward wave coupling, from January to March. However, the climatological wave geometry is not completely bounded because a bounded wave geometry forms as part of the interannual variability.

In both hemispheres, a vertical reflecting surface also forms during early winter; however, there is no high-latitude meridional waveguide and no significant downward wave coupling. In the absence of a high-latitude meridional waveguide, the waves can propagate meridionally toward the subtropics where they are absorbed near the critical surface decreasing the likelihood of downward wave coupling. Thus, the lack of a high-latitude meridional waveguide explains the lack of significant downward wave coupling during early winter. An enhancement of meridional propagation in the absence of a high-latitude meridional waveguide was confirmed in both hemispheres using Eliassen–Palm flux diagnostics.

Perlwitz and Harnik (2003) previously noted the importance of the wave geometry for wave reflection during JFM but emphasized the importance of a vertical reflecting surface. Here, we have shown that a wave geometry involving both a vertical reflecting surface and a high-latitude meridional waveguide is a general condition required for downward wave coupling between the stratosphere and troposphere in both hemispheres. This leads to the distinction between wave reflection in general, which requires a vertical reflecting surface, and downward wave coupling, which requires both a vertical reflecting surface and a high-latitude meridional waveguide.

The analysis of the relative roles of zonal-mean and downward wave coupling during the seasonal cycle revealed that in the Southern Hemisphere downward wave coupling dominates during spring. The intraseasonal zonal-mean coupling maximizes in winter and is very weak. The dominance of downward wave coupling in the Southern Hemisphere is consistent with the shorter time scale of planetary wave activity pulses (Harnik 2009). In the Northern Hemisphere, both downward wave and zonal-mean coupling contribute to intraseasonal stratosphere–troposphere coupling. From December to March, downward wave coupling maximizes in the midtroposphere (at 500 hPa) while zonal-mean coupling maximizes at the surface. The results are consistent with Perlwitz and Harnik (2004), who focused on stratosphere–troposphere coupling during JFM. They also showed that zonal-mean and downward wave coupling dominate during different years.

Our study suggests that a proper representation of the seasonal cycle of the basic-state structure (i.e., the stratification, the position of the jet maximum, and the critical surface, which affect the vertical and meridional wavenumbers) is crucial for obtaining the correct downward wave coupling in an atmospheric general circulation model. The inability of current state-of-the-art climate models to capture the seasonal cycle of annular mode coupling between the stratosphere and troposphere (Gerber et al. 2008, 2010) may be caused in part by biases in the intraseasonal variability of downward wave coupling associated with biases in the wave geometry. An intermodel comparison of downward wave coupling and wave geometry is currently in progress.

Our analysis focused on dynamical coupling between the stratosphere and troposphere on intraseasonal time scales. In the Northern Hemisphere, stratosphere–troposphere coupling is characterized by large variability on the interannual and interdecadal time scales (e.g., Christiansen 2003). In the Southern Hemisphere, a 1-month delay in the breakup of the stratospheric polar vortex (Neff 1999; Waugh et al. 1999; Karpetchko et al. 2005) and a pronounced shift toward a positive SAM index (Thompson and Solomon 2002) have been observed since 1979. These changes are most likely caused by stratospheric polar ozone depletion (Gillett and Thompson 2003; Perlwitz et al. 2008). In addition, Neff et al. (2008) noted nonzonal decadal variability around the coast of Antarctica in the troposphere, and Lin et al. (2009) noted nonzonal trends in stratospheric temperature and ozone from 1979 to 2007. The causes for these nonzonal trends are not well understood. Extending the current analysis to include the effects of interannual and interdecadal variability as well as the influence of anthropogenic climate forcings (such as increases in greenhouse gas concentration and ozone depletion and recovery) on the basic state to better understand their effects on downward wave coupling between the stratosphere and troposphere is work in progress.

Acknowledgments

We thank ECMWF for providing the ERA-40 reanalysis dataset. This research has been supported by the Natural Sciences and Engineering Research Council of Canada through a Post Doctoral Fellowship to the first author. J.P.’s contribution was funded by the NOAA Climate Program Office. N.H.’s contribution was funded by Grant 1370/08 from the Israeli Science Foundation. The authors are grateful to two anonymous reviewers for their helpful comments.

REFERENCES

  • Baldwin, M. P., , and T. J. Dunkerton, 1999: Downward propagation of the Arctic Oscillation from the stratosphere to the troposphere. J. Geophys. Res., 104 , 3093730946.

    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., , and T. J. Dunkerton, 2001: Stratospheric harbingers of anomalous weather regimes. Science, 294 , 581584.

  • Baldwin, M. P., , and D. W. J. Thompson, 2009: A critical comparison of stratosphere–troposphere coupling indices. Quart. J. Roy. Meteor. Soc., 135 , 16611672.

    • Search Google Scholar
    • Export Citation
  • Baldwin, M. P., , D. B. Stevenson, , D. D. W. Thompson, , T. J. Dunkerton, , A. J. Charlton, , and A. O’Neill, 2003: Stratospheric memory and skill of extended-range weather forecasts. Science, 301 , 636640.

    • Search Google Scholar
    • Export Citation
  • Charney, J. G., , and P. G. Drazin, 1961: Propagation of planetary disturbances from the lower into the upper atmosphere. J. Geophys. Res., 66 , 83109.

    • Search Google Scholar
    • Export Citation
  • Christiansen, B., 2001: Downward propagation of zonal mean zonal wind anomalies from the stratosphere to the troposphere: Model and reanalysis. J. Geophys. Res., 106 , 2730727322.

    • Search Google Scholar
    • Export Citation
  • Christiansen, B., 2003: Evidence for nonlinear climate change: Two stratospheric regimes and a regime shift. J. Climate, 16 , 36813689.

    • Search Google Scholar
    • Export Citation
  • Gerber, E. P., , L. Polvani, , and D. Ancukiewicz, 2008: Annular mode time scales in the Intergovernmental Panel on Climate Change Fourth Assessment Report models. Geophys. Res. Lett., 35 , L22707. doi:10.1029/2008GL035712.

    • Search Google Scholar
    • Export Citation
  • Gerber, E. P., and Coauthors, 2010: Stratosphere–troposphere coupling and annular mode variability in chemistry–climate models. J. Geophys. Res., 115 , D00M06. doi:10.1029/2009JD013770.

    • Search Google Scholar
    • Export Citation
  • Gillett, N. P., , and D. W. J. Thompson, 2003: Simulation of recent Southern Hemisphere climate change. Science, 302 , 273275.

  • Harnik, N., 2009: Observed stratospheric downward reflection and its relation to upward pulses of wave activity. J. Geophys. Res., 114 , D08120. doi:10.1029/2008JD010493.

    • Search Google Scholar
    • Export Citation
  • Harnik, N., , and R. S. Lindzen, 2001: The effect of reflecting surfaces on the vertical structure and variability of stratospheric planetary waves. J. Atmos. Sci., 58 , 28722894.

    • Search Google Scholar
    • Export Citation
  • Karpetchko, A., , E. Kyrö, , and B. M. Knudsen, 2005: Arctic and Antarctic polar vortices 1957–2002 as seen from the ERA-40 reanalyses. J. Geophys. Res., 110 , D21109. doi:10.1029/2005JD006113.

    • Search Google Scholar
    • Export Citation
  • Lin, P., , Q. Fu, , S. Solomon, , and J. M. Wallace, 2009: Temperature trend patterns in the Southern Hemisphere high latitudes: Novel indicators of stratospheric change. J. Climate, 22 , 63256341.

    • Search Google Scholar
    • Export Citation
  • Neff, W., 1999: Decadal time scale trends and variability in the tropospheric circulation over the South Pole. J. Geophys. Res., 104 , 2721727251.

    • Search Google Scholar
    • Export Citation
  • Neff, W., , J. Perlwitz, , and M. Hoerling, 2008: Observational evidence for asymmetric changes in tropospheric heights over Antarctica on decadal time scales. Geophys. Res. Lett., 35 , L18703. doi:10.1029/2008GL035074.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , and N. Harnik, 2003: Observational evidence of a stratospheric influence on the troposphere by planetary wave reflection. J. Climate, 16 , 30113026.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , and N. Harnik, 2004: Downward coupling between the stratosphere and troposphere: The relative roles of wave and zonal mean processes. J. Climate, 17 , 49024909.

    • Search Google Scholar
    • Export Citation
  • Perlwitz, J., , S. Pawson, , R. Fogt, , J. E. Nielsen, , and W. Neff, 2008: The impact of stratospheric ozone hole recovery on Antarctic climate. Geophys. Res. Lett., 35 , L08714. doi:10.1029/2008GL033317.

    • Search Google Scholar
    • Export Citation
  • Randel, W. J., 1987: A study of planetary waves in the southern winter troposphere and stratosphere. Part I: Wave structure and vertical propagation. J. Atmos. Sci., 44 , 917935.

    • Search Google Scholar
    • Export Citation
  • Randel, W. J., , D. E. Stevens, , and J. L. Stanford, 1987: A study of planetary waves in the southern winter troposphere and stratosphere. Part II: Life cycles. J. Atmos. Sci., 44 , 936949.

    • Search Google Scholar
    • Export Citation
  • Shaw, T. A., , and T. G. Shepherd, 2008: Raising the roof. Nat. Geosci., 1 , 1213.

  • Thompson, D. W. J., , and S. Solomon, 2002: Interpretation of recent Southern Hemisphere climate change. Science, 296 , 895899.

  • Thompson, D. W. J., , M. Baldwin, , and S. Solomon, 2005: Stratosphere–troposphere coupling in the southern hemisphere. J. Climate, 62 , 708715.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131 , 29613012.

  • Waugh, D. W., , W. J. Randel, , S. Pawson, , P. A. Newman, , and E. R. Nash, 1999: Persistence of the lower-stratospheric polar vortices. J. Geophys. Res., 104 , 2719127201.

    • Search Google Scholar
    • Export Citation
Fig. 1.
Fig. 1.

Vertical–time lag section of the correlation coherence for wave 1 averaged from 45° to 80°S at 500 hPa with vertical levels between 1000 and 1 hPa for 2-month overlapping periods from April–May to November–December and for time lags between −10 and 10 days. Only correlations that are statistically significant at the 99% levels are contoured, and the contour interval is 0.1.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 2.
Fig. 2.

The wave-1 correlation phase in degrees associated with the correlation coherence in Fig. 1 for −4- (solid line) and +4 (dashed line)-day time lags during (left) May–June and (right) October–November.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 3.
Fig. 3.

(top) The climatological seasonal cycle of the vertical wavenumber averaged between 55° and 75°S and (bottom) zonal-mean cross sections of the climatological vertical wavenumber in (left) May–June and (right) October–November in the Southern Hemisphere. The contours (units 10−5 m−1) are shown at 0.01 (thick line); 2, 4 (dashed line); and 6–30 in jumps of three (solid). Shading indicates regions of wave evanescence (m2 < 0).

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 4.
Fig. 4.

(top) The climatological seasonal cycle of the meridional wavenumber averaged between 100 and 30 hPa and (bottom) zonal-mean cross sections of the climatological meridional wavenumber in (left) May–June and (right) October–November in the Southern Hemisphere. The contour interval is 1 rad−1, and the thick line represents the 0.01 contour. Dark shading indicates regions of wave evanescence (l2 < 0). Overlying the climatological seasonal cycle of the meridional wavenumber is the seasonal cycle of the vertical wavenumber averaged between 10 and 1 hPa. The light shading indicates regions of wave evanescence (m2 < 0). The dashed line shows the m = 0.01 m−1 contour.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 5.
Fig. 5.

Ratio of the climatological meridional component of the Eliassen–Palm flux at 35°S integrated between 100 and 10 hPa to the climatological vertical component at 100 hPa integrated between 35° and 90°S (solid line) and the ratio of the vertical component at 10 hPa integrated between 35° and 90°S to the vertical component at 100 hPa integrated between 35° and 90°S (dashed line) from June to December. The integrated flux components were smoothed with a 10-day running average.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 6.
Fig. 6.

As in Fig. 1, but for 45°–80°N October–November to April–May.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 7.
Fig. 7.

The wave-1 correlation phase in degrees associated with the correlation coherence in Fig. 6 for −6- (solid line) and +6 (dashed line)-day time lags during (left) November–December and (right) February–March.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 8.
Fig. 8.

As in Fig. 3, but averaged between 55° and 75°N and (bottom) in (left) November–December and (right) February–March in the Northern Hemisphere.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 9.
Fig. 9.

As in Fig. 4, but for (top) The climatological seasonal cycle of the meridional wavenumber averaged between 100 and 30 hPa; (bottom) (left) November–December and (right) February–March in the Northern Hemisphere. Overlying the climatological seasonal cycle of the meridional wavenumber is the seasonal cycle of the vertical wavenumber averaged between 3 and 1 hPa.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 10.
Fig. 10.

As in Fig. 5, but at 35°N and integrated between 35° and 90°N from October to April.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 11.
Fig. 11.

(left) The climatological seasonal cycle for 2-month overlapping periods of the time-lagged correlation (from −30 to 30 days) of the SAM index between 20 and 1000 hPa (shaded), and the wave-1 cross-coherence correlations between 20 and 500 hPa for wave-1 averaged from 45° to 80°S (solid lines). (right) The time-lagged correlation of the SAM index (shaded) and wave-1 cross coherence (isolines) at the 20-hPa levels with all levels from 1000 to 1 hPa for the September–December period.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

Fig. 12.
Fig. 12.

As in Fig. 11, but for the NAM index (shaded) and the wave-1 cross-coherence correlations for wave-1 averaged from 45° to 80°N (solid lines), and (right) the NAM index and cross-coherence correlations for the December–March period.

Citation: Journal of Climate 23, 23; 10.1175/2010JCLI3804.1

1

We focus on zonal wave 1 because it represents the dominant source of downward wave coupling. We also found evidence for downward wave-2 coupling in the Southern Hemisphere; however, it is much weaker than wave-1 coupling.

2

Perlwitz and Harnik (2003) created an index for downward wave coupling based on the difference of the zonal wind at 2 and 10 hPa averaged between 58° and 74°N. The index was used by Perlwitz and Harnik (2004) to separate years with and without wave coupling. Our results suggest that a wave coupling index should be based on both m and l.

Save