Coupling of Long‐Term Trends of Zonal Winds Between the Mesopause and Stratosphere in Southern Winter

We examine the relationships between the observed long‐term trends of the zonal wind in the mesopause regions at King Sejong Station (KSS), Antarctica, and wind trends in the Southern Hemisphere (SH) middle atmosphere using the 15‐year data set from KSS meteor radar, Aura MLS and MERRA‐2. During July, significant positive trends of zonal winds appear above z = 90 km and near the stratopause over the KSS, while negative trends exist between the two layers. In the SH winter, the observed mesopause winds correlate positively (negatively) with stratospheric (mesospheric) winds in the polar region, while they exhibit opposite correlations with the low‐latitude winds. The positive mesopause trends of zonal winds near KSS are connected, through the thermal wind relationship, to cooling (warming) trends induced by the upward (downward) trends of residual circulation over the high‐latitude mesosphere and low‐latitude stratosphere (high‐latitude stratosphere), which shows vertical coupling throughout the SH winter middle atmosphere.


Introduction
The mesopause region (z ∼ 80-100 km) is a transition layer between the Earth's neutral lower atmosphere and the ionospheric region where the roles of ionized gases become important (Sinnhuber et al., 2012;Vincent, 2015).The Southern Hemisphere (SH) mesopause is one of the most active areas of vertical coupling processes with substantial upward transfer of momentum and energy by planetary waves (PWs) and gravity waves (GWs) associated with the baroclinity, jet stream and steep orography in the Andes and Antarctic Peninsula (Plougonven et al., 2008;Preusse et al., 2002;Song et al., 2021;Yoshiki et al., 2004).In particular, the SH polar mesopause remains one of the least understood regions of the Earth's atmosphere because of the scarcity of observations of wind and temperature with high temporal resolution and wide spatial coverage.In addition to the observational limitations, numerical models have had difficulties in properly simulating circulations around the SH polar mesopause (Becker & Vadas, 2018;Dempsey et al., 2021) due to the lack of understanding of gravity wave processes and their mutual interaction.
To better understand the SH mesopause region, analysis of the variability and long-term changes in dynamical fields (wind and temperature) is needed, and meteor radars (MRs) are equipment optimized for continuous longterm mesopause observations unaffected by weather conditions.The MR installed at King Sejong Station (KSS; 62.22°S, 58.78°W) in the Antarctic Peninsula has been operated nearly continuously since March 2007 (Jee et al., 2014;Song et al., 2023).Recently, Song et al. (2023) analyzed the climatology, short-and long-period variability, and trends of the winds over KSS using the approximately 15-year (March 2007-November 2021) observations made by the MR at KSS (hereafter KSS-MR).The observed mesopause zonal wind at KSS had distinct positive trends above z = 90 km in July (i.e., SH winter).Using the trends of GW drag (GWD) estimated from KSS-MR observations, they demonstrated that the enhanced positive zonal GWD contributed to the observed wind trends to some extent.However, as discussed in their paper, the trend of the observed mesopause wind at KSS was not completely explained by the GWD trend, with a pattern correlation coefficient between the two trends of only 0.35.Hence, in this study, we examine the possibility of effects of the changes in the wind and temperature in the middle atmosphere below the mesopause on the wind trends in the SH polar mesopause region by investigating the coupling processes occurring throughout the middle atmosphere.
In contrast to a warming in the troposphere, anthropogenic emissions of greenhouse gases (mainly CO 2 ) can cause a cooling effect globally in the mid-to-upper atmosphere (Brasseur & Hitchman, 1988;Cicerone, 1990;Roble & Dickinson, 1989).Long-term trends of dynamics and chemistry are considered to be important not only from a scientific but also from a practical point of view because they can affect the weather and climate of certain areas in the troposphere (Baldwin et al., 2007) and affect the space environment (Laštovička, 2023).Studies on the effect of these long-term trends in the mid-to-upper atmosphere in all parts of our atmosphere have been conducted.Most of these studies have used whole-atmosphere models that have a global coverage (Garcia et al., 2019;Qian et al., 2019;Solomon et al., 2019).Meanwhile, observational studies are relatively limited due to the lack of largescale networks of long-term mid-to-upper atmosphere observations (especially in the SH polar region).
Several previous studies have shown teleconnection patterns between the polar mesopause region and below (Becker & Schmitz, 2003;Karlsson et al., 2007;Körnich & Becker, 2010;Liu & Roble, 2002;Pedatella & Harvey, 2022;Smith et al., 2022).Some studies have conducted correlation analyses to investigate intra-or even interhemispheric coupling processes, including dynamics in the polar mesosphere and lower thermosphere (MLT) (Karlsson & Becker, 2016;Karlsson et al., 2009;Smith et al., 2020;Tan et al., 2012;Xu et al., 2009).These studies have demonstrated that the temperatures in the mesopause region over the winter high latitudes have negative (positive) correlations with the temperatures in the mesosphere (stratosphere) over the same latitude range, while the opposite correlation patterns appear at low latitudes.Residual mean meridional circulation contributes mainly to the teleconnection between different altitudes and latitudes (Karlsson & Becker, 2016;Liu & Roble, 2002;Tan et al., 2012).The objective of this study is to investigate the coupling processes between the polar mesopause and the middle atmosphere below in the SH using observational and reanalysis data, examining potential coupling processes that can induce the observed trend of the wind in the SH winter polar mesopause region.The rest of the paper is organized as follows: First, descriptions of the data used in this study are given in Section 2. A methodology for estimating long-term trends is also presented in this section.Then, in Section 3, we present trends of the observed mesopause wind at KSS.The results of the correlation analysis between the mesopause wind at KSS and the winds in the SH middle and upper atmosphere are also shown.In addition, long-term changes in the observed wind and temperature and the effect of the residual circulation that induces the trends are examined.Finally, a summary and discussion are given in Section 4.

KSS-MR Wind Observations
The KSS-MR observational data from March 2007 to November 2021 (∼15 years) are used to investigate the trends of the mesopause winds at KSS. Detailed operating parameters of the KSS-MR are summarized in Table 1 of Lee et al. (2013).Approximately 15,000-40,000 meteor echoes have been detected daily by the KSS-MR.Using the observed radial velocities of the meteor echoes within the zenith angles of 15°-75°, hourly zonal winds with a vertical interval of 2 km from 80 to 100 km are estimated (Hocking & Thayaparan, 1997).

Aura MLS Observations
To investigate the spatiotemporal characteristics of the winds and temperature in the SH middle and upper atmosphere, version 5.1 of the observational data from the Microwave Limb Sounder (MLS) mounted on the Aura spacecraft (Schwartz et al., 2008) are used.Daily temperature and geopotential height (Z) data with a horizontal resolution of 5°× 5°are used, and they are vertically interpolated at 50 isobaric levels from 250 hPa (∼9.7 km) to 0.001 hPa (∼96.7 km) with vertical resolutions of ∼1.3-2.5 km.The zonal component of the geostrophic wind (u g ) can be estimated from the geopotential height using the following equation (Andrews et al., 1987): Here, Φ is the geopotential (≡ gZ, where g is gravity); f is the Coriolis parameter; a is the mean radius of the Earth; and ϕ denotes the latitude.

MERRA-2 Reanalysis Data
The Modern-Era Retrospective Analysis for Research and Applications Version 2 (MERRA-2; Gelaro et al., 2017) data set on 72 vertical model levels from the surface to 0.01 hPa (∼80.6 km) with a horizontal resolution of 0.625°× 0.5°(longitude × latitude) is used to examine the zonal wind and residual mean meridional circulation in the lower and middle atmosphere.We use the three-hourly assimilated data for zonal wind, meridional wind (v), vertical wind (w), and temperature (T).The residual mean meridional (v * ) and vertical (w * ) velocities due to resolved waves are expressed as follows (Andrews et al., 1987): Here, ρ 0 is the reference density given as a function of the log-pressure height (z ≡ -Hln p/ p s ) ), H is the atmospheric-scale height, p is the air pressure, p s is the reference surface pressure (p s = 1,000 hPa), θ is the potential temperature [θ ≡ T p s /p) κ ], and κ is the Poisson constant (κ ∼ 0.2856).The overbar and prime represent a zonal mean and departure from the zonal mean, respectively.

Estimating Long-Term Trends
Multiple linear regression (MLR) analysis is conducted to estimate the trends of the variables used in this study.
As in the MLR model used in Song et al. (2023), the seasonal and intraseasonal variations and well-known natural periodic variations, such as the 11-year solar cycle, Quasi-Biennial Oscillation (QBO), and El Niño-Southern Oscillation (ENSO), which can affect the dynamical fields in the middle and upper atmosphere, are considered.A more detailed methodology for calculating the trends is described in Text S1 in Supporting Information S1.

Climatology and Trends of the Zonal Winds at KSS
Figure 1a shows the climatology of the zonal winds at KSS.Several well-known features of the SH high latitude mesopause winds (Sandford et al., 2010;Smith, 2012)  Figure 1b demonstrates the trends of the zonal winds at KSS.In the mesopause regions (z = 90-100 km) in July, noticeable wind trends appear with a range of 0.5-0.8m s 1 yr 1 , while near the stratopause (z = 45-65 km) in June-July, slightly stronger trends (ranging from 0.9 to 1.1 m s 1 yr 1 ) are found.That is, the eastward winds were strengthened in both the mesopause region above z = 90 km and the stratopause region in July.In contrast, between the two vertical layers, nearly zero or (statistically insignificant) negative trends appeared in July.
Hereafter, we will focus only on the results for July, in which the strongest wind trends occurred.

Correlations With the Winds in the SH Middle Atmosphere
To investigate the relationships between the zonal-mean zonal winds in different vertical layers over KSS, the time series of the monthly zonal-mean zonal wind anomalies averaged for three layers Figure 2b shows the latitude-height cross section of the correlation coefficients between the zonal-mean zonal wind anomalies estimated from the Aura MLS and the zonal wind anomalies observed by the KSS-MR at L3 during July for 15 years.The significant correlations between the zonal wind anomalies in the three layers over the KSS, shown in Figure 2a, are also presented in Figure 2b.Moreover, there are generally opposite correlations poleward and equatorward of approximately 45°S, showing a quadrupole structure below z = 90 km.

Trends of Zonal Winds and Temperature Estimated From Aura MLS
Figure 3a shows the climatology of the zonal-mean zonal geostrophic winds (shading) with their vertical shear (contour) in July estimated from the Aura MLS.The typical characteristics of the zonal-mean zonal winds in the middle atmosphere and mesopause region during the SH winter (Fleming et al., 1990) are shown in Figure 3a: (a) The jet maximum is located at 40°S-60°S and z = 40-60 km, (b) the jet axis is tilted toward the equator with height, and (c) the secondary peak of the westerlies is found in the high-latitude lower mesosphere.Figure 3b shows the trends of the zonal-mean zonal geostrophic winds in July.In the polar region poleward of 50°S , the trends of the mesopause winds above z = 90 km are consistent with those observed from the MR at KSS (∼62°S), as shown in Figure 1b.Significant positive trends for more than 0.5 m s 1 yr 1 appear above z = 90 km, and almost zero or negative trends are observed below z = 90 km.Significant positive wind trends exist in the upper stratosphere and lower mesosphere (USLM; z ∼ 35-70 km).The USLM regions with significant positive wind trends of approximately 0.5-1.7 m s 1 yr 1 are tilted southward with height.In the mid-latitude region equatorward of 50°S, the sign of the wind trend in the USLM is generally negative.Opposite trends are found between the mesopause region over the KSS and the mid-latitude region in the USLM.Specifically, large negative and positive wind trends are observed at z = 30-70 km and z = 75-95 km, respectively, at mid-latitudes, which is almost isomorphic to the correlation pattern shown in Figure 2b.
The structure of the observed zonal wind in the middle atmosphere and mesospheric region is closely related to that of the temperature through the thermal wind balance.Therefore, to understand the wind trends, the spatiotemporal structure of the temperature is also investigated.Figures 3d and 3e show the climatology of the zonalmean temperature and their trends.The trends of the temperature poleward of 30°S are positive at z = 30-60 km, negative at z = 60-90 km, and weakly positive above z = 90 km (Figure 3e).In contrast, in the low-latitude region equatorward of 30°S, significant negative (positive) trends appear at z = 30-55 km and above z = 90 km (at z = 65-75 km).As a result, the trends of the wind and temperature in the SH winter have a quadrupole structure below the mesopause (z ∼ 90 km), which can also be found in the structures of the anomalies of the wind and temperature for anomalous stratospheric polar vortex conditions (Pedatella & Harvey, 2022).The observed temperature trends are strongly associated with the wind trends, given that similar structures are seen between the vertical gradient of the zonal wind and the meridional gradient of the temperature (see contour lines in Figures 3a  and 3d) as well as between their trends (see Figures 3c and 3f).Therefore, understanding the possible mechanisms responsible for the temperature trends is necessary to examine the trends of zonal winds, which will be discussed in the next section.

Residual Circulation and Trends
The quadrupole structures for the trends of the zonal-mean zonal winds and temperature (or correlation) below the SH mesopause region, as shown in Figures 2 and 3, are likely to be due to changes in the large-scale meridional circulations in the middle atmosphere.Liu and Roble (2002) showed that different zonal-mean temperatures before and during a stratospheric sudden warming (SSW) event exhibit a quadrupole structure below z ∼ 100 km owing to the marked changes in the residual mean circulation in the stratosphere-MLT region.Meanwhile, Tan et al. (2012) showed that the teleconnection pattern in the middle and upper atmosphere can be induced by residual mean circulation regardless of the occurrence of SSW.Ball et al. (2016) demonstrated that the quadrupole structure of the zonal-mean temperature anomaly can be formed by the enhanced or weakened Brewer-Dobson circulation (BDC), which extends from the stratosphere into the mesosphere (i.e., from z ∼ 15 to ∼ 65 km).Therefore, to examine the cause of the trends of the observed winds and temperatures, a trend of the residual mean circulations is examined.
Figure 4 shows the trends of the residual mean velocities (vectors) during July, calculated using the MERRA-2 data, overlaid on those of the zonal-mean temperatures (shading) estimated from the MLS.The observed temperature trends are related to adiabatic warming and cooling by vertical motions.Significant trends of the downward motions appear at z = 35-55 km over 30-60°S, where distinct warming trends are observed.The noticeable trends of the downward motion seem to be associated with the enhanced deep branch of the BDC in the upper stratosphere (above z ∼ 40 km).Another enhanced downward motion exists at relatively higher latitudes (∼60-70°S) and higher altitudes (z ∼ 65-75 km), contributing to the formation of poleward-tilted positive temperature trends below z ∼ 70 km.On the other hand, at z = 60-70 km and 35-50°S, a significant upward trend exists, which is associated with significant negative temperature trends.In the regions equatorward of 40°S, statistically significant trends with 85% and 95% confidence levels, respectively.Note that the geostrophic winds are calculated only at latitudes higher than 15°S to avoid errors due to the small Coriolis parameter in the tropics.Panel (d) is the same as (a) except for the zonal-mean temperature (shading) and the meridional gradient of the zonal-mean temperature (contour).Panel (e) is the same as (b) except for the results for the zonal-mean temperature.Panels (c) and (f) show the trends in the vertical gradient of zonal-mean zonal wind and those in the meridional gradient of the zonal-mean temperature, respectively.enhanced clockwise residual circulations above 1 hPa (z ∼ 48 km) and weak counterclockwise circulations below 1 hPa are found.

Summary and Discussion
In this study, the relationship between trends of the zonal winds in the mesopause regions at KSS and those in the SH middle atmosphere is examined using the 15-year data of MR observations, Aura MLS satellite observations and MERRA-2 reanalysis.In July, significant positive trends are observed both above z = 90 km and near the stratopause regions (z ∼ 45-65 km), while either near-zero or negative trends are observed between z = 65-90 km.
At the SH high latitudes, zonal winds in the mesopause regions (especially above z = 90 km) in July are positively and negatively correlated with those at z ∼ 30-60 km and at z ∼ 70-90 km, respectively.In addition to these correlations between the zonal winds at the different vertical layers at high latitudes, significant correlations (with opposite signs) exist between the winds at the high and low latitudes, producing the quadrupole structure of the correlation coefficients below z = 90 km.
The wind trends in the middle and upper atmosphere over the SH high latitudes during July obtained from the Aura MLS are consistent with the results of the KSS-MR and MERRA-2.The quadrupole structure of the trends of the zonal-mean zonal winds appears in the SH during July.Given that the zonal winds are strongly associated with the temperature through the thermal wind relationship, we investigated the trends of the temperatures and trends of residual circulation that can lead to the temperature trends.The enhanced deep branch of the BDC in the upper stratosphere contributes to the warming (cooling) trends in the high (low) latitudes in the stratosphere by inducing strengthened downward (upward) motions.On the other hand, in the mesosphere, counterclockwise (clockwise) circulation trends at high (low) latitudes lead to the observed mesospheric temperature trends.
In this study, the trends of the zonal wind are investigated mostly by temperature and residual-mean circulation.The residual circulations are due to waves resolved in the MERRA2 data set, and the resolved waves include PWs and GWs with relatively large horizontal scales, such as inertia-GWs.As both PW and GW forcings have a strong influence on the changes in the zonal wind in the middle atmosphere (Andrews et al., 1987), investigating the role of each wave forcing that contributes to the trends in the zonal wind is important to fully understand the causes of the long-term changes in the dynamics in the SH middle and upper atmosphere.Although Song et al. (2023) demonstrated that the GW forcing in the mesopause region at the KSS contributes to the trends of zonal winds to some extent, no comprehensive studies have been conducted regarding the trends of each wave forging and the relative effects of PW and GW on the long-term wind changes in the SH mid-to-upper atmosphere.This comprehensive study is underway using MR observations, high-resolution global reanalysis data sets, and a hightop numerical model.This work was supported by the Korea Polar Research Institute (KOPRI, PE23020/PE24020).The authors would like to give special thanks to all overwintering teams at KSS for the undisrupted operation of the KSS-MR over a long time.The authors would like to express our appreciation to the editor and three anonymous reviewers for their valuable comments and suggestions.

•
Trends of zonal winds in the SouthernHemisphere mesopause and stratosphere are investigated using meteor radar and satellite observations • Significant correlations are found between the mesopause wind at high latitudes and the winds in the middle atmosphere• The observed wind trends in austral winter are coupled vertically between the mesopause region and stratosphere across the mesosphere Supporting Information: Supporting Information may be found in the online version of this article.
Figure1ashows the climatology of the zonal winds at KSS.Several well-known features of the SH high latitude mesopause winds(Sandford et al., 2010;Smith, 2012) are found over the KSS: (a) zonal winds change from westward to eastward at an altitude of approximately z = 90 km during summer, (b) weak zonal winds are observed during the equinoxes, and (c) eastward jets are dominant in winter.In the middle atmosphere, westward winds exist above approximately z = 25 km in summer, and a strong eastward polar night jet appears in winter.The polar night jet has a maximum amplitude ranging from z = 25-55 km, and it gradually weakens with height and has the smallest magnitude near z ∼ 90 km.Above z = 90 km, the eastward winds are slightly strengthened with altitude.
[L1 (2-0.25 hPa, corresponding to z = 43.5-58.1 km), L2 (0.025-0.007 hPa, corresponding to z = 74.2-83.1 km), and L3 (z = 90-100 km)] are shown in Figure2a.For L1 and L2, zonal-mean zonal geostrophic winds estimated from the Aura MLS at the latitude of KSS are computed, while for L3, the zonal winds observed from the KSS-MR are used.The correlation coefficients between the zonal wind anomalies in two different layers are shown at the bottom of the panel.Note that a statistically significant correlation coefficient for the 95% confidence level is 0.514.In layer L3, negative wind anomalies occur mainly at the beginning of the observation period, while large positive wind anomalies appear after 2012, resulting in an overall positive wind trend.The temporal variation in the wind anomalies in L3 is similar to that in L1.The correlation coefficient between the zonal wind anomalies in L1 and L3 is 0.840.In contrast, the wind anomalies in L2 and L1 (or L2 and L3) are nearly in the opposite phase.The correlation coefficients between the wind anomalies in L1 and L2 and in L2 and L3 are 0.526 and 0.618, respectively.

Figure 1 .
Figure 1.(a) Time-height cross section of the 15-year mean of the monthly averaged zonal winds at KSS revealed from the meteor radar observations (above z = 80 km) and MERRA-2 reanalysis data (below z = 80 km).(b) Time-height cross section of the trends of the zonal-mean zonal winds.The non-shaded regions (gray dots) indicate statistically significant trends with an 85% (95%) confidence level.

Figure 2 .
Figure 2. (a) Time series of the monthly zonal-mean zonal wind anomalies at KSS averaged in three vertical layers.The zonal winds at Layer 1 (L1; z = 43.5-58.1 km) and Layer 2 (L2; z = 74.2-83.1 km) are obtained from the MLS observations, and the zonal winds at Layer 3 (L3; z = 90-100 km) are observed from the KSS-MR.Correlation coefficients between the zonal wind anomalies at each layer are shown at the bottom of the panel.(b) Latitude-height cross section of correlation coefficients between the zonal-mean zonal wind anomalies estimated from the MLS and the zonal wind anomalies observed by the KSS-MR at L3 during July for 15 years.The solid and dotted contour lines in green (yellow) denote statistically significant positive and negative correlation coefficients, respectively, at the 85% (95%) confidence level.The white dashed-dotted line indicates the latitude of the KSS (62.22°S).

Figure 3 .
Figure3.Latitude-height cross sections of (a) the 15-year climatological mean of the zonal-mean zonal geostrophic winds (shading) with the vertical gradient of the zonal-mean zonal wind (contour) derived from the Aura MLS and (b) their long-term trends during July.The non-shaded regions and gray dots in panel (b) indicate statistically significant trends with 85% and 95% confidence levels, respectively.Note that the geostrophic winds are calculated only at latitudes higher than 15°S to avoid errors due to the small Coriolis parameter in the tropics.Panel (d) is the same as (a) except for the zonal-mean temperature (shading) and the meridional gradient of the zonal-mean temperature (contour).Panel (e) is the same as (b) except for the results for the zonal-mean temperature.Panels (c) and (f) show the trends in the vertical gradient of zonal-mean zonal wind and those in the meridional gradient of the zonal-mean temperature, respectively.

Figure 4 .
Figure 4. Latitude-height cross sections of the trends of the zonal-mean temperature (shading) estimated from the MLS observations and the trends of the meridional and vertical components of the residual mean velocity (v * and w * ; vectors)during July, calculated using the MERRA-2 data.For better illustration, w * is approximately 300 times exaggerated in the residual mean velocity vector.The dark regions indicate statistically nonsignificant trends in the zonal-mean temperature with a 95% confidence level.The vectors are filled in pink, where statistically significant trends of w * appear at the 95% confidence level.