Mesospheric Water Vapor From SABER as a Tracer for the Residual Mean Circulation During SSW Events

The sparse wind observations and reanalysis winds in the mesosphere make it challenging to accurately determine the residual mean meridional circulation (MMC). As winds distribute tracers, an alternative approach is to utilize long‐lived trace species such as water vapor (H2O). The recently released H2O data from Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) version 2.07, combined with simulations by the Whole Atmosphere Community Climate Model eXtension with specified dynamics (SD‐WACCM‐X) and with the Data Assimilation Research Testbed (WACCMX + DART), provide an opportunity to assess the accuracy of using H2O isopleths to derive the vertical component of the residual MMC during the 2009 sudden stratospheric warming (SSW) event. In winter, the impact of photochemistry and diffusion on the distribution of mesospheric H2O is negligible compared to advective processes. H2O poleward of 70°N accurately captures the anomalous ascent that occurs a few days prior to the onset of the SSW and the subsequent descent. The derived vertical velocity in SD‐WACCM‐X and WACCMX + DART is qualitatively consistent with SABER observations. However, the derivation of vertical motion from isopleth analysis has limitations at the beginning of 2009 when the meridional transport is stronger than the vertical transport. While measuring winds in the mesosphere is challenging, satellite observations of mesospheric H2O prove to be an effective dynamical tracer during time periods characterized by strong vertical coupling.


Introduction
The middle atmosphere is an important region that couples the lower atmosphere, where atmospheric waves originate, with the upper atmosphere which is affected by solar and geomagnetic disturbances.Knowledge of middle atmosphere dynamics, chemical processes, composition distribution, and variability is therefore crucial in understanding the behavior of the atmosphere.The mesosphere extends from 50 km to roughly 85 km and is governed by three types of atmospheric waves: planetary waves (PWs), atmospheric tides, and gravity waves (GWs).As the waves propagate upward, their amplitudes grow exponentially to compensate for the decrease in atmospheric background density (Andrews et al., 1987).As a result, the wind field in the mesosphere is largely defined, even dominated, by wave motions.
As the waves become dynamically unstable, they break, deposit momentum, and drive the residual mean meridional circulation (MMC).The momentum deposited maintains a thermal structure that prevents the atmosphere from relaxing to radiative equilibrium.In the winter stratosphere, the breaking of PWs drives a meridional circulation from the equator to the winter pole (e.g., Garcia & Boville, 1994;Holton et al., 1995), while in the mesosphere, GW breaking results in an interhemispheric circulation from the summer to winter pole (e.g., Andrews et al., 1987;Holton, 1983;Matsuno, 1982).The summer to winter meridional circulation in the mesosphere is closed by an ascent at the summer hemisphere and a descent at the winter hemisphere (Holton, 1983), which is Abstract The sparse wind observations and reanalysis winds in the mesosphere make it challenging to accurately determine the residual mean meridional circulation (MMC).As winds distribute tracers, an alternative approach is to utilize long-lived trace species such as water vapor (H 2 O).The recently released H 2 O data from Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) version 2.07, combined with simulations by the Whole Atmosphere Community Climate Model eXtension with specified dynamics (SD-WACCM-X) and with the Data Assimilation Research Testbed (WACCMX + DART), provide an opportunity to assess the accuracy of using H 2 O isopleths to derive the vertical component of the residual MMC during the 2009 sudden stratospheric warming (SSW) event.In winter, the impact of photochemistry and diffusion on the distribution of mesospheric H 2 O is negligible compared to advective processes.H 2 O poleward of 70°N accurately captures the anomalous ascent that occurs a few days prior to the onset of the SSW and the subsequent descent.The derived vertical velocity in SD-WACCM-X and WACCMX + DART is qualitatively consistent with SABER observations.However, the derivation of vertical motion from isopleth analysis has limitations at the beginning of 2009 when the meridional transport is stronger than the vertical transport.While measuring winds in the mesosphere is challenging, satellite observations of mesospheric H 2 O prove to be an effective dynamical tracer during time periods characterized by strong vertical coupling.

Plain Language Summary
The net air motion in the mesosphere is governed by the residual mean meridional circulation.However, understanding this circulation is difficult due to limited wind observations in the mesosphere.To overcome the challenge, we use long-lived trace species-water vapor (H 2 O) from satellite observations and whole atmosphere model simulations to estimate the vertical velocity during the 2009 sudden stratospheric warming (SSW) event.Our results show that H 2 O can effectively capture the vertical motion during the SSW, and the model simulations align well with satellite observations.The method has limitations when horizontal transport dominates over vertical transport.Nevertheless, mesospheric H 2 O measurements poleward of 70°N are a valuable tool for investigating vertical processes in the mesosphere.
The vertical interaction between atmospheric layers becomes extremely strong during sudden stratospheric warmings (SSWs) (e.g., Chandran et al., 2014;Funke et al., 2010;Liu & Roble, 2002;Orsolini et al., 2022).Even though SSWs are an extreme meteorological phenomenon in the stratosphere, the dramatic changes can extend into the mesosphere, which is primarily due to changes in GW drag (Baldwin et al., 2021).During SSW, large amplitude PWs penetrate upward into the stratosphere and the induced wave forcing decelerates the westerly stratospheric winds (Matsuno, 1971).The anomalous westward wind in the stratosphere during SSW inhibits the upward propagation of westward GWs from the troposphere while allowing more eastward propagating GWs to reach the mesosphere (e.g., Siskind et al., 2010).The enhanced eastward forcing induced by wave breaking at mesosphere altitudes changes the high latitude residual MMC from downward to upward leading to mesospheric cooling (Limpasuvan et al., 2016).Moreover, significant changes in the transport and distribution of chemical species, are associated with the circulation changes during SSWs in the upper stratosphere and mesosphere (Baldwin et al., 2021).
To quantify the transport rate in the upper stratosphere and mesosphere, the residual MMC is computed from precise global wind and temperature fields (Andrews et al., 1987) or diagnosed from observed heating rates (Gille et al., 1987;Solomon et al., 1986).The residual MMC represents the net air motion and is used as a proxy for the Lagrangian-mean circulation.However, the sparse wind observations and reanalysis winds at these altitudes make an accurate determination of the residual MMC difficult.Alternate approaches include the use of long-lived trace species like carbon monoxide (CO), methane (CH 4 ), and nitric oxide (NO) (Bailey et al., 2014;Hendrickx et al., 2015;Lee et al., 2011;Manney et al., 2009;Ryan et al., 2018;Siskind et al., 2015).Water vapor (H 2 O) is also a widely used indicator of middle atmospheric transport processes given that its chemical lifetime is long with respect to dynamical processes in the middle atmosphere (Orsolini et al., 2010;Rong et al., 2010;Straub et al., 2012).Descent rates were inferred based on the long-lived trace species isopleth slopes in the studies mentioned above.However, the reliability of this method has been questioned, due to the presence of significant chemical and dynamical processes other than vertical advection (Ryan et al., 2018).
H 2 O measurements provide valuable evidence supporting the presence of the residual MMC in the mesosphere.Satellite instruments, including the Solar Occultation For Ice Experiment (SOFIE), Sounding of the Atmosphere using Broadband Emission Radiometry (SABER), Aura Microwave Limb Sounder (MLS), Halogen Occultation Experiment (HALOE), Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), and Atmospheric Chemistry Experiment-Fourier Transfer Spectrometer (ACE-FTS), play a significant role in measuring mesospheric H 2 O.With the finer vertical resolution and year-round polar coverage, SOFIE H 2 O provides an opportunity to estimate the vertical velocity in the polar mesosphere (Rong et al., 2010).Straub et al. (2012) compared H 2 O from MLS and ground-based microwave radiometer to Whole Atmosphere Community Climate Model with Specified Dynamics (SD-WACCM) and found similar descent rates below 80 km following the 2010 SSW.Recently, the H 2 O errors of unknown origin in SABER were identified and, to some extent, corrected in version 2.07 (Rong et al., 2019).However, little is currently known about the ability of SABER version 2.07 H 2 O to estimate vertical velocity in the mesosphere.
In this study, we evaluate the accuracy of the vertical component of the residual MMC derived from SABER version 2.07 H 2 O isopleths during the 2009 SSW.The SSW event considered is one of the strongest and most prolonged SSWs on record (Manney et al., 2009).The year is characterized by a solar minimum and relatively weak geomagnetic activity.Simulations of the WACCM with thermosphere-ionosphere eXtension (WACCM-X) with specified dynamics (SD-WACCM-X) and with data assimilation provided by the Data Assimilation Research Testbed (WACCMX + DART) have been performed for this SSW event.Model simulations allow us to compute the vertical component of residual MMC directly from wind and temperature fields, as well as estimate it from H 2 O isopleths.The comparison provides an overall assessment of the isopleth based vertical velocity due to the dynamical tracer assumption.In addition, we are able to separate dynamical processes from other effects for a clearer interpretation of the vertical velocity derived from SABER H 2 O.We aim to answer the question "how realistic is the residual MMC derived from observed and simulated H 2 O?" The manuscript is organized as follows.Section 2 describes the SABER version 2.07 H 2 O data, the model framework and simulations, as well as Lagrangian parcel backward trajectory method used to track the origin of the 10.1029/2023JD039526 3 of 17 air.Section 3 presents the estimated vertical component of residual MMC based on H 2 O, compares the results among SABER, SD-WACCM-X, and WACCMX + DART, and discusses the limitation of the isopleth analysis.Section 4 presents the conclusions.

SABER
SABER is a 10-channel broadband limb-scanning infrared radiometer (1.27-17 μm) on board NASA's TIMED (Thermosphere Ionosphere Mesosphere Energetics and Dynamics) satellite that was launched in December 2001 (Esplin et al., 2023;Russell et al., 1999).The routine data collection began on 25 January 2002 and the products include vertical profiles extending from the tropopause region to the lower thermosphere with 2 km vertical resolution.The measurements span latitudes from 82°S to 53°N or 53°S to 82°N daily depending on the 60-day spacecraft yaw cycle that prevents direct solar radiance from entering the SABER telescope.
A positive bias was found in the H 2 O data from the SABER instrument, preventing its release to the scientific community.However, the cause of the errors has been identified and, to some extent, corrected in the new SABER version 2.07 (Rong et al., 2019).Each day, SABER measures approximately 1,400 H 2 O profiles on the ascending and descending nodes of the orbits.Validation of H 2 O against ACE, MIPAS, SOFIE, and MLS has demonstrated good agreement in the mean profile, with differences within 10% in most cases (Rong et al., 2019).The data has been used to investigate the diurnal variability of mesospheric H 2 O (Koushik et al., 2023).Considering the small bias, as shown in Figure 5 in Rong et al. (2019), SABER version 2.07 H 2 O can correctly reflect the vertical velocity in the polar winter.In this study, SABER H 2 O is averaged over all orbits for each day to focus on the large-scale circulation.
It is important to note that SABER version 2.07 H 2 O still contains contamination resulting from out-of-band spectral leakage.In brief, the correction method in version 2.07 involves using MLS Aura H 2 O daily zonal-mean profiles to determine the correction coefficient for out-of-band radiance in SABER (Rong et al., 2019).However, MLS measures H 2 O globally every day at two local times that are 12 hr apart.As a result, potential local time dependence on the spectral contamination is not accounted for in the correction process.Nevertheless, the correction is expected to have minimal errors in the daily zonal-mean H 2 O. Since our analysis is based on the daily zonal-mean H 2 O, any local time variation in SABER H 2 O does not impact the validity of our findings.
For context, as global winds in the upper atmosphere are notoriously difficult to measure, we also compute quasi-geostrophic winds using SABER geopotential height fields, following the methodology used in previous studies (e.g., Oberheide et al., 2002).

SD-WACCM-X
WACCM-X is a comprehensive numerical model that contains chemical, physical, and dynamical processes to simulate the Earth's atmosphere from the surface to the upper boundary, which is located between 500 and 700 km depending on solar and geomagnetic activities (Liu et al., 2018a).The model is built upon WACCM 4 (Marsh et al., 2013) up to the lower thermosphere and adopts ionosphere-thermosphere processes from the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIEGCM; Qian et al., 2014).Important improvements to WACCM are included in WACCM-X, such as the revision of parameterized non-orographic gravity wave forcing (Garcia et al., 2017) and treatment of surface stress due to unresolved topography (Richter et al., 2010).The latter is especially relevant as it significantly improves the SSW frequency (Richter et al., 2010).With the new version, WACCM-X incorporates neutral wind dynamo, ionospheric transport, calculations of ion/ electron energetics and temperatures to better resolve the thermospheric energetics and thermal structure (Liu et al., 2018a).Moreover, the model can self-consistently resolve lower atmospheric processes, and provide a more realistic simulation of upper atmospheric variability due to lower atmospheric forcing (Liu et al., 2018a).
To simulate the 2009 SSW event, we use the specified dynamics configuration in which the model meteorology (winds and temperature) in the troposphere and stratosphere is constrained by NASA Modern-Era Retrospective analysis for Research and Applications version 2 (MERRA-2) reanalysis (Gelaro et al., 2017).The constraint to MERRA-2 is gradually reduced linearly between 50 and 60 km.Above 60 km, the model is fully interactive and free running.The horizontal resolution is 1.9° latitude by 2.5° longitude, with 145 vertical levels (Liu et al., 2018b).The simulation was initialized on 1 October 2008.Hourly model output in 2009 is averaged daily for this study.

WACCMX + DART
By directly incorporating mesosphere and lower thermosphere observations into the model, data assimilation is expected to provide a more accurate estimate of the atmospheric state compared to nudging toward reanalysis fields.Using the DART ensemble Kalman filter (Anderson, 2001), data assimilation has been implemented in WACCM-X by Pedatella et al. (2018a).WACCMX + DART assimilates conventional meteorological observations and refractivity from the Constellation Observing System for Meteorology, Ionosphere, and Climate in the troposphere and stratosphere, as well as temperatures from MLS and SABER from 20 to 100 km.As highlighted in Pedatella et al. (2018a), the representation of large-scale dynamical variability in the stratosphere, mesosphere, and lower thermosphere is improved in WACCMX + DART during 2009 SSW event compared to SD-WACCM-X.As such, WACCMX + DART can better capture the downward transport of chemical species from the mesosphere into the stratosphere following the SSW.
The model horizontal resolution is 1.9° latitude by 2.5° longitude (Pedatella et al., 2018b).WACCMX + DART simulations were performed during the 2009 SSW event with hourly data assimilation cycling and 40 ensemble members.To initialize the simulations, small perturbations were applied to the winds and temperatures in a single-member, free-running WACCM-X simulation on 1 October 2008.Variables are output hourly, from which daily averages are calculated.In order to compare the estimated vertical velocity to the "actual" values, H 2 O in both SD-WACCM-X and WACCMX + DART is not sampled based on the longitudinal locations of the SABER observations.While WACCMX + DART can better simulate the mesospheric circulation, it is also more computationally expensive compared to SD-WACCM-X.Conducting long-term simulations using SD-WACCM-X is more feasible.By comparing SD-WACCM-X and WACCMX + DART, we can assess the accuracy of H 2 O simulations and provide guidance for choosing between them based on computational resource considerations.

Residual Mean Meridional Circulation
The concentrations of trace species are impacted by various processes, including advection, photochemistry, and diffusion (Smith et al., 2011).The continuity equation for zonal-mean H 2 O mixing ratio (   ) is given by: where t, ϕ, z, a, and ρ are the time, latitude, log-pressure altitude, Earth's radius, and atmospheric density, respectively.v and w are the meridional and vertical components of the wind.The overbars denote the zonal-mean values, while the primes indicate the deviations from these zonal-mean values.P, L, X k , X D and E are the chemical production rate, chemical loss rate, chemical tendencies due to eddy diffusion, molecular diffusion, and resolved eddy flux divergence, respectively.In WACCM, eddy diffusion is a parameterization of vertical air parcel mixing processes that are not resolved (Garcia et al., 2007;Richter et al., 2010), while molecular diffusion uses the parameterization of Banks and Kockarts (1973).
To closely approximate the net air parcel displacements, the transformed Eulerian mean (TEM) was introduced (Andrews & McIntyre, 1976).The meridional (   * ) and vertical component (   * ) of the residual MMC are as follows (Andrews et al., 1987): where θ is the potential temperature.In the TEM formulation, part of the circulation resulting from eddy heat transport is subtracted from the Eulerian mean circulation to approximate the average net drift of air parcels (Andrews & McIntyre, 1976).The concept is valuable for understanding and diagnosing the net transport of air masses in the middle atmosphere.
The TEM continuity equation for   is written as (Abalos et al., 2020;Ryan et al., 2018): where The term F denotes the TEM divergence of wave fluxes, which represents the resolved eddy transport of H 2 O (Garcia & Solomon, 1983).Equation 5describes the transport of   due to advection, chemical production, chemical loss, eddy diffusion, molecular diffusion, and resolved eddy transport.It quantifies the relative contributions of each process and their impact on the overall distribution and variability of H 2 O in the atmosphere.
With the global winds and temperatures, we can calculate the "actual" vertical component of the residual MMC (   * ) from Equation 4.However, due to the dearth of global wind observations in the mesosphere, the method cannot be applied directly.An alternate approach to determine the vertical velocity is to use H 2 O isopleths (Rong et al., 2010;Straub et al., 2012).At a given latitude, the descent rate is determined by performing a least squares linear fit to zonal-mean H 2 O isopleths in the time-altitude cross section.The linear fit is derived on a 3-day basis with a 1-day increment.The estimated vertical velocity obtained from this method is denoted as  w * to distinguish it from the "actual" vertical velocity (   * ).

Lagrangian Backward Trajectory
To further understand the origin of   , Lagrangian backward trajectories are computed by solving ) at position  ⃖ ⃗  and time t from SD-WACCM-X and WACCMX + DART.Daily backward trajectories are calculated at each altitude and latitude.Considering that the photochemical lifetime of H 2 O decreases in the upper mesosphere, from months below 70 km to ∼5 days by 80 km (Brasseur & Solomon, 2005b), our study focuses on altitudes below 80 km.H 2 O mixing ratio is assumed to be conserved along each trajectory within 2 weeks.The backward trajectories provide the geographical origin of the air parcels.

Dynamical Overview of 2009 SSW
The 2009 SSW event, characterized as an elevated stratopause event (Manney et al., 2009), occurred on January 21st.For simplicity, important dates related to the 2009 SSW are referred to as the day of year (DOY) in 2009, such as the onset being DOY 21.
Figure 1 illustrates the evolution of zonal-mean zonal wind at 60°N during the 2009 SSW using SABER, SD-WACCM-X, and WACCMX + DART.The gaps in SABER (Figure 1a) are missing data caused by the yaw cycle of the spacecraft, that is, SABER was looking away from the pole before DOY 11 and after DOY 75.The polar vortex is quite strong in early January with the stratopause at around 60 km.Following DOY 12, the eastward wind in the mesosphere weakens and eventually transitions to westward at the onset of the SSW.The onset is also characterized by the rapid descent of the stratopause.Concurrently with the stratospheric warming, the mesosphere undergoes tremendous cooling (not shown), which is consistent with previous studies (e.g., Coy et al., 2011;Siskind et al., 2005).On DOY 34, the eastward wind returns in the mesosphere, and the stratopause reaches its lowest altitude at around 35 km.In the subsequent days, the eastward wind strengthens and a new stratopause reforms near 80 km on DOY 36.The stratopause discontinuity magnitude is approximately 45 km for the 2009 SSW event.The eastward wind reaches its maximum around DOY 60, which corresponds to the end of February.
Note that the westward winds in the mesosphere associated with the SSW precede the wind reversal in the stratosphere by 5 days in SABER, SD-WACCM-X, and WACCMX + DART. Lee et al. (2009) interpreted this structure as the downward propagation of Northern Annular Mode (NAM) anomalies initiated in the upper mesosphere using the MLS instrument on the Aura satellite.The mesospheric wind anomalies were also found in Advanced Level Physics High-Altitude prototype of the Navy Operational Global Atmospheric Prediction System (Coy et al., 2011).
The evolution of the winds at 60°N is generally similar among SD-WACCM-X, WACCMX + DART, and SABER observations.However, some differences emerge in the mesosphere.A comparison between SD-WACCM-X and WACCMX + DART shows that the westward wind is weaker in the latter during the SSW (Figure 1c).Moreover, the descent of the elevated stratopause after DOY 36 is more rapid in SD-WACCM-X (Figure 1b).These features align with the findings of Pedatella et al. (2018a).In general, the simulated evolution of the winds in WACCMX + DART exhibits greater consistency with SABER observations (Figure 1a), which is expected since the winds and temperatures are constrained by observations.
In both SD-WACCM-X and WACCMX + DART simulations, the zonal winds blow westward near the mesopause prior to DOY 10.The westward mesospheric wind is known to be a model bias in WACCM and WACCM-X (e.g., Harvey et al., 2019;Hindley et al., 2022;Zhang et al., 2021).Harvey et al. (2022) found that the model's zonal winds in the upper mesosphere are in closer agreement with the observations when the polar vortex is weak, which is consistent with the patterns shown in Figure 1.One possible explanation is that the models do not account for higher order GWs, which leads to the omission of the eastward drag caused by these waves.

Zonal-Mean Water Vapor Distribution
The latitude-altitude cross sections of zonal-mean H 2 O in January 2009, as measured by SABER and simulated by SD-WACCM-X and WACCMX + DART, are shown in Figure 2. As our study focuses on winter high latitudes, Figure 2 is confined to the Northern Hemisphere.There are missing measurements poleward of 84°N for the entire month of January due to the yaw cycle of SABER (Figure 2a).The distribution of H 2 O exhibits similari ties among SABER, SD-WACCM-X, and WACCMX + DART.Notably, H 2 O maximizes locally at the equator in the upper stratosphere (Figure 2).The vertical gradient is associated with the oxidation of CH 4 in the stratosphere and photo-dissociation in the mesosphere, as described in Brasseur and Solomon (2005a).The horizontal gradient of H 2 O toward the Northern Hemisphere is positive with dry air at higher latitudes and humid air at lower latitudes.The depletion of mesospheric H 2 O in the polar regions is primarily driven by the downward branch of the residual MMC, making mesospheric H 2 O a valuable parameter for determining the vertical velocity.It should be noted that while the variation of mesospheric H 2 O is predominantly driven by vertical transport, horizontal transport also contributes to the overall distribution.The influence of horizontal transport should not be ignored.
A notable difference between the observations and model simulations is that SABER exhibits ∼20% more H 2 O in the mesosphere compared to SD-WACCM-X and WACCMX + DART (Figure 2).In addition, the local H 2 O maximum in the tropical region is simulated at a higher altitude, ∼61 km in SD-WACCM-X, ∼57 km in WACCMX + DART, while at ∼55 km in SABER.The disparity in H 2 O could be attributed to contamination resulting from out-of-band spectral leakage in SABER or potential model bias in WACCM-X.Compared to WACCMX + DART, the latitude-altitude distribution of H 2 O simulated by SD-WACCM-X aligns more closely with SABER.
The distribution of H 2 O in the mesosphere is controlled by various processes, including advection, photochemistry, eddy diffusion, molecular diffusion, and resolved eddy transport, as described in Equation 5. Figure 3 exemplifies the evolution of H 2 O tendencies using SD-WACCM-X due to (a, d) chemical production, (b, e) chemical loss, and (c, f) eddy diffusion, with the unit in parts per million by volume (ppmv) per day.These patterns are not expected to exhibit any significant differences in WACCMX + DART.Molecular diffusion is not shown as its impact is negligibly small below 90 km (Smith et al., 2011).The time-altitude cross section is shown at 80°N (Figures 3a-3c), while the time-latitude cross section is averaged over 60-65 km (Figures 3d-3f).
At high latitudes in the winter hemisphere, non-advective processes affecting mesospheric H 2 O are relatively weak.Among these processes, chemical production is the strongest in the upper mesosphere, peaking at around 0.3 ppmv/day (Figure 3a), while eddy diffusion has a magnitude of around 0.2 ppmv/day (Figure 3c).As the spring equinox approaches in March, the impact of chemical loss becomes more pronounced (Figure 3b).Regarding latitudinal variations, Figures 3d-3f illustrate that the impact of chemical production and loss in the mesosphere decreases with increasing latitude during wintertime.Additionally, the impact of eddy diffusion remains small at all latitudes.Thus, H 2 O distribution in the winter mesosphere poleward of 60°N is primarily governed by advective processes and resolved eddy transport.

Estimated Descent Rate From H 2 O Isopleths
To estimate the vertical component of the residual MMC using H 2 O isopleths, we first examine the time-altitude evolution of zonal-mean H 2 O in the middle atmosphere (Figure 4).80°N is used to eliminate the impact of chemical and eddy diffusive processes.The gaps in SABER before DOY 11 and after DOY 75 are missing data caused by the yaw cycle (Figure 4a).The general evolution of H 2 O, as simulated by SD-WACCM-X, WACCMX + DART and observed by SABER, shows similarities.WACCMX + DART mesospheric H 2 O decreases during DOY 5-DOY 15, which is attributed to the descent of the residual MMC.This feature is not well captured in SABER and SD-WACCM-X.With the reversal of the polar vortex during the SSW, the evolution of H 2 O in the mesosphere is briefly interrupted by an increase.Between DOY 30 and DOY 60, H 2 O shows a general decrease, which is linked to the descent as the polar vortex recovers its original strength.The behavior is evident in both observations and simulations.After DOY 60, H 2 O tends to increase in altitude, known as the H 2 O inversion.The occurrence of the H 2 O inversion in March is induced by meridional transport as the polar night jet weakens (Figure 1), along with chemical and eddy diffusive processes (Figure 3).
There are notable differences in mesospheric H 2 O among SABER, SD-WACCM-X, and WACCMX + DART (Figure 4).Both SD-WACCM-X and WACCMX + DART show a smaller mesospheric H 2 O maximum, which is consistent with the findings from Figure 2. To quantitatively compare mesospheric H 2 O among SABER, SD-WACCM-X, and WACCMX + DART, selected H 2 O isopleths from Figure 4 are combined in Figure 5.There are differences in the temporal evolution of H 2 O between the observations and simulations during DOY 20-DOY 30 (Figure 5).The increased upper mesospheric H 2 O persists for approximately 5 days in SABER, whereas in SD-WACCM-X and WACCMX + DART, the period is about 10 days longer with a second peak on DOY 30.Thus, models might have limitations in capturing the precise timing and magnitude of the H 2 O variation.Compared to SD-WACCM-X, the subsequent descent from the mesosphere into the stratosphere following the SSW is more rapid in WACCMX + DART (dotted red lines).In addition, the magnitude and evolution of mesospheric H 2 O during wintertime simulated in SD-WACCM-X (dotted blue lines) are more similar to SABER (solid black lines).
The left column of Figure 6 displays the H 2 O isopleths at 80°N.Due to the difference in the observed and simulated magnitudes of mesospheric H 2 O, the range of H 2 O isopleths varies among the datasets to ensure coverage of the altitude range from 45 to 85 km.Isopleth analysis cannot be conducted during periods when inversions occur, such as in March.Therefore, isopleths are not included for those periods.
The vertical velocity determined from H 2 O isopleths (  w * ; details in Section 2.4) are shown in the right column of Figure 6.The estimated vertical rates from SABER, SD-WACCM-X, and WACCMX + DART exhibit good agreement.Starting from DOY 12, there is ascent in the mesosphere lasting around 10 days (red in Figures 6d-6f).Following the onset (DOY 21), the mesospheric air descends into the stratosphere.These are typical features of SSW events (Matsuno, 1971).However, there are differences in the magnitude and timing of the estimated vertical rates among the three datasets.The peak ascent rate during the SSW is greater in SD-WACCM-X (2.12 cm/s) and WACCMX + DART (2.24 cm/s) compared to SABER (1.58 cm/s).In SABER, the peak occurs at a higher altitude of 74 km.Interestingly, there is a second peak of ascent on DOY 25 at around 70 km in SD-WACCM-X and WACCMX + DART, which corresponds to the second peaks of H 2 O isopleths as shown in Figures 4 and 5. On the other hand, the peak decent rate following the onset of SSW is weaker in SD-WACCM-X (−0.97 cm/s) compared to WACCMX + DART (−1.44 cm/s) and SABER (−1.60 cm/s).The peak descent rate simulated in WACCMX + DART is closer to SABER, which is attributed to the better simulation of the elevated stratopause, as discussed in previous studies (e.g., Meraner et al., 2016).However, the evolution of the descent after DOY 30 in SD-WACCM-X is more similar to SABER.

Evaluate the Accuracy of Vertical Velocity Derived From H 2 O Isopleths
The residual MMC is the Eulerian mean circulation transformed according to Equations 3 and 4 to approximate the average net drift of air parcels (Andrews & McIntyre, 1976;Matsuno, 1980).To assess the accuracy of  w * , the vertical component of the residual MMC (   * ) is calculated using Equation 4. Figure 7 shows the evolutions of   * during the 2009 SSW at 80°N using SD-WACCM-X (a) and WACCMX + DART (b).The models are generally consistent in terms of the temporal variability of   * . However,   * is stronger in WACCMX + DART compared to SD-WACCM-X.The downward motion persists in the polar mesosphere during the first few days of 2009, which aligns with the typical wintertime circulation.As the SSW onset approaches, the downward pattern tends to descend with the zero-wind line, coinciding with the falling warm stratopause layer (Figure 1).
Figure 8 depicts the altitude-time evolution of the wave amplitude in geopotential height of zonal wavenumber 1 (WN1) and wavenumber 2 (WN2), as well as the wave forcing (Eliassen-Palm flux divergence) using SD-WACCM-X (a-c) and WACCMX + DART (d-f).The amplitudes of WN1 are larger than WN2 before DOY 10, indicating the significant role of preconditioning (Figures 8a and 8d).Starting from DOY 10, WN2 amplifies and peaks in the upper stratosphere (Figures 8b and 8e).The breaking of the waves induces westward wave forcing (Figures 8c and 8f) and weakens the eastward wind.The diminished eastward wind allows more eastward propagating GWs to reach the mesosphere, leading to eastward GW drag that surpasses the westward wave forcing (not shown).Consequently, mesospheric motion reverses to ascent during the SSW (Figure 7), resulting in anomalously cold mesospheric air.As the H 2 O distribution in the winter mesosphere poleward of 60°N is primarily governed by advective processes and resolved eddy transport (Figure 3), isopleths analysis based on H 2 O at these latitudes has also been examined.We find that H 2 O poleward of 70°N is able to accurately capture the vertical component of the residual MMC H 2 O during wintertime.

Meridional Advection of H 2 O
A discrepancy between  w * and   * arises during the first few days of 2009.The positive  w * (Figures 6e and 6f) is associated with the increase in H 2 O before DOY 8 (Figures 4b and 4c).However, it contradicts the expected wintertime descent motion (Figure 7) driven by GW drag.The ascent determined from H 2 O isopleths suggests that during DOY 1-DOY 8, H 2 O transport is predominantly controlled by meridional processes rather than vertical processes, which requires further investigation.
To illustrate the impact of meridional advection, Figure 9 shows the evolution of the meridional component of the residual MMC (   * ) during the 2009 SSW at 80°N.The temporal variability and magnitude of   * in WACCMX + DART (panel b) are similar to the SD-WACCM-X (panel a).Prior to and after the SSW onset, the mesospheric meridional circulation is strongly poleward driven by GW drag.For a brief period of around 15 days starting from DOY 15, the mesospheric motion reverses to equatorward, accompanied by ascending motion as shown in Figure 7.The motions are associated with the breaking of PWs (Figures 8c and 8f).It is important to note that during DOY 1-DOY 8, the poleward motion can transfer moist air flow from lower latitudes to higher latitudes, thereby increasing the amount of H 2 O in the polar region.As a result, the decrease in H 2 O caused by the descending motion is offset.Trajectories launched at 60°N, 70°N, and 80°N are indicated by blue, green, and red, respectively.The start (end) points of the backward trajectories are marked with stars (triangles).For example, in Figure 10a, triangles indicate the location of air parcels on DOY 2, while stars represent their location on DOY 11.In general, the features agree with the residual MMC depicted in Figures 7 and 9. From DOY 2 to DOY 11 (Figure 10a), air parcels from the upper mesosphere in mid-latitudes are transported to higher latitudes and lower altitudes.Prior to the SSW onset (dots with darker colors in Figure 10b), the circulation in the mesosphere becomes disrupted, with weak equatorward and upward transport of air masses.The meridional transport of mesospheric air becomes less pronounced, while strong upward motion transports humid air from below, leading to an increase in H 2 O.The lower mesospheric circulation undergoes a complete reversal, becoming equatorward and upward during DOY 22-DOY 31 (Figure 10c).After DOY 31 (not shown), the pre-warming circulation starts to reestablish.The mesospheric air is pushed toward high latitudes and lower altitudes, resembling the pattern before the SSW (Figure 10a).Trajectories for WACCMX + DART (not shown) exhibit similar behavior to SD-WACCM-X.
Trajectories originating at 60°N show strong meridional transport in the mesosphere (Figure 10).Even though the photochemical, eddy diffusive, and molecular diffusive processes are weak at mid-latitudes in wintertime (Figures 3d-3f), it is important to note that mesospheric H 2 O poleward of 70°N is a suitable tracer for accurately estimating vertical velocity.Prior to the SSW, the origin of the mesospheric air masses is mostly in mid-latitudes (Figures 11a and 11c) and upper mesosphere (Figures 11b  and 11d).The pattern is consistent with the winter mean circulation characterized by poleward and downward motions (Figures 7 and 9).The increase  in H 2 O before DOY 8 (Figure 4) is a sign of strong meridional transport from mid-latitudes.During the SSW, the origin of mesospheric air shifts to the polar region and lower mesosphere.As the reformed stratopause descends after DOY 36, the mesospheric air originates from lower latitudes and higher altitudes, suggesting the reestablishment of the winter mean circulation.Since the transport is predominantly driven by vertical processes during and following the SSW, the  w * determined based on H 2 O isopleths can effectively capture the characteristics of   * .

Conclusions
Direct measurement of mesospheric circulation is difficult due to the small meridional winds and significantly vertical winds compared to tidal amplitudes.To overcome this challenge, an alternative approach is to utilize long-lived trace species.In this study, we used the corrected H 2 O dataset from SABER version 2.07 to derive the vertical component of the residual MMC during the 2009 SSW event.The "actual" vertical component of the residual MMC was computed using the global wind and temperature fields from the SD-WACCM-X and WACCMX + DART.Through a comparison between the estimated and actual vertical component of the residual MMC, we assessed the accuracy of the H 2 O isopleth-based approach.Furthermore, the Lagrangian backward trajectories provided valuable insights into the meridional and vertical transport of H 2 O.The main conclusions of the study are as follows: Mesospheric H 2 O poleward of 70°N is an effective dynamical tracer during time periods characterized by strong vertical coupling.The impact of photochemistry and diffusion on the distribution of mesospheric H 2 O is negligible compared to advective processes in the winter polar region.H 2 O poleward of 70°N accurately captures the anomalous ascent that occurs a few days prior to the SSW onset, as well as the subsequent descent following the SSW.The derived vertical velocity in both SD-WACCM-X and WACCMX + DART is qualitatively consistent with SABER observations.
The horizontal transport of mesospheric H 2 O is a significant factor that cannot be overlooked.At the beginning of 2009, the increased mesospheric H 2 O simulated in SD-WACCM-X and WACCMX + DART is attributed to meridional transport, which mixes midlatitude air into the polar regions.Lagrangian backward trajectories also indicate strong meridional transport during that time period.It is important to note that the derivation of vertical motion from isopleth analysis has limitations when there is intense meridional transport.
The dynamics of the 2009 SSW event in SD-WACCM-X and WACCMX + DART models show good agreement with SABER observations.Between the models, the subsequent descent from the mesosphere into the stratosphere following the SSW onset is more rapid in WACCMX + DART compared to SD-WACCM-X.The The observations of water vapor in the mesosphere will greatly enhance our understanding of the circulation in the upper atmosphere.

Figure 1 .
Figure 1.Time-altitude evolution of the zonal-mean zonal wind (m/s) at 60°N from (a) SABER, (b) SD-WACCM-X, and (c) WACCMX + DART.In SABER, zonal wind is the zonal component of geostrophic wind.The x-axis is days from 1 January 2009.Red (blue) is eastward (westward) zonal winds with zero-value contour in black.The approximate location of the stratopause is shown by the black plus symbol.

Figure 2 .
Figure 2. Meridional cross-sections of middle atmospheric water vapor (ppmv) through January 2009 from (a) SABER measurement, (b) SD-WACCM-X, and (c) WACCMX + DART simulations.Red indicates relatively higher values and purple relatively lower values.

Figure 3 .
Figure 3. Time-altitude evolution at 80°N (a-c) and time-latitude evolution averaged over 60-65 km (d-f) of photochemical production rate (top panel), photochemical loss rate (middle panel), and chemical tendencies due to eddy diffusion (bottom panel) from SD-WACCM-X simulation.Red (blue) is positive (negative) values in ppmv/day.The x-axis is days from 1 January 2009.

Figure 4 .
Figure 4. Time-altitude evolution of the daily zonal-mean water vapor mixing ratio (ppmv) at 80°N from (a) SABER measurement, (b) SD-WACCM-X, and (c) WACCMX + DART simulations.The water vapor mixing ratio is smoothed by a 5-day running mean.The x-axis is days from 1 January 2009.The vertical gray lines mark the SSW onset.

Figure 6 .
Figure 6.Time-altitude evolution of 80°N water vapor isopleths (a-c) and the estimated vertical component of residual mean circulation (d-f) based on SABER measurement (top panel), SD-WACCM-X (middle panel), and WACCMX + DART simulations (bottom panel).Water vapor isopleths are from 1.5 ppmv to 5.5 ppmv with a 0.1 ppmv interval (a), 2.6 ppmv to 5.5 ppmv with a 0.05 ppmv interval (b), and 3.7 ppmv to 5.5 ppmv with a 0.05 ppmv interval (c).The black contours in (d)-(f) are ±1 cm/s vertical velocity.

Figure 7 .
Figure 7. Time-altitude evolution of the vertical component of residual mean circulation (

Figure 8 .
Figure 8. Time-altitude evolution of the wave amplitude in geopotential height (m) of zonal wavenumber 1 (top panel), wavenumber 2 (middle panel), and wave forcing by the resolved waves (bottom panel; m/s/day) at 80°N in (a-c) SD-WACCM-X and (d-f) WACCMX + DART simulations.All values are smoothed by a 5-day running mean.The vertical gray lines mark the SSW onset.
The meridional and altitudinal origins of air parcels, determined from 3-day backward Lagrangian trajectories, are shown in Figure11.The trajectories are launched at 80°N (a, c) and 60 km (b, d).Air origins from latitudes southward (northward) of 80°N and altitudes below (above) 60 km are represented by blue (red) colors.The timing of the meridional and vertical transport is similar in SD-WACCM-X and WACCMX + DART, though the magnitude is stronger in WACCMX + DART.

Figure 9 .
Figure 9. Time-altitude evolution of the meridional component of residual mean circulation (   * ) at 80°N in (a) SD-WACCM-X and (b) WACCMX + DART simulations.Red (blue) is northward (southward) motions in m/s.The black contours show ±1 m/s.All values are smoothed by a 5-day running mean.

Figure 11 .
Figure 11.Latitudinal (top panel) and altitudinal (bottom panel) origin of water vapor determined using Lagrangian 3-day backward trajectory calculations in (a and b) SD-WACCM-X and (c and d) WACCMX + DART simulations.The backward trajectory was launched at 80°N (top panel) and 60 km (bottom panel).Blue (red) indicates latitudes southward (poleward) of 80°N and altitudes below (above) 60 km.The vertical gray lines mark the SSW onset.