On the Abnormally Strong Westward Phase of the Mesospheric Semiannual Oscillation at Low Latitudes During March Equinox 2023

Different meteor radars at low latitudes observed abnormally strong westward mesospheric winds around the March Equinox of 2023, that is, during the first phase of the Mesospheric Semiannual Oscillation. This event was the strongest of at least the last decade (2014–2023). The westward winds reached − 80 m/s at 82 km of altitude in late March, and decreased with increasing altitude and latitude. A considerable increase in the diurnal tide amplitude was also observed. The Whole Atmosphere Community Climate Model with thermosphere‐ionosphere eXtension constrained to meteorological reanalysis up to ∼ 50 km does not capture the observed low‐latitude behavior. Additionally, these strong mesospheric winds developed during the westerly phase of the Quasi‐Biennial Oscillation, in accordance with the filtering mechanism of


Introduction
The Semiannual Oscillation (SAO) in the stratosphere and lower mesosphere was first reported in the 1960s (Reed, 1965(Reed, , 1966;;Reed & Rogers, 1962).In particular, Reed (1966) reported a pronounced SAO between the altitudes of 30 and 65 km using rocketsonde data at Ascension Island (8°S, 14°W).His findings revealed that winds are westward near the solstices and eastward near the equinoxes.Also, Reed (1966) inferred that the maximum amplitude of the SAO occurs at the equator around the stratopause (∼50 km), and that the amplitude subsequently decreases with altitude and latitude.This oscillation is usually referred to as the "stratopause SAO," or SSAO.Hirota (1978) extended the analysis of the SAO from 30 to 90 km.He reported the existence of a second SAO in the mesopause (MSAO), which is out of phase with respect to the SSAO.The MSAO eastward (westward) peaks appear around the solstices (equinoxes), whereas SSAO eastward (westward) peaks occur around the equinoxes (solstices).Garcia et al. (1997) summarized the climatology of the SAO using different instruments such as satellite, groundbased, and rocket measurements.They showed the asymmetry of the MSAO, whose westward phase during the first cycle is considerably stronger than during the second cycle.Furthermore, they observed that the westward phase of the MSAO experiences significant interannual variations.In the context of interannual variability, strong mesospheric westward winds have been reported around the March equinox with radars (e.g., Kishore Kumar et al., 2014) and satellites (Fig 6 in Smith et al. (2017)).At around 82 km, these winds reached 80 m/s, which is significantly more intense than the usual values below 20 m/s in the climatology at low latitudes.
Observations have shown a correlation between the occurrence of the abnormally strong westward winds in the first cycle of the MSAO and the westerly (eastward) phase of the stratospheric Quasi-Biennial Oscillation (QBO) (Burrage et al., 1996;Garcia et al., 1997;Mengel et al., 1995).A possible explanation of this is the filtering in the stratosphere of eastward vertically propagating small-scale gravity waves (Antonita et al., 2008;Dunkerton, 1982) and inertia-gravity waves (intermediate-scale waves) (Sassi & Garcia, 1997;Garcia & Sassi, 1999;R. Lieberman et al., 2006) during the westerly phase of the QBO, allowing the selective upward propagation of westward gravity waves, which could then deposit momentum and contribute to the westward background zonal wind observed in the mesosphere.Another possible explanation for the strong first cycle of the MSAO is the breaking of the migrating diurnal tide (DW1) due to convective instabilities in the mesosphere and lower thermosphere (MLT) region at and above 85 km (Akmaev et al., 1996;Garcia, 2023;Gurubaran & Rajaram, 2001;R. S. Lieberman & Hays, 1994).Under this mechanism, the DW1 that is undergoing dissipation will impart an acceleration on the zonal-mean zonal wind (e.g., Andrews et al., 1987;R. S. Lieberman, 1997;Lindzen, 1981).Recently, Garcia (2023) confirmed this mechanism by using temperature observations made by the SABER infrared radiometer, and quantified the breaking events, which are more frequent around the equinoxes at low latitudes when the amplitude of DW1 is larger.
It has been observed that the westerly phase of the QBO is a necessary but not sufficient condition for the occurrence of the strong first cycle of the MSAO.In a long-term study, Kishore Kumar et al. (2014) proposed a criterion to explain the exceptions.That is, the occurrence of strong mesospheric westward winds is more likely to happen during the westerly QBO phase, except when there is a strong sudden stratospheric warming (SSW) in the previous (northern hemisphere) winter, which could change the filtering conditions in the stratosphere.Simulations conducted by Zülicke and Becker (2017) supported this criterion.The classification of strong and weak SSWs can be found in Kishore Kumar et al. (2014).
In this work, we investigate the abnormally strong first cycle of the MSAO that occurred around the March equinox of 2023.We study this global event using zonal and meridional winds from five meteor radars at low latitudes (±18°), SABER temperatures, and simulations made with the Specified Dynamics Whole Atmosphere Community Climate Model with thermosphere-ionosphere eXtension (SD-WACCM-X) constrained to an atmospheric reanalysis up to ∼50 km.The paper is structured as follows.In Section 2, we describe the data set and methodology.In Section 3, we present the results.First, we show large-scale dynamics (background winds and solar diurnal tide amplitudes) from 2020 to June 2023 to contextualize the abnormally strong first cycle of the MSAO of March equinox 2023.Second, we show the global morphology of this abnormal cycle using five meteor radars at low latitudes and supplement our results with the SD-WACCM-X simulations.Third, we report zonal meteor winds from 2014 to 2023 between 80 and 100 km to pinpoint the strongest cycles of the last decade.Finally, we present temperature amplitudes of the migrating diurnal tides and other quantities to show the role of tidal breaking in the generation of the strongest cycles between 2014 and 2023.To summarize, in Section 5 we present the concluding remarks.

Specular Meteor Radars
We utilized horizontal hourly winds estimated from SIMONe Jicamarca radar located at 11.9°S, 76.8°W from January 2020 to June 2023.This data was used to introduce and highlight the unusual behavior of the westward winds and the diurnal tide around the March equinox in 2023.For more information on the SIMONe Jicamarca radar, please refer to Chau et al. (2021).
Additionally, we have used monthly winds from Tirupati from January 2014 to June 2023 to analyze the MSAO around the March equinox during the last ten years.

WACCM-X
WACCM-X is a whole atmosphere model that extends from the surface to the upper thermosphere (∼500-700 km), and includes the necessary dynamics, chemistry, and physics to simulate the whole atmosphere and ionosphere (Liu et al., 2018).For the present investigation, we have used the Specified Dynamics (SD) configuration, where the model meteorology is constrained by MERRA-2 up to ∼50 km (Smith et al., 2017).This allows for the model state in the troposphere-stratosphere to closely follow the true state of the atmosphere.At higher altitudes the model is unconstrained, and the accuracy of the dynamical variability in the mesosphere is thus partly dependent upon processes internal to the model itself.
We have used two subsets of zonal winds from SD-WACCM-X to study the MSAO.The first data set was used to compare with observations.It covers the period January 2010 to December 2021 and is provided for each of the five radar latitudes (12°S, 7°S, 5°S, 14°N, and 18°N), every 3 hr, 5°in longitude (from 0°to 355°), and ∼2 km altitude resolution at MLT altitudes.The second data set was used to study the reproducibility of the past abnormally strong first cycles of the MSAO, and it covers the interval from 2000 until 2021.

Mean Winds and Diurnal Tide Estimation
The background (mean) winds, as well as the amplitude and phase of the dominant tides were estimated using a least squares method with a running window of 21 days and 2 km, shifted by 1 day and 1 km, respectively.The fitted periods correspond to 8, 12, 12.42, 24, and 48 hr.The 21-days window was chosen to separate the contributions from the 12-hr and 12.42-hr waves.The last one corresponds to the lunar tide, which could have relatively weak but non-negligible amplitudes (e.g., Sandford and Mitchell (2007)).

Migrating Diurnal Tide (DW1) From SABER Temperatures
We have estimated the amplitude of the solar migrating diurnal tide (DW1) from temperature observations made by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) infrared radiometer (Russell et al., 1994) on NASA's Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics (TIMED) satellite (Yee et al., 2003).Following the methodology of Garcia (2023), the amplitudes were synthesized from the temperature spectrum between frequencies 0.99-1.006cpd (periods of 0.99-1.01days).This range of periods includes the diurnal harmonic itself and all the annual, semiannual and quasi-biennial variability of DW1; see Fig. 2 in Garcia (2023) for details.Moreover, we have followed the methodology presented in Section 5 of Garcia (2023) to study the breaking of DW1.After estimating the potential temperature (θ), two parameters were estimated.Those proxies allow us to measure the intensity of the breaking.The first one is the number of breaking events (where the vertical gradient of potential temperature, d(θ)/dz, is negative) per pentad (in a window of 5 days).The second one is the breaking layer depth, which is the average depth of the atmospheric region where d (θ)/dz is negative.

Planetary-Scale Dynamics Over Jicamarca
In Figure 1, we can see the background winds and the amplitudes of the total diurnal tide in the MLT region over Jicamarca (12°S, 77°W) from January 2020 to June 2023.Additionally, Figure S1 in the Supporting Information S1 shows the amplitudes of the quasi-two-day wave (Q2DW) and the 12-hr and 8-hr tides.It is important to emphasize that a single radar measures total tides, meaning it cannot separate the contribution of migrating (e.g., DW1) and non-migrating tides.

10.1029/2024GL110331
These results indicate that from 2020 to 2023, the large-scale dynamics in general follow the climatology previously reported over Jicamarca by Suclupe et al. (2023).That is, the amplitude of the meridional component of the diurnal tide is more intense than that of the zonal one.The 24-hr tide dominates (see Figure S1 in the Supporting Information S1 for the full picture).The meridional diurnal component shows its largest peak in August-September and the second one in April-May, while the zonal diurnal component only shows a clear maximum in August-September.The Q2DW reaches maximum amplitudes in January.The meridional semidiurnal component peaks in April-May (similar to the diurnal tide).The terdiurnal tide generally shows a maximum above 90 km in March-April and September-October (see Figure S1 in the Supporting Information S1).The SAO in the zonal mean wind is observed between 80 and 93 km.The meridional mean winds are mostly southward with weak interannual variability.
The striking feature in these observations is the strong mesospheric westward wind around March equinox 2023, which reveals an abnormally strong first cycle of the MSAO.Also, the diurnal tide presents larger amplitudes around the same period (see the orange dash ovals in Figure 1).In addition, the meridional components of the 12hr and 8-hr tides also exhibit larger amplitudes around March 2023 over Jicamarca (see Figure S1 in the Supporting Information S1).

Abnormally Strong First Cycle of the MSAO in March Equinox 2023 at Low Latitudes
Figure 2a shows the zonal background winds from January 2022 to June 2023, measured at five low-latitude radar stations: Jicamarca (12°S, 77°W), Cariri (7°S, 36°W), Piura (5°S, 81°W), Tirupati (14°N, 72°E), and Ledong (18°N, 109°E).Figures 2b and 2c display the zonal background winds for Jicamarca and Ledong at three different altitudes (82, 88, and 94 km) from 2020 to 2023 and 2019 to 2023, respectively.The light-blue shaded region shows the altitudinal average (±1σ) between 80 and 100 km, of the climatology given by SD-WACCM-X from 2010 to 2021. Figure 2d shows the zonal background winds at 82 km over the five low-latitude stations obtained from 2-a.The light-blue shaded region represents the latitudinal average (±1σ) between 18 and 18°, of the climatology given by SD-WACCM-X from 2010 to 2021.
Besides the different annual behavior of the zonal winds between the northern and southern hemisphere sites, strong westward winds are observed over all the stations during the March equinox of 2023 (Figure 2a).Even at 30°S over central Chile, an enhancement of the mean zonal wind in the westward direction can be appreciated (not shown here).The strength of these westward winds decreases with altitude (Figures 2b and 2c) and latitude (Figure 2d).The strong westward winds start developing in February, to vanish almost simultaneously at the end of April (same slope, Figure 2d).The westward peak occurs first in the southern stations, from West to East (81°W to 36°W), at the end of March, and then in the northern hemisphere stations, first over Tirupati (72°E) and later over Ledong (109°E) at the beginning of April.SD-WACCM-X simulations reproduce the MSAO, but with considerably smaller amplitudes.The relatively lower variability of SD-WACCM-X winds observed in Figure 2b-2d, indicates that the model does not capture the abnormally strong first cycle of the MSAO between 2010 and 2021.Furthermore, we investigated the reproducibility of strong westward background winds in the model from 2000 to 2021 at 5°S at 82 km.Then we examined whether strong winds, previously reported in the literature during March-April 2002, 2008, 2011, and

Geophysical Research Letters
10.1029/2024GL110331 2017 (Kishore Kumar et al., 2014), were reproduced by SD-WACCM-X. Figure S2 in the Supporting Information S1 shows that, although the SD-WACCM-X winds in three of the years exceeded 1 σ from the mean of March and April, they never reached 3σ, and they showed smaller amplitudes than in the observations.

Mesospheric Zonal Winds Over Tirupati
Monthly zonal winds from 2014 to 2023 over Tirupati at 82 km are shown in Figure 3.The light-blue shaded region represents the region of ±1σ with respect to the mean.The data shows that there was significant variability around the March equinox.The strongest westward winds were observed in 2023, followed by 2017.We applied the classification criterion suggested by Kishore Kumar et al. (2014) to this data set and found that 8 of our 10 cases match their criterion (see the list in Figure 3 for further details).That is, an abnormally strong first cycle of the MSAO, referred to as mesospheric spring equinox enhancement (MSEE) by Kishore Kumar et al. (2014), is favored under the following conditions: when there is a westerly (W) QBO phase, and there was no "Strong" SSW (i.e., "None" or "Weak" class, according to Kumar et al.'s classification).Note that the QBO phase was classified, as in Kishore Kumar et al. (2014), based on the strongest stratospheric westward winds within the region 100-5 hPa (∼16-35 km), from the Singapore radiosonde observations, being "Easterly QBO" when it is less than 30 m/s, and "Westerly QBO" in the other cases.
As we can see, the criterion based on the filtering mechanisms of gravity waves in the stratosphere, works from 2014 to 2023, except for 2014 and 2016.Some years are shown in parentheses in Figure 3, since they must be further investigated.For example, in 2014, the QBO was slightly above the threshold; it could have also been easterly.In 2021, the QBO was slightly below the threshold; it could have also been westerly.In 2023, the SSW

Geophysical Research Letters
10.1029/2024GL110331 was formally classified as "Weak," but after some days with weak eastward winds, another westward wind phase began, making the classification unclear.

DW1 Breaking Observed by SABER
Figure 4 illustrates the behavior of DW1 and its breaking from temperature measurements obtained by SABER.
Figure 4a shows the amplitude of DW1 at 12 scale height (about 85 km).As expected, DW1 presents higher amplitudes at low latitudes (±15°).Moreover, the DW1 tide is very large in March equinox 2023.It is also large in 2017, the second year with strong westward zonal winds in March equinox, as indicated by observations over Tirupati (Figure 3).
Figures 4b and 4c show two measures of the tidal breaking over the equator.The first one shows the number of breaking events per pentad; and the second one shows the breaking layer depth, which is indicative of the strength of breaking.The year 2023 stands out on both counts (and so does 2017).

Discussion
Zonal momentum fluxes estimated over SIMONe Jicamarca and SIMONe Piura exhibit a clear enhancement in the westward direction during March 2023 (not shown here), suggesting that mesoscale gravity waves can partially account for the abnormally strong westward winds.Moss et al. (2016) observed over Ascension Island (8°S, 14°W) that strong westward fluxes of high-frequency gravity waves are not always sufficient to produce abnormally strong westward winds during the March equinox.They suggested that other forcing should be involved in order to fully explain the appearance of those strong westward winds.Garcia (2023) showed that the migrating diurnal tide breaks due to convective instabilities at and above ∼85 km around the equinoxes, but more frequently around the March equinox.Analysis of SABER temperatures revealed a peak in DW1 amplitude at ∼84 km in March 2023 (results not shown here) with a breaking layer depth of around 5 km.Furthermore, our analysis indicates that there were outstanding and strong DW1 breaking events around the March equinox 2023.This supports the breaking of DW1 as another driver of the abnormally strong westward mesospheric winds.
It has been expected that the meridional advection of angular momentum from higher latitudes enhances the equatorial westward flow related to the SSAO near the stratopause (∼50 km) in the solstitial seasons (Smith et al., 2023;Tomikawa et al., 2008).The meridional advection is carried out by the deep branch of the Brewer-Dobson circulation in the stratosphere and can cause large interannual variability, because the speed of the deep branch reflects the Eliassen-Palm (EP) flux divergence due to planetary waves, which mainly determines the presence or absence of sudden stratospheric warmings at winter high latitudes.However, it is unclear how important the meridional advection of angular momentum is in the upper mesosphere, where the strong westward winds are observed.This issue is left for future studies.
The comparison of SD-WACCM-X simulations with previously reported abnormally strong westward winds (in March 2002(in March , 2008(in March , 2011(in March , and 2017) ) shows that the model reproduces winds that exceed the mean by over one standard deviation, but with much weaker amplitudes than the events reported in the observations.This could be because the MSAO is generally too weak in WACCM (Gettelman et al., 2019;Richter & Garcia, 2006), which is likely related to poor representation of the gravity wave activity.Since SD-WACCM-X shares the same physics/ dynamics in the mesosphere with WACCM, it would have the same issue.Inadequate representation of the wave forcing may also account for the weaker westward winds during the strong cycles of the MSAO.
The strongest westward winds, which reached values larger than 80 m/s, were reported at low latitudes (Day & Mitchell, 2013;Garcia et al., 1997;Kishore Kumar et al., 2014) in 1993and 2002.In our study, we report a new case in 2023.The occurrence of these strongest westward winds in March 2023 satisfies the conditions of the filtering mechanisms in the stratosphere proposed by Kishore Kumar et al. (2014): they developed during the Westerly phase of the QBO, and there was no Strong SSW class in the previous northern hemisphere winter.Lossow et al. (2008) reported the climatology between 2002 and 2006 of mesospheric water vapor measurements taken by the Sub-Millimetre Radiometer (SMR) instrument aboard the Odin satellite.They showed an asymmetric variability around 75 km over low latitudes during the equinoxes, with a higher peak around the March equinox.Considering that strong westward winds were also reported in March 2002, this suggests that the events with abnormally strong westward phases of the mesospheric semiannual oscillation could affect water vapor transport around this altitude.Moreover, it is expected that the strong westward flow itself modulates the filtering of the upward propagation of gravity waves and affects the circulation in the region of the lower thermosphere and above so that this process could modulate the transport of minor constituents at this region.

Concluding Remarks
In this study, we have reported the strongest westward winds during the first phase of the mesospheric SAO in the last ten years (2014)(2015)(2016)(2017)(2018)(2019)(2020)(2021)(2022)(2023).They developed around the March equinox of 2023, at low latitudes.Meteor radar observations show that the westward winds reached 80 m/s (more than three standard deviations with respect to a three-year climatology) at 82 km of altitude, and decreased with altitude and latitude.On the other hand, the abnormally strong westward winds are not reproduced in SD-WACCM-X simulations.
Temperature measurements made by SABER suggest that the breaking of the diurnal tide plays a significant role in accelerating the zonal mean winds.Besides, these abnormally strong westward winds took place during the

Geophysical Research Letters
10.1029/2024GL110331 westerly phase of the QBO, which supports the filtering mechanism of eastward-propagating gravity waves in the stratosphere.Thus, the here reported abnormally strong westward winds must be the result of momentum deposition by both westward-propagating gravity waves and the DW1 tide.

Figure 1 .
Figure 1.Planetary-scale dynamics in the MLT region obtained by SIMONe Jicamarca, located at 12°S, 77°W.The figure displays the zonal (left column) and meridional (right column) components of the mean wind (first row) and the amplitude of the total 24-hr tide (second row).The orange dashed ovals highlight the unusual winds and tidal amplitudes around March equinox 2023.Data gaps are indicated in black (first row) and white (second row).

Figure 2 .
Figure 2. (a) Zonal mean winds from January 2022 to June 2023 obtained by five meteor radars at low latitudes.(b) Yearly zonal mean winds over Jicamarca (12°S) from 2020 to June 2023 at 82, 88, and 94 km.(c) The same as (b) but over Ledong (18°N) from 2019 to June 2023.(d) Zonal mean winds obtained from (a) at 82 km.The lightblue shaded region shows the climatology (average ±1σ) given by SD-WACCM-X from 2010 to 2021.See the text for more details.Data gaps are indicated in black.

Figure 3 .
Figure 3. Monthly zonal winds from 2014 to 2023 over Tirupati at 82 km.The light-blue shaded region represents the region of ±1σ with respect to the mean.The list shows the classification criterion suggested by Kishore Kumar et al. (2014) applied to this data set.Westerly (W) QBO phase, SSW "None" or "Weak," with both resulting in "Abnormal" flag, indicated in bold.The cases indicated in parenthesis must be further investigated.

Figure 4 .
Figure 4. Analysis of DW1 breaking by following the methodology of Garcia (2023).(a) Amplitude of DW1 at 12 sh (about 85 km) obtained from SABER temperatures, (b) number of breaking events per pentad, and (c) the average of the breaking layer depth over the equator.