Contributions of equatorial planetary waves and small-scale convective gravity waves to the 2019/20 QBO disruption

. In January 2020, unexpected easterly winds developed in the downward-propagating westerly quasi-biennial oscillation (QBO) phase. This event corresponds to the second QBO disruption in history, and it occurred four years after the first disruption that occurred in 2015/16. According to several previous studies, strong midlatitude Rossby waves propagating from the Southern Hemisphere (SH) during the SH winter likely initiated the disruption; nevertheless, the wave forcing that finally led to the disruption has not been investigated. In this study, we examine the role of equatorial waves and 10 small-scale convective gravity waves (CGWs) in the 2019/20 QBO disruption using MERRA-2 global reanalysis data. In June–September 2019, unusually strong Rossby wave forcing originating from the SH decelerated the westerly QBO at 0°– 5°N at ~50 hPa. In October–November 2019, vertically (horizontally) propagating Rossby waves and mixed Rossby–gravity (MRG) waves began to increase (decrease). From December 2019, contribution of the MRG wave forcing to the zonal wind deceleration was the largest, followed by the Rossby wave forcing originating from the Northern Hemisphere and the 15 equatorial troposphere. In January 2020, CGWs provided 11% of the total negative wave forcing at ~43 hPa. Inertia–gravity (IG) waves exhibited a moderate contribution to the negative forcing throughout. Although the zonal-mean precipitation was not significantly larger than the climatology, convectively coupled equatorial wave activities were increased during the 2019/20 disruption. As in the 2015/16 QBO disruption, the increased barotropic instability at the QBO edges generated more MRG waves at 70–90 hPa, and westerly anomalies in the upper troposphere allowed more westward IG waves and CGWs to 20 propagate to the stratosphere. Combining the 2015/16 and 2019/20 disruption cases, Rossby waves and MRG waves can be considered the key factors inducing QBO disruption.

calculated. Second, the GW momentum flux and drag are calculated based on Lindzen's saturation scheme (Lindzen, 1981) based on columnar propagation. It should be noted that, in order to constrain the magnitude of the CGW momentum flux obtained from an offline parameterization to prevent over-or under-estimation of the CGW forcing, we use GWs observed from super-pressure balloons in the tropical region (Jewtoukoff et al., 2013) (c.f., Kang et al., 2017). The small-scale CGWs considered in this study have small horizontal wavelengths smaller than 100-200 km. The details of the parameterization 110 scheme of the CGWs can be found in KCG20.
As a key source of the equatorial waves, convective activity is investigated using the precipitation data provided by MERRA-2. In addition, barotropic instability at the QBO edges is investigated as a potential source of the MRG waves The negative regions of ̅ indicate baroclinic/barotropic instability.

Results
3.1 Morphology of the zonal wind and each type of wave 120 Figure 1 shows the latitude-height cross section of the zonal-mean zonal wind from July 2019 to January 2020 with the corresponding monthly climatology (Fig. 1a) and the vertical profile of the zonal-mean zonal wind averaged for 5°N-5°S from July 2019 to January 2020 overlaid with the climatology (Fig. 1b). As early as July 2019, the northern side of the WQBO jet starts to be deformed. In September 2019, the westerly jet becomes weak at the altitude range of 40-50 hPa by more than 1 (Fig. 1b). Thereafter, the westerly wind at 43 hPa begins to decelerate, changing into the easterly in January 125 2020. The 2019/20 QBO disruption period shows a weaker westerly wind at altitudes near 30 hPa and a shallower WQBO jet compared to that in the 2015/16 QBO disruption period ( Fig. 1 of KCG20). As in the 2015/16 QBO disruption, positive wind shear anomaly and westerly anomaly compared to the climatology are observed in the upper troposphere  in July-December 2019 and January 2020, respectively. Figure 2 shows the EPF and EPFD for each equatorial wave and CGWs in a latitude-height cross section in January 130 2020. The EPF and EPFD are each multiplied by a factor of 2, except for the Rossby waves, to suitably represent the morphology of each wave. The P-CGWs (Fig. 2a) exhibit a positive (negative) forcing at 60-80 hPa (20-30 hPa and ~50 hPa), which is the strongest at 20 hPa over 5°N -5°S. Close to the equator, the negative CGW forcing is anomalously strong at 50-60 hPa.
In the lower stratosphere (60-100 hPa), Kelvin waves exert positive forcing on the QBO jet, thereby maintaining the 135 westerly jet below the easterly wind development (Fig. 2b). However, the Kelvin wave forcing at 20-30 hPa is considerably smaller than that in February 2016 ( Fig. 2b of KCG20); this is because the upper jet is very weak. The Kelvin waves propagating from the troposphere are larger than the climatology (Fig. S1), though the increase is lesser than that in January-February 2016.
MRG waves provide a strong negative forcing to the zonal wind at 25-100 hPa, concentrated at the equator (Fig. 2c). 140 The negative MRG wave forcing at 40-50 hPa, which is critical for inducing the QBO disruption, is anomalously strong at 2°-5°N/S compared to the climatology. The MRG waves seem to be mainly generated at the location with positive EPFD in 5°-10°N/S and 60-90 hPa, as in the 2015/16 QBO disruption (Fig. 2c of KCG20).
IG wave forcing (Fig. 2d) shows negative values at 10°N-5°S, with an anomalously large magnitude located at 60-80 hPa and 5-15 hPa. In addition, Rossby wave forcing (Fig. 2e) exhibits large negative values at 0°-5°N, and they appear to 145 propagate from the NH extratropics. Figure 3 presents the monthly evolution of the zonal wind, zonal wind tendency, vertical advection (ADVz), required wave forcing (REQ), and each wave forcing averaged for 5°N-5°S from May 2019 to April 2020 at 10-70 hPa. In order to calculate the REQ, both the meridional and vertical advection terms are subtracted from the zonal-mean zonal wind tendency in Eq. (1). From June to September 2019, the magnitude of the WQBO is reduced, without any significant downward propagation (Fig. 3a), compared to the climatology (Fig. 3k). Comparison between the zonal-mean zonal winds in the 2019/20 QBO disruption and climatology (Fig. 3b) suggest an anomalous weakening of the zonal wind from July 2019, which is maximized at 40-60 hPa. The negative zonal wind tendency near 43 hPa is evident from June to August 2019 ( Fig.   3c), which can be mainly attributed to the Rossby wave forcing (Fig. 3j).
The WQBO that maintains its depth without any significant downward propagation in June-September 2019 seems to 155 be related to the strong ADVz (Fig. 3d). ADVz values at 20 hPa in June, July, August, and September 2019 are 9.6, 12.5, 13.3, and 11.3 m s month -1 , respectively, and these values are considerably larger than those for the climatology (2.8, 4.4, 5.8, and 6.3 m s -1 mon -1 , respectively; Fig. 3m). In particular, the ̅ * values (Fig. S2) in July and September 2019 are 0.7 and 0.9 mm s -1 , respectively, which are 1.6 and 1.5 times larger than that for the climatology, respectively. In this period, midlatitude Rossby wave forcing is extremely large and induces a minor SSW (Anstey et al., 2020;Eswaraiah et al., 2020;Shen et al., 160 2020), possibly resulting in the enhanced vertical upwelling of the Brewer-Dobson circulation (BDC) and, thereby, a large magnitude of the ADVz. This implies that the ADVz can help QBO disruption by retarding the downward propagation of the WQBO jet. Although the 2019 SSW is classified as a minor SSW in that the zonal-mean zonal wind at 10 hPa at 60°S does not undergo a reversal, the zonal wind at ~32 km at 72°S shows an easterly wind (Eswaraiah et al., 2020), which implies a strong Rossby wave forcing in the SH. 165 Climatologically, REQ (Fig. 3n) exhibits a negative (positive) sign in negative (positive) wind shear zone, and the sign of P-CGW forcing (Fig. 3f) generally follows that of the REQ. The larger the magnitude of the vertical wind shear, the more the P-CGWs explain the REQ. However, in June-July-August (JJA) 2019 (Fig. 3e), a negative REQ is observed at 30-60 hPa without negative vertical wind shear; this seems to be unusual. The P-CGWs start to contribute to the deceleration of the QBO jet after the negative vertical wind shear is generated at ~50 hPa (i.e., October 2019). In contrast to the strong Kelvin 170 wave forcing in the 2015/16 QBO disruption, Kelvin wave forcing (Fig. 3g) in the 2019/20 QBO disruption is smaller than or comparable to the climatology (Fig. 3p). This weak Kelvin wave forcing could be one of the reasons why the upper jet at 20-30 hPa is not maintained after the QBO disruption.
During the 2019/20 QBO disruption, the momentum forcing by the MRG waves ( Fig. 3h) is considerably stronger than its climatology (Fig. 3q). For instance, from October 2019 to January 2020 the MRG wave forcing at 43 hPa is dominant 175 among that of the equatorial waves, largely explaining the REQ. This result suggests that MRG waves play a role in reversing the sign of the zonal in the later stages. IG wave forcing (Fig. 3i) shows strong negative values in May 2019 above 43 hPa and after July 2019 following the negative wind shear zone. Rossby wave forcing (Fig. 3j) is strong from June to September 2019 below ~20 hPa. At 40-50 hPa, Rossby waves continue to provide a negative wave forcing until February 2020. Figure 4 shows the monthly evolution of zonal wind, zonal wind tendency, and wave forcing of each wave type from May 2019 to April 2020 at 43 hPa; their exact values and percentages are summarized in Table 1. As early as May 2019, the zonal wind tendency (dotted line in Fig. 4a) becomes negative, while, in January 2020, the zonal wind (solid line in Fig. 4a) becomes easterly. The negative wind tendency is weakened until October 2019 although it intensifies again in November 185 2019. The negative wind tendency in May 2019 is mainly explained by the Rossby (-0.62 m s -1 mon -1 ) and IG (-0.57 m s -1 mon -1 ) waves, with contributions of 48% and 45%, respectively. The momentum forcing by the Rossby waves becomes dominant from June to November 2019. The maximum contribution is 82% (in July 2019), and it decreases subsequently. In December 2019 and January 2020, the MRG wave forcing accounts for 44% and 41% of the total negative wave forcing, respectively, which are larger than any other equatorial wave forcing. During the same period, the Rossby wave forcing is 190 the second largest, with contributions of 33% and 38%, respectively. In January 2020, parameterized CGWs start to contribute to the easterly development (11%), and they provide large negative forcing in February 2020 with a percentage of 44%.

Contributions of each wave type at 43 hPa
The contribution from the parameterized CGWs is smaller than that in the 2015/16 QBO disruption. As shown in Fig.   13, the magnitude of the source-level westward CGW momentum flux is not significantly larger than that of the climatology; 195 this is the probable cause of the smaller magnitude of the negative CGW forcing during the 2019/20 disruption than that during the 2015-2016 disruption. The smaller CGW forcing is also explained by the vertical wind shear at ~40 hPa in wave forcings in December 2019-January 2020. IG waves decelerate the WQBO jet with a moderate magnitude throughout, and the P-CGWs contribute 11% of the negative forcing in January 2020. indicated by the gray shading. In July 2019 (Fig. 5a), the EPFD for the Rossby waves is unusually strong at the northern flank of the QBO at 40-60 hPa. The meridional EPFD dominates the total EPFD at 40-60 hPa in the NH. They are most likely to propagate from the SH based on the large northward EPF at 10°S. Moreover, vertical EPF at 70 hPa is larger than 215 the climatology at 10°N-10°S; accordingly, a large negative EPFD-z can be observed at 30-50 hPa. In August 2019 (Fig.   5b), there is evident deceleration of the WQBO jet by the Rossby waves propagating from the SH; however, the negative wave forcing becomes stronger at the jet core. It is also found that the EPF-z at 70 hPa in August 2019 is larger than that in July 2019 at 5°N-20°S.

Rossby waves
In October 2019 (Fig. 5c), the shape of the zonal wind is significantly deformed by the anomalously strong negative 220 forcing in the WQBO jet, mainly attributed to the strong meridional Rossby wave forcing originating from the SH. In January 2020 (Fig. 5d), when the QBO disruption occurs, the Rossby wave forcing is generally weaker than that shown in Figs. 5a-c; consequently, the EPFD in Fig. 5d is multiplied by a factor of two. The Rossby waves laterally propagating from the NH decelerate an isolated small westerly jet at 30-40 hPa, while the vertically propagating Rossby waves provide an anomalously strong easterly forcing below the altitude of 40 hPa at 25°N-15°S, except close to the equator. The EPF-z at 225 70hPa, which is larger during the disruption period than the climatology at 0°-20°S and 10°-20°N, confirms the presence of the strong Rossby waves propagating from the equatorial region.
In summary, Rossby wave forcing and flux during the austral winter of 2019 have a dominant meridional component propagating from the SH. However, a relatively small magnitude of the Rossby wave forcing is found with comparable meridional and vertical components in January 2020. The strong EPF-z at 70 hPa mostly propagates from the equatorial 230 troposphere and the NH, when the EPF is traced back to the troposphere (Fig. S3).
As mentioned previously, a minor SSW took place in the SH in September 2019, which was an exceptionally rare event.
This implies that Rossby wave flux and forcing in the midlatitude stratosphere was above average during the austral winter of 2019. Figure 6 shows the latitude-height cross section of the EPF overlaid with the zonal-mean zonal wind (Fig. 6a), vertical EPF at 100 hPa (Fig. 6b), and zonal wind at 15°S (Fig. 6c) in JJA. The red line represents the 2019 case, and the 235 black line represents the climatology. The waves are generally vertically propagating, while a part of the waves propagates into the Tropics. The vertical EPF penetrating the stratosphere is considerably larger than the climatology by ~2 (Fig. 6b).
An excessively large EPF in the midlatitude stratosphere could also propagate into the equator because the zonal-mean zonal wind in the SH subtropics at 40-80 hPa exhibits stronger westerly winds than the climatology (Fig. 6c). 5°-10°N, (ii) 40-80 hPa near 10°N, and (iii) ~40 hPa near 10°S. This is supported by considerably stronger vertical EPF at 245 70 hPa at 0°-10°N and meridional EPF at 20-50 hPa at 10°N/S. In November 2019 (Fig. 7b), similar features as in October 2019 are shown but with a reduced vertical range for the negative wave forcing near the equator.

MRG waves 240
In December 2019 (Fig. 7c), westerly winds at 30-50 hPa are weakened. The negative MRG wave forcing becomes unusually strong at 50 hPa in the 5°-10°S range, although the increase in the EPF-z at 70 hPa is smaller than those in October and November 2019. In January 2020 (Fig. 7d), MRG wave forcing at 43 hPa is the largest among all the equatorial 250 wave forcings. Not only the equatorward waves at 10°N/S at 30-50 hPa but also the equatorward and upward waves at 10°N/S at 70 hPa are much stronger than the climatology by more than 1 . In particular, the upward and equatorward EPF vectors starting from 5°-10°S at 70 hPa appear to exhibit the maximum contribution to the negative forcing observed at 43 hPa. Figure 7 shows that the MRG waves weaken the QBO jet and finally reverse the wind sign in the later period (e.g., 255 December 2019 and January 2020). The negative MRG wave forcing is exerted on the jet core not only at the 43 hPa altitude but also at the altitude range of 25-50 hPa, resulting in an excessive weakening of the upper jet (~30 hPa) during the 2019/20 QBO disruption. In addition, MRG waves are strongly generated in regions with a large horizontal wind curvature, coincident with the location of the positive EPFD. Therefore, in order to investigate whether the MRG waves are generated by barotropic/baroclinic instability, we select a region (boxed region in Fig. 8) with small positive ̅ values. 260 Figure 8 shows the monthly-averaged ̅ and the daily time series of the number of grids with the negative ̅ at the boxed region in December 2019 (Fig. 8a,c) and January 2020 (Fig. 8b,d), along with the climatology. Note that the total number of grids in the boxed region is 33. The monthly mean ̅ in the boxed region shows small positive values in December 2019 and in January 2020; however, the number of negative ̅ in the boxed region based on the daily-mean values (Figs. 8c-d) is generally much larger during the disruption period compared to that of the climatology. The barotropic 265 term [first two terms on the right-side of Eq. (3)] dominates the ̅ value in the boxed region; on that basis barotropic instability at the boxed region is likely to generate anomalously strong MRG waves.  statistically significant wave signals at the 95% confidence level. The spectrum more than 1 stronger than the climatology (blue-stippled pattern) starts to widen in December 2019, although the area is smaller than that in the 2015/16 QBO disruption. Generally, the strong power is evident in the spectrum related to the Kelvin and IG waves. In the symmetric 280 spectrum, statistically significant Rossby wave signals ( = -16-19, = 0.06-0.1 cpd) are shown, which are stronger than the climatology by more than 1 in November 2019-January 2020 (Figs. 10b-d). The enhancement of the Rossby waves in the troposphere in January 2020 probably affects the large vertical EPF at 70 hPa (Fig. 5d). Kelvin wave signals ( = 0-8 and = 0-0.25) are statistically significant throughout and are more than 1 stronger than the climatology after November 2019. It is likely that these waves propagate to the stratosphere and, thereby, contribute to the strong EPF-z at 70 hPa (see 285 Fig. S1). In the antisymmetric spectrum, the MRG wave signals in the antisymmetric spectrum ( = -10-0 and = 0.2-0.32) are stronger than the climatology by more than 1 in December 2019-January 2020. Therefore, the enhanced convective activity in the MRG wave spectrum in December 2019-January 2020, together with the barotropic instability at the QBO edges, may affect the anomalously strong MRG wave forcing near 43 hPa. Overall, convectively coupled equatorial waves are slightly enhanced in the later period of the 2019/20 QBO disruption, although the zonal-mean precipitation is not 290 significantly increased. Figure 11 shows the EPF and EPFD as a function of latitude and height, and latitudinal distribution of the vertical EPF by the IG waves at 70 hPa from October 2019 to January 2020. Given that the IG waves generally propagate upward in the stratosphere, the upward-directed EPF vectors inside the WQBO jet at 5°N-5°S indicate a larger magnitude of the westward 295 IG waves compared to that of the eastward IG waves. The negative IG wave forcing is exerted on the jet core throughout, with a significant magnitude located at the altitude range of 60-90 hPa. However, the magnitude of the EPF-z at 70 hPa is slightly larger than that of the climatology in December 2019 and January 2020, differing from the case in the 2015/16 QBO disruption. Figure 12 illustrates the 10°S to 10°N averaged phase-speed spectrum of the precipitation for the IG wave ranges, 300 which approximately represents the source spectrum of the IG waves in December 2019 (Fig. 12a) and January 2020 (Fig.   12b) along with the climatology. Generally, the disruption period shows a larger IG wave source spectrum by ~1 compared to the climatology. The zonal wind speed at 140 hPa is approximately 2.6 m s -1 and 4.9 m s -1 in December 2019 and the climatology, respectively. Therefore, the IG source spectra during both the disruption period and climatology exhibit dominant westward components, although the climatology exhibits additional westward waves in the phase speed of 2.6-4.9 305 m s -1 . However, the additional westward waves of the climatology in 2.6-4.9 m s -1 are dissipated by the critical-level filtering (-0.2-5.4 m s -1 ), and this range is wider than that (1.1-4.2 m s -1 ) of the disruption period. Thus, the remaining westward waves at 70 hPa are stronger in December 2019 than the climatology. The narrower critical-level filtering range is related to the westerly anomalies and easterly anomalies at 70-100 hPa and 100-140 hPa, respectively (Fig. S4). In January https://doi.org/10.5194/acp-2021-85 Preprint. Discussion started: 1 March 2021 c Author(s) 2021. CC BY 4.0 License. competition between the two factors results in an increased eastward momentum flux (bottom panel of Fig. 13). In our CGW parameterization, we obtain ℎ by averaging the zonal wind below 700 hPa, which is related to the propagation speed of the gust front (Choi and Chun, 2011). Therefore, the westerly anomalies in the ℎ are caused by the westerly anomalies in the zonal wind below 700 hPa. The westerly anomalies in the lower troposphere frequently occur under El Niño conditions and 345 in future climate simulations (Lu et al., 2008;Collins et al., 2010;Kawatani et al., 2019). Therefore, the high surface temperature during the 2019/20 QBO disruption likely led to the westerly anomalies in the lower troposphere. There is a need for further study on the cause and significance of the westerly anomalies in a warmer climate, although doing so is beyond the scope of this study. Although the magnitude of the westward CGWs at the source level is similar to that in the climatology, the eastward shift of the zonal winds at 100-200 hPa (Fig. 13) resulted in more westward waves propagating 350 into the stratosphere compared to those in the climatology. Overall, the increase in the CGW momentum flux in January 2020 is considerably smaller than that in February 2016, and no significant increase is observed in the westward momentum flux. Together with the weaker negative vertical wind shear at 43 hPa, this results in a small magnitude of the negative CGW forcing near 43 hPa.

Summary and Conclusions 355
In this study, we examined the role of each equatorial planetary wave mode and parameterized convective gravity waves (CGWs) in the 2019/20 QBO disruption and compared with the results from the 2015/16 QBO disruption (KCG20).
Using MERRA-2 model-level data, we separated each equatorial wave mode (Kim and Chun, 2015) and obtained smallscale CGW forcing by performing an offline CGW parameterization (Kang et al., 2017). The main results are summarized schematically in Fig. 15 and in the following text: 360 • From June to September 2019, unusually strong Rossby wave forcing at ~50 hPa decelerated the westerly QBO jet at 0°-5°N. The strong Rossby wave flux propagated mostly from the SH midlatitudes due to the large wave activity associated with the 2019 minor SSW in the SH and the westerly anomalies in the SH subtropics. MRG and IG wave forcing partly contributed to the wind deceleration.
• From October to November 2019, laterally propagating Rossby wave flux from the SH was weakened, with the 365 vertically propagating Rossby wave flux from the Tropics being enhanced. MRG wave forcing increased with nearly the same contribution as that from the latitudinally propagating Rossby waves. Furthermore, the IG wave forcing began to increase, albeit with a smaller magnitude than that of the MRG wave forcing. In this period, the oval structure of the zonal wind was significantly deformed.

•
From December 2019 to January 2020, the momentum forcing by the MRG waves was stronger than that by any 370 other equatorial waves, mainly due to the strong barotropic instability at the QBO edges at 70-90 hPa, and partly due to the enhanced convective activity, as in the 2015/16 QBO disruption. Rossby waves propagating from the NH midlatitudes also decelerated the QBO jet. In January, the QBO westerly was changed to easterly at 43 hPa.  Table 1. Monthly-averaged momentum forcing by each wave type (m s -1 month -1 ) at 43 hPa averaged for 5°N-5°S from June to January for the disruption period (2019/20) and the climatology. The ratio of each wave forcing to the total negative forcing is given in the parenthesis only when the wave forcing is negative. 505      monthly climatology (black) and ±1 standard deviation (gray shading). The spectrum with a negative sign represents the westward-propagating waves. Double-sided arrows in the upper part of each panel indicate the zonal wind ranges between 140 hPa (i.e., source level) and 70 hPa for the QBO disruption period (red) and climatology (black).  for 5°N-5°S in January 2020 and those for the climatology (black solid and black dashed, respectively) with ± 1-standard deviation (dark-gray and light-gray shading, respectively). (Bottom) Zonal-mean zonal CGW momentum flux spectrum at the cloud top averaged for 5°N-5°S in January 2020 (red) and its climatology (black) with ± 1-standard deviation (gray 605 shading).