Ozone Changes Due To Sudden Stratospheric Warming‐Induced Variations in the Intensity of Brewer‐Dobson Circulation: A Composite Analysis Using Observations and Chemical‐Transport Model

We quantify the changes in the intensity of Brewer‐Dobson Circulation (BDC) during sudden stratospheric warming (SSW) and its impact on the tropical stratospheric thermal structure and ozone distribution by composite analysis using observations and a chemical‐transport model. An increase in the planetary wave activity and enhancement in BDC intensity before the central date of SSW is noticed. A positive ozone anomaly is observed in the tropical upper stratosphere. The tropical lower stratosphere shows a cooling (1–2 K) and negative ozone anomaly (∼0.1 ppmv) after ∼10 days from the central date. The polar stratosphere experiences a positive ozone anomaly, whereas the upper stratosphere shows ozone depletion due to the downwelling of NOx‐rich mesospheric air. The cold‐point tropopause temperature shows a cooling of ∼0.5 K for major warming which in turn dries the lower stratosphere.

The changes in BDC would also affect the water vapor composition in the stratosphere. The stratospheretroposphere exchange processes are modulated and water vapor concentrations vary in response to various atmospheric oscillations (Das & Suneeth, 2020; and references therein). Evan et al. (2015) have reported elevated tropopause and a decrease in water vapor concentration by 1.5 ppmv following the 2013 SSW. Tao et al. (2015b) have suggested that an increasing number of major warmings could lead to a lower stratospheric water vapor after the 2000s.
The present study aims to investigate the changes in BDC intensity and its impact on ozone and water vapor distribution using a composite analysis of the Northern Hemisphere's major and minor SSW events. We further compare the results obtained for SSW events from 2005 to 2020, the time period for which Aura-Microwave Limb Sounder (MLS) data are available, with the quiet period to delineate the observations from the inter-annual variation of BDC.

Microwave Limb Sounder
Ozone, water vapor, and temperature measurements used in the study are from the Microwave Limb Sounder (MLS, version 5.0x) onboard the Aura satellite. Aura is a near-polar orbiting satellite at 705 km and has about 14.6 orbits per day with an equator crossing time at ∼1:30 a.m./p.m (Local time) (Schoeberl et al., 2006). Ozone is measured from a 236 GHz ozone line with a precision of 15%-20% and an accuracy of 8%-15%. Water vapor profiles were obtained from a 190 GHz line with a precision of 4%-15% and an accuracy of 4%-19% in the stratosphere. Temperature is primarily obtained from bands near 118 and 239 GHz and has a precision of 0.7-1.2 K in the stratosphere. The data were gridded into 5° latitude × 20° longitude for the daily mean. A detailed description and data quality can be found in Livesey et al. (2020).

ERA5 Reanalysis
We use daily data of diurnal mean for horizontal and vertical winds and temperature from ERA fifth generation (ERA5) reanalysis, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF). We used data with a grid resolution of 2.5° (latitude) × 2.5° (longitude). ERA5 features an improved representation of atmospheric parameters, especially over the tropical regions of the troposphere. The details are available in Hersbach et al. (2020).

GEOS-Chem Model
We used Goddard Earth Observing System-Chemistry (GEOS-Chem) model output (version 10-01i; https:// geos-chem.seas.harvard.edu/) to analyze ozone during the warming. GEOS-Chem is a global chemical transport model with a fully coupled tropospheric NO x -O x -hydrocarbons-aerosol chemistry. The ozone mixing ratios in the model agree within 10 ppbv out of 30-60 ppbv with ozonesonde values and within 6% with satellite observations over the Indian subcontinent (David et al., 2019). The model output with a grid resolution of 2°(latitude) × 2.5°(longitude) up to ∼80 km was used for the study. The emissions and model output details are given in David et al. (2019).

Ozonesonde
There were regular (weekly to a fortnight) ozonesonde launches by India Meteorological Department (IMD) over Thiruvananthapuram (8.48°N,76.95°E). The ozonesondes use an Electrochemical Concentration Cell that measures vertical height profiles of atmospheric ozone by electrochemical oxidation of potassium iodide by ozone in an aqueous solution. The ozone mixing ratios are obtained from the surface up to altitudes typically 35 km with a height resolution of 500 m. An uncertainty of 5%-10% exists in the troposphere (Das et al., 2019 and references therein). Six days (1, 3, 18 and 25 January and 1 and 15 February) of ozone profiles were available during the SSW event of 2006, which covers the life cycle of SSW.

COSMIC and CHAMP
Temperature and mean sea level altitude profiles were obtained from three GPS Radio Occultation (GPS-RO) missions, that is, CHAMP (Challenging Minisatellite Payload), COSMIC-1 and COSMIC-2 (Constellation Observing System for Meteorology, Ionosphere and Climate) missions. The GPS-RO measurement technique has the advantage of a high vertical resolution of 100 m (50 m for COSMIC-2 up to tropopause) and global coverage under all weather conditions. It has a temperature accuracy of 0.5 K (0.1 K) for individual (averaged) profiles (Hajj et al., 2004;Kishore et al., 2011, Suneeth et al., 2017Veenus et al., 2022 and references therein).

Classification of SSW
The zonal-mean zonal wind and temperature obtained from ERA5 reanalysis were used to classify SSW events (Liu et al., 2022). Figure 1 shows the zonal-mean temperature at 60°N and 90°N at 10 hPa during quiet, minor warming and major warming periods. Zonal-mean zonal wind at 60°N at 10 hPa is also superimposed on the  figures. When the temperature at 90°N suddenly exceeds 210 K compared with the temperature at 60°N, and the zonal-mean zonal wind at 60°N is reversed from climatological wintertime eastward to westward direction, it is identified as major warming, whereas if no reversal of wind is observed it is classified as minor warming. A period of 45 days was selected based on minimum wave activity for a quiet period. For major SSW, the central date was identified as the date of wind reversal, whereas for minor, it is the date of maximum temperature, as shown in Figure 1 (central dates are tabulated in Table S1 in Supporting Information S1). 2013 and 2019 were early warmings, 2008 had multiple warmings and 2010 had a short period of wind reversal. Thus, the results were evaluated excluding these years and presented in the supplementary materials ( Figures S2 and S3 in Supporting Information S1).

Methodology
The composite profiles of BDC strength, temperature and ozone distributions were generated with super-epoch analysis for the first 90 days of a year. The results obtained for SSW events were compared with quiet period to delineate the observations from the inter-annual variation of BDC. The Bootstrap technique (Efron, 1992) was used to estimate statistical significance, where samples were generated by rearranging the years in the composites. We examined if the difference between the original composites was larger than the difference between the Bootstrap samples in 90% of the cases. The following BDC metrics were computed for minor, major warming, and quiet periods.

Eliassen-Palm Flux
Eliassen-Palm flux (EP flux) acts as a proxy for the wave-driven torque in the atmosphere. The components of EP flux are estimated using Equation 1 following Andrews et al. (1987): Here, the overbar denotes the zonal means and the prime denotes the deviation from the zonal mean. ρ is the atmospheric density, a is the radius of the Earth, φ is the geographical latitude, θ denotes the potential temperature, and the subscript z implies partial derivative with respect to z (log-pressure height). u and v denote the zonal and meridional winds, respectively, from ERA5 reanalysis.

Residual Mean Meridional Stream Function
We estimate the residual mean meridional stream function (RMMSF, ψ*), which provides an approximation to the mean advective transport of trace substances using the following Equation 2 (Andrews et al., 1987): where * denotes the meridional component of residual velocity and a is the radius of the Earth.

Residual Vertical Velocity
The residual vertical velocity averaged from 15°S to 15°N is used to analyze the tropical upwelling. The vertical component of residual velocity is defined (Equation 3) as given in Andrews et al. (1987); ω denotes pressure velocity. The unit of residual vertical velocity in log pressure coordinates is Pa s −1 .

Diabatic Heating Rate
Zonal-mean diabatic heating rate (DHR) averaged from 15°S to 15°N and 100-10 hPa is calculated from ERA5 reanalysis using the following Equation 4 (Andrews et al., 1987): here, Q denotes the total heat absorbed, c p is specific heat capacity, T denotes temperature, t is time, p is pressure, v and w are meridional and vertical winds, respectively. κ is R/c p , with a value of 0.286, where R is the universal gas constant. DHR is used to analyze the strength of the tropical upwelling branch of BDC.

Dynamical Changes During Major and Minor Warming
The EP flux divergence was evaluated for 90 days for each SSW event (Table S1 in Supporting Information S1). A super-epoch analysis was carried out to analyze the wave-driving during the SSW period based on the central date, as shown in Figure 2a. The quiet period had an oscillating pattern of EP flux with divergence and convergence of waves occurring periodically. The value of EP flux for the quiet period was ∼1 ms −1 day −1 in the 8-15 hPa, where wave breaking occurs. For the minor warming, the increased negative values of EP flux divergence show enhanced wave breaking in the stratosphere before the warming. After the warming, a reduction in wave activity throughout the stratospheric height was observed. For major warming, the wave breaking itself was greatly reduced after the central date and followed a downward propagating pattern. The zonal-mean zonal wind at midlatitude (averaged between 32.5°N and 62.5°N) is plotted for a similar period (Figure 2b), which also showed a similar downward propagating pattern.
Figures 2c and 2d show the vertical (at 40°N) and latitudinal distribution at 10 hPa of RMMSF, respectively. The RMMSF during the quiet period showed low values compared to the disturbed years. A peak in RMMSF was observed along the central date of warming. For minor warming, the RMMSF peaked for a short duration (∼5 days), it extended for a longer duration (∼10 days) and higher heights (∼1 hPa) for major warming and reached values up to 400 kg m −1 s −1 . The latitudinal extent reaches a peak on the central date and reduced afterward ( Figure 2d). Within 10 days the latitudinal reduction occurred for major warming, and a gradual decrease was observed for minor warming. Figure 2e shows the residual vertical velocity averaged between 15°S and 15°N. Similar to EP flux, super epoch analysis is applied. The upwelling inferred from the residual vertical velocity is below 2 Pa s −1 throughout the middle stratosphere for quiet periods and minor warming. But for major warmings, increased upwelling to higher pressure levels is observed. Major warming upwelling reduced to lesser values after the central date than minor warming. Figure 2f shows the zonal-mean DHR averaged between 15°S and 15°N from 100 to 10 hPa. A super-epoch analysis is applied for DHR similar to EP-flux. The DHR was below ∼0.60 K day −1 for quiet periods that increased during minor and major warming. The enhancement in BDC persisted for more than 15 days after the central date for major warming and the magnitude of DHR also increased above 0.60 K day −1 . The DHR gradually relaxed to normal values 15 days after the central date.

Stratospheric Ozone Distributions
The stratospheric ozone mixing ratios (OMR) and temperature from Aura MLS were taken for all SSW years from 1 January to 31 March. A composite analysis was carried out on the first 90 days for all years. For leap years, 31 March was not considered. The time mean was removed using the average of 90 days. The seasonal mean was removed by selecting non-SSW years from 2005 to 2020 for a similar period of 90 days. The super-epoch analysis was carried out based on the central date of SSW. A similar analysis was done using the output from the GEOS-Chem model. The resulting temperature structure (MLS) and OMR (MLS and GEOS-Chem) at the tropics and poles are shown in Figure 3 for the major and minor warming. Over the tropics, cooling in the upper stratosphere was observed during and after a few days from the central date, with a magnitude of 1-2 K for minor warming (Figure 3a). For major warming, the temperature showed a 2-3 K decrease in the upper stratosphere around the 6 of 11 central date and a propagating negative anomaly of ∼1 K up to the tropopause levels, 5-10 days after the central date ( Figure 3b) that persisted for next 25 days.
For minor warming, the tropical upper stratosphere showed regions of ozone enhancement. The ozone reduction was observed in the lower stratosphere for a narrower part than for major warming (Figure 3a). Similar analyses were also carried out using ERA5 climatological data set 1979-2020 ( Figures S4 and S5 in Supporting Information S1; central dates are in Table S3 in Supporting Information S1) and the features remained same. The GEOS-Chem model captured the stratospheric ozone changes but differs in lower stratospheric ozone anomalies, the anomalies were displaced to a lower vertical level (8-9 hPa). The OMR from MLS showed an increase in ozone above 10 hPa for major warming. Over the tropical region, an increase of 0.2-0.3 ppmv was observed on and around the central date in the upper stratosphere. In the lower stratosphere, a decrease was observed, starting At the poles, the minor warming showed a temperature increase of ∼5-10 K. The downward propagation seen in major warming was not observed in minor warming (Figure 3a). The OMR exhibited a similar pattern as in major warming, but the anomalies were confined to smaller regions for minor warming. During major warming, the downwelling raised the temperature to 30 K, reaching the lower stratosphere within 5 days (Figure 3b). The OMR showed an increase of more than 0.4 ppmv over a vast region and for a longer duration in the lower stratosphere for major warming. The upper stratosphere showed discrete regions of ozone depletion on the central date of magnitude ∼0.4 ppmv.

Effect on UTLS Region
The upper tropospheric OMR for 2006 major SSW were analyzed from the in-situ ozonesonde measurements from IMD as shown in Figure 4a. It was observed that after the warming, OMR in the troposphere was lower than the climatological mean, calculated for 1969-2014 from IMD ozonesonde measurements (Figure 4a). The OMR anomalies obtained from MLS at 56 hPa are shown in Figure 4b. Further the CPT temperatures and height were detected from the individual temperature profiles obtained from CHAMP and COSMIC-1/2 from December to April. The CPT temperature and height were averaged zonally between the latitude range of 15°N-15°S for the quiet periods, major warming, and minor warming. The water vapor at 82 hPa was obtained from Aura MLS in the tropics. Before removing the time mean, the values from the non-SSW period were subtracted from the major and minor warming observations. A cooling of ∼0.5 K in the CPT temperature was recorded for major warming ∼10 days after the central date. The CPT height showed an increase after ∼10 days, and a reduced water vapor mixing ratio was observed along with a decrease in CPT temperature.

Discussion and Concluding Remarks
In this study, a composite analysis was performed using satellite observations from Aura MLS and simulation from the GEOS-Chem model along with the improved reanalysis product, ERA5 to quantify the thermal, ozone and water vapor variations in the tropical and polar stratosphere in response to the intensity changes in BDC. We analyzed major and minor SSWs, along with the quiet period. The study is limited by only having high temporal resolution ozone measurements from a relatively small number of years, thus interannual variability could play a role. In addition, Quasi-Biennial Oscillation (QBO) and El Niño -Southern Oscillation (ENSO) phases will influence BDC. The different phases of QBO and ENSO used for the present study are listed in the supplementary Figure S1 and Table S2 in Supporting Information S1. The composite analysis has all phases of QBO and ENSO, thus their influence has been controlled.
The EP flux divergence plotted around the central date of SSW clearly showed an increase in wave activity as SSW approached (Figure 2a), after which EP flux divergence substantially reduced, which is in line with the previous study by Tao et al. (2017). After the central date, the region of wave-breaking showed a downward propagation for major warming, which correlated with the zonal wind propagation (Figure 2b) due to the critical layer control. Since wind reversal was not observed in minor warming, the wave breaking continued after the central date. The RMMSF plotted for stratospheric heights also denoted a reduction in residual transport after the wind reversal in major warming (Figure 2c). The RMMSF showed increased residual circulation that peaked around the central date and subsided in the following days. The minor warming periods did not observe the sudden reduction in RMMSF as in major warming, owing to the persisting wave driving. The mass continuity requirements would lead to an increase in upwelling over the tropical region. The upwelling was observed to be strengthened in major warmings from the residual vertical velocities (Figure 2e) which also extend to higher pressure levels.
The enhanced upwelling was reflected in the increased DHR around and after the central date (Figure 2f). The upwelling resulted in stratospheric cooling in line with the warming observed in the polar regions. The observed substantial ozone enhancement (Figure 3) over the tropical region could be due to the decrease in temperature over the tropical stratosphere. The temperature over the tropical stratosphere decreases by the strengthening upwelling branch of BDC. The tropical stratosphere temperature dropped from 2 to 3 K simultaneously with the polar warming. The upper stratosphere ozone distribution is strongly affected by photochemical reactions, where the nitrogen cycle (R1-R3) constitutes ∼70% of the ozone destruction (Lary, 1997).
As Flury et al. (2009) mentioned, the ozone destruction by NO x is strongly linked to temperature and suggested an anti-correlation between ozone and temperature at 4 hPa. From a chemical box model simulation, the authors suggest that an increased temperature would aid ozone destruction by NO x at mid-latitudes. The downward propagation of the positive ozone anomaly observed in Figure 3b is in line with the temperature pattern over the stratosphere. The observed interannual variability in the tropical stratospheric ozone can be up to 400 ppbv (e.g., Tummon et al., 2015), whereas the effect of QBO and ENSO can be up to 200 and 100 ppbv, respectively (e.g., Xie et al., 2020). The present analysis shows SSW induced ozone changes are in the range of 300-600 ppbv.
The GEOS-Chem model reproduces some aspects of the pattern of enhancement in the tropical upper stratospheric ozone. But the magnitude is less than the MLS observations. Whereas, the dynamically driven lower stratospheric ozone reduction is barely seen in the model output for both minor and major warming. Again, the ozone reduction in the polar upper stratosphere, which is due to the amplified NO x destruction cycle, is captured in the model. Thus, being a chemical-transport model, GEOS-Chem reproduces the elements of the chemically driven features but is unable to match the magnitudes.
During the SSW lifespan, deviations in CPT temperature and height were also seen. The tropopause is a region, which is in the state of convective and radiative equilibrium. The CPT height is mainly controlled by tropical convection, whereas the CPT temperature is controlled by stratospheric ozone (Das et al., 2012 and references therein). The tropopause height was further raised by the upwelling that peaked during SSW (Figure 4), whereas there is an ozone loss in the lower stratosphere, causing the CPT temperature to cool after ∼10 days from the central date of SSW. The decrease in temperature affects the freeze-drying mechanism and would obstruct the transfer of water vapor to the stratosphere. Earlier studies have shown that the extratropical wave drives the entire tropical upwelling with an equatorward propagation of about 10 days (e.g., Ueyama et al., 2013), thus, a delayed cooling was observed in CPT. The reduction in CPT could also be a result of the upwelling's direct influence on heating rates, which was shown by Evan et al. (2015).
Previous studies have only shown the changes in CPT during major SSWs. In contradiction to the decreasing CPT in major warming, CPT increased after minor SSWs. In addition, during minor warming, CPT showed instantaneous cooling whereas in major SSW warming was observed. The water vapor also decreased after the central date in major warming, as the freeze-drying mechanism at the CPT region hinders the passage of water vapor to the lower stratosphere. The increasing height in the CPT could be associated with enhanced transport in the tropical stratosphere. The positive anomaly observed in the polar region agrees with the OMR analysis done by Denton et al. (2019) using ground-based ozonesondes at four sites in the Northern Hemisphere. The downward propagating positive ozone anomaly was evident in the observations. Also, the downwelling from the vortex edges increased OMR at lower stratospheric heights (Callaghan & Salby, 2002). The disruption of the polar vortex would also allow more ozone-rich air from the subtropics to reach polar latitudes. This could be the reason for a reduced magnitude in ozone anomaly for minor warming since such vortex disruption is not observed for minor warming. The polar region showed depleting regions of ozone in the upper stratosphere. This feature is common for major and minor warming. The intrusion of NO x -rich air from the mesosphere that is propagating down into the stratosphere during the warming would lead to more ozone depletion around the central date. The enhanced temperatures would also aid in the chemical destruction of ozone.
To summarize, the present study has attempted to quantify the changes in stratospheric thermal structure and ozone compositions occurring as a result of varying intensities of BDC during major and minor warming in the Northern Hemisphere, simultaneously in the tropical as well as in the polar stratosphere. When the concentration of ozone, which also balances the CPT temperature, is altered by SSW-induced changes in BDC, a dry stratosphere is also observed. In addition, we also separate the effects of minor and major SSW on the intensity of BDC. The changes in stratospheric minor composition have a vital impact on the stratospheric chemistry which finally results in altering the global radiation budget.