Nighttime Mesospheric Ozone During the 2002 Southern Hemispheric Major Stratospheric Warming

A Sudden Stratospheric Warming (SSW) affects the chemistry and dynamics of the middle atmosphere. The major warmings occur roughly every second year in the Northern Hemispheric (NH) winter, but has only been observed once in the Southern Hemisphere (SH), during the Antarctic winter of 2002. Using the National Center for Atmospheric Research’s (NCAR) Whole Atmosphere Community Climate Model with specified dynamics (WACCM-SD), this study investigates the ef5 fects of this rare warming event on the ozone layer located around the SH mesopause. This secondary ozone layer changes with respect to hydrogen, oxygen, temperature, and the altered SH polar circulation during the major SSW. The 2002 SH winter was characterized by three zonal-mean zonal wind reductions in the upper stratosphere before a fourth wind reversal reaches the lower stratosphere, marking the onset of the major SSW. At the time of these four wind reversals, a corresponding 10 episodic increase can be seen in the modeled nighttime ozone concentration in the secondary ozone layer. Observations by the Global Ozone Monitoring by Occultation of Stars (GOMOS, an instrument on board the satellite Envisat) demonstrate similar ozone enhancement as in the model. This ozone increase is attributable largely to enhanced upwelling and the associated cooling of the altitude region in conjunction with the wind reversal. Unlike its NH counterpart, the secondary ozone 15 layer during the SH major SSW appeared to be impacted more by the effects of atomic oxygen than hydrogen.


Introduction
In the winter hemisphere, a large temperature difference can be found between the dark polar region and the sunlit extratropics.Combined with the Coriolis effect, this meridional temperature gradient 20 leads to the formation of the eastward Polar Night Jet (PNJ) at high latitudes in the stratosphere.The PNJ acts as a barrier against meridional transport, so the polar region confined by the jet is isolated The reactions rates (c1 − 3) and the air density M are all temperature dependent, with c 1 = 6 • 10 −34 (300/T ) 2.4 , c 2 = 1.4•10 −10 exp(−470/T ), c 3 = 8•10 −12 exp(−2060/T ) cm 3 molecules −1 s −1 , and M = p/kT (where p is pressure and k is Boltzmann's constant) Sander et al. (2006).Because of the short lifetime (on the order of minutes), nighttime ozone cannot be transported.Rather, it rapidly 60 responds to transport of its longer-lived sources and sinks.Daytime ozone is quickly destroyed by sunlight, and is an order of magnitude smaller than in nighttime.Tweedy et al. (2013) showed that a brief enhancement in nighttime ozone occurs in the secondary layer at high latitudes during the Northern Hemispheric major SSWs.This enhancement is mostly 65 driven by cooling and the decrease in atomic hydrogen during the mesospheric ascent.To date, no studies have documented the variation in the secondary ozone during the lone major SSW in the Southern Hemisphere.To this end, the aim of this study is to document and explain, for the first time, a similar ozone enhancement during the Southern Hemispheric SSW that took place in 2002.WACCM is a chemistry-climate model that reaches up to 150 km altitude, and is part of the Community Earth System Model (CESM) Marsh et al. (2013).In this study, WACCM is used with specified dynamics (hereafter referred to as WACCM-SD), such that horizontal winds, temperature and surface 75 pressure are constrained to analyses from NASA Global Modeling and Assimilation Office Modern-Era Retrospective Analysis for Research and Applications (MERRA) Rienecker et al. (2011), by the method described in Kunz et al. (2011).This nudging is applied from the surface to about 50 km altitude (0.79 hPa), and the model is free running above 60 km (0.19 hPa), with a linear transition in between.Simulated from 1990 until 2010, the model has a horizontal resolution of 1.9 o latitude and 80 2.5 o longitude, and a variable vertical resolution with 88 vertical levels.The model output is only once a day (00:00 GMT).We examine in detail the austral winter and spring 2002, and compare it to the climatology over the years 1990-2010 (excluding 2002).
Although we are interested in ozone variability in this paper, hence model biases are not so crit-85 ical, it is worth noting that WACCM tends to underestimate the mesospheric ozone abundance.In comparing ozone from WACCM4 with measurements by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) on the Thermosphere-Ionosphere-Mesosphere Energetics and Dynamics (TIMED) satellite, Smith et al. (2014) showed that ozone volume mixing ratios were 50 percent lower than observed.They pointed out that this difference is likely due to negative 90 bias in atomic oxygen.Too little atomic oxygen is transported down from the MLT region into the mesosphere as a result of weak vertical diffusion and eddy mixing.Bertaux et al. (2010) for an overview of the mission).
Ozone profiles are retrieved from measurements in the UV-visible spectral range, using the stellar occultation method, from the tropopause to the lower thermosphere.The vertical resolution in the mesosphere is 3 km.The retrieval technique is described by Kyrölä et al. (2010) and an analysis of 100 retrieval errors is presented by Tamminen et al. (2010).We use only nighttime measurements in our study.In order to ensure a good quality of the mesospheric ozone profiles, it is important to filter the data based on the star effective temperature.Following European Space Agency's recommendation ESA (2007), we screen out all profiles from cold stars (temperature lower than 6000 K).
105 Finally, our results will be compared to the similar study by Tweedy et al. (2013) of the Northern Hemisphere SSWs.The goal is to determine if the Southern Hemispheric SSW affects the secondary ozone layer in a different way than in the Northern Hemisphere.

Results and Discussion
3.1 The sudden stratospheric warming's effect on the secondary ozone layer 110 The zonal-mean zonal winds at 60 o S in 2002 can be compared to the climatology over the period 1990-2010 (Fig. 1, top row) (Fig. 1, top row).The climatological eastward PNJ stretches from 10 2 hPa to 10 −2 hPa in August, while westward winds prevail above.The eastward winds progressively weaken as spring transitions to summer in September and October, while westward winds descend to the lower mesosphere.In 2002, the zonal-mean zonal wind is reduced in strength three times in the mesosphere 115 (down to 1 hPa), before a fourth stronger wind reversal propagated further down, reaching 10 hPa.This marks the onset of the major SSW (based on the WMO definition), on 25th of September 2002.
In the upper left panel of Fig. 1, all four wind reversals are highlighted with vertical lines.After the onset, the PNJ recovery to eastward direction is only found above 10 −1 hPa, while between this level and 10 1 hPa, westward winds persisted into the summer.We also note that the mesospheric zonal summertime transition around early October.In 2002, regions of ascent can be seen at several pressure levels between 10 −1 hPa to 10 −3 hPa for each of the three first zonal wind reduction episodes.
During the fourth reversal, the ascent is strongest and covers a large vertical range (pressure between 10 0 hPa and above at 10 −2 hPa).The strong descent in the upper mesosphere above is also enhanced 130 after the SSW onset (solid vertical line), as seen by Kvissel et al. (2012).
These strong vertical motions will in turn affect the zonal-mean temperature, as shown in the third row of Fig. 1 for the same latitude band average.Repeated cooling episodes (associated with adiabatic ascent) can be seen below the cold mesopause region at the time of the wind reversals, showing 135 a downward propagation with time.The SSW onset gives rise to the strongest and most persistent cooling, starting at 10 −3 hPa and propagating down to 10 −1 hPa.The strong upper mesospheric descent that begins at the same time above 10 −2 hPa causes a warm anomaly in this region and limits the cooling.

140
The bottom row in Fig. 1 shows zonal-mean ozone volume mixing ratio for the same latitude band.
The secondary ozone layer is situated at the mesospause around 10 −3 hPa (or an altitude of about 90-95 km), as seen in 2002 (left) and the climatology (right).Since the ozone in the mesopause region has a much higher volume mixing ratio at night compared to daytime, the climatological ozone decreases towards spring as the nights get shorter and insolation increases.At the time of the wind 145 reversals and mesospheric coolings in 2002, increases can be seen in the secondary ozone layer.By early October 2002, the upper mesosphere warming associated with the strong descent contributes to lower ozone, since the latter is anti-correlated with temperature.At these austral high latitudes, the volume mixing ratio of the zonal-mean ozone in the secondary layer is around 2 ppmv, while the nighttime values are of the same order of magnitude as in the stratospheric ozone layer.This will be 150 illustrated below as we consider the nighttime sector only.
The zonal wind reversals mark strong ozone departures from the climatology in the mesopause region.Fig. 2 show ozone, temperature, atomic oxygen (the main ozone source, but also sink) and atomic hydrogen (the main ozone sink) at the mesospause in 2002 (solid line).The corresponding In order to better represent spatial variations, polar maps at the mesopause altitude (pressure 10 −3 hPa) of ozone, temperature, atomic oxygen and hydrogen are shown for a few days before the SSW onset Fig. 3, and for the onset day (September 25th) in Fig. 4. Before the SSW, high values of 170 atomic oxygen and hydrogen can be seen in the regions with high temperature.When the temperature decreases, at the SSW onset, so does the atomic oxygen and hydrogen.In the nighttime sector at latitudes 55 o -75 o S, the enhancement in ozone can clearly be seen in conjunction with this low temperature, and the low values of atomic hydrogen (the main ozone sink) and atomic oxygen (the main ozone source, but also sink).Only nighttime ozone is shown, with masked daytime region values.

175
The ozone volume mixing ratio reaches 8 ppmv in the nighttime, but is close to zero in the daytime, explaining why the the zonal mean is around 3 ppmv (see Figure 2).In the next section, the focus will be on just the nighttime sector.Although our focus is on the high latitudes, the polar maps also show ozone enhancements at mid latitudes, associated with cold temperatures and low abundances of atomic hydrogen and oxygen.

Mechanisms behind the nighttime ozone increase
As discussed in the introduction, nighttime ozone in the MLT region is in chemical equilibrium.
Its concentration is only dependent on temperature, on the concentrations of atomic and molecular oxygen, and of atomic hydrogen, as given by Equation 1.We can use the local temperature and densities of hydrogen and oxygen from WACCM-SD to calculate the nighttime ozone, and determine 185 if the model ozone is in chemical equilibrium.Upper panel in Fig. 5 shows nighttime ozone from WACCM-SD (black line), and nighttime ozone calculated according to Equation 1, (gray line), after averaging zonally and over the 55 o S-75 o S latitude band.We see an excellent agreement between this calculated chemical equilibrium nighttime ozone, and the model nighttime ozone.

190
This relation can be used to illuminate which factors in the photochemical equilibrium (temperature, hydrogen, or atomic oxygen) are most important.We let one of them vary at the time, while the other factors are fixed to their mean values over the whole period, following Tweedy et al. (2013).To see the effect of temperature upon nighttime ozone, Equation 1 is used with a time-mean value for hydrogen and atomic oxygen for the period, while only the nighttime temperature is varied.The re-195 sulting nighttime ozone estimate is shown as the blue line in Fig. 5.The same procedure is followed to see the effect from varying hydrogen (green line) and from atomic oxygen (red line).The result is that the brief changes in temperature (coolings) during the episodic wind reversals account for most of the increase in ozone.The oxygen increases have however a counteracting effect: the tem-Several peaks in ozone number density clearly appear in Figure 6.35% and 40% increases, compared to the 2003-2004 average, are observed around August 24 and September 1, respectively.The largest enhancement is observed in the end of September, when the measured ozone number density was approximately 60% higher in 2002 than in the other years.The ozone peak, as observed by GOMOS, varies between 87 and 91 km during this period (not shown).GOMOS was then measuring mostly 225 before sunrise, between 3AM and 6 AM local time.The varying local time sampling throughout the period may cause aliasing of tides, which modulate the height of the secondary ozone layer.After September 22nd, the number of GOMOS measurements is quite small, fluctuating between 2 and 15.This may also contribute to the lack of exact day-to-day correspondence with the WACCM-SD zonal and meridional means of ozone (e.g. on Fig. 2, or on WACCM-SD ozone densities, not shown).

230
To make a direct comparison with the model results, we would need to consider temporal and geographical collocations with GOMOS measurements.However, as explained in the Methodology, WACCM-SD outputs used in this study are available only once a day (00:00 GMT).Furthermore, a direct comparison of WACCM-SD and GOMOS with temporal and geographical collocations, while 235 Atmos.Chem. Phys. Discuss., doi:10.5194/acp-2016-758, 2016 Manuscript under review for journal Atmos.Chem.Phys.Published: 29 September 2016 c Author(s) 2016.CC-BY 3.0 License. in principle possible, would hence also fold tidal signals in the ozone evolution.However, Figure 6 shows that the phenomenon described in our model study has also been observed by a satellite remote sensing instrument: increased ozone densities in the secondary ozone layer have indeed been measured by GOMOS at southern high latitudes during the winter 2002 around the time of the wind reversals.Values were highest during the last week of September, when the middle atmosphere was 240 affected by a major SSW.

Conclusions
In their study of nighttime mesospheric ozone during Northern Hemisphere SSWs with WACCM-SD, Tweedy et al. (2013) concluded that the leading factors contributing to the ozone peak were the decrease in temperature, followed by the decrease in hydrogen.In comparison, our study on the 245 Southern Hemisphere case finds that atomic oxygen plays a larger role than hydrogen.This difference may be attributed to the influence of the seasonal cycle, given that the Southern Hemispheric SSW occurred later in the year than the mid-winter warmings studied in Tweedy et al. (2013).Also, the seasonal cycle in atomic oxygen is more pronounced in the Southern Hemisphere with a more rapid decline in late winter/spring, lowering the production of ozone.

250
This paper focuses on the behavior of the secondary ozone layer during the Southern Hemispheric sudden stratospheric warming (SSW) in 2002.Using the Whole Atmosphere Community Climate Model with specified dynamics (WACCM-SD), we conclude that the polar mesospheric cooling above the stratospheric warming contributes mainly to the short-duration peaks in nighttime ozone.

255
The concurrent decrease of atomic oxygen can reduce the effect of temperature change, while changes in hydrogen play a minor role.Our results agree with studies of the nighttime mesospheric ozone in the northern hemisphere.However, the effect from atomic oxygen is much stronger than what is found for the Northern Hemisphere, where Tweedy et al. (2013) concluded that hydrogen played a much stronger role.

2
Methods, model and instrumentation 70 This study uses the National Center for Atmospheric Research (NCAR) Whole Atmosphere Community Climate Model (WACCM 4) to examine the secondary ozone layer during the major SSW.
We furthermore examine the secondary ozone layer during the Antarctic winter and spring 2002 using observations of the Global Ozone Monitoring by Occultation of Stars (GOMOS).GOMOS 95 is a stellar occultation instrument on board the European platform ENVISAT, launched in March 2002 and in operation until April 2012 (see

120
winds returned to the eastward direction following the three previous reductions.Associated with the PNJ reversal, we expect a change in the vertical transport.The second row of panels in Fig.1shows the the residual vertical velocity, averaged between 55 o S and 75 o S, again for 2002 (left) and for the climatology (right).At altitudes around pressure 1 hPa to 10 −2 hPa, a 125 strong climatological descent is dominant until mid September, followed by a weak ascent at the Atmos.Chem.Phys.Discuss., doi:10.5194/acp-2016-758,2016   Manuscript under review for journal Atmos.Chem.Phys.Published: 29 September 2016 c Author(s) 2016.CC-BY 3.0 License.
155 climatology (dashed line, with the gray lines indicating one standard deviation) is also illustrated.The plots reveal how the mesospheric composition during the repeated strong coolings (and ascents) in 2002 differs from other years.The ozone volume mixing ratio is correspondingly higher than the climatological condition in August and September of 2002, with short-duration peaks at the time of the wind reversals (shown again as vertical lines).On the contrary, temperature shows pronounced 160 decreases at the time of the wind reversals.Atomic oxygen shows weak decreases, within the standard deviation of the climatology.Hydrogen exhibits weak decreases superposed on a background seasonal decline, and its abundance is lower than the climatology through the entire period.TheAtmos.Chem.Phys.Discuss., doi:10.5194/acp-2016-758,2016   Manuscript under review for journal Atmos.Chem.Phys.Published: 29 September 2016 c Author(s) 2016.CC-BY 3.0 License.largest decrease in temperature, atomic oxygen and hydrogen occurs the 26th of September, one day after the SSW onset.Following the decrease, there is a strong increase in temperature, oxygen, and 165 hydrogen at 10 −3 hPa, associated with the strong upper mesospheric descent. 180

260Figure 1 .Figure 2 .
Figure 1.Zonal-mean of zonal wind U (m/s), residual vertical wind W (mm/s), temperature T (K) and ozone O3 (ppmv).Panels to the left are from the year of the sudden stratospheric warming in the southern hemisphere (2002), where the three wind reversals at high altitudes are highlighted as dashed vertical lines, and the wind reversal at 10 hPa is indicated by a solid vertical line.The panels to the right are the climatologies from 1990-2010 for comparison.The plots of residual vertical wind are 10 day moving averages.The zonal wind is at 60 o latitude to capture the polar night jet, while the others are a mean of latitudes 55 o -75 o because this is where the secondary ozone layer has it's peak in mixing ratio.

Figure 6 .
Figure 6.Daily mean of nighttime ozone (molecules/cm 3 ) from the instrument GOMOS onboard Envisat.Measurements are taken at the altitude corresponding to the peak of the secondary ozone layer, averaged over latitudes poleward of 55 o S.