Chemical ozone loss in the Arctic winter 1991-1992

Chemical ozone loss in winter 1991–1992 is recalculated based on observations of the HALOE satellite instrument, Version 19, ER-2 aircraft measurements and balloon data. HALOE satellite observations are shown to be reliable in the lower stratosphere below 400 K, at altitudes where the measurements are most likely disturbed by the enhanced sulfate aerosol loading, as a result of the Mt.~Pinatubo eruption in June 1991. Significant chemical ozone loss (13–17 DU) is observed below 380 K from Kiruna balloon observations and HALOE satellite data between December 1991 and March 1992. For the two winters after the Mt. Pinatubo eruption, HALOE satellite observations show a stronger extent of chemical ozone loss towards lower altitudes compared to other Arctic winters between 1991 and 2003. In spite of already occurring deactivation of chlorine in March 1992, MIPAS-B and LPMA balloon observations indicate that chlorine was still activated at lower altitudes, consistent with observed chemical ozone loss occurring between February and March and April. Large chemical ozone loss of more than 70 DU in the Arctic winter 1991–1992 as calculated in earlier studies is corroborated here.


15
The Arctic winter 1991-1992 was a climatological moderately warm winter. For this winter, chemical processes in the polar vortex were strongly influenced by the enhanced burden of sulfate aerosols after the eruption of Mt. Pinatubo in June 1991. Strong chlorine activation and significant chemical ozone loss was observed (e.g., Waters et al., 1993;Toohey et al., 1993;Proffitt et al., 1993;Salawitch et al., 1993;Brandt-20 jen et al., 1994). The amount of chemical ozone loss was quantified in several previous studies (e.g., Proffitt et al., 1993;Rex et al., 1998;Müller et al., 2001). These studies were based on different data sets: ER-2 aircraft measurements from the Airborne Arctic Stratospheric Expedition II (AASE-II) (Anderson et al., 1991;Toohey et al., 1993), data from the European Arctic Stratospheric Ozone Experiment (EASOE) (Pyle et al., EGU Further ozone loss was derived using satellite observations from the Halogen Occultation Experiment (HALOE) aboard the UARS satellite (Russell et al., 1993) and from the UARS microwave limb sounder (MLS) measurements (Tilmes et al., 2004;Manney et al., 2003) These studies consistently found ozone loss of ≈25% in mid-winter (up to the end 5 of January) and large ozone loss rates during January (Salawitch et al., 1993;von der Gathen et al., 1995). Further, model studies for January indicated largest chemical ozone loss in the outer part of the vortex due to a longer solar exposure (Lefèvre et al., 1994). Müller et al. (2001) and Tilmes et al. (2004) derived chemical ozone loss from HALOE satellite observations between 72 and 90 DU in February, March 10 and April, using tracer-tracer correlations. Largest loss in column ozone was found below 450 K. However, especially below 400 K the burden of aerosol particles in this year was strongly enhanced due to the volcanic eruption of Mt Pinatubo in June 1991. Retrieved HALOE O 3 mixing ratios were strongly overestimated in situations of heavy aerosol loading before a correction had been applied. After a correction of the data, the 15 uncertainty of O 3 mixing ratios in the peak aerosol layer is 25% (Hervig et al., 1995). It is well established that enhanced stratospheric sulphate aerosol leads to greater halogen induced chemical ozone destruction (e.g., Cox et al., 1994;Tabazadeh et al., 2002;Rex et al., 2004;Tilmes et al., 2004;WMO, 2007). Model simulations by Tabazadeh et al. (2002) predicted a large increase of chemical ozone loss values after 20 a strong volcanic eruption, especially at lower altitudes (below 17 km). For the Arctic winters 1991Arctic winters -1992Arctic winters and 1992Arctic winters -1993, after the eruption of Mt. Pinatubo, Tilmes et al. (2006) reported strong chemical ozone loss based on an analysis of HALOE satellite observations. These values are outliers from the compact empirical relationship between chemical ozone loss and the PSC (polar stratospheric cloud) formation potential 25 (PFP) 1 for winters between 1991 and 2005, as shown in Fig. 1

EGU
ter. Based on this relationship, the influence of enhanced sulfate aerosols on chemical ozone loss is about 40 DU for altitudes between 380-550 K potential temperature and about 20 DU between 400-500 K potential temperature in winter 1991-1992 (Tilmes et al., 2006). However, up to now it is not proved that the enhanced chemical ozone loss derived in that study is a result of chemical processes, or if observations are 5 wrongly effected by the enhanced sulfate aerosols. Rex et al. (2004) reported the impact of enhanced sulphate aerosol on chemical ozone loss to be smaller based on ozone sonde data than in Tilmes et al. (2006) (about 10 DU between 400-550 K) for this winter. For winter 1992-1993, Rex et al. (2006) do not find a significant impact of enhanced sulphate aerosol on ozone loss.
As described above, the presence of the sulphate aerosol in the stratosphere caused by the eruption of Mt. Pinatubo has also severely affected the remote sensing measurements of the radiometer channels of the HALOE experiment (Hervig et al., 1995).
Here, we will address the question whether the large chemical ozone loss values during spring 1992 based on HALOE observations (Tilmes et al., 2006) are an artifact 15 caused by uncertain observations, or if this much ozone loss can result from chemical processes not included in the PFP value used for the linear relationship in Fig. 1.
The goal of this paper is to derive a reliable value of chemical ozone loss for the winter 1991-1992. For the first time for this winter, we combine all relevant available insitu observation, which are balloon-borne tracer measurements 20 Müller et al., 2001) and high-altitude aircraft ER-2 measurements (Proffitt et al., 1993) to validate the HALOE satellite observations. We will show that HALOE observations in winter 1991-1992 are reliable, especially at lower altitudes. This finding should be also conferable for measurements in winter 1992-1993, a winter that also shows a strongly enhanced aerosol burden. In this way, the impact of sulfate aerosols on chemical ozone 25 loss can be further discussed elsewhere.
Tracer-tracer correlations are used to derive chemical ozone loss as described in Sect. 2. Using this technique transport processes within the polar vortex are accounted for. Compared to other methods, this has the advantage that a possible mis-calculation Chemical ozone loss is a result of chlorine activation in connection with sunlight in the polar vortex. Starting in late December 1991, strong chlorine activation and enhanced ClO were observed until late February in the Arctic lower stratosphere (Waters et al., 1993;Toohey et al., 1993). Enhanced OClO is reported for as late as 11 March (Brandtjen et al., 1994). In Sect. 7, we use balloon measurements in mid-March 1992 15 (von Clarmann et al., 1993;Oelhaf et al., 1994;Wetzel et al., 1995) to further scrutinize the vertical structure of chlorine activation in March 1992 in the polar vortex.

Method
In this study, the tracer-tracer correlation method is used to derive chemical ozone loss in the Arctic polar vortex. The relationship between ozone and a long-lived tracer 20 will not change in absence of heterogeneous chemistry to cause chlorine activation, because of a sufficiently long life-time of the tracers. Further, if the polar vortex is isolated and mixing across the vortex edge can be neglected, changes from the relation between ozone and a long-lived tracer in the early winter (the early winter reference function) can be assumed to be a result of chlorine activation (e.g. Proffitt et al., 1993; the conditions considered here. In the conceptual model used by Plumb et al. (2000) diffusivities for transport across the vortex edge are employed that are very likely too high by more than an order of magnitude (Müller et al., 2005). Moreover, the conceptual model is formulated in terms of artificial, chemically inert tracers χ 1 and χ 2 . Salawitch et al. (2002) have demonstrated that the model results for χ 1 and χ 2 should not be 10 applied to the interpretation of the O 3 /N 2 O relation in the Arctic vortex. In the model used by Plumb et al. (2000), the development of the χ 1 -χ 2 relation is driven primarily by supply of air at the top of the vortex with near zero mixing ratios of both species. However, ozone mixing ratios at the top of the vortex are typically greater than those in the vortex after chemical ozone loss occured. Even though air low in ozone exists in the 15 mesosphere, photochemical model calculations indicate that O 3 is quickly regenerated to mixing ratios of 3-4 ppm by normal gas phase photochemistry as these air parcels descend to lower than ≈40 km (Salawitch et al., 2002). Indeed ozone mixing ratios ranging between 3.6-5.6 ppm have been measured in the mesospheric air-masses that have intruded into the Arctic stratosphere in early (Müller et al., 2007 ). 20 To apply the technique carefully (Tilmes et al., 2004;Müller et al., 2005), the location of profiles have to be discussed with respect to the polar vortex edge to understand if they are influenced by air masses from outside the vortex. Profiles which are located within or outside the boundary region of the polar vortex show different characteristics. A detailed description of the technique and a discussion of the uncertainties due to 25 mixing processes is given in Tilmes et al. (2004) andMüller et al. (2005) EGU vortex edge and the poleward edge of the vortex (vortex core) are calculated as defined by Nash et al. (1996), based on the PV gradient as a function of equivalent latitude. A more precise selection of vortex profiles is performed for aircraft observations with a large horizontal resolution, described in Sect. 5. The area of the vortex outside the vortex core is defined as the vortex boundary region.

Observations
In this study, we use satellite observations taken from the HALOE instrument (Russell et al., 1993). HALOE measured during January, February, March and April 1992 were within the polar vortex core (vortex boundary region for January). HALOE CH 4 and HF are suitable long-lived tracers to apply to the tracer-tracer correlation method 10 as described in Tilmes et al. (2004). The satellite instrument uses gas filter channels to measure these long-lived tracers, different than ozone, which is observed using radiometer channels. The gas filter channels are only weakly affected by aerosol particles and, therefore, a correction of these species was not necessary (Hervig et al., 1995). Ozone mixing ratios have an uncertainty of 25% in the peak aerosol layer (caused by 15 the Mt. Pinatubo eruption) after a correction of the data (Hervig et al., 1995). Additionally, HCl/tracer relations were investigated in Tilmes et al. (2004) to estimate possible chlorine activation in the polar vortex. Balloons were launched in Kiruna, Sweden, during the Arctic winter between December 1991 and March 1992 employing cryogenic (Schmidt et al., 1987) and whole 20 air grab sampling techniques . Ozone observations were taken by standard electrochemical-concentration-cell sondes (Pyle et al., 1994) and N 2 O abundance were measured with balloon-borne whole-air samplers . Further for this study, we use ER-2 aircraft observations of N 2 O and CH 4 taken by the ALIAS instrument (Webster et al., 1994)  EGU balloon and ER-2 observations of ozone mixing ratios are not influenced by enhanced sulfate aerosols in the lower stratosphere and can be used as a comparison to HALOE satellite observations. We compare O 3 and CH 4 profiles for the different measurements as well as the relationship between O 3 /CH 4 . N 2 O balloon and aircraft observations were converted 5 to CH 4 mixing ratios using a CH 4 /N 2 O tracer relation derived using whole air sampler measurements, reported in Tilmes et al. (2006).

Meteorology of the Arctic vortex 1991-1992
The polar vortex in winter 1991-1992 was cold between November and January and 15 disturbed by several warming pulses (Naujokat et al., 1992). Temperatures were below 195 K (favorable conditions for chlorine activation by PSCs) only during January (Newman et al., 1993;Manney et al., 2003). Owing to the enhanced sulfate aerosol densities in the lower stratosphere, the potential of chlorine activation also exists during the first part of February and the first half of March 1992 below 450 K. At the end of 20 January, a major warming resulted in weaker, westerly zonal winds at 60 Naujokat et al., 1992). Transport of air into the vortex was reported by Grooß and Müller (2003

Location of the observations
The characteristic of the distribution of long-lived tracers in the polar region strongly depends on the location of observations. In the polar vortex, air masses descend most significantly at the beginning of the winter. Descent results in larger ozone and lower CH 4 mixing ratios within the polar vortex than outside the vortex. Additionally, chemical 5 ozone loss occurs during winter and spring within the Arctic vortex, if vortex temperatures reach values to allow halogen activation, resulting in a decrease of ozone mixing ratios. In the vortex boundary region airmasses are influenced by air from outside the vortex as a result of isentropic mixing and show different distributions compared to airmasses within the polar vortex core. In order to discuss the reliability of satellite data, it 10 is important to compare similar airmasses. We therefore define the location of various observations, based on whether they were observed within the vortex core, in the vortex boundary region or outside the vortex. In Fig. 2 tinguish observations that are taken within the polar vortex core, the vortex boundary region and outside the vortex. In the following, only satellite profiles in the vortex and the vortex boundary region (denoted as "entire vortex") are considered. For ER-2 observations, only the part of the flight path within the entire polar vortex is considered. We use the potential vorticity P from NMC, interpolated on the flight path to 20 distinguish between measurements that have being taken inside and outside the polar vortex, as described in the following. The modified potential vorticity (Lait, 1994;Müller and Günther, 2003) Π=P · (θ 0 /θ) −ε is employed, where θ is the potential temperature, θ 0 =475 K is a reference potential temperature and the exponent ε may be chosen to adjust the scaling to the prevailing temperature profile (Müller and Günther, 2003). 25 Here, we use ε=4.5, based on the approximately isothermal temperature profiles measured by radiosonde and by the Microwave Temperature Profiler (MTP) aboard the ER-2 in January/February 1992. The modified potential vorticity Π is scaled to θ 0 =475 K.

EGU
The potential vorticity at the edge of the vortex is calculated to be P =29 PVU (potential vorticity units) and at the vortex core to be P =34 PVU during January and February 1992, using MetO data based on the Nash et al. (1996) criterion. Measurements with Π larger 29 PVU are denoted to the entire vortex (dotted lines) and measurements with Π larger than 34 PVU are denoted to the vortex core (solid lines in Fig. 2).

5
Some of the available Kiruna balloon profiles represent the airmass composition at various locations relative to the vortex boundary as described by Bauer et al. (1994) and Schmidt et al. (1994) and discussed below.
In Fig. 2, CH 4 and O 3 mixing ratios are shown for different periods and different observations. In each panel, a to f, two Kiruna balloon data are shown as black and 10 red dashed lines. The profile taken on 5 December 1991, black line, was located well inside the vortex for all altitudes. The profile taken on 12 December 1991, red line, was partly located inside the polar vortex, for altitudes above 475 K, and partly outside but close to the vortex edge for altitudes at and below 475 K (compare Fig. 3  Smaller CH 4 mixing ratios compared to the reference profiles is a result of descent of the polar vortex between December and January. Profiles taken on 22 and 31 January 1992, over Kiruna, are located outside the polar vortex at altitudes below 500 K (see Fig. 3, bottom panels). Above 500 K, the measurements show vortex characteristics with smaller CH 4 mixing ratios compared to the reference profiles. The flight direction 25 of the balloon, especially for the 31st of January, has possibly moved toward the vortex core. Additionally, ER-2 aircraft observations were taken at locations between outside the vortex and deep inside the polar vortex on 20 January (Fig. 3 EGU CH 4 mixing ratios (dotted lines in Fig. 2, panel a). ER-2 CH 4 mixing ratios and their deviation from the reference in the vortex core are in good agreement with the balloon observations over Kiruna on 18 January 1992. One HALOE satellite profile, taken on 14 January within the vortex boundary region (panel a and b, cyan diamonds), does not show lower CH 4 mixing ratios and was possibly influenced by air from outside the 5 vortex. The corresponding O 3 profiles observed in January 1992 indicated minor deviations from the reference profile (see Fig. 2, panel b). ER-2 ozone mixing ratios and the balloon profile on January 18, observed within the vortex core, are slightly larger compared to the reference function due to the descent of vortex air masses. Kiruna balloon 10 data on January 22 and 31 and HALOE observations in January show smaller ozone mixing ratios below 500 K. These profiles were possibly influenced by air masses with smaller ozone mixing ratios from outside the vortex, as also explained by Müller et al. (2001). The smaller ozone mixing ratios on 18 January above 550 K might be the result of the influence of outer vortex air.

15
In Fig. 2 panel c to f, we compare CH 4 and O 3 mixing ratios of HALOE satellite observations taken within the vortex with Kiruna balloon observations (panel c and d) and ER-2 observations (panel e and f) during February to April 1992. All HALOE observations show smaller CH 4 mixing ratios compared to the reference profile (panel c and e) and therefore indicate the descent of airmasses within the polar vortex between 20 December and March/April. One HALOE profile was partly observed within the vortex core (below 500 K) on February 8 and several profiles were observed at the end of March and the beginning of April in the polar vortex, as shown in Fig. 4.
Only the Kiruna balloon profile observed in 12 March 1992 (Fig. 2, panel c, green squares) indicates a similar characteristic as the HALOE observations between March 25 and April between 350 and 420 K, at 475 K and above 550 K. During its flight, the balloon moved towards the edge of the vortex for the other altitudes as described by Bauer et al. (1994) and Schmidt et al. (1994). The balloon profile taken over Kiruna on 5 March was located in the vortex boundary region (Fig. 2,

EGU
does not show significant deviations from the reference profiles. Ozone mixing ratios of balloon and satellite observations are in agreement, see panel d. ER-2 observations are only available within the polar vortex during January and February. All ER-2 observations during March were obtained outside the polar vortex (see Fig. 4). As for January, February observations were taken at locations between 5 the outer vortex and the vortex core. Profiles showing smallest CH 4 mixing ratios and largest O 3 mixing ratios at a certain theta levels were taken deep inside the vortex core. CH 4 profiles taken within the vortex core are in good agreement with HALOE observations. ER-2 ozone mixing ratios (Fig. 2 panel f) are slightly larger in February compared to HALOE observations in March and April for the same potential temperature level. This is in agreement with possible chemical ozone loss that has taken place between February and March/April.
In summary, CH 4 and O 3 mixing ratios of the considered observations agree well if observed within the polar vortex core. Profiles observed outside the polar vortex core show larger CH 4 mixing ratios and smaller O 3 mixing ratios as a result of the influence 15 of outer vortex air. It is shown that HALOE O 3 and CH 4 observations are reliable between 350 and 700 K. Observations are in good agreement with ER-2 observations below 550 K potential temperature and with Kiruna balloon observations taken within the polar vortex core in January and March.

5
The standard deviation of the derived reference function is σ=0.29 ppmv. This value is used to calculate the uncertainty of the loss in column ozone. During January, ER-2 aircraft observations indicate significant deviation from the early winter reference function, (Fig. 5, middle panel). Kiruna balloon observations taken within the vortex core on January 18, as discussed above, agree well with ER-10 2 observations. Kiruna profiles that are influenced by outer vortex air scatter above the reference function. HALOE observations in January show less deviation from the reference function than ER-2 observations due to the location outside the polar vortex core.
During February, ER-2 observations indicate a larger deviation from the early winter 15 reference function compared to January. Therefore, further chemical ozone loss has occurred. Later in March and April, ER-2 aircraft observations are not considered, because they were taken far outside of the polar vortex (see Fig. 4). HALOE satellite observations during February are in agreement with ER-2 data ( EGU balloon data (red squares). In the vortex boundary region, one HALOE profile shows less ozone loss (up to 0.8 ppmv) in January. The observed chemical ozone loss during January is in agreement with the large ozone loss values observed in mid-winter (e.g. von der Gathen et al., 1995). Further, HCl/CH 4 profiles observed by HALOE in the vortex boundary region indicate significant chlorine activation (Tilmes et al., 2004) and 5 strongly enhanced ClO mixing ratios (Waters et al., 1993;Toohey et al., 1993). Between January and February a significant increase of local ozone loss is obvious only for one HALOE profile measured within the polar vortex core. Later in March and April, HALOE observations show local ozone loss values between 1.6 and 2.1 ppmv between 420 and 460 K. Additional ozone loss is possible during March and April due to pos-

Chemical loss in column ozone
Chemical ozone loss in winter 1991-1992 is summarized in Figure 8. All derived ozone loss values have an uncertainty of up to 20 DU, due to the uncertainty of the early 25 winter reference function used (see Figure 5, black dotted lines). Between 400-500 K, chemical ozone loss increases from 41 DU in January (observed by ER-2 within the entire vortex) towards 65 DU in April (observed by HALOE), as shown in Fig. 8 EGU columns. Between 380 and 550 K, ozone loss reached 74 DU in March and 88 DU in April (blue columns). As discussed above, Kiruna profiles are partly located outside the polar vortex core and show less chemical ozone loss, especially between 400 and 500 K. Nevertheless, balloon data indicate significant chemical ozone loss between January and March below 400 K in agreement with HALOE observations in April.

7 Chlorine activation deduced from MIPAS-B and LPMA measurements
Chemical ozone loss during February and March in the lower polar stratosphere is only possible if chlorine is still activated. Information about the two major chlorine reservoir species, ClONO 2 and HCl, is available from the balloon-borne measurements MIPAS-B and LPMA on 14 March 1992. By that time, chlorine deactivation in the Arctic vortex 10 had proceeded where the majority of the active chlorine, ClO x (Cl + ClO + 2 × Cl 2 O 2 ), had been converted to ClONO 2 (e.g., Toohey et al., 1993;Müller et al., 1994). The observed sum of HCl and ClONO 2 , Cl y , can be calculated from the balloonborne measurements, to estimate the abundance of inorganic chlorine Cl y in March 1992. We derived the Cl y /CH 4 relationship, using the concurrently measured methane 15 mixing ratios and compared it to the Cl y /CH 4 relation reported by Grooß et al. (2002) for deactivated conditions. Cl y has been reduced by 12% to account for the lower chlorine loading in 1992 than in 2000 (WMO, 2007). The increase in methane (≈2%) from 1992 to 2000 (Simpson et al., 2002) has only a minor effect but is also taken into account (Fig. 9). The maximum Cl y of about 3 ppb at the lowest CH 4 mixing ra-20 tios is in very good agreement with Cl y deduced independently from whole air sampler measurements for winter 1991. Clearly, for methane mixing ratios greater than ≈0.8 ppm the sum of ClONO 2 and HCl is much lower than the estimated Cl y . This indicates that chlorine is not yet completely deactivated at this time.
5 Figure 11 shows the vertical profile of active chlorine on 14 March found as the difference between the estimated Cl y for deactivated conditions and the observed Cl y (the sum of the measured ClONO 2 and HCl mixing ratios) on 14 March 1992. Maximum active chlorine of ≈1 ppb occurred around 380 K, which could lead to rapid ozone loss due to the long period of sunlight present in March.

Discussion
The comparison of different observations in the winter 1991-1992 results in a consistent picture of chemical ozone loss derived using tracer-tracer correlations. The detailed discussion about the location of profiles in Sect. 5 was performed to distinguish between profiles observed within the vortex core and the vortex boundary region. Pro-15 files located within the polar vortex core show largest chemical ozone loss values, as observed by ER-2 during January and February and by HALOE between February and April. Kiruna balloon observations confirm larger local ozone loss values if the observations were taken within the polar vortex core. Profiles that were observed in the vortex boundary region indicate less chemical ozone loss. 20 During March, Kiruna balloon measurements indicate chemical ozone loss of 24 DU below 380 K. By mid-March 1992, chlorine deactivation had substantially proceeded, dominated by the formation of ClONO 2 . Nevertheless, balloon-borne measurements (Fig. 11) demonstrate that chlorine was still activated in the altitude range between 345 and 400 K potential temperature, with a maximum active chlorine of ≈1 ppb oc-25 curring at ≈380 K. These findings are in agreement with very low HCl mixing ratios observed by HALOE in the polar vortex between January and February below 450 K EGU (Tilmes et al., 2004). The large amount of ozone loss between 350 and 380 K potential temperature was not found using HALOE observations in March, due to the lack of observations at these altitude levels (not shown). Nevertheless, the comparison of averaged ozone loss profiles of Arctic winters between 1991-1992 and 2002-2003, derived using HALOE observations, shows significantly larger ozone loss values at 5 altitudes below 400 K for the years shortly after the Mt. Pinatubo eruption. Between 400 and 500 K, chemical ozone loss values of 59±20 DU are in agreement with values derived by Rex et al. (2004) for altitudes between 400 and 550 K. Therefore, the influence of enhanced sulfate aerosols between 400-550 K described by Rex et al. (2004) is in agreement with the result found here. Kiruna balloon data show less ozone loss, 10 because of their location within the edge of the polar vortex.
The comparison between ER-2, balloon observations that are not influenced by enhanced aerosol loadings and HALOE observations show a good agreement around and below 400 K. This is the altitude where the uncertainty of HALOE was presumed to be largest. Further, column ozone loss between 400-500 K is in agreement for ER-15 2 and HALOE observations. Therefore, we conclude the influence of large aerosol loadings on HALOE measurements is small and ozone loss values calculated between February and April are reliable. Chemical ozone loss in winter 1991-1992 is significantly larger than comparable moderately warm winters.