Quantification of transport across the boundary of the lower stratospheric vortex during Arctic winter 2002/2003

Strong perturbations of the Arctic stratosphere during the winter 2002/2003 by planetary waves led to enhanced stretching and folding of the vortex. On two occasions the vortex in the lower stratosphere split into two secondary vortices that re-merged after some days. As a result of these strong disturbances the role of transport in and out of the vortex was stronger than usual. An advection and mixing simulation with the Chemical Lagrangian Model of the Stratosphere (CLaMS) utilising a suite of inert tracers tagging the original position of the air masses has been carried out. The results show a variety of synoptic and small scale features in the vicinity of the vortex boundary, especially long filaments peeling off the vortex edge and being slowly mixed into the mid latitude environment. The vortex folding events, followed by re-merging of different parts of the vortex led to strong filamentation of the vortex interior. During January, February, and March 2003 flights of the Russian high-altitude aircraft Geophysica were performed in order to probe the vortex, filaments and in one case the merging zone between the secondary vortices. Comparisons between CLaMS results and observations obtained from the Geophysica flights show in general good agreement. Several areas affected by both transport and strong mixing could be identified, allowing explanation of many of the structures observed during the flights. Furthermore, the CLaMS simulations allow for a quantification of the air mass exchange between mid latitudes and the vortex interior. The simulation suggests that after the formation of the vortex was completed, its interior remaind relatively undisturbed. Only during the two re-merging events were substantial amounts of extra-vortex air transported into the polar vortex. When in March the vortex starts weakening additional influence from lower latitudes becomes apparent in the model results. In the lower stratosphere export of vortex air leads only to a fraction of about 5% polar air in mid latitudes by the end of March. An upper limit for the contribution of ozone depleted vortex air on mid-latitude ozone loss is derived, indicating that the maximum final impact of dilution is on the order of 50%.

out of the vortex was stronger than usual. An advection and mixing simulation with the Chemical Lagrangian Model of the Stratosphere (CLaMS) utilising a suite of inert tracers tagging the original position of the air masses has been carried out. The results show a variety of synoptic and small scale features in the vicinity of the vortex boundary, especially long filaments peeling off the vortex edge and being slowly mixed into the 10 mid latitude environment. The vortex folding events, followed by re-merging of different parts of the vortex led to strong filamentation of the vortex interior. During January, February, and March 2003 flights of the Russian high-altitude aircraft Geophysica were performed in order to probe the vortex, filaments and in one case the merging zone between the secondary vortices. Comparisons between CLaMS results and observations lived leading to the production of vortex "fossils" that remain noticeable until early June 2000, whereas the dilution of vortex air in mid-latitudes in 2000/2001 was much faster.
On several occasions remnants of vortex air in the Northern Hemisphere midlatitudes have been observed and have been reproduced in model simulations (Orsolini, 2001;Konopka et al., 2003;Manney et al., 2005;Grooß et al., 2007). Model 10 studies have been conducted to quantify this effect (e.g., Knudsen and Grooß, 2000;Hadjinicolaou and Pyle, 2004;WMO, 2007). In the Arctic, there is a large year-to-year variability of the impact of polar processes on mid-latitudes, but contributions of vortex air to mid-latitude ozone loss of up to 25 DU have been reported .
Here we employ simulations with the Chemical Lagrangian Model of the Stratosphere 15 (CLaMS) (McKenna et al., 2002b,a) to investigate transport through mixing in the dynamically very active situation in winter 2002/2003. The permeability of the Arctic polar vortex edge has been studied previously with CLaMS by using artificial tracers that are initialized in the vortex as 100% and zero outside (Steinhorst et al., 2005). Here we extend this concept by considering a whole set of tracers of air mass origin. The mixing 20 scheme in CLaMS allows to follow the evolution of the spectrum of air mass origin for every single air parcel in the model. An advantage of this technique is that averages of air-mass origins over selected regions may be deduced that are based on the information of the mixture of air-mass origin in each single air parcel in the model. In so far, our model concept is similar, but 25 not identical to the concep of air mass origin put forward by Morgenstern et al. (2003).
During winter 2002/2003 the European field campaign "European Polar Stratospheric Cloud and Lee Wave Experiment" (EUPLEX) was carried out as a part of VINTERSOL ("Validation of International Satellites and Study of Ozone Loss"). The ob-Introduction EGU jectives of EUPLEX were to study ozone loss in the Arctic stratosphere in general and the effects of PSCs formed in lee waves in particular. The campaign took place from mid January to mid February 2003 in Kiruna, Sweden, utilising the high flying Russian aircraft Geophysica for stratospheric in-situ measurements. Directly connected to this campaign was an ENVISAT validation phase, carried out between the end of February 5 and mid March with very similar instrumentation. The measurements obtained during both, EUPLEX and the ENVISAT validation, showed indications of strong filamentation in and outside the lowest part of the Arctic polar vortex. To help with the interpretation of the observations, to investigate the influence of the observed small scale structures on the composition of the lower vortex and to estimate the cross boundary flux, a numerical model is used, which will be described in the next section. Information on air mass origin and mixing from observations is provided by the measurements of long lived tracers. During EUPLEX in-situ tracer measurements were performed by the gas chromatograph HAGAR (High Altitude Gas Analyzer, Riediger et al., 2000). Methane (CH 4 ) observations used in this study were made with a time 15 resolution of 90 s and a mean precision of 2%. In-flight calibrations were carried out with two standards linked to within 1% of the NOAA-CMDL scale.
To aid the detection and characterisation of small scale dynamical features that are not resolved with the HAGAR time resolution, in-situ ClO measurements are used in addition. These were made by the HALOX instrument described in detail by von Hobe 20 et al. (2005). It employs the chemical conversion resonance fluorescence technique (Brune et al., 1989) and measures ClO with an accuracy of 16%, a precision of 8% and a time resolution of 10 s under conditions of high ClO caused by chlorine activation in the vortex.
Because ClO is affected by chemical processes on much shorter timescales than 25 the rather inert long lived tracers, it is not suited as a quantitative tracer. However, as long as significant and not too inhomogeneous chlorine activation prevails within the polar vortex and under daylight conditions, the contribution of vortex air to the probed air masses can be identified qualitatively by elevated ClO levels. This was the Introduction CLaMS leads to merging of adjacent air masses or the creation of new air parcels, so that the direct identification of the origin of air masses as in a pure trajectory model is not possible. This problem can be obviated by the use of artificial tracers, that mark particular regions in the atmosphere (e.g. polar or tropical air) at the beginning of the simulation. These artificial inert tracers are then advected and mixed through the course of the simulation by the CLaMS transport scheme. This procedure does not only allow the tracking of air masses as in a trajectory model, but goes beyond as it allows to investigate the contribution of air mass origin in every individual CLaMS air parcel.
Thus, to study and quantify both, advection and mixing, a whole suite of inert tracers are introduced: Five tracers were initialised according to the modified Potential Vorticity value at the original position (see Table 1 for details). Modified PV (mPV), first described by Lait (1994) and later generalised by Günther (2003, 2005) was defined to reduce the vertical dependency of the PV dependent tracers from the exponential increase of PV with height. The mPV values used to distinguish between 20 the different parts of the vortex and mid latitudes or tropics were chosen because they roughly indicate the boundaries between these different compartments of air over the considered altitude range. The tracers referred to in the following as "nert P-Tracers" were initialised with a value of 1, if the conditions according to their original positions were fulfilled, and with a value of 0 elsewhere. When mixing is triggered in CLaMS,

EGU
With the use of these tracers it is now possible to analyse every individual air mass simulated in CLaMS with respect to the mixing it underwent in the course of the simulation. Since the simulation presented here is relatively short (4 months), no timedependent tracers were used. This could be introduced for longer simulations to obtain further information about the life-cycle of the considered air masses.

3 Meteorological overview and evolution of the polar vortex
The northern hemisphere winter of 2002/2003 started with very low temperatures below 190 K in the Arctic lower stratosphere (Fig. 2). During the first half of December, the polar vortex was only slightly disturbed by planetary waves. During the second half of December, in mid January and mid February the perturbation by planetary waves 10 increased strongly (see Fig. 3), leading to a warming of the Arctic lower stratosphere (see Fig. 2). While in December the perturbation was characterised mainly by a zonal wave number one pushing the vortex off the north pole, in January and February the dynamics was dominated additionally by a zonal wave number 2 event. Especially during the strong wave breaking events around the 21 January and the 15 February the 15 polar vortex split into two secondary vortices which re-merged within the next few days (for the splits in January and February see Figs. 4 and 5, respectively). Besides causing vigorous mixing in the vicinity of the vortex edge the repeating splitting, folding and re-merging led to strong filamentation of the interior of the lower stratospheric vortex with mid latitudinal air masses being folded into the vortex. In turn, the mid latitudes 20 were enriched with air originating from inside the vortex.
The wave number 2 forcing vanished almost completely in March 2003, while the wave number 1 forcing remained as strong as in February, leading to continuous distortion of the lower vortex. As a consequence, the vortex was eroded strongly: big parts of it were cut off the main vortex and mixed into their mid latitudinal environment, 25 thus reducing the vortex-covered area of the northern hemisphere (Fig. 6). The transport barrier at the vortex edge weakened significantly, leading to enhanced transport Introduction EGU of extra vortex air into the vortex interior. The evolution of the vortex in the lower stratosphere from mid to end of January is shown in Fig. 4 by means of the modified PV (mPV) on the 440 K level for several days. On 19 January the polar vortex was elongated strongly by the above mentioned perturbations caused by enhanced planetary wave activity. Over northern Canada 5 and Greenland the mPV gradient was strongly eroded, thus providing a region for air masses from the mid latitudes to enter the vortex area. Air stripped off the outer vortex edge can be seen to wrap around an anticyclonic structure positioned over the northern Pacific, indicating vortex influence in an otherwise mid latitudinal environment.
The lower stratospheric part of the vortex (below 500 K) was split into two lobes afterwards (21 January), which re-merged after two days, generating a small longstretched zone in between them. According to the ECMWF operational analyses this re-merging zone was then eroded within a few days. Two Geophysica flights (on 23 and 26 January) were carried out during this episode. Observations obtained on these flights are discussed in the next section. 15 At the end of January another wave breaking event led to a deformation of the vortex with a wave number 3 characteristic, transporting low mPV air deep into the polar regions. The dynamics of the lower stratosphere during February 2003 showed similar features (see Fig. 5). Again the vortex was strongly distorted during the first two weeks of the month, broken up into two secondary vortices on the 17 February, which re-20 merged after a few days. In the beginning of this episode, on the 2 February, a flight was performed when the polar vortex was located over Kiruna. During this flight filaments of extra vortex air inside the vortex were encountered (see next section).
Unfortunately the second vortex split and re-merging event took place in between the EUPLEX campaign and the ENVISAT validation campaign. Therefore no observations 25 are available for this period of time. The first observations available from the ENVISAT validation campaign were made on 28 February.
During the first half of March the vortex was less disturbed than during the preceeding episode (see Fig. 6 EGU number one forcing perturbed the vortex leading to air masses being drawn from the vortex into the mid latitudes, but not being separated completely from the vortex interior. In the framework of the ENVISAT validation campaign five flights were performed during this late stage of the decaying vortex. In the next section we show the observations for the flight on 28 February.

5
All observations made during EUPLEX suggest that the base of the lower stratospheric vortex was eroded rapidly during the episodes of strongest perturbation during January and February 2003. In between these episodes the vortex seemed to recover, extending its base further downward. The interactions between the vortex and the planetary waves resulted in the generation of flow patterns of small scale both inside 10 and outside the vortex. Some of these were observed during the EUPLEX and the ENVISAT validation campaign.

Observations
As already mentioned in the former section, the strong perturbations in mid January led to a vortex split with subsequent re-merging of the lower stratospheric vortex. During 15 this episode two EUPLEX flights were carried out. Figure 7 shows the observed mixing ratios of the long-lived tracer CH 4 and ClO along the flight path of the Geophysika on 23 January, 2003. This flight was designed to probe the re-merging zone of the lower stratospheric vortex. The top panel of this Figure displays the evolution of modified PV along the flight path (in PVU and K, respectively), 20 revealing that during the first part of the flight the Geophysika reached the inner vortex edge region and later the lower part of the vortex core (as defined in Table 1). This is in accordance with the observed values of CH 4 (center panel EGU from the vortex already during ascent. This is not noticeable in the mPV values derived from the meteorological data set. Maximum ClO values were about 500 ppt after about 14:00, for about 30 min when also enhanced mPV values and lower methane mixing ratios were observed. After 14:30 the solar zenith angle along the flight path is greater than 100 • . Therefore the ClO variability is not due to photolysis.

5
The second part of the flight reached the vortex core, where the CH 4 values were as low as 800 ppb. Again the structures observed in the ClO distribution are more pronounced than in the CH 4 distribution. The structures visible in the ClO measurements during the second (return) flight leg resemble those in the mPV, that is ClO decreases as the flight path leads into air masses of lower PV (see also Fig. 4). Consistently, 10 the CH 4 observations show slightly increasing values. However, as for the beginning of the flight, the observations suggest the Geophysika approaching a filament of chlorine activated air, which is not consistent with the comparably low mPV values in the meteorological data and the relatively high (≈1.2-1.3 ppmv) CH 4 mixing ratios.
The observations on the flight on the 26 January are shown in Fig. 8. Most parts of 15 the flight were carried out in vortex core air masses (see top panel of Fig. 8). Apart from some small scale features the observed values of CH 4 are in agreement with the modified PV calculated from the ECMWF data. The slow decrease in mPV with increasing flight altitude around 12:00 indicates an approach to the vortex edge, which is however not apparent in the CH 4 observations. 20 The increase of ClO at the same time is again a combined effect of the vertical gradient of Cl y and the decrease of the solar zenith angle (not shown here) along the flight path enhancing the conversion of Cl 2 O 2 into ClO.
Around 12:15 the Geophysika encounters a filament of chlorine deactivated air with still very low CH 4 around 12:15, again a feature which can not be seen in the meteoro-25 logical data. After having left the vortex air masses (12:40) a very thin filament with a short dip in the CH 4 values and an indication of small chlorine activation is observed.
At the end of January strong perturbations led to a deformation of the vortex with a wave number 3 characteristic. During this episode, on the 2 February, a flight was per- EGU formed when the polar vortex was located over Kiruna. The observations obtained on this day regarding the tracer CH 4 and ClO are shown in Fig. 9. Although the horizontal position of the vortex was perfect for a flight inside the vortex core, the Geophysica only reached the base of the vortex core during the first flight leg on 420 K. Between 09:30 and 10:15 UT pronounced filaments of air from the vortex core and the vortex edge 5 are encountered. Later during the second part of the flight the Geophysica was able to penetrate deeper into the vortex core, where another filament with air of significantly higher tracer values and lower ClO mixing ratios was met. This filament seemed to consist at least partly of air from the vortex edge. The distribution of mPV on the 440 K isentropic level displayed in Fig. 5 suggests that the filaments encountered during this 10 flight were part of a larger structure several hundred kilometers long. This and other occurences of filaments will be investigated in the next section.
In the framework of the ENVISAT validation campaign five flights were performed during the late stage of the decaying vortex. The observations obtained on the flight on 28 February are shown in Fig. 10. Most of the flight was clearly inside the polar 15 vortex core with CH 4 values dropping down to 1000 ppbv and below at the highest flight levels. Moderate chlorine activation can be observed at all times inside the vortex, with ClO values between 100 pptv and more than 400 pptv. While CH 4 values inside the vortex core (defined by the 21 PVU contour) are low in general, strong variablity can be observed in these areas.

Simulations with CLaMS
The most interesting dynamical features of the Arctic winter 2002/2003 besides the minor warming in December were two splits of the vortex in the lower stratosphere, which were accompanied by several large-scale roll-ups of air masses stripped from the vortex edge. Re-merging zones were formed on several occasions between secondary vortices or between the main vortex and large cut-offs merging back, thus creating filamentary structures inside the vortex core. Simultaneously filaments of vortex air were Introduction EGU peeled off the vortex edge and moved into the mid-latitudes. The latter processes have been observed far more often than the creation of filaments by re-merging of parts of a split vortex. All of these features have in common, that their horizontal extension in one direction may be of several hundreds of kilometers, while the scale in the other direction with distances of clearly below hundred kilometres is in the order of the horizontal 5 resolution of the ECMWF grid used. Therefore they can hardly be seen in horizontal PV distributions based on these data like in Figs. 4 and 5. In the following some of these observed features will be investigated with CLaMS. The analyses presented here are all based on the simulation described in Sect. 2.
Since the CLaMS simulation was carried out with a horizontal resolution of about 10 70 km, the re-merging zone, stretching from northern Siberia across Spitsbergen and central Greenland to northern Canada, can be observed much better than in ECMWF data. The results of the simulation for the 23 January for the inert P-tracers again on the 440 K isentropic surface are shown in Fig. 11. The distributions of the tracers P4 and P3, originally initialised in the vortex core and the inner vortex edge respectively 15 (see Table 1), show that the strong mixing inside the vortex has led to an almost homogeneous distribution within the 7 weeks simulated. Only some islands of rather pure vortex core air have survived. Small traces of air from the outer vortex edge (Tracer P2) can be detected inside the vortex, while no air masses from the mid latitudes and tropics have entered the vortex. 20 Overall 66% of the air north of 70 • N equivalent latitude in the layer between 430 K and 450 K is still vortex air. Air masses peeled off the vortex edge and wrapped around the vortex while being transported equatorwards with substantial amounts of vortex air (more than 40% of tracer P4 and P3 within some filaments) can be identified too. These filaments make up as much as 6% of the mid-latitude air outside the vortex in a 25 layer 20 K thick centered on 440 K.
The re-merging zone is clearly visible in the distribution of all tracers. It is almost completely free of vortex core air, with contributions of mainly air from the inner and outer vortex edge (P2) and the mid latitudes (P1). Even some influence of air originat- EGU ing in the tropics (P0) can be observed. In the central area of the re-merging zone over Greenland substantial influence of vortex air is noticeable, while mid latitudinal and tropical air dominate its eastern and western part over Siberia and Canada, respectively. This is because in this central area of the re-merging zone the two lobes came closest and started to re-connect, thus showing the strongest horizontal wind shear, 5 which in turn enhances the mixing intensity.
On several days during this unusual dynamical situation Geophysica flights were carried out, thus giving the opportunity to study some dynamical features of the lower part of the polar vortex more closely.
The center panel of Fig. 7 shows observations of CH 4 made on-board the Geo- The flight on this day was aimed at probing the re-merging zone. As for all the other flights for the EUPLEX and ENVISAT validation campaign, the Geophysica only reached the lowest parts of the polar vortex. These can easily be identified by the 20 areas, where CH 4 (center panel) shows low values, indicating air transported downward by diabatic descent inside the vortex from the middle stratosphere, where the abundances of these tracers are generally lower.
On this flight, the Geophysica encountered air from inside the vortex, i.e. the vortex core and the inner vortex edge, already during ascent and shortly after having reached 25 the 440 K level. This is in good agreement with the CLaMS results, while the next encounter with vortex air around 14:00 seems to be shifted to earlier times in the model results and is not as apparent as in the observations. The mPV derived from ECMWF data shows no indication of vortex air during the first ClO peak. The following encoun- EGU ters with air masses showing low CH 4 values are again in good temporal alignment with the distribution of P-tracers. The same holds true for the increase of outer vortex influences around 16:00 as well as for the filament of chlorine activated air and its fine-scale structure encountered during the final descent. The influence from air originating from the vortex core and the inner vortex edge 5 simulated by CLaMS in a vertical cross-section along the flight path is displayed in Fig. 12. After summing up the amount of tracers P4 (vortex core) and P3 (inner vortex edge) in every air parcel in a curtain with a maximum horizontal distance of 60 km from the flight path, these values are binned in 60 s flight time and 10 K vertical range, thus creating a measure for the vortex influence. The filament of vortex influenced air encountered during ascent and descent, which is very prominent in the observations of ClO, has a vertical extent of about 50 K and consists of up to 80% vortex air. This feature can not be seen in the the mPV distribution derived from ECMWF data.
In general, there is good agreement between observation and simulation regarding temporal alignment and vortex influence. Exceptions may be due to the fact, that (i) 15 the CLaMS results shown in the lower panel of Fig. 7 are derived from an interpolation onto the flight path by the averaging processes described, and that (ii) the observations are at a comparatively higher resolution, vertically and horizontally, than the CLaMS simulation. Nevertheless, the strong filamentation observed can be reproduced by the model. The amount of original vortex air inside the vortex has decreased from 66% 20 on 19 January to 51% in the 440 K layer, showing that strong mixing has been going on during the previous four days. These values are calculated by integrating the tracer distributions of P3 (inner vortex edge) and P4 (vortex core) inside the 18 mPV units contour line (see Fig. 11).
The CLaMS results for the next flight on the 26 January (see Fig. 8 EGU filament with higher amount of extra vortex air observed around 10:00 is not as sharp in the model results as in the observations. The increase of extra vortex air later on the second flight leg (shortly after 12:00) and the small filament of vortex air met shortly afterwards agree very well with the observations. Even the very fine filament of chlorine activated air encountered during descent is represented in the model results as a 5 sharp increase in influence from vortex air. The averaged distribution of the tracers tagging air from the vortex core and the inner vortex edge in a curtain along the flight track displayed in Fig. 13 again gives a better overview about the vertical structure of the simulated vortex. The base of the vortex is characterized by both, erosion and filamentation. The small filaments encountered 10 during ascent and descent (both visible in the observations, but not in the mPV derived from the meteorological data) are prominent in the model results. There are additional shallow structures of air dominated by vortex influence only apparent rudimentally in the observations. Although their vertical extent is only about 10 K, they consist of 40 to 50% air originally located inside the vortex boundary. 15 The flight on 2 February (see Fig. 9) is especially interesting, since during this flight it was possible to meet an aged re-merging filament created during a vortex re-merging process twice (around 09:30 and 12:00). The history of the filament can be tracked back for about 4 days. The eastern part of the filament, which is probed first, is located more to the edge of the vortex, while the western is already drawn more into the re-20 merging vortex. Especially the second encounter with the old filament is clearly visible in both model results and observations. During the first part of the flight, the encounter is smeared out in the model results and also not visible in the ClO measurements. This may be due to the fact that the first part of the flight was on a lower level and in an area where the erosion of the vortex and the filaments seems to be stronger, while the 25 second part of the flight took place 30 K higher up and deeper inside the newly reunited vortex. According to CLaMS results, the vortex at the 440 K layer at this time consisted of about 60% air of original vortex origin, meaning that its interior had been filled up with vortex core air moving downward due to radiative cooling. Introduction  EGU time, and in March, when the final warming and the vortex breakdown is about to start, influence from the outer vortex edge, the mid latitudes and subtropics is increasing to values of about 35%. In Fig. 15 the latitudinal variation of the average of the simulated air mass origin for lower stratospheric air between 400 K and 850 K is shown. The air mass origin 5 is initialized on 1 December 2002 according to 1. On 19 January 2003 (top panel), after about six weeks of simulation and right before the first vortex split, the air in the polar vortex remains largely isolated from mid-latitude air. Only 8% of the air masses north of 70 • equivalent latitude in the altitude range between 400 K and 850 K are of extra vortex origin. Indeed, the gradient between air masses of different origin at the 10 polar vortex edge is very strong as it is apparent as a very narrow range of equivalent latitudes where the air shows a mixed composition between mid-latitude and vortex air. A small fraction of vortex air (≈6% between equivalent latitudes of 30 • N and 60 • N) has been transported out of the vortex. Further noticeable is that tropical and mid-latitude air masses are mixed to some extent. 15 The situation on 11 February 2003 (the last flight day of the EUPLEX campaign) is remarkably similar to mid January (Fig. 15, center panel). The gradient at the vortex edge has steepened compared to mid January. There is almost the same fraction of polar air at mid-latitudes (≈6% between 30 • N and 60 • N equivalent latitudes) than in mid-January and there is a slighly lower fraction of pure vortex core air within the vortex 20 (60%) that now also contains about 12% of air masses of extra vortex air, some even of tropical origin.
Even by the end of March (see Fig. 15, bottom panel) the transport barrier at the edge of the vortex is clearly noticeable and the vortex contains about 64 % pure vortex air (indicated by the blue colour). Transport of vortex air to mid-latitudes has increased 25 compared to the situation in January and February (≈7% equivalent mid-latitudes). The influence of vortex air is strongest at high equivalent latitudes, that is close to the polar vortex, and decreases equatorwards.
For a precise estimate of the impact of the transport of polar air out of the vortex on EGU mid-latitude ozone, simulations including the transport and chemistry of ozone would be required. However, assuming that the air in the polar vortex at the beginning of the winter has ozone mixing ratios about as large as the mid-latitude air and assuming that ozone has been completely chemically destroyed in the polar vortex during winter, one would estimate that the midlatitude ozone levels by the end of March have been 5 reduced by about 7% due to the export of polar vortex air. In reality, ozone in the Arctic vortex is not destroyed completely, so that this value constitutes an upper limit.
The results indicate that as long as a polar voxtex exists, the impact of ozone loss on mid-latitude ozone trends will be small. This impact should be even smaller in Arctic winters with a stronger vortex than in 2002/2003 or in the Antarctic. In the mid-latitudes the air masses which have been drawn from the vortex core and edge by the above described dynamical situations are slowly mixed into the background air. The model calculations suggest this transport from the polar vortex into the midlatitudes to result in a fraction of the order of 7% for the lower stratosphere until the end of March. This can serve as estimate for the maximum impact of Arctic ozone depletion