Transport of methane associated with the polar vortex and its breakdown was observed by the data from the Improved Limb Atmospheric Spectrometer (ILAS) during November and December of 1996. The zonal mean mixing ratio in the middle stratosphere at high latitudes in the Southern Hemisphere increased rapidly during the period, and the vertical gradient of methane sometimes showed an unusual reversal. The reversal of the vertical gradient confirms that the erosion of the vortex and the subsequent breakdown occurs in the upper stratosphere first and descends with time. The variation of the tracer mixing ratio was investigated with respect to the boundary of the polar vortex. The mixing ratios inside and outside of the polar vortex evolved with distinctive patterns with time. The small difference in methane between inside and outside of the vortex in the upper stratosphere is presumably a result of the vortex erosion prior to November. Also, the small difference between inside and outside of the vortex in the lower stratosphere confirms the weak barrier effect of the polar vortex on that altitude. High concentrations of methane observed inside the vortex show mixing of lower-latitude air into the vortex. An RDF calculation confirms the behavior of methane transport into the vortex.
 Since the discovery of the ozone hole in the mid-1980s [Farman et al., 1985] various instruments onboard satellites, airplanes and balloons have been used to observe the Antarctic polar stratosphere. These observations have helped scientists to understand the mechanisms of the formation of the ozone hole [Solomon, 1988], and more generally various features about this polar region. ILAS onboard the Advanced Earth Observing Satellite (ADEOS) is another instrument that observed the stratospheric chemical species. The characteristic features and performance of the ILAS are described in detail by Nakajima et al. . ILAS observed mainly the high latitude regions in both hemispheres and hence provided very useful data for investigation of tracer variations in the polar regions.
 In section 2, the ILAS data are described briefly. In section 3, we show that the temporal variation of methane concentration was dependent on mixing of inside and outside air masses and the breakdown of the polar vortex. In section 4, we investigate fluctuations of zonal mean methane mixing ratio and show that the phenomenon is associated with the shape of the vortex, or the wave number pattern of the planetary wave. In section 5, we consider transport across the vortex edge. By looking at the region near the specified edge, it is possible to show an increase in methane as well as potential vorticity inside the vortex near the vortex edge. In the final section, observed features are summarized and discussed.
2. ILAS Data
 ILAS is an instrument onboard the ADEOS which was launched by the NASDA of Japan on 17 August 1996. It used the solar occultation method and observed only the high latitude regions. During the 8 months of operation from November 1996 until June 1997, ILAS observations covered the Northern (58°N–73°N) and Southern (65°S–88°S) Hemisphere high latitudes. During any day, observations took place up to 14 times following the latitude circle. Thus the best horizontal resolution is about 25° in longitude. ILAS observed temperature, pressure, and molecular species such as O3, HNO3, NO2, N2O, CH4, H2O, and aerosol extinction coefficients [Yokota et al., 1998]. The effective vertical resolutions are 1.9 km at 15 km, 2.6 km at 25 km, 3.2 km at 35 km, 3.5 km at 45 km, and 3.7 km at 55 km and above due to the smoothing numerical filter, the size of the instantaneous field of view and the atmospheric refraction [Hayashida et al., 2000; Yokota et al., 2002]. Validation of ILAS data and results of intercomparisons with other data were reported by Lee et al.  and Sasano et al. [1999b]. Those validation and intercomparisons showed that the ILAS data include errors whose ranges are acceptable. In this study, Version 3.47 methane concentrations are used. The methane concentrations from Version 3.47 are generally similar to those from Version 5.2. A validation of Version 5.2 is reported by Kanzawa et al. . The mixing ratios of methane are interpolated onto the 32 levels of isentropic surfaces from 400 K to 1419 K by the interval of about 0.8 km in altitude.
3. Influence of the Vortex Breakdown on the Methane Distribution
 The time-altitude cross-section of zonally averaged mixing ratio following the ILAS sunset observation is shown in Figure 1a during November and December 1996. During the time period the latitude of ILAS observation moved from 74.6°S on 1 November to 64.0°S on 31 December 1996 [e.g., Sasano et al., 1999a]. Figure 1a is the time series that represents partly atmospheric change and partly latitude progression of the ILAS instrument. Since the observation latitude changes with time, interpretation of Figure 1 should be done carefully. During the vortex breakdown, the timescale of change in zonal mean methane mixing ratio seems to be about a week. Over that timescale, the dominant planetary wave pattern as well as the vortex shape changes between wave number 1 and 2. The zonal mean methane mixing ratio also changes in the same time period, which is clearly seen between the 22- and 32-km levels. In a week the ILAS latitude changes by only 1–2 degrees, which seems to be small enough for the ILAS latitude being considered as fixed. Thus change in methane concentration up to a week seems to be due to change of the vortex shape rather than due to change of the observation latitude. As long as the observation latitude can be assumed to be constant, the zonal mean time series is a useful method to understand the temporal change of methane in the global scale. The mixing ratio of methane increases rapidly in the altitude range of 22–32 km level. However, below 20 and above 35 km, changes are relatively small. Since this time of year corresponds to the transition period from the winter to summer circulation pattern, which is associated with breakdown of the polar vortex, the variation of methane with time evidently shows the sign of the transition. Thus the characteristic features shown in methane variation must be associated with the vortex and its breakdown. Erosion of vortex can start as early as September [Tuck, 1989] and continues through October [Russell et al., 1993; Bithell et al., 1994; Rosenlof et al., 1997]. The features seen in the early November in Figure 1 therefore are considered the state of methane concentration after considerable mixing of air inside and outside of the vortex.
 One of the distinctive features in increasing mixing ratio is its fluctuation in time, and its timescale is about 7 to 8 days. This fluctuation seems to be associated with the variations in planetary wave activities. The vortex shape changes following the dominant planetary wave pattern, and each observation point subsequently moves in and out of the vortex. This movement can be observed in Figure 2. Harwood , Hartmann , Mechoso et al. , and Shiotani et al.  found the existence of the quasi-stationary wave number 1 and traveling wave number 2 planetary waves in the Antarctic winter and spring. Manney et al.  found that MLS O3 mixing ratio is affected by the planetary wave activities in winter, and probably in spring too. Although the ILAS data are not available in the earlier months, we suspect that methane concentration was increasing prior to November based on the findings by previous studies [e.g., Godson, 1963; Russell et al., 1993]. The second feature is an occasional occurrence of a larger mixing ratio at the higher altitude. This fact is better observed in the vertical gradient of mixing ratio in Figure 1b. Normally the vertical gradient of methane is negative because of its vertical stratification since the source of methane is the troposphere. Therefore the appearance of the vertical-gradient reversal shown in Figure 1b is unusual. The vertical gradient reversal seems to be associated with transport of methane-rich middle-latitude air into the high latitude region at the higher altitude. A similar mechanism was shown to be operative by Murphy et al. . At a given location weakening and the subsequent breakdown of the vortex occurs earlier at the higher altitude region. This feature was also shown from the shape of the polar barrier [Nakamura and Ma, 1997]. The polar barrier that they obtained by calculating equivalent length for early November 1992 is close to the pole in the upper stratosphere, and thus is characterized by a distinctive “tipped-over” orientation. Thus more effective transport of middle latitude air into the high-latitude upper region than at the lower region gives rise to the reversal of the gradient for the short period of time. The regions of gradient reversal descend with time, and this fact agrees with descent of polar vortex breakdown although its descending speed (see Figure 3) is not identical to the speed of the descending gradient reversal. Nakamura and Ma  reported that the polar barrier moves downward and poleward as winter becomes spring, and this is consistent with descent of polar vortex breakdown.
 The most prominent meteorological feature in the spring hemisphere stratosphere is the existence of the polar vortex and its breakdown. The polar vortex can be well described in the isentropic distributions of Ertel's potential vorticity (EPV). The EPV is sometimes scaled for an easy interpretation of the vertical distribution because of its exponential increase with height [Dunkerton and Delisi, 1986; Lait, 1994; Manney et al., 1994, 1996]. In this study, a modified potential vorticity (MPV) is used following Lait . The edge of the polar vortex is must be determined carefully. Dynamically, the edge is usually defined as the region of the maximum meridional gradient of MPV or the region of strong zonal wind. Nash et al.  defined the vortex edge as the equivalent latitude [Butchart and Remsberg, 1986] of the maximum gradient in EPV constrained by the proximity of a reasonably strong westerly jet. Objectively, the equivalent latitude of the vortex edge can be obtained by the maximum of average wind multiplied by MPV gradient [Nash et al., 1996]. We compared the edges obtained by the maximum wind, the maximum MPV gradient, and by the method of Nash et al. . The vortex edges defined in three different methods were in good agreement, and the third method is used in this study. The chemically defined vortex edge [Tao and Tuck, 1994] is hard to employ in this study since the ILAS observations are separated by farther than 25° in longitude following the latitude circle.
 The distributions in MPV on the 738 K isentropic surface (about 26–27 km level) on 4 days in December are shown in Figure 2. The MPV distribution was calculated by using pressure and temperature from the United Kingdom Meteorological Office (UKMO) data [Swinbank and O'Neill, 1994]. The solid dots represent the locations of ILAS observation on each day and the thick solid line is the vortex edge. The unit of MPV is the potential vorticity unit (PVU) and 1 PVU is 10−6 K m2 kg−1 s−1. The vortex edge plays a role of transport barrier, and the methane concentration can be much different outside of the vortex from inside. The polar vortex disappears as the season changes from winter to summer. The polar vortex in Figures 2a–2b becomes highly elongated in Figure 2d. Zonal mean mixing ratio in Figure 2d at the ILAS latitude should represent methane-rich middle-latitude air since most of the ILAS observation occurs outside of the vortex on that day. This methane increase by outside air is also shown in Figure 1a.
 The breakdown of polar vortex is easily observed from distributions of potential vorticity on isentropic surfaces. Waugh et al.  examined several methods to determine the date of vortex breakdown and obtained similar results by all the methods. In this study, following the definition by Nash et al. , we define the date of vortex breakdown as the date when the maximum wind speed averaged along the MPV isolines falls below 15.2 m s−1. The choice of the value of 15.2 m s−1 is somewhat arbitrary, but it seems to be not unreasonable as discussed for Figure 3. The maximum average wind was obtained from the UKMO data. The winds data were interpolated onto levels with an 0.8-km interval, and the daily wind maximum was averaged by a 3-day running mean to reduce the noise appeared in the day-to-day variation. Based on the above definition the polar vortex on the 738 K level is “present” until 5 December. On 6 December the vortex is no longer “present”, and the date is defined here as the day of vortex breakdown. After the breakdown the vortex shrinks rapidly but its remnant remains for a considerable time.
 The day of vortex breakdown on each level is shown by white dots in Figure 3, which is observed later at the lower altitudes. This descending breakdown of the polar vortex seems to be the reason of the descending region of the vertical gradient reversal. Descent of the vortex breakdown with time was shown in terms of the movement of the polar barrier by Nakamura and Ma . The variation of mixing ratio inside and outside of the vortex can be estimated by calculating the standard deviation. Figure 3 shows the standard deviation of the methane mixing ratio normalized by the zonal mean. The normalized standard deviation does not change much at the 20-km level, but above the 25-km level it shows decrease with fluctuation of the period of several days. The smaller standard deviation implies that mixing ratio difference around a latitude circle is small and thus concentration around the latitude circle is relatively uniform. The dates of vortex breakdown in Figure 3 are located between the regions of large and small standard deviation, which shows choice of 15.2 m s−1 for the vortex-breakdown criterion is acceptable. After the breakdown of the vortex, the normalized standard deviation is generally less than 0.3. There is an exception at 22 km in mid-December. The isentropic distributions of MPV in that period (not shown) show regions of a remnant of high MPV, in which very small concentrations of methane compared to outside of the remnant region are also found.
 The temporal variation of zonal mean mixing ratio can be better understood by separating those observation points by inside and outside of the polar vortex [e. g., Kent et al., 1985; Russell et al., 1993; Bevilacqua et al., 1995, 1997; Abrams et al., 1996a, 1996b]. In this study, the location of each ILAS observation before vortex breakdown is classified as inside, outside or in the boundary region of the vortex based on the definitions of the equatorward and poleward boundaries. The equatorward and poleward boundaries of the vortex are determined as the equivalent latitudes at which the values of MPV values show the maximum convex and concave curvatures [Nash et al., 1996]. The inside region is defined as poleward of the poleward boundary, and the outside region is defined as equatorward of the equatorward boundary. The boundary region is defined as between the two boundaries. Although separation of the observation points by the dynamically defined vortex edge is a common practice, it is not a perfect method to distinguish two regions of much different methane concentrations. The analyzed MPV values can be inaccurate due to errors in meteorological variables [Fairlie et al., 1997]. Also, the dynamically defined vortex edge does not coincide with the chemically defined vortex edge [Tao and Tuck, 1994].
 The difference between inside and outside the vortex can be observed in Figure 4. The vertical profiles of methane are averaged each for inside and outside of the vortex and for 5 days from 6 November until the end of November, and they are shown up to the height of polar-vortex breakdown. The inside and outside profiles are distinctively different. Differences between those two profiles are the largest at about 25 km, but small in the lower and upper stratosphere. Erosion of vortex prior to November in the stratosphere [Godson, 1963; Russell et al., 1993; Nakamura and Ma, 1997] occurs in the upper stratosphere earlier than in the lower stratosphere. Mixing by this erosion seems to have made small differences between the inside and outside profiles in the upper stratosphere. In the lower stratosphere, air inside the vortex is much less isolated than in the middle stratosphere [Bithell et al., 1994; Chen et al., 1994; Rosenlof et al., 1997], and this seems to be the reason of small differences between the inside and outside profiles. If the vortex breaks down at the altitude of strong contrast in mixing ratio, mixing of both air masses would change the zonal mean mixing ratio rapidly. However, mixing of both air masses after vortex breakdown at the 35-km level would not change the zonal mean mixing ratio significantly. Thus the altitude-dependent change of zonal mean mixing ratio in Figure 1a can be understood both by the effectiveness of mixing of two air masses across the polar vortex and by the contrast in mixing ratios in those two air masses.
 The outside methane profiles do not change much in all of November, but inside profiles show significant increase above the 22-km level. The local maximum at the 31-km level for the 6–10 November period progresses downward with time. Although there is mixing within the vortex [Pierce et al., 1994], the change inside the vortex is due to middle latitude air transport into the vortex. The increase inside the vortex from the upper to lower region is consistent with descent of vortex breakdown shown in Figure 3.
4. Fluctuation of Zonal Mean Mixing Ratio With Vortex Shape
 The increase in methane mixing ratio is not gradual but, as shown in Figure 1, fluctuates with a period of about a week. This increase is also shown in Figure 4. The increase inside the vortex occurs from the first to the second 5-day periods and from the 3rd to the 4th 5-day periods, but changes from the second to third and from the 4th to 5th periods are relatively small. To understand this difference in sporadic increase, distributions of potential vorticity on the 738 K isentropic surface for 2 days, which represent the third and 4th periods, are shown in Figure 5. The latitudes of observation on those 2 days are similar and their observation points are relatively well distributed in the latitude circle. However, the methane concentration inside the vortex as well as its zonal mean increased significantly from 17 to 22 November. The shape of the polar vortex and the MPV distribution are also different. The shape of the vortex on the 17th and the 22nd are wave number 1 and 2 type, respectively. The number of observation points inside the vortex and their relative positions with respect to the vortex edge are dependent on the evolution of the vortex shape. Those two aspects influence zonal mean concentration most strongly.
 The differences in methane concentrations inside the vortex on 17 and 22 November also appear in Figure 6. As shown by Schoeberl et al. , the relationship between MPV distribution and methane mixing ratio is easily seen in Figure 6, which is a scatter diagram on 738 K surface showing methane concentration and MPV from 6 November until the date of the vortex breakdown (5 December by our definition). The relationship between MPV and methane is very close and its correlation on each day is higher than 0.8. Outside values are generally uniform and higher than those at inside the vortex and the boundary region. However, in some cases as shown in Figure 6b, mixing ratios inside the vortex can be as large as those at outside. The values inside the vortex are fairly different from each other and dependent on MPV on the point. This implies that the individual mixing ratio is dependent on the location relative to the vortex edge. Among the observation points denoted by open circles in Figure 6b, four points inside the vortex with unusually high values happen to be close to the vortex edge (solid triangles in Figure 5b) and thus could be influenced by the methane-rich air outside the vortex. This kind of transport across the vortex edge has been previously observed and discussed in detail [Russell et al., 1993; Bithell et al., 1994; Harries et al., 1995; Rosenlof et al., 1997]. The value of the MPV at the vortex edge is about −10 PVU on both days. If we defined the chemically defined vortex edge based on the chemical gradient [Tao and Tuck, 1994] from Figure 6, the MPV value at the chemical edge would be close to −11 PVU on the 17th and −12 on the 22nd. Thus the region inside the chemical edge on both days is smaller than the region inside the dynamical edge.
5. Transport Into the Vortex
 The increase in zonal mean mixing ratio is associated with the shape of the vortex and number of observation points. The vortex edge is not a perfect barrier and material can be occasionally mixed into the vortex interior and also can be transported out as the vortex erodes [Atkinson et al., 1989; Russell et al., 1993; Bithell et al., 1994; Plumb et al., 1994; Tao and Tuck, 1994; Harries et al., 1995; Atkinson and Plumb, 1997; Rosenlof et al., 1997; Tuck and Proffitt, 1997]. We have defined the date of vortex breakdown for convenience; however, the vortex weakens before that time. The evolution of profiles inside the vortex in Figure 4 implies transport across the vortex edge before breakdown of the vortex. To understand transport across the vortex, the longitude-time section of methane concentration on the 738 K isentropic surface is shown in Figure 7. The locations of the vortex edges are also shown. There are usually two edges, but four edges appear when the vortex is distorted, which are on 15, 16, 23, 24, 25, and 29 November and 4 December. For the future purpose, we have defined the vortex edges in the Western and Eastern hemispheres as the “Western edge” and the “Eastern edge,” respectively. During November and December, the methane concentration increases significantly. This variation in the 738 K surface shows a distinctive difference before and after the vortex breakdown, which is 6 December on this level. The distribution is quite uniform after breakdown. One of the distinctive features is eastward migration of constant mixing ratio and the “Eastern edge.” This propagation is associated with phase propagation of wave number 2 pattern.
 Methane transport across the vortex edge in Figure 7 is difficult to see because the position of the vortex edge changes with time. In Figure 8 we therefore concentrate on the “Western edge” on 738 K level. The “Western edge” moves fast sometimes, which was shown in Figure 7, but does not move at other times. Figure 8b shows the concentration only near the “Western edge.” The zero-degree longitude in Figure 8b corresponds to the “Western edge.” In this figure mixing ratio distribution near the vortex edge clearly shows continuous transport across the edge and into the inside of the vortex region. On the 738 K isentropic surface, the vortex had broken down on 6 December by our definition. As the contour line of 1.2 ppmv shows, there is a significant increase of methane just before the vortex breakdown.
 Because of the low resolution of the ILAS observation, Figure 8 does not provide the fine scale behavior of tracer movement. For the purpose of understanding the fine scale transport, observing potential vorticity is more useful. Figure 9 shows a 738 K 7-day reverse domain fill (RDF) calculation of the potential vorticity field and the actual analyzed field for 24 November 1996. The RDF calculation is performed by running the Goddard trajectory model backward 7 days starting with regularly spaced points on a 0.5° longitude and 0.5° latitude grid. The PV values at the end of the trajectory calculation (7 days earlier than the date shown in the figure) are then copied forward to the regular grid points. This technique developed by Sutton et al. , Newman and Schoeberl , and Schoeberl and Newman  produces an equivalent high-resolution picture of the potential vorticity field. Filaments such as those generated by the RDF technique shown in Figure 9a have been observed by aircraft [Plumb et al., 1994]. The Lagrangian forecast by RDF was discussed in detail by Fairlie et al. .
Figure 9 shows the mechanism by which the vortex erodes. A long filament has been pulled off the vortex stretches across West Antarctica and bends toward Africa. This is the usual vortex erosion process seen in both hemispheres. The figure also shows a small zone of vortex inward intrusion of air near the South Pole which suggests the mechanism for the increase of methane near the vortex edge. The UKMO analyzed field in Figure 9b shows that the fine scale features indicated in the RDF are only hinted at with the lower resolution of the UKMO data.
 Using the RDF and analyzed fields at 738 K, the time series of the “Western edge” MPV field, equivalent to Figure 8b, is shown in Figure 10. Both the analyzed and RDF PV time series show the same structure as that revealed in Figure 8b, and Figure 10 shows the clear correlation between low MPV values within the vortex and the low methane. The strength of the gradient in PV and methane decreases rapidly after 25 November. An examination of the RDF fields show that the number of filaments increases after that date as well as the number of inward mixing vortex intrusions from midlatitudes. This explains the rapid decrease in vortex integrity after 25 November.
6. Summary and Discussion
 In this study, methane observed by ILAS during November and December 1996 was investigated associated with change of the polar vortex in the Southern Hemisphere. The vortex broke down at the upper stratosphere first and proceeded to the lower stratosphere, and the remnants of the vortex existed for considerable time. As the season changes in this period from spring to summer, the temperature goes through a transition period and finally reaches at the summer pattern in late December. The distributions of stratospheric methane also change rapidly, and these variations are confined to the altitude range of 25–35 km. The primary cause of such variations is the vortex breakdown. The temporal fluctuations in the tracer mixing ratios are associated with the rapid change of zonal mean mixing ratio, and those are assumed to be effects of periodic activities of the planetary waves. The influence of periodic planetary waves can be confirmed in the daily distributions of quasi-conservative tracers such as nitrous oxide and water vapor and even in the distribution of ozone. The mixing ratio of methane increased rapidly between the 22- and 32-km levels with fluctuations of periods of several days. The increase is dependent on the evolution of the polar vortex before breakdown of vortex. After the vortex breakdown, the distribution following the latitude circle is fairly uniform as indicated by the standard deviation. The normalized standard deviation shows good agreement with the altitude of vortex breakdown. The date of vortex breakdown is defined by a condition, which is that the maximum zonal wind following MPV isolines is below 15.2 m s−1. However, separation between the high and low standard deviation in Figure 3 implies that there might be a better definition of breakdown of vortex. The interior of the vortex is well isolated from the outside until the vortex breakdown occurs, but the mixing ratio inside the vortex near the boundary is considerably influenced by the mixing ratio of the outside or boundary of the vortex. These distinctive differences between near the vortex center and near the edge have been reported [e. g., Russell et al., 1993; Bithell et al., 1994]. To deduce the fine scale motion of methane transport, potential vorticity distribution was analyzed by the RDF method. Observation of the potential vorticity transport into the polar vortex confirms the behavior of methane transport into the vortex.
 We are grateful to the ILAS Project team for providing the ILAS data. Helpful discussions with William J. Randel at NCAR and Paul A. Newman, Lynn C. Sparling, and Eric R. Nash at NASA/GSFC are greatly appreciated. Mijeong Park, Dong Joon Kim, and Daeok Youn helped draw the figures. The comments and suggestions of two anonymous referees are greatly appreciated. The ILAS data were processed at the ILAS Data Handling Facility, National Institute for Environmental Studies (NIES). This study was supported by the Climate Environment System Research Center sponsored by the SRC program of the Korean Science and Engineering Foundation, and by the BK21 program of the Ministry of Education of Korea.