We report Rayleigh lidar measurements of nightly temperature profiles in the 40–80 km altitude region and 30 min relative density profiles in the 40–50 km altitude region at Chatanika, Alaska (65°N, 147°W) in December, January, February, and March over two winters (2007–2008, 2008–2009). We characterize the gravity wave activity in terms of the measurements of buoyancy period and relative density fluctuations and estimate the gravity wave potential energy density. We compare these measurements with measurements at Kangerlussuaq, Greenland (67°N, 51°W) and Kühlungsborn, Germany (54°N, 12°E). We use satellite and global meteorological data to analyze the synoptic structure of the stratospheric vortex and the Aleutian anticyclone, the planetary wave activity, and the mean winds. Major stratospheric warmings with displacement of the vortex and splitting of the vortex occurred in 2007–2008 and 2008–2009, respectively. We find a positive correlation between the gravity wave activity in the upper stratosphere and the winds in the stratosphere at all three sites. During January and February 2008, we attribute the lower average potential energy density (1.6 J/kg) at Chatanika (relative to 4.7 J/kg at Kangerlussuaq and 2.6 J/kg at Kühlungsborn) to the blocking of gravity waves by the lower winds in the Aleutian anticyclone, while the higher value at Kangerlussuaq (where the winds are similar in strength to those at Kühlungsborn) may reflect stronger sources of gravity waves. During February and March 2009, we attribute the lower average potential energy density (1.1 J/kg) at both Chatanika and Kühlungsborn to the seasonal decrease of the middle atmosphere winds. In general the gravity wave activity is lowest when the wind is weak at the lowest altitudes. We compare the gravity wave activity and winds in these winters at Chatanika with the winter of 2003–2004, when an extreme warming event occurred resulting in an elevated stratopause and major reduction of gravity wave activity. We find that the 2004 warming had a stronger influence on the gravity wave activity.
 Gravity waves are known to play a crucial role in the general circulation of the middle atmosphere. These small-scale waves drive a pole-to-pole mesospheric circulation with ascending flow in the summer hemisphere and descending flow in the winter hemisphere [Houghton, 1978]. These vertical motions result in the cooling of the polar summer mesopause region and the warming of the polar winter stratopause region (e.g., see reviews by Holton and Alexander , Fritts and Alexander , and references therein). The propagation of gravity waves is strongly modulated by spatially varying winds. Several studies have shown that gravity wave propagation is modulated by the stratospheric vortex and Aleutian anticyclone, with higher gravity wave activity associated with the stronger winds in the edge of the polar vortex, and lighter winds in the anticyclone blocking the upward transmission of orographic gravity waves [e.g., Dunkerton and Butchart, 1984; Duck et al., 1998].
 In a recent study, Thurairajah et al.  used Rayleigh lidar data from three winters (2002–2003, 2003–2004, 2004–2005) over Chatanika, Alaska (65°N, 147°W) to document the reduced gravity wave activity at the upper stratosphere during the formation of an elevated stratopause in late January 2004. A major stratospheric warming occurred in early January 2004 [Manney et al., 2005]. This warming was a vortex displacement event that led to a two month long disruption of the middle and lower stratospheric vortex. This disruption of the vortex led to the formation of an isothermal atmosphere. The upper stratospheric vortex reformed into a robust, colder cyclone and subsequently led to the formation of an elevated stratopause. Drawing on the modeling study of Siskind et al. , Thurairajah et al.  concluded that the weak stratospheric winds associated with the warmings suppressed the upward propagating gravity waves, disrupted the meridional circulation and adiabatic heating, allowing the region to radiatively cool, and the elevated stratopause to form at ∼80 km.
 In this paper we extend the study of Thurairajah et al.  at a single site at Chatanika, Alaska, to include Rayleigh lidar data from more recent stratospheric warming events during the Arctic winters (2007–2008 and 2008–2009) of the International Polar Year (IPY). We also document the geographic variability in gravity wave activity by comparing the gravity wave activity at Chatanika with that at Kangerlussuaq, Greenland (67°N, 51°W), and Kühlungsborn, Germany (54°N, 12°E). The 2007–2008 and 2008–2009 Arctic winters were meteorologically different. During the 2007–2008 winter, four pulses of warming have been recorded with reversal of wind and temperature gradient on 25 January and 2, 16, and 23 February 2008 [Wang and Alexander, 2009]. The fourth warming on 23 February 2008 was a major warming where the vortex was displaced off the pole as the anticyclone strengthened. Wang and Alexander  also reported a significant geographical variability in gravity wave activity during this winter with higher gravity wave activity in the Eastern Arctic and lower activity in the North Pacific and Alaska. During the 2008–2009 winter, a major warming occurred with reversal of zonal wind and temperature gradient on 24 January 2009 [Manney et al., 2009]. During this warming the polar vortex split in two and the stratopause reformed at ∼80 km during the first week of February and remained elevated until the first week of March 2009 [Manney et al., 2009]. Charlton and Polvani  have analyzed vortex splitting and vortex displacement events in major midwinter stratospheric warming events from 1958 to 2002. Charlton and Polvani  reported that during vortex splitting events the wind reversals are longer, stronger, and extend deeper into the lower atmosphere. We thus expect that the vertical propagation of gravity waves to be more significantly blocked during vortex splitting events.
 In this paper we present Rayleigh lidar measurements of gravity wave activity in the 40–50 km altitude, from Chatanika, Alaska (65°N, 147°W), during both the 2007–2008 and 2008–2009 winters. These direct measurements provide an opportunity to study gravity wave variability under different types of stratospheric warming events. The lidar measurements at Kangerlussuaq, Greenland (67°N, 51°W), and Kühlungsborn, Germany (54°N, 12°E), allows us to investigate the geographical variability in gravity wave activity in the upper stratosphere during stratospheric warming events.
 This paper is organized as follows. In section 2 we review the Rayleigh lidar technique. In section 3 we present lidar measurements of nightly temperature profiles and gravity wave activity from all three sites during the 2007–2008 and 2008–2009 Arctic winters. In section 4 we present the meteorology of the 2007–2008 and 2008–2009 winters in the northern hemisphere in terms of the synoptic structure, evolution, and variability of the Arctic stratospheric vortex and Aleutian anticyclone using meteorological global analyses data, and planetary wave activity and winds using satellite data. In section 5 we discuss the evolution of the vortex and anticyclone and the wind structure over Chatanika, Kangerlussuaq, and Kühlungsborn. In section 6 we discuss the variability in the gravity waves at Chatanika between the 2007–2008, 2008–2009, and 2003–2004 winters. In section 7 we discuss the geographical variability in gravity wave activity during the 2007–2008 and 2008–2009 winters. In section 8 we present a summary and our conclusions.
2. Rayleigh Lidar Data and Analysis
 Rayleigh lidars work on the principle that in the absence of aerosols the scattered light from neutral molecules in the atmosphere is directly proportional to the atmospheric density. Lidar observations yield temperature measurements of the stratosphere and mesosphere (∼30–90 km) under assumptions of hydrostatic equilibrium. In this paper we use lidar measurements from three different sites at, Poker Flat Research Range (PFRR), Chatanika, Alaska, USA (65°N, 147°W); Sondrestrom Upper Atmospheric Research Facility (SUARF), Kangerlussuaq, Greenland (67°N, 51°W); and Leibniz-Institute for Atmospheric Physics (IAP), Kühlungsborn, Germany (54°N, 12°E).
 The lidar systems are zenith pointing with Rayleigh temperature measurements obtained using a Nd:YAG laser operating at 532 nm and a photon counting receiver system. Further technical details about the Rayleigh lidar at Chatanika, Alaska; the Rayleigh/Mie lidar at Kangerlussuaq, Greenland; and the Rayleigh/Mie/Raman lidar at Kühlungsborn, Germany can be found at Collins et al. , Thayer et al. , and Alpers et al. , respectively. Specific details of photon count acquisition and temperature retrieval for each of the lidar systems are described by Thurairajah et al. , Gerrard et al. , and Alpers et al. , respectively. The Rayleigh lidar at Chatanika has been operated since 1997. The lidar at Kangerlussuaq has been operated since 1993. The lidar at Kühlungsborn has been operated for noctilucent cloud measurements since 1997 and year-round temperature measurements, from the troposphere to the mesosphere lower thermosphere (MLT), since 2002. These lidars have supported a variety of middle atmospheric investigations that include studies of temperature variations, mesospheric inversion layers, gravity wave activity, noctilucent clouds, and aerosols [e.g., Collins et al., 2009; Cutler et al., 2001; Murayama et al., 2007; Rauthe et al., 2006; Gerding et al., 2008; Thayer et al., 2003; Gerrard et al., 2000].
 The lidar measurements at Chatanika from October to March over the two winters (2007–2008 and 2008–2009) have yielded a total of 39 nights of measurements lasting between 4 and 13 h for a total of ∼320 h of observations. The lidar measurements at Kangerlussuaq during January and February of 2008 have yielded a total of 5 nights of measurements lasting between 4 and 9 h for a total of ∼34 h of observations. The lidar measurements at Kühlungsborn during January and February of 2008 and October to March of 2009 have yielded a total of 14 nights of measurements lasting between 4 and 13 h for a total of ∼100 h of observations. Individual temperature profiles are presented as nighttime averages. For comparison of gravity wave activity between the 2007–2008 and 2008–2009 winters we use the data from December to March at Chatanika. For the purpose of comparison between the three sites we use data only from January and February of 2008 at Chatanika, Kangerlussuaq, and Kühlungsborn and February and March of 2009 at Chatanika and Kühlungsborn.
 We characterize the gravity wave activity in the upper stratosphere (40–50 km) in terms of the background stability (buoyancy period) of the atmosphere, the RMS relative density, the RMS vertical displacement, and the wave potential energy density (Ep) calculated from 30 min resolution density data. The mean square vertical displacement (ξ2) and the potential energy density (Ep) are given by [e.g., Gill, 1982],
where N is the buoyancy frequency, ξ is the vertical displacement of air parcel, and ρ′/ρo the RMS relative density fluctuation. The RMS relative density perturbations, the buoyancy frequency, and the constant of acceleration due to gravity are independently averaged over the 40–50 km altitude. A detailed description of the gravity wave analysis method is given by Thurairajah et al. . To reduce noise and obtain short period waves, the density perturbations are spatially band-limited between vertical wave numbers 0.5 km−1 and 0.1 km−1 (vertical altitude range of 10 km, i.e., between 40 and 50 km) and temporally band-limited between the Nyquist frequency of 1 h−1 and the low frequency 0.25 h−1 (time period between 30 min and 4 h). The signal-to-noise ratio (SNR) is calculated as the ratio of the variance of the relative density fluctuations to the uncertainty in the estimate of the variance of the statistical fluctuations. Since the Rayleigh lidar measurements are made under clear sky conditions and lidar signal levels are relatively constant, the statistical variance is relatively constant while the variance of the relative density fluctuations, and hence the SNR, represents the variations in the wave activity. The raw photon count data from all three sites has been processed consistently to avoid inter-site biases in the gravity wave activity due to different data processing methods.
 In Figure 1 we plot examples of relative density perturbation from all three lidar sites. The perturbations over Chatanika (Figure 1, top) were taken over ∼9 h on the night of 22 January 2009, over Kangerlussuaq (Figure 1, middle) were taken over ∼9 h during the early morning hours of 9 January 2008, and over Kühlungsborn (Figure 1, bottom) were taken over ∼12 h on the night of 14 February 2008. The relative density perturbations show a downward phase progression at all three sites typical of upward propagating gravity waves. The different tilt of the relative density perturbations reflects the different vertical wavelength of the dominant waves. The relative density perturbations have a dominant vertical wavelength of 10 km, 10 km, and 21 km and time period of 3.6 h, 3.4 h, and 4.0 h at Chatanika, Kangerlussuaq, and Kühlungsborn, respectively. We tabulate the buoyancy period and gravity wave activity during these 3 days in Table 1.
Table 1. Buoyancy Period and Gravity Wave Activity at 40–50 kma
Chatanika 22 January 2009
Kangerlussuaq 9 January 2008
Kühlungsborn 14 February 2008
Measured at Chatanika, Alaska (65°N, 147°W), Kangerlussuaq, Greenland (67°N, 51°W), and Kühlungsborn, Germany (54°N, 12°E).
Calculated from nightly average temperature profile.
 To illustrate the response of the thermal structure of the stratosphere and mesosphere to the stratospheric warmings that occurred during the 2007–2008 and 2008–2009 winters, we plot in Figure 2a, individual nighttime mean and monthly mean profiles for January, February, and March of 2008 and 2009 at Chatanika, Alaska. We average the nighttime profiles to form the monthly mean profile. In Figure 2b we compare the monthly mean temperature profile to the monthly mean profile calculated from 22 (January), 19 (February), and 29 (March) nights of lidar measurements at Chatanika from 1998 to 2005 and reported by Thurairajah et al. , and to the zonal mean temperature climatology from the Stratospheric Processes And their Role in Climate (SPARC) reference atlas [Randel et al., 2004; SPARC, 2002].
 The individual profiles from January 2008 are from late January and thus the monthly mean profile (Figure 2b) indicates a mesospheric cooling associated with the first pulse of January 2008 warming. The individual profiles from January 2009 span the entire month and illustrate the change in temperature from warmer and higher stratopause in early January to mesospheric cooling in late January. For example the two profiles in early January (6 and 8 January 2009) have stratopause altitudes above 50 km and stratopause temperatures warmer (by more than 20 K) than the monthly mean January 2009 profile (stratopause at 57.5 km with temperature of 276.8 K on 6 January and stratopause at 54 km with temperature of 274.3 K on 8 January). After the major warming in late January 2009, the temperature (for example at 60 km) cooled from 264.4 K on 6 January 2009 to 222.8 K on 29 January 2009. This change in thermal structure is also associated with the major warming in late January 2009. Similar mesospheric cooling associated with stratospheric warmings has been reported in previous studies [e.g., Holton, 1983; Walterscheid et al., 2000]. The monthly mean January 2009 profile is similar to the SPARC reference climatology but is warmer than the Chatanika January average, in the upper stratosphere and lower mesosphere. In Figure 2b we also plot the January 2004 monthly mean temperature. The elevated stratopause observed in January 2004 is not observed in either January 2008 or 2009.
 The February 2008 profiles show a higher degree of variability than the February 2009 temperature profiles. The profile on the night of 4 February 2008 has maximum temperature of 263 at 69.5 km, but this is an isolated observation and the other profiles do not exhibit the long characteristics of a pronounced elevated stratopause recorded over several weeks in 2004 [Thurairajah et al., 2010]. The profile on the night of 9 February 2008 has the coldest mesospheric temperature. For example, the temperature at 69.5 km is 202.7 K, more than 20 K colder than the monthly mean February 2008 profile. The monthly mean temperature profile in February 2008 is similar to the Chatanika average but colder than the SPARC climatology in the upper stratosphere and lower mesosphere. In February 2009 the monthly temperature has the least pronounced stratopause and is consistently colder than the February 2008, Chatanika, and SPARC average profiles. As in February 2008 the one profile on the night of 24 February 2009 has a maximum temperature near 74 km and is an isolated observation of the stratopause at elevated altitudes.
 In March 2008 and 2009 the individual nighttime profiles are all similar in structure with the stratopause located between ∼50 and 55 km and stratopause temperatures varying from 240 to 260 K in 2008 and from 245 to 270 K in 2009. The one profile on 15 March 2009 appears to have a broad stratopause from 49 to 65 km with an average temperature of 245 K. The monthly mean March temperature profiles are similar in structure to both the SPARC climatology and Chatanika average and the values in the upper stratosphere and lower mesosphere are within the range of these average profiles.
 In Figure 3 we plot the individual nighttime mean and monthly mean profiles for January and February 2008 at Kangerlussuaq and Kühlungsborn (Figure 3, top and middle), and February and March 2009 at Kühlungsborn (Figure 3, bottom). We average the individual nighttime profiles to form the monthly mean profile. We also plot the zonal mean temperature climatology (average of January and February for 2008 and average of February and March for 2009) from the SPARC reference atlas for comparison. The Rayleigh lidar measurements from both Kangerlussuaq and Kühlungsborn show a high degree of variability in 2008. At Kühlungsborn, such variability in temperature has been reported previously by Rauthe et al. . Two nighttime profiles from Kühlungsborn (22 January and 20 February) coincide with the first and last pulse of warming and show stratopause temperatures warmer than 280 K (stratopause at 55 km with temperature 286.5 K, and stratopause at 38 km with temperature 304.2 K, respectively). The mean Kangerlussuaq temperature is warmer than the SPARC average while the mean Kühlungsborn temperature is similar to the SPARC average. The Rayleigh lidar measurements from Kühlungsborn in 2009 are all similar except the nighttime profile from 12 February that is colder than the 2009 average and the SPARC average.
 In Figure 4 we plot buoyancy period averaged over the 40–50 km altitude as a function of day during December to March of 2007–2008 and 2008–2009 at Chatanika (Figure 4, top) and during January and February 2008 at Kangerlussuaq and Kühlungsborn, and February and March 2009 at Kühlungsborn (Figure 4, bottom). At Chatanika, during both the 2007–2008 and 2008–2009 winters the background atmosphere decreased in stability (increasing buoyancy period) from early December to late January. In late January, the buoyancy period increased to a maximum of 347 s on 24 January 2008 and 343 s on 29 January 2009. The decrease in stability in late January reflects the occurrence of stratospheric warmings during this time period. After this time the atmosphere became more stable and remained more stable through the end of March. At Kangerlussuaq and Kühlungsborn, the 40–50 km altitude region is less stable than Chatanika with buoyancy period between 280 s and 320 s. The two “outliers” with lower stability on 13 January and 20 February 2008 at Kühlungsborn have a buoyancy period of 349 s and 403 s, respectively, and are associated with negative temperature gradient in the 40–50 km altitude region.
3.2. Gravity Wave Activity
 In Figure 5 we plot the gravity wave activity at Chatanika in terms of RMS relative density fluctuation, RMS vertical displacement fluctuation, and potential energy density averaged over 40–50 km as a function of day during December to March in the 2007–2008 and 2008–2009 winters. We tabulate the average and range of values for December to March in Table 2. The gravity wave fluctuations exhibit a high degree of night-to-night variability but in general indicate a broad annual cycle with a maximum in late December and early January and decreasing values through February and March. There are no systematic differences in night-to-night gravity wave activity during the two winters with similar range of values in the RMS density fluctuations, RMS vertical displacement, and potential energy densities. In Figure 6 we plot the gravity wave activity at Kangerlussuaq and Kühlungsborn. As in Chatanika, the gravity wave activity exhibits a broad annual cycle with decreasing gravity wave activity toward March. We tabulate the average and range of values for January and February 2008 and February and March 2009 in Table 3.
Table 2. Buoyancy Period and Gravity Wave Activity at 40–50 km in December to March of 2007–2008 and 2008–2009a
Measured at Chatanika, Alaska (65°N, 147°W).
Calculated from nightly average temperature profile.
Mean value and uncertainty in mean (i.e., standard error).
Table 3. Buoyancy Period and Gravity Wave Activity at 40–50 km in February and March of 2009a
Measured at Chatanika, Alaska (65°N, 147°W), Kangerlussuaq, Greenland (67°N, 51°W), and Kühlungsborn, Germany (54°N, 12°E) in January and February of 2008, and at Chatanika, Alaska (65°N, 147°W) and Kühlungsborn, Germany (54°N, 12°E).
Calculated from nightly average temperature profile.
Mean value and uncertainty in mean (i.e., standard error).
 Since the potential energy density varies as a function of both the RMS relative density fluctuation, ρ′/ρo and the buoyancy frequency, local stability conditions could affect the gravity wave fluctuations. We find no significant correlation (i.e., r < 0.5) of RMS relative density fluctuation with buoyancy period, and we conclude that increases in wave activity are not due to local instability. The range of values is larger in potential energy densities than the relative density fluctuations, which reflects the increase due to mean square, but there is no systematic difference due to interannual variations in the buoyancy period.
 In summary the Rayleigh lidar observations from these three-sites illustrate the thermal variability of the upper stratosphere and lower mesosphere associated with stratospheric warmings and mesospheric cooling. The background atmosphere is in general less stable in late January during the occurrence of stratospheric warming. While there is considerable night-to-night variability in gravity wave activity, there is no pronounced interannual variation in the gravity wave activity and the wave activity exhibits an annual cycle with maximum in early January and decreasing values in February and March.
4. Synoptic View and Planetary Wave Activity During the IPY Winters
 The two winters of 2007–2008 and 2008–2009 were characterized by different meteorological conditions with four warming events in late January (25), early (2), middle (16), and late (23) February 2008 [Wang and Alexander, 2009] and a major strong prolonged warming in late January (24) 2009 [Manney et al., 2009]. In this section we present an overview of the dynamic structure of the middle atmosphere in terms of the position and evolution of the polar vortex and anticyclones (Figures 7 and 8), planetary wave activity and winds (Figures 8 and 9) similar to that presented by Thurairajah and colleagues in describing earlier winters [Thurairajah et al., 2010]. We use the United Kingdom Meteorological Office (MetO) global analyses data to calculate the characteristics of the vortices [Harvey et al., 2002]. Harvey et al.  identify anticyclones and vortex in terms of evolving three-dimensional air masses, based on stream function analysis and Q diagnostics (Q, a scalar quantity is a measure of the relative contribution of strain and rotation in the wind field). We take the zonal and meridional winds from the MetO global analyses data. We calculate the horizontal winds as the total magnitude of the zonal, u and meridional, v winds (). The planetary wave activity is described in terms of the wave geopotential amplitude and Eliassen Palm (EP) flux divergence calculated from temperature, pressure, and geopotential measured by the Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) instrument data (Level 2A version 1.07) (http://saber.gats-inc.com/) aboard the Thermosphere Ionosphere Mesosphere Energetics Dynamics (TIMED) satellite [Mertens et al., 2004; J. M. Russell III et al., An overview of the SABER experiment and preliminary calibration results, paper presented at 44th Annual Meeting, SPIE, Denver, Colorado, 1999]. The EP flux values have been scaled by the basic density to give the zonal force per unit mass exerted by eddies [Holton, 2004]. The gradients winds are also calculated from the SABER data [Lieberman, 1999].
4.1. Arctic Winter of 2007–2008
 The stratospheric vortex and Aleutian anticyclone were dynamically active during the second half of January and February 2008. Four pulses of warming events have been recorded during this time with the fourth warming recorded as a major warming event [Wang and Alexander, 2009]. This major warming was a wave-one vortex displacement event. In Figure 7 we show the 3-D structure of the stratospheric vortex and anticyclones from late December 2007 through March 2008. The amount of intertwining between the vortex (colored) and anticyclones (black) indicates the level of interaction between them and the level of planetary wave breaking. On 25 December there is some interaction, though the vortex is large and robust. In late January and February, there is strong interaction between the traveling anticyclones and the stratospheric vortex, with the vortex beginning to elongate and tilt westward with height. The Aleutian high strengthened, thus displacing the vortex toward the North Atlantic during the first three events, while the stronger anticyclone, weakened and distorted the vortex during the last event leading to a major stratospheric warming.
 The planetary wave analysis (Figure 9, left) indicates the four periods of strong planetary wave-one amplitudes with maximum amplitudes on 23 January 2008, 4, 15, and 22 February 2008. The amplitudes vary between 1800 m and 2600 m at altitudes between 6.3 (∼44 km) and 9.5 (∼66 km) scaled heights. (A scaled height is defined as a pressure scale height, sc = -ln(P/Po), where Po = 1000 mb. Approximate altitudes are obtained by multiplying sc by a typical middle atmosphere scale height of 6.5 km). There are also four periods of large wave-two amplitudes (Figure 9, lower left) during the same time period (with maximum amplitudes on 24 January 2008, 2, 20, and 27 February 2008) and the maximum geopotential amplitudes vary between 700 m and 800 m, between altitudes of 6.9 (∼48 km) and 8.6 (∼60 km) scaled heights. We observe weakening and reversal of the gradient winds (Figure 9, upper right) and convergence of EP flux (Figure 9, lower right) during the period when the planetary wave amplitudes increase. During the major warming in late February 2008, the winds reversed to a maximum of −34.6 m/s (negative implies easterly winds) at 6.0 (∼42 km) scaled heights. The reversals were on 23 January 2008, the same day as the maximum planetary wave-one amplitude, and in February (7, 17, and 24) a few days after the occurrence of the maximum planetary wave-one amplitudes.
 Large values of EP flux divergence precede the maximum values of easterly winds, except during the first warming. Maximum values were recorded on 23 January, 4, 14, and 22 February 2008. The maximum flux convergence values vary between −172 m/s/day to −136 m/s/day at 6.0 (∼42 km) scaled heights. The flux convergence (i.e., negative EP flux divergence) is a measure of the westward zonal force exerted by eddies on the atmosphere and is linked to the entwining of the upper stratospheric vortex around the Aleutian anticyclone on 22 February 2008 in Figure 7. After the late February major warming, the mid and lower stratospheric vortex remained weak and quasi stationary as seen on 30 March 2008 (Figure 7, lower right) until the transition from winter to summer conditions in April.
4.2. Arctic Winter of 2008–2009
 During the 2008–2009 winter, a major stratospheric warming occurred during the third week of January 2009 when the Aleutian high strengthened and the vortex split in two. In Figure 8 (top) we show the nearly pole centered, undisturbed polar vortex on 25 December 2008 and 5 January 2009. The vortex first split in the upper stratosphere around 19–20 January (Figure 8, middle, left), continued to split downward to the midstratosphere (800 K, ∼30 km, ∼10 hPa) on 22 January (Figure 8, middle, right), and was split through the entire stratosphere by the 24 January. The vortex remained split for almost 3 weeks until ∼6 February (Figure 8, lower left) when the upper stratospheric vortex recovered with colder temperatures. During the major warming, planetary wave analysis at 65°N shows maximum planetary wave-two geopotential amplitude of 1400 m at altitude of 5.9 scaled heights (∼ 41 km) on 19 January 2009 (Figure 10 left). Maximum planetary wave-one amplitude of 1200 m and 1000 m at higher altitude of 8.1 (∼57 km) and 9.6 (∼67 km) scaled height are also observed on 21 January and 10 February 2009, respectively.
 A strong and deep wind reversal extending from the mesosphere to the lower stratosphere in mid January 2009 is seen in Figure 10 (right). The wind reversed to a maximum of −53.8 m/s at scaled height of 8 (∼49 km) on 23 January 2009, 2 days after the maximum planetary wave-one amplitude was observed. Such wind reversals in the mesosphere (also seen during the 2008 warming events) have been reported previously [e.g., Hoffmann et al., 2007, and references therein] and show the downward progression of disturbances from the MLT to the stratosphere. The altitude extend of the reversal from the mesosphere to the lower stratosphere has been noted by Charlton and Polvani  as a feature of a vortex split warming events that is not observed in vortex displacement events (see winds in February 2008 in Figure 9). The maximum EP flux convergence of −138 m/s/day at 6.0 scaled heights (∼42 km) occurred on 20 January 2009, 3 days before the maximum wind reversal was observed. This large EP flux convergence indicates the strength of the westward zonal force exerted by eddies on the atmosphere.
 By the second week of February the upper stratospheric vortex had completely recovered and strengthened while the middle and lower stratospheric vortex remained weak (22 February 2009, Figure 8 lower right). The temperature inside the vortex was anomalously cold and lead to the formation of an elevated stratopause in February 2009. The formation of the elevated stratopause has been documented in zonal mean temperature measurements at 70°N [Manney et al., 2009]. The lower stratospheric vortex remained weak until late March 2009.
5. Winters of 2007–2008 and 2008–2009 at Chatanika, Kangerlussuaq, and Kühlungsborn
 In this section we document the temporal evolution of the stratospheric vortex and anticyclone over the three lidar sites during the 2007–2008 and 2008–2009 winters. In Figures 11 and 12 we plot the temporal evolution the stratospheric vortex (green) and anticyclone (red) from 400 K to 2000 K (∼15 km to ∼ 47 km) altitude region during the 2007–2008 and 2008–2009 winters. The vertical yellow lines indicate those dates when Rayleigh lidar measurements were made. The 800 K horizontal wind speeds are plotted along the bottom. Distinct differences existed in the temporal evolution of the stratospheric vortex and anticyclone at the three sites and over the two winters. During the 2007–2008 winter, the Aleutian anticyclone was dominant over Chatanika in January and February (Figure 11, top) when the planetary wave activity is the strongest. At Kangerlussuaq the vortex was present at all stratospheric altitudes from December to mid-February (Figure 11, middle). At Kühlungsborn the vortex appeared periodically overhead throughout the winter (Figure 11, middle). Over the 4-month period the horizontal winds at 800 K were highest over Kühlungsborn, where the winds varied between 0.9 m/s and 97.7 m/s with a median value of 39.2 m/s and an average value of 37.7 m/s. At Kangerlussuaq the winds varied between 1.9 m/s and 94.8 m/s with a median value of 33.0 m/s and average value of 33.9 m/s. At Chatanika the winds varied between 2.0 m/s and 88.6 m/s with a median value of 20.8 m/s and average value of 27.0 m/s. As expected, peaks in the winds are associated with the appearance and disappearance of the vortex when the vortex edge passes overhead. The evolution of the winds at Kangerlussuaq and Kühlungsborn is similar with local maxima in the winds in early January, mid-January, early February, and mid-February reflecting the common influence of the vortex at both sites. The strongest winds were in December, January, and February, and the winds in March were consistently weak at all three sites.
 The temperature profiles measured during the 2007–2008 winter and reported in section 3 are also associated with the movement of the vortex and anticyclone over the three lidar sites. For example, at Chatanika, the temperature profile on the night of 4 February 2008 with the elevated stratopause was measured when the upper stratospheric vortex was over Chatanika. The colder mesosphere reported on 9 February was observed when the anticyclone was over Chatanika. Similarly, at Kühlungsborn the temperature profiles on 22 January and 20 February 2008 with the warmer stratopause (during the first and last pulse of warming) occur when the vortex edge was over Kühlungsborn.
 During the 2008–2009 winter, the major stratospheric warming during the third week of January is evident as the disappearance of the vortex at all sites in late January and early February and the presence of anticyclones at Chatanika and Kühlungsborn in the second half of January (Figure 12). Over the 4-month period the horizontal winds at 800 K were highest over Kühlungsborn, where the winds varied between 0.0 m/s and 67.1 m/s with a median value of 30.1 m/s and an average value of 20.4 m/s. At Kangerlussuaq the winds varied between 0.0 m/s and 83.9 m/s with a median value of 27.0 m/s and average value of 20.4 m/s. At Chatanika the winds varied between 0.0 m/s and 93.5 m/s with a median value of 28.6 m/s and average value of 19.2 m/s. The pronounced reduction of the winds at Kühlungsborn and Kangerlussuaq in 2008–2009 relative to 2007–2008 reflects the reduction of the winds during February 2009 associated with the major stratospheric warming event. Again, peaks in the winds are associated with the appearance and disappearance of the vortex when the vortex edge passes overhead. The strongest winds were in December and January and the winds in February and March were consistently weaker at all three sites.
 As in the 2007–2008 winter the temperature profiles observed during the 2008–2009 winter are also associated with the movement of the vortex and anticyclone over the three lidar sites. The two warm profiles on 6 and 8 January 2009 at Chatanika occur when the vortex edge is overhead at all stratospheric heights. The elevated stratopause measured on 24 February 2009 coincides with the presence of a robust vortex in upper stratosphere. The colder profile on 12 February 2009 reported at Kühlungsborn was measured inside an anomalously cold upper stratospheric vortex. In general, we observe warmer stratospheric temperatures while sampling at the vortex edge, colder stratospheric temperatures in the vortex core, and colder mesospheric temperatures in the anticyclone.
 In summary the evolution of the stratospheric vortex and anticyclones shows distinct differences between the three sites and between the two IPY winters. In 2007–2008 there are several warming events characterized by appearance and disappearance of the anticyclone over Chatanika and the vortex over Kangerlussuaq and Kühlungsborn. The major stratospheric warming event in late February 2008 is observed as growth of the anticyclone over Chatanika and the disruption of the vortex over Kühlungsborn and Kangerlussuaq. The vortex reforms in the upper stratosphere over all three sites in early March. The stratospheric winds at Chatanika are lower than the other two sites reflecting the presence of the Aleutian anticyclone. In 2008–2009 there is a major warming event in January with disruption and disappearance of the vortex over Kangerlussuaq, disappearance of the anticyclone over Chatanika, and appearance of an anticyclone over Kühlungsborn. The winds at all three sites are similar to the values over Chatanika in 2007–2008. The vortex reforms in the upper stratosphere over all three sites in mid-February.
6. Interannual Variability of Gravity Wave Activity at Chatanika
 In order to investigate the effect of winds on the gravity wave activity in the upper stratosphere we plot the zonal winds and gravity wave potential energy densities at Chatanika for the winters of 2007–2008, 2008–2009, and 2003–2004 (Figure 13). As noted earlier, there was a major stratospheric warming in early January 2004 that was associated with a vortex displacement event, large-amplitude planetary waves, and negative EP flux divergence from mid-January through early March. Thurairajah et al.  documented both the formation of an elevated stratopause and reduced gravity wave activity over Chatanika in January 2004. The potential energy density is lower during January and February of 2004, compared to 2008 and 2009, with an average value of 0.81 J/kg (± 0.14) in 2004, 1.6 J/kg (± 0.3) in 2008, and 1.9 J/kg (± 0.5) in 2009. The winds are also lower during January and February of 2004 compared to 2008 and 2009, with median values of 6.3 m/s at 500 K (∼19 km, ∼60 hPa), 10.8 m/s at 800 K (∼30 km, ∼10 hPa), and 37.7 m/s at 1600 K (∼44 km, ∼1 hPa). In 2008 the median winds were 25.0 m/s at 500 K, 23.2 m/s at 800 K, and 40.0 m/s at 1600 K. In 2009 the median winds were 15.2 m/s at 500 K, 19.4 m/s at 800 K, and 35.8 m/s at 1600 K.
 From Figure 13 it is clear that the gravity wave potential energies are reduced when the middle atmosphere winds are reversed and the reversal extends to the lower stratosphere. This behavior is observed consistently in March 2008 (Figure 13, top), in February 2009 (Figure 13, middle), and in the winter of 2003–2004 (Figure 13, bottom). We calculate the linear correlation between the potential energy density of the gravity waves at 40–50 km and the horizontal wind speed at each of the 22 altitudes over the December to March period of 2007–2008 and 2008–2009. We find that the maximum correlation coefficient is 0.64 at 900 K in 2007–2008, and the correlation coefficient is above 0.60 from 450 K to 1200 K, with maximum correlation of 0.81 at 900 K in 2008–2009 (Figure 14). The correlation coefficient is above 0.60 from 450 K to 900 K with a maximum correlation coefficient of 0.85 at 700 K in 2004. Compared to 2007–2008 and 2008–2009, this decrease in altitude of the maximum correlation between potential energy density of the gravity wave at 40–50 km and horizontal wind speed is consistent with the pronounced weakening of winds at the lower stratosphere during the winter of 2003–2004.
7. Geographic Variability in Gravity Wave Activity
 In this section we analyze the geographical variability in gravity wave activity during the 2007–2008 and 2008–2009 Arctic winters using Rayleigh lidar measurements from Chatanika, Kangerlussuaq, and Kühlungsborn. As noted earlier, we process the data from all three locations uniformly to avoid intersite biases in the values of the gravity wave potential energies due to differences in data processing methods.
 Owing to limitations in data availability at all locations at the same time period we compare the buoyancy periods, gravity wave activity, and meteorological conditions for the three lidar sites only during January and February 2008 and at Chatanika and Kühlungsborn during February and March 2009. We tabulate the values of the buoyancy period and gravity wave activity at all three sites for these two periods in Table 3. In 2008 the buoyancy period was highest at Kühlungsborn and Chatanika and lowest at Kangerlussuaq. The potential energy densities were lower at Chatanika with an average value of 1.6 J/kg (± 0.3), and higher at Kangerlussuaq with an average value of 4.7 J/kg (± 1.1). The value of average potential energy density at Kühlungsborn (2.6 J/kg ± 0.6) is between the high average value at Kangerlussuaq and low average value at Chatanika. In 2009 the buoyancy period was highest at Kühlungsborn and lowest at Chatanika. The potential energy densities at Kühlungsborn with an average value of 1.1 J/kg (± 0.2) are similar to those at Chatanika with an average value of 1.1 J/kg (± 0.3).
 We plot the zonal winds and gravity wave potential energy densities at Kangerlussuaq and Kühlungsborn in Figure 15. As in Figure 13, the zonal winds generally decrease in springtime as the middle atmospheric circulation changes from winter to summer conditions and the vortex weakens and disappears. In January–February 2008 the horizontal wind speeds at Kangerlussuaq have median values of 29.2 m/s at 500 K, 45.3 m/s at 800 K, and 56.1 m/s at 1600 K. The wind speeds at Kühlungsborn have median values of 26.2 m/s at 500 K, 47.6 m/s at 800 K, and 98.0 m/s at 1600 K. The 800 K wind speeds at these two sites are about twice as strong as those at Chatanika reflecting the fact that Chatanika is under the Aleutian anticyclone (Figure 11). Despite the similarity of the wind speeds at Kangerlussuaq and Kühlungsborn the gravity wave potential energy densities are nearly twice as high at Kangerlussuaq. This difference may reflect enhanced generation of gravity waves over Kangerlussuaq during the repeated warming events. During such events increased generation of gravity waves is expected as unbalanced flows are restored to balanced states (termed “spontaneous imbalance” by Wang and Alexander ). We note the higher gravity wave activity over Greenland has also been reported during the summer months [Chandran et al., 2009]. In February–March 2009 the horizontal wind speeds at Kühlungsborn have median values of 6.3 m/s at 500 K, 14.4 m/s at 800 K, and 49.7 m/s at 1600 K. The wind speeds at Chatanika have median values of 9.5 m/s at 500 K, 12.9 m/s at 800 K, and 39.1 m/s at 1600 K. As at Chatanika, we find a positive correlation between the gravity wave potential energy densities at 40–50 km and the stratospheric wind speeds.
8. Summary and Conclusions
 We have used Rayleigh lidar observations at Chatanika, Alaska, Kangerlussuaq, Greenland, and Kühlungsborn, Germany to characterize the gravity activity in the upper stratosphere lower mesosphere (40–50 km) during the 2007–2008 and 2008–2009 Arctic winters. These two winters are meteorologically different. In 2007–2008 there are four stratospheric warming events from late January to late February, where the fourth warming in late February 2008 was a major warming with displacement of the vortex. In 2008–2009 there was one major warming event in late January 2009 with splitting of the vortex.
 At Chatanika, we find considerable night-to-night variability in gravity wave potential energy density during the two winters, but we do not observe any systematic differences in gravity wave potential energy between the vortex displacement stratospheric warming of 2008 and the vortex splitting stratospheric warming of 2009. The potential energy densities are positively correlated with the mean winds in the lower stratosphere. The lowest gravity wave activity is observed when the zonal winds reverse in the middle and lower stratosphere.
 We find the gravity wave activity at Chatanika in these IPY winters is not as low as measured in 2003–2004 when an elevated stratopause was observed for over a month and the reversal of the zonal winds routinely extended downward into the lower stratosphere. While the January 2009 warming has been documented as the strongest and longest-lasting on record we conclude (from this observational study and Thurairajah et al. ) that the 2004 warming had a stronger influence on the gravity wave activity in the 40–50 km altitude range at Chatanika.
 We find lower gravity wave activity at Chatanika (1.6 J/kg) compared to the wave activity in Kangerlussuaq (4.7 J/kg) and Kühlungsborn (2.6 J/kg) during January and February 2008. We find a positive correlation between the gravity wave activity in the upper stratosphere and the winds in the stratosphere at all three sites. We attribute the lower values at Chatanika to blocking of gravity waves by the lower winds in the Aleutian anticyclone. The higher values at Kangerlussuaq may reflect stronger sources of gravity waves over Greenland.
 We do not observe a significant difference in gravity wave activity at Chatanika (1.1 J/kg) and Kühlungsborn (1.1 J/kg) in February and March 2009. We attribute these lower values to the seasonal decrease in both the amplitude and geographic variation of the middle atmosphere winds associated with the breakdown of the vortex in the transition from the winter circulation pattern to the summer circulation pattern.
 The authors thank the staff at Poker Flat Research Range, Sondrestrom Upper Atmospheric Research Facility, and the Leibniz-Institute of Atmospheric Physics for their ongoing support of the lidar programs. The lidar observations were conducted as part of the Arctic Observing Network, a program of the International Polar Year. The authors thank Agatha Light and Brita Irving for their assistance in making the lidar observations at Chatanika. The authors acknowledge the SABER science and data processing teams for providing the SABER data presented in the paper. The authors acknowledge support from the United States National Science Foundation under grants ATM 0334122, ARC 0632387, and ATM 0640340. PFRR is a rocket range operated by Geophysical Institute of the University of Alaska Fairbanks with support from the United States National Aeronautic and Space Administration. The observations at Kühlungsborn were conducted as part of the International Leibniz Graduate School for Gravity Waves and Turbulence in the Atmosphere and Ocean with support from the Mecklenburg-Vorpommern Ministry of Education, Science and Culture and the German Federal Ministry of Education and Research. The authors thank two anonymous reviewers for their valuable comments.