Mesopause Structure from Thermosphere, Ionosphere

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 112, D09102, doi:10.1029/2006JD007711, 2007

Mesopause structure from Thermosphere, Ionosphere, Mesosphere, Energetics, and Dynamics (TIMED)/Sounding of the Atmosphere Using Broadband Emission Radiometry (SABER) observations Jiyao Xu,1 H.-L. Liu,1,2 W. Yuan,1 A. K. Smith,3 R. G. Roble,2 C. J. Mertens,4 J. M. Russell III,5 and M. G. Mlynczak4 Received 29 June 2006; revised 14 November 2006; accepted 17 December 2006; published 2 May 2007.

[1] Thermosphere, Ionosphere, Mesosphere, Energetics, and Dynamics (TIMED)/ Sounding of the Atmosphere Using Broadband Emission Radiometry (SABER) temperature observations are used to study the global structure and variability of the mesopause altitude and temperature. There are two distinctly different mesopause altitude levels: the higher level at 95–100 km and the lower level below 86 km. The mesopause of the middle- and high-latitude regions is at the lower altitude in the summer hemisphere for about 120 days around summer solstice and is at the higher altitude during other seasons. At the equator the mesopause is at the higher altitude for all seasons. In addition to the seasonal variation in middle and high latitudes, the mesopause altitude and temperature undergo modulation by diurnal and semidiurnal tides at all latitudes. The mesopause is about 1 km higher at most latitudes and 6–9 K warmer at middle to high latitudes around December solstice than it is around June solstice. These can also be interpreted as hemispheric asymmetry between mesopause altitude and temperature at solstice. Possible causes of the asymmetry as related to solar forcing and gravity wave forcing are discussed. Citation: Xu, J., H.-L. Liu, W. Yuan, A. K. Smith, R. G. Roble, C. J. Mertens, J. M. Russell III, and M. G. Mlynczak (2007), Mesopause structure from Thermosphere, Ionosphere, Mesosphere, Energetics, and Dynamics (TIMED)/Sounding of the Atmosphere Using Broadband Emission Radiometry (SABER) observations, J. Geophys. Res., 112, D09102, doi:10.1029/2006JD007711.

1. Introduction Zahn et al., 1996; Yu and She, 1995; She et al., 1993] but exist at only a few sites. [2] The mesopause, defined as the temperature minimum [3] A 10-year climatology from lidar measurements indi- that separates the middle atmosphere from the thermosphere, cates that there are two distinct mesopause altitudes, one at is the coldest region in the terrestrial atmosphere. The mean about 100 km during winter and the other at 86 km during mesopause location and temperature are established by summer [von Zahn et al., 1996; She et al., 1993; Yu and radiative and dynamical processes [e.g., Leovy, 1964; Holton, She, 1995; Leblanc et al., 1998; She and von Zahn, 1998]. 1983] and also display large variability due to gravity waves, These observations, however, were from separate local sites planetary waves and tides. Both lower atmospheric and based on a limited number of observation days, and most thermospheric processes can contribute to the variability. were taken during nighttime. The temperature profile is the most important parameter for [4] The NASA Thermosphere, Ionosphere, Mesosphere, characterizing this transition from the mesosphere to the Energetics, and Dynamics (TIMED) satellite was launched thermosphere; two ground-based methods that are currently in December 2001, carrying four instruments for investi- used for determining it are in situ rocket measurement [e.g., gating the region between 60 and 180 km. The Sounding of Lu¨bken and von Zahn, 1991; Fritts et al., 2004; Lu¨bken and the Atmosphere Using Broadband Emission Radiometry Mu¨llemann, 2003; Lu¨bken, 1999] and sodium lidar measure- (SABER) instrument obtains global profiles of temperature ments. The lidars have very high vertical and temporal and other parameters [Russell et al., 1999; Mertens et al., resolution in the altitude region of 80–110 km [e.g., von 2001]. In this paper we analyze the global altitude and temperature of the mesopause and their latitudinal, 1State Key Laboratory of Space Weather, Chinese Academy of seasonal and diurnal variations as measured by SABER. Sciences, Beijing, China. Simulations with the thermosphere-ionosphere-mesosphere 2High Altitude Observatory, National Center for Atmospheric Research, electrodynamics–global climate model (TIME-GCM) help Boulder, Colorado, USA. 3 in the interpretation of the results and the implications for Atmospheric Chemistry Division, National Center for Atmospheric mesospheric dynamics. Research, Boulder, Colorado, USA. 4NASA Langley Research Center, Hampton, Virginia, USA. 5Department of Physics, Hampton University, Hampton, Virginia, USA. 2. Data Processing

D09102 1of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 1. The global distribution of mesopause (a) altitude (units are km) and (b) temperature (units are K), for a 60-day period centered around March based on 4-year-averaged Sounding of the Atmosphere Using Broadband Emission Radiometry (SABER) temperature. files from the SABER instrument onboard the TIMED marked as missing were used in this study. In other words, satellite. The height of the TIMED orbit is about 625 km, we did no extra screening to remove profiles or individual the inclination is 74.1°, and the period is about 1.6 h. Because temperature measurements that were outliers. TIMED is a slowly precessing satellite, it takes about 60 days [8] In order to analyze the mesopause altitude for the full to complete a full 24-hour coverage of local time. 24-hours local time, we apply 60-day windows on the [6] SABER measures temperature from the lower strato- SABER data. Averages over shorter periods include only sphere to the lower thermosphere by limb observation of a subset of local times, which can affect the mean altitude CO2 infrared emission. The inversion takes into account and temperature of the mesopause. Analysis labeled by a departures from local thermodynamic equilibrium (LTE) particular month includes 30 days data in that month plus and variations in CO2 mixing ratio (measured by SABER 15 days before and after the month. The data used extend during daytime). Mertens et al. [2004] present comparisons from February 2002 through February 2006, with averaging of preliminary SABER observations to in situ falling sphere periods centered on each month. measurements at high latitudes. The comparisons indicate [9] For averaging over all local times, the mesopause that the mesopause temperatures are similar between the two temperatures and altitudes were first sorted into the 24 local but that the SABER mesopause altitude is lower than that time bins and averaged. Then values for all local time bins determined by the falling sphere method by about 2–4 km. that were not empty were averaged. Note that a few bins This difference persists in more recent versions of the around local noon were never filled. This introduces a bias SABER retrieval (Version 1.06). The exact cause of this into the diurnal means, especially in the tropics where the discrepancy is not clear, though recent study suggests that diurnal temperature perturbations are largest. non-LTE retrieval with additional CO2 v-v exchange processes can bring the SABER summer mesopause altitudes 3. SABER Mesopause Altitude and Temperature closer to ground base observations [Kutepov et al., 2006]. The present study uses Version 1.06 Level 2A data. Four [10] In this study we are going to focus only on the years of SABER data from February of 2002 to February of climatological mean temperature structure and its variation 2006 are used in this paper. with solar local time. We thus average the temperature along [7] The SABER temperature inversion provides profiles the latitude circle for each local time hour and each latitude on pressure levels. The effective field of view of the bin over the 60-day period. This averaging will remove or SABER instrument gives a vertical resolution of about reduce the nonmigrating tidal components, but the migrat- 2km[Mertens et al., 2001]. Altitude information for each ing components will remain. The confidence level of the temperature measurement is a standard data product that averaged temperature can be estimated. At June solstice at was obtained by integrating the temperature hydrostatically the mesopause for the local time bin 0–1 hour, for example, from a reference position in the stratosphere, using NCEP the analysis error estimates for a confidence level of 95% analyses. In our analysis, the mesopause location for each are 2.4 K, 2.4 K, and 5.8 K for the latitudes of 0°,45°N, and profile is identified by determining the minimum tempera- 75°N, respectively. The mesopause altitude and temperature ture. Mesopause temperature and altitude data are binned in are then calculated from these averaged profiles. It should 10 degrees of latitude (from 90°S–90°N) and 1 hour of be noted here that, in this analysis, we take the minimum local time, and the binning is performed every 5 degrees . temperature of the vertical profile as the mesopause. When There are two hours of missing data around noon in almost more than one local minimum is present in a profile, as can every latitude [Zhu et al., 2005]; at high latitudes, the length happen in the presence of tides or other waves, the smallest of the missing midday measurements becomes longer: up to among these local minima determines the mesopause. 3–4 hours varying with season. The SABER data not Figures 1 and 2 give the results for the March and

2of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 2. (a, b) Similar to Figure 1 but for September.

September equinoxes for the 4-year-averaged results. The night. The nighttime mesopause positions are in good mesopause altitude is between 95 and 100 km around qualitative agreement with the bimodal mesopause altitudes equinoxes, but dips below 90 km for a few hours at low derived from nighttime potassium lidar measurements in latitudes during the morning and at middle latitudes during June of 1996 [vonZahnetal.,1996],exceptthatthe night. The range of mesopause temperature is from 160 to summer mesopause altitude is lower in SABER than in 185 K; it is cooler in the equatorial region and warmer the lidar observations by about 2–4 km. This discrepancy is toward both poles. similar to that between the SABER results and rocket [11] The temporal and spatial variations of the mesopause measurements as mentioned in section 2. Note from Figure 4 altitude and temperature suggest that these variations are that the lowest mesopause altitude (83 km) is not at the probably associated with atmospheric tides, particularly the highest latitude of the observation, but at around 40–50°N, diurnal tide. Both mesopause altitude and temperature have while the lowest temperature is at the highest latitude a 24-hour periodicity at low latitudes and midlatitudes. The sampled (124–135 K, depending on local time, at 80°N). rapid latitudinal change of the mesopause altitude and temperature at around 20° is consistent with the phase changes of the diurnal (1, 1) mode. The strongest modulation of both the altitude and temperature occurs at the equator (peak-to-peak 15 km and 30 K, respectively, for March) and also at latitudes between 30–40°, again consistent with the diurnal (1, 1) mode. The modulation is larger in March than in September, agreeing with the variations of the diurnal tide amplitude [McLandress, 2002]. The semidiurnal signature is also more evident in September (Figure 2) than in March (Figure 1). The tidal modulation of mesopause height and temperature is also consistent with the model study by Berger and von Zahn [1999]. These features are similar for every year from 2002 to 2005. [12] Figure 3 demonstrates the rapid shift of mesopause altitude due to tidal perturbation. The vertical distance between the two low-temperature values (at 30°N in March) is 20 km, similar to the vertical wavelength of the observed migrating diurnal tide [McLandress et al., 1996]. At LT 0.5 hour, the minimum temperature occurs at 84.7 km, that is, at the lower elevation of the two temperature minima. In the following hour (LT 1.5 hour), the higher elevation minima is now colder, and therefore the mesopause altitude jumps to 101 km. [13] For the solstice period at June (Figure 4), there are two distinct ranges of mesopause altitude, with the higher Figure 3. The temperature profiles at four local times at range between 95–102 km and the lower between 82 and the latitude of 30°N in March of the 4-year average. Note 85 km. The boundary between the two altitude ranges is that LT 23.5, 0.5, 1.5, and 2.5 hours refer to hourly means located between 30–40°N during the day and 25–30°Nat between LT 23–24, 0–1, 1–2, and 2–3 hours, respectively.

3of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 4. (a, b) Similar to Figure 1 but for June.

At middle and low latitudes, both mesopause altitude and signal. From May to August poleward of 40°N, the diur- temperature display modulations with a 12-hour period and nally averaged mesopause altitude is between 83 and a clear northward progression with strongest modulation 84.5 km. The equatorward extension of the region of low near 40°S (peak-to-peak 5 km and 25 K, respectively). mesopause altitude is especially prominent in May–July (to These characteristics suggest that the modulation is closely about 35°N). In August, the lowest mesopause altitude of tied to the migrating semidiurnal tide, which peaks around 82 km is found between 45°N and 55°N. During the boreal solstice. summer, the lowest diurnally averaged mesopause temper- [14] The general features are similar in December ature is 126 K (at 80°N in July). This is in general (Figure 5), although the minimum summer mesopause in agreement with a set of falling sphere temperature profiles the polar region (133–143 K) is warmer than that in June measurement at Longyearbyen (78°N) by Lu¨bken and (124–135 K), and the lowest mesopause altitude is about Mu¨llemann [2003], although the diurnal average mesopause 1 km higher. A thorough comparison of winter mesopause altitude from SABER is again slightly lower than those altitude and temperature between December and June is not from the falling sphere measurements. In the austral winter, possible because there are no observations at most local the mesopause altitude increases to 98 km and the temper- times poleward of 50° in the winter hemisphere. With the ature minimum reaches as high as 188 K in the polar region. data available, it appears that the winter mesopause tem- In November, December, January and February the region perature in December is generally warmer than that in June. of low mesopause altitude extends poleward from 40°S, [15] Figure 6 shows the latitudinal and annual variation of with the minimum value of about 82 km at 55°S. The lowest mesopause altitude and temperature, where the mesopause mesopause temperature of about 135.5 K is found at 80°Sin is determined from temperature profiles averaged over the January. 24 hours of local time for each month. The modulation by [16] Hemispheric asymmetry is evident in Figure 6: the tides is thus in principle removed although a tide that varies summer mesopause altitude at high latitudes is higher in the significantly during the 60-day period can leave a residual SH than in the NH by about 1 km, and the summer

Figure 5. (a, b) Similar to Figure 1 but for December.

4of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 6. The diurnally averaged global distribution from the 4-year-averaged results of (a) the mesopause altitude and (b) the mesopause temperature. Each monthly data point represents 60 days of observation centered on the specific month. Contour intervals are 1 km (Figure 6a) and 5 K (Figure 6b). mesopause temperature in the polar region is higher in the are quite sensitive to water vapor, nucleation processes, and SH than in the NH by about 5–10 K. To better illustrate this vertical wind. asymmetry, we overplot the latitudinal distribution of the [17] It is evident from Figure 7 that around December mesopause altitude (Figure 7a) and mesopause temperature solstice in both hemispheres, the mesopause altitude is (Figure 7b) derived from the 60-day average around higher at most latitudes, and that the mesopause is warmer December solstice (at day 350) and June solstice (at day 170). at middle and high latitudes. The results for high latitudes in It is clearly seen from Figure 7 that the mesopause altitude the winter hemisphere should be treated with caution during austral summer (solid line) is 1 km higher than that because of very poor local time coverage there (Figures 4 during the boreal summer (dashed line), and the summer and 5). At middle-low latitudes, the mesopause altitude is mesopause temperature in the polar region is warmer in the between 96 and 100 km for the whole year, and the SH than in the NH. Figure 8 shows that the austral summer distribution of mesopause temperature is relatively uniform. is warmer not only at the mesopause but also at all altitudes [18] The transition time between the two mesopause between 20 and 90 km at the polar region. The hemispheric altitudes is also studied using SABER data. Figure 9 shows difference in summer mesopause temperature agrees with the daily mesopause altitude and temperature at 5 latitudes that found by Huaman and Balsley [1999] from analysis of for 4 years. From Figure 9a we can see that at the middle- HRDI temperature data. Hemispheric differences of about and high-latitude regions there are about 120–130 days 1 km are also found in the polar mesospheric cloud (PMC) centered around the summer solstice when the mesopause altitudes from lidar measurement [Chu et al., 2003, 2006]; the altitude is at the lower level, which is around 85 km in both PMC altitudes are strongly affected by temperature but also hemispheres. The summer mesopause altitude in the SH

Figure 7. Latitudinal distribution of (a) mesopause altitude and (b) mesopause temperature, from the diurnally averaged temperature profiles for December (solid line) and June (dashed line). The June curves have been flipped so that left to right on the latitude axis indicates south to north for the December curves and north to south for the June curves.

5of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 8. Profiles of diurnally averaged temperature from SABER at 80° for both hemispheres (solid line for Southern Hemisphere and dashed line for Northern Hemisphere) and two summer months (left profile shows December and June; right profile shows January and July). shows more variability than in the NH. In other seasons the (Figure 7) at and below the altitude of the summer meso- mesopause altitude is at the high level in middle and high pause altitude (Figure 8). Possible mechanisms for this latitudes. At the equator the mesopause altitude is at the hemispheric asymmetry are discussed in this section. high level in all seasons. Owing to the missing data at the [22] The thermal structure of the mesosphere/mesopause latitudes of 80°N and 80°S, we cannot determine the time of is affected by radiative processes, by heating through the transition between the two mesopause levels so we will exothermic chemical reactions, and by dynamical processes. investigate this at 50°N and 50°S. We take the variation of Adiabatic warming and cooling due to vertical motion is the temperature at 50°N in 2005 as an example (Figure 10). responsible for the strong departure from the radiative Figure 10 shows that from day 125 to day 130, the equilibrium [e.g., Brasseur and Offermann, 1986; Mlynczak mesopause altitude jumps from the higher mesopause level and Solomon, 1993; Berger and von Zahn, 1999; Holton, to the lower level. During the period from day 250 to 1983]. Variation in any of the processes could lead to the day 255, the mesopause altitude jumps back from the high hemispheric asymmetry of the mesopause temperature and level to the low level. Therefore the transitions between the altitude. Chu et al. [2003] showed that at most altitudes the two levels of mesopause are fast (several days) and they summer mesosphere is warmer at the South Pole (SP) than occur 40 days after the spring equinox and 15 days that at the North Pole (NP) in simulations made with the before the fall equinox. The temperature transitions, on the TIME-GCM numerical model. It was determined from the other hand, are more gradual (Figure 9b). simulations that the difference is mainly caused by [19] We recognize that the summer mesopause altitudes the greater solar flux in January than in July due to the derived from SABER temperature profiles may be system- Earth’s orbital eccentricity. Therefore the warmer meso- atically low by 2–4 km, especially in the high latitude, as sphere during austral summer is a consistent feature in mentioned earlier. This bias, however, should affect the multiple-year TIME-GCM simulations. determination of summer mesopause in both hemispheres [23] Figure 11 shows the latitudinal and seasonal varia- similarly. tion of the mesopause altitude and temperature derived from [20] Larger error may occur at higher latitudes (>52°)in monthly mean temperature profiles of TIME-GCM simula- the analysis because the polar regions are only alternatively tion for the year 2004. The variation is in general agreement observed half of a year owing to the yawing of the TIMED with those of the SABER measurements (Figure 6). The satellite. The yaw cycle is 60 days, and only one polar mesopause temperature from the simulation is generally region is observed in each yaw cycle. warmer at most latitudes in December/January than in June/ July. Similar to the SABER observations (Figure 8), the 4. Asymmetry Between the Northern and temperature below the mesopause is also higher during Southern Hemispheres austral summer in the simulation. Detailed comparisons, however, show disagreement between the observation and [21] As shown in section 3, the SABER measurements simulation. For example, in the simulation the temperature indicate that the mesopause is warmer around December is generally lower at low latitudes and the polar mesopause solstice than June solstice at middle and high latitudes is colder in summer and warmer in winter than those in the

6of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 9. Daily (a) mesopause altitude and (b) mesopause temperature, at five latitudes during 2002 to February of 2006.

7of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 10. The variation of temperature at 50°N around (left) March equinox and (right) September equinox. The dashed lines show the mesopause altitudes. observations. The meridional temperature gradients are single level to change the sign of the thermal gradients, and generally larger at high latitudes in the simulation that in (2) most models indicate the bulk of the gravity wave the observations. It is also difficult to resolve the 1–2 km momentum driving and associated adiabatic warming hemispheric difference of mesopause altitude with certainty occurs below 100 km, which will accentuate rather than in the model because the vertical resolution of the model alter the temperature minimum above. near the mesopause is about 2 km. [25] In the summer hemisphere, however, the mesopause [24] An asymmetry in the gravity wave forcing of the is determined by the lower temperature from two competing circulation is another possible cause of the hemispheric cooling mechanisms: the net radiative balance and adiabatic asymmetry of the mesopause altitude and temperature, cooling [Chu et al., 2005]. From both observations and especially in the summer hemisphere. As is well established model simulations, it is clear that the summer mesopause in previous studies, the gravity wave breaking in the altitude at high latitudes is always at lower altitudes (below mesosphere drives a circulation from the summer to the 90 km) and thus that the adiabatic cooling is reliably strong. winter hemisphere; this circulation generates adiabatic cool- Model studies [e.g., Garcia and Solomon, 1985] also ing and warming, respectively, in the summer and winter indicate that the summer mesopause altitude coincides with mesosphere [e.g., Andrews et al., 1987]. The winter meso- the strongest upward vertical wind of the residual circula- pause location is not controlled by gravity wave forcing in tion (Figure 1 in that paper) and occurs near the altitude of any significant way, but rather tied to the minimum tem- the strongest forcing by gravity wave breaking (Figure 6 in perature near 100 km owing to weak solar heating. This is that paper). Therefore the summer mesopause altitude and because (1) the adiabatic warming associated with the temperature are dependent closely on the strength of the gravity wave driven circulation is spread over a deeper gravity wave sources in the troposphere and on the wave vertical layer because of the broader spectrum of waves filtering by the wind below the mesopause. Studies by [Garcia and Solomon, 1985] and is not sufficient at any Vincent [1994] and Dowdy et al. [2001] suggested that the

Figure 11. (a, b) Similar to Figure 6 but from thermosphere-ionosphere-mesosphere electrodynamics– global climate model (TIME-GCM) simulation of the year 2004.

8of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Figure 12. Zonal mean temperature at December solstice from three otherwise identical TIME-GCM simulations, with the gravity wave momentum fluxes at the lower boundary in the (left) ‘‘weak’’ gravity wave case and the (right) ‘‘strong’’ gravity wave cases 50% and 200% of that in the (center) base case. The dashed lines indicate the mesopause heights at corresponding latitudes. Contour interval is 10 K. gravity wave activity is weaker in the austral summer; the more pronounced with a narrower gravity wave phase-speed weaker gravity wave driving and warmer temperature in the spectrum. mesosphere are consistent with their measurements of a [28] The sensitivity of the mesopause to gravity wave smaller mean vertical wind. source strength is tested here through three TIME-GCM [26] Large uncertainties in gravity wave sources, gravity simulations. In all three simulations, the gravity wave wave launch level, the parameterization of gravity waves, spectra of phase speed specified at the lower boundary and limited vertical resolution make it difficult to accurately of the model (10 hPa) are anisotropic with the spectral peak determine the summer mesopause and its hemispheric at 20 msÀ1 (eastward). As discussed by Liu and Roble asymmetry in global models. For example, the disagreement [2002], the gravity wave spectra was specified to be between the observed summer mesopause structure anisotropic in order that the jet reversal in the summer (Figure 6) and the simulation (Figure 11) may be due to hemisphere occurred at a lower altitude than that in winter, the uncertainty in the latitudinal distribution of gravity as observed. Observational support for such an anisotropy wave. This is because gravity wave drag plays a determin- in the equatorial region was provided recently by Kovalam ing role in the departure of the summer mesopause from et al. [2006]. The magnitudes of the gravity wave momen- radiative equilibrium; the radiative and chemical processes tum fluxes are different for the three cases: that in the are of secondary importance there. In contrast, the winter ‘‘weak’’ gravity wave case is half of the ‘‘base’’ case and mesopause structure is mainly determined by the balance that in the ‘‘strong’’ gravity wave case is twice as large. The between radiative cooling, chemical heating and diffusive simulations are otherwise identical. It is shown in Figure 12 transport of heat from the thermosphere. The uncertainty that within this range of gravity wave fluxes, a stronger flux may also explain why the summer mesopause altitudes vary leads to a lower breaking altitude for the waves while a among a range of global models with different gravity wave weaker flux leads to a higher breaking altitude. The position parameterization schemes [e.g., Garcia and Solomon, 1985; of the mesopause altitudes will change accordingly: lower Berger and von Zahn, 1999; Siskind et al., 2003]. The for stronger gravity wave fluxes. The mesopause altitudes at winter mesopause, at the same time, has been consistently at high latitudes (poleward of 40°) in the summer hemisphere 100 km in all these models. are highest in the weak gravity wave case (87–90 km) and [27] The sensitivity of mesopause temperature to the lowest in the strong gravity wave case (at 80–83 km). The gravity wave filtering was demonstrated by Siskind et al. mesopause altitudes are between 81–86 km in the base [2003], who showed that the greater eastward wind in the case. With the scale height near the summer mesopause Southern Hemisphere troposphere during summer leads to around 4 km, the change of the mesopause altitudes in these stronger breaking of westward gravity wave components in three numerical experiments suggest that the mesopause the troposphere and lower stratosphere and greater filtering altitude decreases by three quarters to one scale height when of the eastward wave components. As a result of the greater the momentum flux of the wave source doubles (increased filtering, there are fewer eastward wave components that by 100%). can propagate into the mesosphere. The eastward forcing is [29] Accompanying the lowering of the summer meso- thus weaker in the summer mesosphere and the summer pause altitudes, the summer mesopause temperature becomes mesopause in the SH is warmer. Because the tropospheric warmer. This can be associated with the increasing air density wind mainly affects the gravity wave components with at lower altitudes. With the doubling of gravity wave relatively lower phase speed, the hemispheric difference is momentum flux, the wave breaking altitude (and the summer

9of11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102 mesopause altitude) decreases by about 0.75–1 scale temperature retrieval due to uncertainties in parameters; height from z0 (the base mesopause height) to zd (the slow precession in local time of the TIMED satellite; and mesopause height with doubled gravity wave fluxes). At zd, missing data, especially at high latitudes due to yawing of the momentum deposition nearly doubles while, with r(zd)= the TIMED satellite. Further comparisons with other obser- r(z0) exp((zd À z0)/H), the air density is about 2.11 to vations are thus desirable to better quantify the mesopause 2.72 times as large as the air density at z0. Therefore the structure and variability. increase of the momentum flux deposition is offset by the increase of the air density at the breaking altitude zd and [33] Acknowledgments. We would like to thank Xinzhao Chu and Qian Wu for helpful discussions. This research was supported by the the acceleration/deceleration rate at zd is 94%–74% of National Science Foundation of China (40523006, 40225011, 40336054) that at z0. The strength of the residual circulation and the and the National Important Basic Research Project (2006CB806306). This adiabatic cooling rate are directly related to the acceleration/ work was also supported in part by the International Collaboration Research deceleration rate and will be smaller at the lower altitude. At Team Program of the Chinese Academy of Sciences. H.L.L.’s effort is in 70°S, the increase of the mesopause temperature is about part supported by the NASA Sun Earth Connection Theory Program. Support for A.K.S. and partial support for J.Y.X. to visit NCAR was 5 K for each doubling of the momentum flux. provided by the NASA Office of Space Science. The National Center for Atmospheric Research is operated by the University Corporation for Atmospheric Research under the sponsorship of the National Science 5. Conclusions Foundation. [30] In this study we use TIMED/SABER temperature References observations to analyze the global distribution of the me- Andrews, D. G., J. R. Holton, and C. B. Leovy (1987), Middle Atmosphere sopause altitude and temperature. The results indicate that Dynamics, 489 pp., Elsevier, New York. there are two distinct ranges of mesopause altitude, a higher Berger, U., and U. von Zahn (1999), The two-level structure of the meso- one between 95 and 100 km and a lower one below 86 km. pause: A model study, J. Geophys. Res., 104, 22,083–22,093. Brasseur, G., and D. Offermann (1986), Recombination of atomic oxygen These results are in general agreement with the mesopause near the mesopause: Interpretation of rocket data, J. Geophys. Res., 91, structure observed by lidar [von Zahn et al., 1996; She et al., 10,818–10,824. 1993] although the summer mesopause altitude is lower in Chu, X., C. S. Gardner, and R. G. Roble (2003), Lidar studies of inter- SABER than either lidar or falling sphere data. The meso- annual, seasonal, and diurnal variations of polar mesospheric clouds at the South Pole, J. Geophys. Res., 108(D8), 8447, doi:10.1029/ pause temperature varies from 126 K in the summer polar 2002JD002524. region to 190 K in the winter polar region. Around Chu, X., C. S. Gardner, and S. J. Franke (2005), Nocturnal thermal structure equinox, the diurnally averaged distribution of mesopause of the mesosphere and lower thermosphere region at Maui, Hawaii (20.7°N), and Starfire Optical Range, New Mexico (35°N), J. Geophys. altitudes is quite uniform and mainly in the higher altitude Res., 110, D09S03, doi:10.1029/2004JD004891. range and the mesopause temperature is between 180– Chu, X., P. J. Espy, G. J. Nott, J. C. Diettrich, and C. S. Gardner (2006), 190 K. Around solstice the mesopause position at middle Polar mesospheric clouds observed by an iron Boltzmann lidar at Rothera (67.5°S, 68.0°W), Antarctica from 2002–2005: Properties and implica- and high latitudes in the summer hemisphere is at the lower tions, J. Geophys. Res., 111, D20213, doi:10.1029/2006JD007086. altitude while it is at the higher altitude at other latitudes and Dowdy, A., R. A. Vincent, K. Igarashi, Y. Murayama, and D. J. Murphy in the winter hemisphere. Therefore there is strong seasonal (2001), A comparison of mean winds and gravity wave activity in the variation of the mesopause altitude and temperature in the northern and southern polar MLT, Geophys. Res. Lett., 28,1475– 1478. middle- and high-latitude regions, while at low latitudes the Fritts, D. C., B. P. Williams, C. Y. She, J. D. Vance, M. Rapp, F.-J. Lu¨bken, mesopause altitude and temperature are quite uniform A. Mu¨llemann, F. J. Schmidlin, and R. A. Goldberg (2004), Observations throughout the year. In addition to seasonal variation, the of extreme temperature and wind gradients near the summer mesopause during the MaCWAVE/MIDAS rocket campaign, Geophys. Res. Lett., 31, mesopause altitude and temperature are also strongly mod- L24S06, doi:10.1029/2003GL019389. ulated by the migrating diurnal tide at equinox and the Garcia, R. R., and S. Solomon (1985), The effect of breaking gravity waves semidiurnal tide at solstice. on the dynamics and chemical composition of the mesosphere and lower thermosphere, J. Geophys. Res., 90, 3850–3868. [31] SABER temperature measurements from 2002 to Holton, J. R. (1983), The influence of gravity wave breaking on the general February of 2006 show that the mesopause altitude is higher circulation of the middle atmosphere, J. Atmos. Sci., 40, 2497–2507. at most latitudes by 1–2 km and the mesopause temperature Huaman, M. M., and B. B. Balsley (1999), Differences in near-mesopause warmer by about 7–8 K at middle and high latitudes around summer winds, temperatures, and water vapor at northern and southern latitudes as possible causal factors in inter-hemispheric PMSE differ- December solstice than June solstice (136.8 K and 129.2 K ences, Geophys. Res. Lett., 26, 1529–1532. at 80°). The temperature is also warmer at most altitudes Kovalam, S., R. A. Vincent, and P. Love (2006), Gravity waves in the between 20 km and the mesopause during austral summer equatorial MLT region, J. Atmos. Sol. Terr. Phys., 68, 266–282. Kutepov, A. A., A. G. Feofilov, B. T. Marshall, L. L. Gordley, W. D. than that during boreal summer. This difference is likely Pesnell, R. A. Goldberg, and J. M. Russell III (2006), SABER tempera- caused at least in part by the greater solar flux in December/ ture observations in the summer polar mesosphere and lower thermo- January than in June/July due to the Earth’s orbital eccen- sphere: Importance of accounting for the CO2 n2 quanta V-V exchange, Geophys. Res. Lett., 33, L21809, doi:10.1029/2006GL026591. tricity, as discussed by Chu et al. [2003]. The mesopause Leblanc, T., I. S. McDermid, P. Keckhut, A. Hauchecorne, C. Y. She, and altitude and temperature, especially during summer, are also D. A. Krueger (1998), Temperature climatology of the middle atmo- sensitive to troposphere gravity wave sources and gravity sphere from long-term lidar measurements at middle and low latitudes, wave filtering below the mesopause, and they could con- J. Geophys. Res., 103, 17,191–17,204. Leovy, C. B. (1964), Simple models of thermally driven mesospheric cir- tribute to the hemispheric asymmetry of the mesopause. culations, J. Atmos. Sci., 21, 327–341. There are still large uncertainties in quantifying the hemi- Liu, H.-L., and R. G. Roble (2002), A study of a self-generated strato- spheric differences of gravity waves from observations and spheric sudden warming and its mesospheric–lower thermospheric im- pacts using the coupled TIME-GCM/CCM3, J. Geophys. Res., 107(D23), numerical studies. 4695, doi:10.1029/2001JD001533. [32] We identify three sources of error that could affect Lu¨bken, F.-J. (1999), Thermal structure of the arctic summer mesosphere, the analysis of the SABER data: errors in the non-LTE J. Geophys. Res., 104, 9135–9150.

10 of 11 D09102 XU ET AL.: MESOPAUSE STRUCTURE D09102

Lu¨bken, F.-J., and A. Mu¨llemann (2003), First in situ temperature measure- Siskind, D. E., S. D. Eckermann, J. P. McCormack, M. J. Alexander, and ments in the summer mesosphere at very high latitudes (78°N), J. Geo- J. T. Bacmeister (2003), Hemispheric differences in the temperature of phys. Res., 108(D8), 8448, doi:10.1029/2002JD002414. the summertime stratosphere and mesosphere, J. Geophys. Res., 108(D2), Lu¨bken, F.-J., and U. von Zahn (1991), Thermal structure of the mesopause 4051, doi:10.1029/2002JD002095. region at polar latitudes, J. Geophys. Res., 96, 20,841–20,857. Vincent, R. A. (1994), Gravity-wave motions in the mesosphere and lower McLandress, C. (2002), The seasonal variation of the propagating diurnal thermosphere observed at Mawson, Antarctica, J. Atmos. Terr. Phys., 56, tide in the mesosphere and lower thermosphere. part I: The role of gravity 593–602. waves and planetary waves, J. Atmos. Sci., 59, 893–906. von Zahn, U., J. Ho¨ffner, V. Eska, and M. Alpers (1996), The mesopause McLandress, C., G. G. Shepherd, and B. H. Solheim (1996), Satellite altitude: Only two distinctive levels worldwide?, Geophys. Res. Lett., 23, observations of thermospheric tides: Results from the Wind Imaging 3231–3234. Interferometer on UARS, J. Geophys. Res., 101, 4093–4114. Yu, J. R., and C. Y. She (1995), Climatology of a midlatitude mesopause Mertens, C. J., M. G. Mlynczak, M. Lopez-Puertas, P. P. Wintersteiner, region observed by a lidar at Fort Collins, Colorado (40.6°N, 105°W), R. H. Picard, J. R. Winick, L. L. Gordley, and J. M. Russell III (2001), J. Geophys. Res., 100, 7441–7452. Retrieval of mesospheric and lower thermospheric kinetic temperature Zhu, X., J.-H. Yee, E. R. Talaat, M. Mlynczak, L. Gordley, C. Mertens, and from measurements of CO2 15-um Earth limb emission under non-LTE J. M. Russell III (2005), An algorithm for extracting zonal mean and conditions, Geophys. Res. Lett., 28, 1391–1394. migrating tidal fields in the middle atmosphere from satellite measure- Mertens, C. J., et al. (2004), SABER observations of mesospheric tempera- ments: Applications to TIMED/SABER–measured temperature and tidal ture and comparisons with falling sphere measurements taken during the modeling, J. Geophys. Res., 110, D02105, doi:10.1029/2004JD004996. 2002 summer MaCWAVE campaign, Geophys. Res. Lett., 31, L03105, doi:10.1029/2003GL018605. ÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀÀ Mlynczak, M. G., and S. Solomon (1993), A detailed evaluation of the H.-L. Liu and R. G. Roble, High Altitude Observatory, National Center heating efficiency in the middle atmosphere, J. Geophys. Res., 98, for Atmospheric Research, P.O. Box 3000, Boulder, CO 80307-3000, USA. 10,517–10,541. ([email protected]) Russell, J. M., III, M. G. Mlynczak, L. L. Gordley, J. Tansock, and R. Esplin C. J. Mertens and M. G. Mlynczak, NASA Langley Research Center, (1999), An overview of the SABER experiment and preliminary Hampton, VA 23681, USA. calibration results, Proc. SPIE Int. Soc. Opt. Eng., 3756, 277–288. J. M. Russell III, Department of Physics, Hampton University, Hampton, She, C. Y., and U. von Zahn (1998), Concept of a two-level mesopause: VA 23668, USA. Support through new lidar observations, J. Geophys. Res., 103, 5855– A. K. Smith, Atmospheric Chemistry Division, National Center for 5864. Atmospheric Research, Boulder, CO 80307, USA. She, C. Y., J. R. Yu, and H. Chen (1993), Observed thermal structure of a J. Xu and W. Yuan, State Key Laboratory of Space Weather, Chinese midlatitude mesopause, Geophys. Res. Lett., 20, 567–570. Academy of Sciences, P.O. Box 8701, Beijing 100080, China.

11 of 11