JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114, D06106, doi: /2008jd011089, 2009

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114,, doi: /2008jd011089, 2009 Observation of local tidal variability and instability, along with dissipation of diurnal tidal harmonics in the mesopause region over Fort Collins, Colorado (41 N, 105 W) Tao Li, 1,2 C.-Y. She, 3 Han-Li Liu, 4 Jia Yue, 3 Takuji Nakamura, 5 David A. Krueger, 3 Qian Wu, 4 Xiankang Dou, 1,2 and Shui Wang 1,2 Received 3 September 2008; revised 26 December 2008; accepted 5 January 2009; published 19 March [1] During the 9-day continuous campaign in September 2003, the Colorado State University sodium lidar observed significant short-term tidal variability in both diurnal and semidiurnal tides above 85 km on days Both diurnal and semidiurnal amplitudes dramatically increased on day 267 with a continuous phase advance in diurnal tidal harmonics, causing local atmosphere to become dynamically unstable. Following the dynamical instability associated with tides, we observed an equally dramatic decrease in diurnal amplitude, which was accompanied by rapid and continuous phase retardation at 87 km on day 268; the accompanying diurnal phase profiles changed from propagating mode to evanescent mode. Since the time scale of the observed variability during days is less than 1 day, gravity wave/tidal interaction at least is partially responsible for the observed variability. The observed changes in tidal wind amplitudes and phases were observed to correlate with gravity wave (GW) activities and the direction of GW momentum, exhibited and discerned concurrently by an OH all-sky imager at nearby Yucca Ridge station, consistent with well-known models of tidal/gw interactions. The stability analysis in the night of 267, when both diurnal and semidiurnal reached the maximum amplitudes, revealed that the running daily tidal waves alone, superimposed with the associated mean state, could force the atmosphere into local dynamical instability near km. The eddy diffusion associated with the instability is believed to have caused a strong dissipation of diurnal tide as observed on day 268. Citation: Li, T., C.-Y. She, H.-L. Liu, J. Yue, T. Nakamura, D. A. Krueger, Q. Wu, X. Dou, and S. Wang (2009), Observation of local tidal variability and instability, along with dissipation of diurnal tidal harmonics in the mesopause region over Fort Collins, Colorado (41 N, 105 W), J. Geophys. Res., 114,, doi: /2008jd Introduction [2] Atmospheric solar tides are the global-scale oscillations with periods of subharmonics of a solar day [Forbes, 1995]. For upward propagating waves, tidal amplitudes grow with altitude to compensate the exponential decrease of air density but also vary from one day to the next because of interactions with planetary waves (PWs) and/or local gravity waves (GWs). Tidal waves are the most persistent and dominant wave phenomena in the mesosphere and lower thermosphere (MLT). Their interactions with GWs [Walterscheid, 1981; Fritts and Vincent, 1987] and PWs 1 School of Earth and Space Sciences, University of Science and Technology of China, Hefei, China. 2 Mengcheng National Geophysical Observatory, Mengcheng, China. 3 Department of Physics, Colorado State University, Fort Collins, Colorado, USA. 4 High Altitude Observatory, National Center for Atmospheric Research, Boulder, Colorado, USA. 5 Research Institute for Sustainable Humanosphere, Kyoto University, Uji, Japan. Copyright 2009 by the American Geophysical Union /09/2008JD [Mayr et al., 2005] lead to significant atmospheric variability in general and particularly in tides themselves. The tidal/gw interactions usually affect tides in two different ways: GW momentum deposition and eddy diffusion. The momentum deposition by breaking GWs into mean flow could either damp or amplify the tides, depending on wavelength and phase speed of the wave [Walterscheid, 1981; Fritts and Vincent, 1987], as well as the direction of wave momentum carried by GWs [Ortland and Alexander, 2006]. If the GW momentum forcing is in phase (less than 90 ) with tidal wind, then it will amplify the tidal amplitude, and vice versa. As suggested, the strong GW forcing at tidal periods will alter the observed tidal amplitudes and phases [Fritts and Vincent, 1987; Ortland, 2005], and such changes, if significant, then imply the presence of tidal/gw interactions. The eddy diffusion generated by GW breaking is believed to damp the tide [Lindzen, 1981] and increase the vertical wavelength [Ortland and Alexander, 2006]. [3] On the other hand, tidal waves are thought to provide enhancement in background temperature gradient and/or wind shear, facilitating GW breaking [Hecht et al., 1997; Liu et al., 2004]. Although global tidal waves will not become critical or unstable, waves at tidal periods, resulting 1of10

2 Figure 1. (left) Amplitudes and (right) phases of zonal wind (a) diurnal and (b) semidiurnal on day 265 (dot), 266 (dash), 267 (dash dot), 268 (solid), and 9-day mean (black solid). The short solid lines indicate the phase slope with corresponding vertical wavelengths. from interactions with GWs and/or PWs, superimposed with mean flow, may become locally strong enough to cause local atmospheric instability or near instability [Sherman and She, 2006]. Consequently, turbulence could be possibly generated within the unstable region to prevent the tidal amplitude from increasing [Lindzen, 1981]. Indeed, a recent analysis with a general circulation model showed that superposition of all tidal components (both migrating and nonmigrating) and mean state could induce convective instability in some regions of the atmosphere [Ward et al., 2005]. [4] Although cases of tidal/gw [Fritts and Vincent, 1987] and tidal/pw [Oberheide et al., 2002] interactions have been observed, a time evolution of tidal wave variability, enhancement followed by subsequent decay, to our knowledge, has not yet appeared in the literature. This is mainly due to observational difficulties, since most lidars operate only at night, and thus are unable to define diurnal tide because of data gap aliasing [Crary and Forbes, 1983]. The Colorado State University (CSU) sodium lidar system located in Fort Collins, CO (40.6 N, W) is capable of simultaneously observing temperature, zonal and meridional winds in mesopause region on a 24-h weather permitting basis. During September 2003, the CSU sodium lidar conducted a 9-day continuous campaign between UT day 264 and 272. This unusually long data set revealed dramatic short-term tidal variability, with a substantial temperature inversion accompanied by strong wind shear on day 267, coupled with planetary wave activities [She et al., 2004]. In addition to a continuous record of tidal variability, this same data set revealed a dramatic tidal amplitude enhancement followed by subsequent decay, reported for the first time in this paper. [5] We thus present the short-term tidal variability, dramatic enhancement in tidal harmonics followed by dramatic dissipation in diurnal tide observed by the CSU sodium lidar between days 265 and 268 in 2003 in section 2, and discuss the tidal/gw interactions and tidal dissipation in section 3 with collaborative evidences from the OH all-sky imager observations at Yucca Ridge Field Station (40.7 N, W) 2of10

3 Figure 2. The vertical profiles of (left) temperature and (right) zonal wind between 0500 and 1100 UT on day 267, The raw data profiles are in Figures 2a and 2c, and the reconstructed data from the sum of running daily mean plus daily 24, 12, and 8 h tides are in Figures 2b and 2d presented at 15-min intervals, corresponding to 10 K in temperature or 25 m/s in wind speed. The vertical dotted lines denote the center of scale at 190 K for temperature and 0 m/s for zonal wind. The black solid line denotes the adiabatic lapse rate, while the dashed lines indicate downward progression of temperature inversion and wind shear. We also note here that the red and blue colors are only used alternatively to make the vertical profiles easier to be identified and do not mean different data sets. in Colorado (20 miles northeast of the lidar site). A conclusion follows in section Sodium Lidar Observations [6] The description of lidar system can be found in a recent paper [Li et al., 2007]. In brief, during the campaign in September 2003, the system receiver employed two 35-cm diameter Celestron telescopes pointing eastward and northward, both at 30 from zenith. The lidar transmitter directed two laser beams into the atmosphere; each aligned parallel to one telescope, 0.5 m from its axis. The signals received by each telescope are binned in 150-m-range bins; after 2-min integrations they are stored as raw photon count profiles. Depending on the signal-to-noise ratio of the data and the nature of application, the analysis programs further average the raw photon profiles over a longer time interval and over a larger vertical range. To compute the running daily mean and tidal amplitudes and phases for each hour, we decomposed the time series of raw data set within a 24 h window at each altitude, centered at the hour in question, into the mean plus 24, 12, and 8 h tidal period waves. The tides were evaluated at each hour for 4 consecutive days between days 265 and 268 to reveal the evolution of a significant tidal variability. [7] Figure 1 shows the vertical profiles of amplitudes (Figure 1, left) and phases (Figure 1, right) of daily zonal wind diurnal (Figure 1a) and semidiurnal (Figure 1b) tides for UT days , centered at 1200 UT, along with the associated 9-day mean. The diurnal amplitudes on days and 268 are typical, and generally the same as the 9-day mean amplitude, but with more wavy patterns. The zonal wind diurnal amplitude above 85 km dramatically increased by a factor of 2 3 from day 266 to 267, reaching 60 m/s above 94 km, and dramatically decreased, returning to normal on day 268. The vertical profiles of diurnal phase for days and 9-day mean clearly show the downward phase progression. The phases between 85 and 95 km on day 267 shifted forward by 3 h from day 266 with the vertical wavelength below 93 km (20 km, comparable to 9-day mean), much shorter than that above (60 km, much longer). However, there was no clear downward phase progression on day 268 for most of altitudes, suggesting the possible dominance of an evanescent mode accompanying the dramatic amplitude decrease. [8] Clearly different from the diurnal tide, the semidiurnal tides (Figure 1b) illustrate the downward phase progression for all individual days and the 9-day mean. Above 90 km, the phases on day 266 led those on day 265 by 3 h, whereas it was reversed for the phases below 90 km. There was not much phase shift from day 266 to 268 with the vertical wavelength below 91 km (30 km), clearly shorter than that above (70 km, comparable to 9-day mean). The semidiurnal amplitudes above 95 km in all individual days ( ) were stronger than that of 9-day mean. Between 87 and 93 km, the semidiurnal amplitude increased dramatically from less than 20 m/s on day 265 to 70 m/s on days 266 and 267, and then decreased considerably to 40 m/s on day 268. The amplitudes and phases of diurnal and semidiurnal in both temperature and meridional wind were 3of10

4 Figure 3. The vertical profiles on day 267, 2003, of (a) Brunt-Väisällä frequency square N 2, (b) Richardson number Ri calculated from the raw data profiles, and Ri profiles calculated from the superposition of (c) running daily mean plus daily 24 and 12 h tides, and (d) daily mean plus daily 24, 12, and 8 h tides presented at 15-min intervals, corresponding to s 2 in N 2 or 0.5 in Ri. The vertical dotted lines represent the instability threshold of 0 for N 2 or 0.25 for Ri. Note that the absolute values larger than s 2 in N 2 and 0.75 in Ri are not shown. The red and blue colors are only used alternatively to make the vertical profiles easier to be identified and do not mean different data sets. similarly examined (not shown), and the similar variations are also seen. Compared to diurnal and semidiurnal tides, the terdiurnal tide was generally weak. [9] Figure 2 shows temperature and zonal wind at 15-min intervals, between 0500 and 1100 UT on day 267 when both diurnal and semidiurnal amplitudes reached their maxima. Figures 2a and 2c show 15-min averaged raw data profiles, while Figures 2b and 2d show the reconstruction from the fitted running daily tides. In Figures 2a and 2c, between 90 and 95 km, we clearly observed correlated strong temperature inversion and strong zonal wind shear. Between 0600 and 0700 UT, the temperature lapse rate above the inversion layer was adiabatic (marked with black solid line) or near adiabatic. The downward progression (marked with dashed line) of the strong temperature inversion and strong zonal wind shear are clearly seen. While the strong temperature inversion gradually decreased with time, the zonal wind shear increased, reflecting the difference between temperature and zonal wind tidal phases on day 267. The reconstructed profiles show smooth tidal patterns similar to the raw data profiles. Comparing the vertical profiles of mean temperature and zonal wind with those reconstructed from tides only (not shown), we found that the tidal period harmonics were the dominant features in the observation, and that the strong temperature inversion and wind shear were clearly associated with the tidal waves, at least locally. [10] The strong wind shear and low convective stability above the temperature inversion layer could lead to dynamical instability in this region. Figure 3 shows the Brunt- Väisällä frequency square N 2 and Richardson number Ri for the same duration on day 267. The atmosphere in the mesopause region was generally convectively stable (N 2 >0), except between 0600 and 0630 UT near 93 km. Within the error bars, the downward progression of dynamical instability (0 < Ri < 0.25) clearly associated with tides was found between 0630 and 0900 UT near km in both raw (Figure 3b) and reconstructed Richardson Numbers fromrunningdailymeanplus24,12,and8hcomponents (Figure 3d). After 0900 UT, the Richardson numbers were very close to dynamical instability threshold of 0.25 and continued to progress downward. Even the Richardson numbers reconstructed from mean plus 24 and 12 h without 8 h components (Figure 3c) were also very close to the dynamical instability near km, though downward phase progression was not as clear. The coherent linear superposition of strong running daily tidal period waves alone with the running mean could cause dynamical instability in a certain altitude range, consistent with recent model analysis [Ward et al., 2005] of tidal waves that lead to convective instability. In contrast, the Richardson numbers determined from the reconstruction of the 9-day mean and tidal fields (fitting the entire 9-day data set), were found to be greater than 0.75 at all altitudes. 3. Discussion [11] Earlier, She et al. [2004] suggested that the dramatic tidal amplitude increase from day 265 to 267 could be the result of tidal interactions with PWs and/or GWs. A recent TIME-GCM simulation [Liu et al., 2007] showed that the 4of10

5 Figure 4. Temporal evolution of running daily zonal wind tidal (left) amplitudes and (right) phases at 87 km of (a) diurnal tide and (b) semidiurnal tide. The horizontal solid line in each figure represents the 9-day mean tidal amplitude or phase at 87 km. nonlinear interaction between migrating diurnal tide and a transient s = 1 quasi stationary planetary wave could modulate the migrating diurnal tide and generate strong nonmigrating diurnal tides, and further induce the shortterm diurnal variability that led to tidal amplitude increase similar to that observed on day 267. However, the increase of simulated diurnal zonal wind amplitude from day 266 to 267 is only about half of the observed values. The enhanced temperature inversion and wind shear due to tidal/pw interactions could strongly impact the propagation of shortperiod local GWs. As a result, the interactions of GWs with the tidal modulated mean state may in turn contribute significantly to tidal amplitude variability, in addition to tidal/pw interactions Evidences of Tidal/GW Interactions [12] Since the observed variability of the running daily tidal amplitudes occurred at a time scale shorter than one day, GWs should have played a significant role in the observed tidal modulation and variability. To illustrate the time scale of such variability, we show the time series of zonal wind amplitude and phase of diurnal and semidiurnal tides at 87 km in Figure 4. It is clear that the diurnal tidal amplitude and phase varied considerably more than the semidiurnal tidal amplitude and phase. The diurnal amplitude varied between 12 m/s and 44 m/s in the first 3 days accompanied by a phase advance from 17 h to 10 h between 1200 UT, day 265 and 0000 UT, day 268. It then dramatically decreased from 40 m/s on day 267 to as low as 5 m/s on day 268 with a diurnal phase increase from 10 h to 23 h continuously from the beginning to the end of day 268. The evolution of semidiurnal tide is simpler; it increased from 20 m/s on day 265 to 70 m/s at 0000 UT, day 267, and then decreased significantly to 40 m/s on day 268. The corresponding phase increased (retardation) from 3 h before 1300 UT, day 256 to 6 hat0000 UT, 266, and stayed more or less at this value for days [13] The observed behaviors shown in Figures 4 and 1 are consistent with the expectation of tidal/gw interaction reported in the literature. Walterscheid [1981] showed that the interactions of semidiurnal tidal harmonics with GWs via the mean wind filtering process could induce amplitude 5of10

6 enhancement, and that the resulting semidiurnal tide has shorter vertical wavelength in the region below the maximum amplitude and longer above. Our observations of semidiurnal tide on days with much shorter vertical wavelength (30 km) below 91 km than that above (70 km, comparable to 9-day mean value) are clearly consistent with his model prediction. This further suggests that in addition to semidiurnal/pw interactions, the observed increase in semidiurnal amplitude below 95 km from day 265 to 266 and 267 as shown in Figure 1b could be partially due to semidiurnal tidal harmonics/gw interactions. Walterscheid [1981] also stated that similar vertical wavelength changes should occur as the result of diurnal tidal harmonic/gw interaction. In Figure 1a (right), we observed 2 3 times longer vertical wavelength of diurnal tide above 91 km than that below 91 km (comparable to the 9-day mean value) on days 266 and 267. More dramatic is the observed diurnal phase change from propagating mode on day 267 to nearly evanescent mode on day 268, accompanied by a dramatic diurnal tidal amplitude decrease as shown in Figure 1a (left), suggesting that the dynamical instability associated with tides observed on day 267 as shown in Figure 3d most likely triggered the dramatic dissipation of local diurnal tide. [14] Radar observations of GW momentum flux at Adelaide, Australia [Fritts and Vincent, 1987] revealed a strong diurnal modulation in the momentum flux and flux divergence, which in turn can act to alter the observed tidal structure; Fritts and Vincent [1987] found diurnal tidal amplitude decrease accompanied by a phase shift of 6 hat90 km during the last 3 days of the 8-day campaign when the observed GW momentum flux increased 5 fold inan 8-h period centered at midnight. A simple model of tidal-gw interaction imposing westward zonal wind acceleration due to GW momentum flux divergence peaked at local midnight was employed, and it adequately explained the observed features [Fritts and Vincent, 1987]. Generally speaking, since the GWs sense the tidal harmonics as part of the background wind system, the resulting feedback on the change of tidal amplitudes (attenuation or amplification) and phases (advances or retardations) will depend on the GW amplitudes and phase speeds, and their forcing relative to the tidal structure and the mean local environment. Thus, though a quantitative prediction of the resulting effects in local tidal changes requires further knowledge on the amplitude and spectrum of the GWs involved, the observation of variability in the time scale of hours accompanied by dramatic tidal amplitude and phase changes is a possible indication of GW/tide interaction. In this connection, we note the increase in semidiurnal tidal amplitude at 87 km from day 265 to day 267, as shown in Figure 4b, is accompanied by a phase retardation of 3 h in the second half of day 265. More dramatic in Figure 4b is the considerable variation in diurnal tidal amplitude at 87 km from day 265 to a maximum of over 40 m/s on day 267, accompanied by a phase advancement of 3 h per day. This progressive phase advancement between day 265 and 267 can also be seen to exist between 86 and 93 km as shown in Figure 1a. Again, the more eye-catching decrease of diurnal tidal amplitude accompanied by dramatic phase changes (from propagating mode on day 267 to evanescent mode on day 268) as shown in Figures 1a and 4a (right) is likely related to the breaking of local tidal harmonics triggered by the observed local dynamical instability as will be discussed further below GW Activities Revealed by Collocated OH All-Sky Imager and Sodium Lidar [15] A recent model study [Ortland and Alexander, 2006] suggested that strong GW momentum forcing can result in the reduction of tidal vertical wavelength, with the resultant tidal amplitude either amplified or reduced by GW momentum forcing depending on the direction of breaking shortperiod GWs relative to the tidal wind. In Figure 1, we did observe the short vertical wavelength in an altitude region with clear phase advance, consistent with their prediction. Since the majority of momentum flux in the MLT region is carried by GWs with period shorter than 1 h [Fritts and Vincent, 1987], we need to identify these short-period GWs and their propagation direction; unfortunately, because of signal-to-noise limitations, we are not able to deduce shortperiod (less than 30 min) GW activities from the lidar observations. However, a collocated OH all-sky imager was in operation during this lidar campaign, which could reveal not only the short-period (down to near buoyancy period, 5 min) GWs in a large horizontal area, but also the small-scale ripples, known as the signature of local GW instability [Hecht, 2004; Li et al., 2005a; Li et al., 2005b]. Though the wave patterns in the OH all-sky imagers are often complex, the propagation direction of dominant GWs can usually be discerned as depicted in Figure 5, facilitating discussion of the observed tidal variability in relation to this recent study [Ortland and Alexander, 2006]. [16] During the lidar campaign between day 265 and 268, the OH all-sky imager located at Yucca Ridge Field Station (20 miles northeast of lidar site) observed the significant GW activities as well as ripple events. Shown in Figure 5 are the 2-min OH flat-fielded images taken near 0700 UT on days (left, from top to bottom), and their corresponding tidal wind hodographs (right) at 87 km from 0230 to 1230 UT for each day. On day 266, for example, we clearly see that the dominant GWs with period of 8 min and horizontal wavelength of 70 km propagated to northeast (blue arrow) with a number of ripples aligning either 45 or 90 with wavefronts. The lidar wind hodograph suggests that the diurnal and semidiurnal tidal horizontal wind at 87 km near 0700 UT (local midnight) during ripple presence were near westward and northward (red and blue dashed arrows) respectively. As a result, the GW momentum forcing (black arrow on right panel or blue arrow on left panel) is out of phase (>90 ) with the diurnal tidal wind and in phase (<90 ) with the semidiurnal tidal wind, leading to suppression of the diurnal amplitude and amplification of the semidiurnal amplitude, respectively, if these waves were breaking because of either wind filtering or instability. [17] Although the diurnal amplitude increased quite dramatically from day 266 to 267, the GW activities in the following night (267) were observed to be weaker than those on days 266 and 268. To closely examine the GW activities in the night of 267, we plotted in Figure 6 the OH flat-fielded images at 0254 (Figure 6a), 0346 (Figure 6b), 0500 (Figure 6c), 0600 (Figure 6d), 0700 (Figure 6e), and 0800 UT (Figure 6f) on day 267. It s clear that the dominant GWs with period of 6 min and the horizontal wavelength of 60 km propagated to the southwest; and the GW 6of10

7 Figure 5. The OH flat-fielded images near 0700 UT on day (a) 265, (b) 266, (c) 267, and (d) 268 with (left) 2-min integration and (middle) 10-min integration, and (right) their corresponding tidal wind hodographs at 87 km from 0230 to 1230 UT deduced from running daily tides shown in Figure 4 for each day. The red lines in OH images denote wavefront, and blue arrows denote wave propagation direction. In the right panels, the black arrows denote the direction of GW momentum forcing. The blue and red dashed lines with arrows denote the wind directions for semidiurnal and diurnal tides, respectively. Note that there were ripples near 0700 UT on days 266 and of10

8 Figure 6. The OH flat-fielded images at (a) 0254, (b) 0346, (c) 0500, (d) 0600, (e) 0700, and (f) 0800 UT on day 267. The short green bars denote the observed ripple. The red lines in OH images denote wavefront, and blue arrows denote wave propagation direction. activities gradually decreased with time. The ripples with lifetime less than 10 min and horizontal scale of 10 km were observed near the CSU lidar north beam in both frames at 0254 and 0346 UT, similar to those observed in the September 2003 cases over northern Colorado [Li et al., 2005b]. The momentum carried by this wave packet was in the same direction as both diurnal and semidiurnal wind (Figure 5c, right). As a result, the saturation or breaking (as evidenced by the presence of ripples) of this GW packet would amplify both diurnal and semidiurnal tides before 0700 UT. On the other hand, the weaker GWactivities in the night of 267 suggest smaller eddy diffusion compared to other nights; as such, the tidal amplitudes may not be attenuated much by eddy diffusion, leading to increased tidal amplitudes on day 267, in a situation similar to a recent model simulation of tidal/gw interactions [Liu et al., 2008]. We are unable to ascertain whether the lack of short-period GWs after 0800 UT on day 267 was the result of filtering process by the enhanced tidal winds, or the enhanced tidal wind was the result of weaker GWactivities and smaller eddy diffusion. Also, since the OH imager does not operate in the daytime; we are not able to observe interaction of tidal harmonics with GWs which may occur during daytime. [18] The GW forcing generated by dominant waves in the night of 268 was out of phase with both diurnal and semidiurnal tides near 0700 UT when the ripples appeared (as shown in Figure 5d, right), indicating that the tidal/gw interactions on day 268 contributed to the decrease of both diurnal and semidiurnal amplitudes. Figures 5a 5d (middle) show the 10-min averaged OH flat-fielded images taken near 0700 UT on days (from top to bottom). They revealed a lack of wave activities with periods longer than 20 min and horizontal wavelengths shorter than 500 km around 0700 UT. The exception is on night 266, when the dominant GW, with period 20 min and horizontal wavelength of 120 km, propagated in phase (out of phase) with the semidiurnal (diurnal) tidal wind. [19] Table 1 summarizes GW activities in the 4 nights observed by the OH imager, showing the complexity of short-period GWs. The relative propagation directions between GWs and tidal winds suggest that energy can flow one way or the other, though the majority of GWs propagated in phase with tidal wind between day 265 and 267, leading to tidal amplitude amplification, while after the dynamical instability associated with local tidal period waves on day 267, the observed GWs generally propagated out of phase with the tidal winds, helping to carry away some of the excess momentum and energy arising from local tidal wave dissipation. We also note that to quantitatively study the GW forcing and horizontal tidal wind acceleration, we need to measure the vertical profile of momentum flux and horizontal wind simultaneously using sodium lidar. In this 8of10

9 Table 1. Dominant GW Activities as Revealed by OH All-Sky Imager a UT Day Duration Dominant GW Propagation Direction Period (min) Horizontal Wavelength (km) In or Out of Phase With DT b In or Out of Phase With SDT b UT 30 West of 4 50 In Phase Out of Phase South UT North 6 40 Out of Phase Out of Phase UT Northeast 8 70 Out of Phase In Phase from significant Near north UT on ripples appear from 0600 UT UT East 6 50 Out of Phase In Phase UT Southwest In Phase In Phase before UT Southwest 6 40 In Phase before 0500 UT, out of phase after 0700 UT In Phase before 0600 UT, out of phase after UT Southeast 6 50 Out of Phase Out of Phase a Measured at 87 km. DT denotes diurnal tide; SDT denotes semidiurnal tide. b In (out of phase) phase if the angle between tidal wind and GW propagation direction is less (more) than 90. paper, we only present the qualitative study on the GW momentum flux obtained from OH imager measurements. The quantitative study of momentum flux divergence and associated mean wind change will be the next step. [20] To shed some light on the activities of GWs with periods between 30 min and 4 h, we calculate the averaged GW wind variances within this frequency band for each day from the nighttime lidar data. As shown in Figure 7, we see that the GW wind variances decreased from day 265 to 266 and to 267 as tidal amplitudes increased accordingly. Similar to the higher-frequency components revealed by the OH imager, the GW wind variances with periods between 30 min and 4 h showed clear increase in response to the strong tidal dissipation on day 268. The hodograph method is able to determine the horizontal propagation direction of inertia-gravity waves [Tsuda et al., 1994], but only works well for the data set embedded with single dominant wave. Unfortunately our lidar data set indicated multiple waves with various periods and vertical scales (not shown), so the results derived from this method may not be correct Tidal Period Wave Induced Dynamical Instability and Dissipation of Diurnal Tide [21] Our observation also found an evanescent diurnal phase on day 268, clearly different from the downward phase progression on days , while, the semidiurnal phase plot on day 268 was almost the same as that on days and also in (not shown). There is a possibility that the locally observed evanescent phase reflects the local phase of the wave source (planetary wave-tidal interaction or solar heating). Since the dramatic dissipation occurred within one day, however, it is most likely that the turbulence generated by dynamical instability within the tidal period wavefield dissipates much of the tidal energy, more so in the diurnal tide because of its shorter vertical wavelength as compared to the semidiurnal tide. By superimposing all tidal components (both migrating and nonmigrating) on the mean state from a model simulation, Ward et al. [2005] revealed the existence of convective instability in some local regions, resulting from tidal harmonics alone. Our lidar observation shown in Figure 3 revealed the existence of dynamical instability with downward progression near km for many hours resulting from the linear superposition of running daily mean plus daily tidal period waves. The diurnal tide with a much shorter vertical wavelength [Forbes, 1995] could generate larger negative temperature gradient and stronger wind shear when superimposed on the mean flow, and thus is more susceptible to dissipation in the mesopause region than the semidiurnal tide with a longer vertical wavelength. Similar to inertiagravity wave dissipation [Fritts and Alexander, 2003], the turbulent mixing generated within the local instability region by tidal wave saturation could further reduce the wind shear and wave amplitudes, and thus recover the atmosphere stability, resulting in much smaller diurnal Figure 7. The averaged GW wind variances for the band between 30 min and 4 h (hu 02 i + hv 02 i)/2.0 for 4 consecutive nights from day 265 to 268 calculated with 15 min and 2 km resolution profiles. 9of10

10 tidal amplitude in the mesopause region as was observed on day Conclusion [22] The unusually long data set of mesopause region temperature and horizontal wind, observed by the CSU sodium lidar during September 2003, revealed significant short-term tidal variability with dramatic enhancement of both diurnal and semidiurnal tides above 85 km on day 267, followed by a rapid decrease on day 268. Although tidal/ PW interactions may partially contribute to the short-term tidal variability observed, we have confirmed, using GW observations by an OH imager and employing simple models from the literatures, the importance of tidal/gw interactions to these dynamics. [23] High-frequency GW activity with significant ripples were observed by a collocated OH all-sky imager at nearby Yucca Ridge Field Station on both days 266 and 268, whereas weaker activity was observed on day 265, and weaker yet activity was seen on day 267, when ripples were not apparent after 0800 UT. The propagation direction of observed dominant GWs relative to that of the diurnal and semidiurnal tidal wind suggests an overall tidal amplitude increase on days , and an overall tidal amplitude reduction on day 268, consistent with the lidar-observed dramatic tidal variability during this period. [24] The instability analysis on day 267, when both diurnal and semidiurnal period waves reached maximum amplitudes, suggests that the running daily tidal period waves alone, superimposed with the associated mean state, were able to push the atmosphere into local dynamical instability near km. The eddy diffusion generated in the local instability region and out of phase GW forcing corroborated well with significant dissipation of the tidal period harmonics, and were thereby responsible for the dramatic decreases in tidal period wave amplitudes observed on day 268, particularly in the shorter vertical wavelength diurnal component. [25] Acknowledgments. The work described in this paper was carried out at the University of Science and Technology of China, under support of One Hundred Talent program from Chinese Academy of Sciences (CAS) and CAS KIP Pilot Projects kzcx2-yw-123. The work at the Colorado State University was supported by the National Aeronautics and Space Administration under grant NNX07AB64G and the National Science Foundation under grants ATM and ATM The work at Kyoto University is supported by grants For Scientific Research (B) and from NEXT, Japan. The National Center for Atmospheric Research is operated by the University Corporation for Atmospheric Research under the sponsorship of the National Science Foundation. The authors acknowledge the contribution of Tao Yuan, Bifford P. Williams, Takuya D. Kawahara, Joe D. Vance, and Phil Acott who participated in data acquisition during the September 2003 lidar campaign. The authors would also like to thank three anonymous reviewers for their constructive comments and suggestions. References Crary, D. J., and J. M. Forbes (1983), On the extraction of tidal information from measurements covering a fraction of a day, Geophys. Res. Lett., 10, , doi: /gl010i007p Forbes, J. M. (1995), Tidal and planetary waves, in The Upper Mesosphere and Lower Thermosphere: A Review of Experiment and Theory, Geophys. Monogr. Ser., vol. 87, edited by R. M. Johnson and T. L. Killeen, pp , AGU, Washington, D. C. Fritts, D. C., and M. J. Alexander (2003), Gravity wave dynamics and effects in the middle atmosphere, Rev. Geophys., 41(1), 1003, doi: /2001rg Fritts, D. C., and R. A. Vincent (1987), Mesospheric momentum flux studies at Adelaide, Australia: Observations and a gravity wave tidal interaction model, J. Atmos. Sci., 44, , doi: / (1987)044<0605:mmfsaa>2.0.co;2. Hecht, J. H. (2004), Instability layers and airglow imaging, Rev. Geophys., 42, RG1001, doi: /2003rg Hecht, J., R. Walterscheid, D. Fritts, J. Isler, D. Senft, C. Gardner, and S. Franke (1997), Wave breaking signatures in OH airglow and sodium densities and temperatures 1. Airglow imaging, Na lidar, and MF radar observations, J. Geophys. Res., 102, , doi: / 96JD Li, F., A. Z. Liu, and G. R. Swenson (2005a), Characteristics of instabilities in the mesopause region over Maui, Hawaii, J. Geophys. Res., 110, D09S12, doi: /2004jd Li, T., C. Y. She, B. P. Williams, T. Yuan, R. L. Collins, L. M. Kieffaber, and A. W. Peterson (2005b), Concurrent OH imager and sodium temperature/wind lidar observation of localized ripples over northern Colorado, J. Geophys. Res., 110, D13110, doi: /2004jd Li, T., C.-Y. She, H.-L. Liu, and M. T. Montgomery (2007), Evidence of a gravity wave breaking event and the estimation of wave characteristics from sodium lidar observation over Fort Collins, CO (41 N, 105 W), Geophys. Res. Lett., 34, L05815, doi: /2006gl Lindzen, R. S. (1981), Turbulence and stress owing to gravity wave and tidal breakdown, J. Geophys. Res., 86, , doi: / JC086iC10p Liu, A. Z., R. G. Roble, J. H. Hecht, M. F. Larsen, and C. S. Gardner (2004), Unstable layers in the mesopause region observed with Na lidar during the Turbulent Oxygen Mixing Experiment (TOMEX) campaign, J. Geophys. Res., 109, D02S02, doi: /2002jd Liu, H.-L., T. Li, C.-Y. She, J. Oberheide, Q. Wu, M. E. Hagan, J. Xu, R. G. Roble, M. G. Mlynczak, and J. M. Russell III (2007), Comparative study of short-term diurnal tidal variability, J. Geophys. Res., 112, D18108, doi: /2007jd Liu, X., J. Xu, H.-L. Liu, and R. Ma (2008), Nonlinear interactions between gravity waves with different wavelengths and diurnal tide, J. Geophys. Res., 113, D08112, doi: /2007jd Mayr, H. G., J. G. Mengel, E. R. Talaat, H. S. Porter, and K. L. Chan (2005), Mesospheric non-migrating tides generated with planetary waves: I. Characteristics, J. Atmos. Sol. Terr. Phys., 67, , doi: / j.jastp Oberheide, J., M. E. Hagan, R. G. Roble, and D. Offermann (2002), Sources of nonmigrating tides in the tropical middle atmosphere, J. Geophys. Res., 107(D21), 4567, doi: /2002jd Ortland, D. A. (2005), Generalized Hough modes: The structure of damped global-scale waves propagating on a mean flow with horizontal and vertical shear, J. Atmos. Sci., 62, , doi: /jas Ortland, D. A., and M. J. Alexander (2006), Gravity wave influence on the global structure of the diurnal tide in the mesosphere and lower thermosphere, J. Geophys. Res., 111, A10S10, doi: /2005ja She, C. Y., et al. (2004), Tidal perturbations and variability in mesopause region over Fort Collins, CO (41N, 105W): Continuous multi-day temperature and wind lidar observations, Geophys. Res. Lett., 31, L24111, doi: /2004gl Sherman, J. P., and C.-Y. She (2006), Seasonal variation of mesopause region wind shears, convective and dynamic instabilities above Fort Collins, CO: A statistical study, J. Atmos. Sol. Terr. Phys., 68, , doi: /j.jastp Tsuda, T., Y. Murayama, H. Wiryosumarto, S. W. B. Harijono, and S. Kato (1994), Radiosonde observations of equatorial atmosphere dynamics over Indonesia 2. Characteristics of gravity waves, J. Geophys. Res., 99, 10,507 10,516, doi: /94jd Walterscheid, R. L. (1981), Inertio-gravity wave induced accelerations of mean flow having an imposed periodic component: Implications for tidal observations in the meteor region, J. Geophys. Res., 86, , doi: /jc086ic10p Ward, W. E., V. I. Fomichev, and S. Beagley (2005), Nonmigrating tides in equinox temperature fields from the Extended Canadian Middle Atmosphere Model (CMAM), Geophys. Res. Lett., 32, L03803, doi: / 2004GL X. Dou, T. Li, and S. Wang, School of Earth and Space Sciences, University of Science and Technology of China, Hefei , China. (litao@ustc.edu.cn) D. A. Krueger, C.-Y. She, and J. Yue, Department of Physics, Colorado State University, 200 West Lake Street, Fort Collins, CO , USA. H.-L. Liu and Q. Wu, High Altitude Observatory, National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO , USA. T. Nakamura, Research Institute for Sustainable Humanosphere, Kyoto University, Uji , Japan. 10 of 10