By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 Extreme longitudinal disturbances in the O(1S) emission rate and horizontal winds in the mesosphere and thermosphere (90–180 km) are identified on three March days in 1994, 1996, and 1997 in the database of the Wind Imaging Interferometer (WINDII) on the Upper Atmosphere Research Satellite (UARS). This type of disturbance has a wave structure with apparent zonal wavenumber 1 and a very long vertical wavelength. It generates longitudinal amplitudes of 70 photons cm−3 s−1 and 120 ms−1 in the O(1S) volume emission rate and winds, respectively, and is strong enough to break the structures of migrating tides in the E-region. In the same longitudinal zone for day and night the zonal wind driven by this disturbance has the same phase, suggesting a wave with s = −1 and a very long period, but the meridional wind has opposite phases, suggesting a standing non-migrating diurnal tide. This super disturbance has not been observed before, and fundamental questions about it await future investigation.
 Measurements by a space-based instrument on a single day at a given latitude correspond in most cases to two local solar times (LT), one in daytime and the other at night. Since longitudinal measurements are normally at regular intervals the longitudinal average for each LT is refereed to as the zonal mean, which is the composite of mainly the background atmosphere and the migrating tide. The values after removing the zonal mean provide longitudinal variations, which are generated by processes other than the migrating tide.
 The dynamics of the background winds and the migrating tides in the mesosphere and lower thermosphere (MLT) have been intensively studied using the zonal mean winds measured by the Wind Imaging Interferometer (WINDII) [Shepherd et al., 1993a] and the High Resolution Doppler Imager (HRDI) [Hays et al., 1993] on the Upper Atmosphere Research Satellite (UARS) launched on September 12, 1991. The climatology of monthly mean winds and diurnal and semidiurnal migrating tides in the region of 90–120 km and 40°S–40°N using WINDII winds has been recently published [Zhang et al., 2007]. Normally the migrating tides are robust and dominate the MLT region of 40°S–40°N. The structures of the emission rate and winds generated by the migrating tide are dependent on latitude. For example, at equinox around midnight the atomic oxygen 557.7 nm (O(1S)) emission rate has an equatorial minimum; the zonal wind is westward in the equatorial region and eastward at mid-latitudes; the meridional wind has opposite directions in the two hemispheres, northward in the northern hemisphere and southward in the southern hemisphere [i.e., Zhang and Shepherd, 1999; McLandress et al., 1996a; Zhang et al., 2007].
 Longitudinal variations of the airglow emission rate and horizontal winds seen in the measurements of WINDII and HRDI have also drawn attention to the effects of non-migrating tides, planetary waves, gravity waves, and solar storms in the MLT region. From a climatological point of view Forbes et al. [2003a, 2003b] studied non-migrating diurnal tides using HRDI winds at 95 km and model simulations. They found three most prominent non-migrating diurnal tides: the eastward and westward propagating tides with zonal wave numbers 3 and 2 (DE3 and DW2), respectively, and the standing tide (D0), all with maximum amplitudes of 7–15 ms−1. Oberheide et al.  reported a maximum amplitude of DW2 of about 20 ms−1 observed by the TIMED Doppler Interferometer on the Thermosphere Ionosphere Mesosphere Energetics and Dynamics satellite (TIMED). Planetary waves were also frequently observed. Talaat et al.  reported the 6.5-day wave with zonal wavenumber 1 and amplitudes of about 30 ms−1 in both the zonal and meridional components observed by HRDI at 95 km.
 Events of longitudinal fluctuations larger than the tidal variations have also been observed. Shepherd et al. [1993b] reported that the O(1S) emission rate measured by WINDII varied from about 30 Rayleighs (R) at the minimum to 300 R at the maximum at 40°N on January 16, 1992. Also from WINDII measurements Zhang et al.  presented longitudinal max/min ratios (10–20) of the nighttime O(1S) zenithal column emission rate that were much larger than the tidal ratios (2–3) in the tropical region. Hays et al.  reported that the O2 atmospheric band nightglow emission measured by HRDI varied from a few kR to 20 kR on single days in March–April, 1998. Ward et al.  reported on a two-day wave with a meridional amplitude of 50 ms−1 during January, 1993.
 WINDII measurements from the O(1S) emission on some days provide the volume emission rate (VER) in photons cm−3s−1 and horizontal winds for both day and night. The daytime data have a vertical range of 90–180 km and the nighttime data mostly fall within altitudes of 90–110 km. Measurements of both emission rate and wind for both day and night provide wider information in a broader region than most works on longitudinal disturbances mentioned above. The focus of this study is on a global phenomenon of extreme longitudinal variations observed with WINDII on March 18, 1994, March 13, 1996, and March 4, 1997.
2. Extreme Longitudinal Disturbance
 Data used in this study are from single days and are binned in intervals of 5 degrees of latitude, 24 degrees of longitude, and 2 km and 5 km of altitude below and above 120 km, respectively, measured by WINDII.
 Extreme longitudinal variations on a global scale are observed on March 18, 1994 from both daytime and nighttime measurements. Similar disturbances are also observed on March 13, 1996 and March 4, 1997 (not shown). It should be kept in mind that the migrating diurnal tide has an annual maximum around the March equinox [i.e., McLandress et al., 1996a; Zhang et al., 2007]. As well, it is the time of a sudden transition from winter to summer in the northern hemisphere and vice versa in the southern hemisphere [Shepherd et al., 2004].
Figure 1 displays horizontal structures of the VER and winds, as well as the corresponding data after removing the zonal means, which are refereed to as the residuals herein, at 95 km on March 18, 1994. The vertical structures of the residual VER and winds at 90–110 km on March 18, 1994 are shown in Figure 2 at 25°N, the equator, and 25°S.
 The daytime VER increases from about 130 photons cm−3 s−1 at 40°N to 660 photons cm−3 s−1 at 40°S due to the change of the solar zenith angle [Zhang and Shepherd, 2005], since the LT of the data changes from dawn at 40°N to noon at 40°S on that day. The distribution of the VER during nighttime maintains some tidal features as mentioned above but are severely distorted. Significant disturbances are shown in both the zonal and meridional winds, in which the tidal wind patterns are broken and the longitudinal fluctuation has a magnitude of 140 ms−1. The disturbance is particularly strong in the zonal wind; there is a clear change of wind direction from westward to eastward and vice versa at all latitudes (Figures 1b).
 In a local time fixed frame, the apparent zonal wavenumber of a planetary wave or a non-migrating tide is given by s − n, where s is the actual zonal wavenumber and n = 24/period (hours). After removing the zonal mean, the residual winds of both daytime and nighttime show a wave structure with apparent zonal wavenumber ±1 (i.e., ∣s − n∣ = 1) in both components (Figures 1e, 1f, and 2). The magnitudes of the longitudinal fluctuations are about 110 ms−1 in both wind components and about 70 photons cm−3 s−1 in the emission rate, which are significantly larger than the averaged migrating tidal amplitudes of about 60 ms−1 in winds and about 50 photons cm−3 in the O(1S) VER for the March equinox period [Zhang et al., 2007; Zhang and Shepherd, 2008].
 The phase of the wave apparently changes with latitude (indicated by the solid lines in Figures 1e and 1f), which should not be mistaken as the real phase change. It is most likely caused by the change of the local time of the measurements. On that day from 40°N to 40°S the LT of the data increases from 6.2 to 11.6 LT during daytime and decreases from 4.1 through midnight to 22.6 LT during nighttime. The phase moves eastward with increasing LT, indicating s−n = −1 (i.e., s = 0 and n = 1 for a diurnal non-migrating tide, or s = 1 and n = 2 for a semidiurnal non-migrating tide).
 What is striking is the behavior of the phases of the two wind components. The zonal component has similar phases day and night. The local time difference of the daytime and nighttime data is about 11.8 hours at 25°S, indicating a semidiurnal non-migrating tide, but it is about 6.5 hours at 25°N, showing that it cannot be a semidiurnal non-migrating tide (Figures 1e and 2). Overall, the phases of the residual zonal wind suggest that it is not a tide but a wave with s = −1 and a very long period (n ∼ 0). On the other hand, the meridional component has opposite phases day and night (Figures 1f and 2). From the local time differences of daytime and nighttime data the phases of the residual meridional wind suggest a standing non-migrating diurnal tide (D0, i.e., s = 0, n = 1). The independent phase behaviors of the residual zonal and meridional winds are also repeatedly observed on March 13, 1996 and March 4, 1997 in almost the exactly same manner. It is hard to conceive that the perturbations in the two wind components are generated by different forcings.
 The longitudinal variations in the emission rate are not as well structured as in the winds. From Figure 1d, during daytime there is a wave with apparent zonal wavenumber 1 at mid-latitudes in both hemispheres in agreement with the winds, but in the tropical region, the variation is small without a clear pattern; at night there are bright and dim patches with similar sizes, which give different longitudinal variations at different latitudes.
 The vertical wavelength seen in both wind components is long, almost evanescent (Figure 2). The daytime VER exhibits vertical variations, which are not seen in the winds and indicate a vertical wavelength of about 25–30 km. Similar daytime vertical variations of the residual VER are also observed on March 13, 1996 and March 4, 1997 (not shown). The longitudinal variations of the nighttime VER at the three latitudes have no resemblance. At 25°N below 100 km the fluctuation is large with apparent zonal wavenumber 1 in agreement with the winds, but at the equator and 25°S there is not a clear pattern. It should be noticed that the apparent structure of the residual VER is not the actual wave behavior, because the emission rate is affected by the densities of atomic oxygen, molecular oxygen and nitrogen, temperature, and winds. During the daytime, it is also affected by solar irradiance. Wave effects on different elements have different phases, therefore, the observed structure of the emission rate is the combination of all the elements involved in producing the airglow so should be different from the structure of each single element, such as the zonal or meridional wind.
 A broader vertical picture extending to 180 km can be obtained from the daytime measurements. Figure 3 displays distributions of the measured and the residual VERs, and zonal and meridional winds at 40°S and 11.5 LT in the region of 90–180 km. There are two O(1S) emission layers during daytime, the E-layer and the F-layer, peaking at about 100 and 150 km, respectively. A wave with apparent zonal wavenumber 1 and a long vertical wave length is seen in the F-region in the emission rate as well in the two wind components. The maximum amplitude of the residual VER is about 70 photons cm−3 s−1 at about 160 km, and the maximum amplitudes of the residual zonal and meridional winds are about 100 ms−1 and 60 ms−1, respectively. Horizontal structures at 155 km are shown in Figure 4. Normally meridional winds in the F-region have opposite directions with respect to the equator [McLandress et al., 1996b], and on March 18, 1994 this pattern is broken (Figure 4c). The residual VER at 20°S–40°S is large and also shows a wave with apparent zonal wavenumber 1, but it is weak to the north of 20°S. The structures of the residual VER and winds at 155 km (Figure 4) are very much like those at 95 km (Figure 1), suggesting that the longitudinal variations in the E- and F-regions may be related. Similar characteristics of the longitudinal variations of the VER and winds in the F-region are also observed on March 13, 1996 and March 4, 1997.
 It is hard to know how long this wave lasts because of lack of continuing measurements. WINDII data on March 14, 1996 are also available, but the longitudinal variations of the winds are back to normal with a fluctuation of about 50 ms−1 compared to 120 ms−1 on March 13, 1996.
 The structures of longitudinal variations as seen on the three days are found on many other days in different months and years in the WINDII database but with much smaller magnitudes. The extreme longitudinal perturbation might be a phenomenon existing only in the March equinox period.
 The exceedingly large longitudinal variations in the emission rate and the horizontal winds presented in this article possess characteristics as follows: (1) the longitudinal variations have amplitudes of 70 photons cm−3 s−1 in the O(1S) VER and 120 ms−1 in the zonal and meridional winds, which break the migrating tidal pattern in the MLT region; (2) both the zonal and meridional wind components show a wave with apparent zonal wavenumber 1 in the local time fixed frame at all latitudes, but this occurs only at some latitudes for the emission rate; (3) the zonal winds are almost in-phase during day and night, suggesting a planetary wave with s = −1 and a long period; but the meridional winds have opposite phases then, suggesting a standing non-migrating diurnal tide; (4) the longitudinal variations of the daytime emission rate and the winds in the F-region also show a wave with apparent zonal wavenumber 1; and (5) the wave has a long vertical wavelength seen in winds indicating an evanescent nature.
 The very existence of this super wave on the three March days for three different years cannot be accidental. Its characteristics and magnitude do not agree with any of the well studied non-migrating diurnal tides or planetary waves. Fundamental questions about the source of this super wave, the correlations between the zonal and meridional winds, its impact on other physical and chemical precesses await a future solution.
 Financial support for this research work is provided by the Natural Sciences and Engineering Research Council of Canada. The WINDII project was sponsored by the Canadian Space Agency and the Centre National d'Etudes Spatiales, France. The authors are grateful to many colleagues for providing WINDII data.