Simultaneous observations of the electron density and the zonal wind obtained by the CHAMP satellite at 400 km are used to study systematic longitudinal variations. The time period selected is August–September 2004 allowing observations at pre-noon and post-sunset hours. The equatorial ionization anomaly (EIA) and the zonal delta-wind (deviation from zonal average) show a persistent and dominant 4-peaked longitudinal variation. We interpret this structure as caused by the wavenumber-3 nonmigrating diurnal tide (DE3). The EIA and the zonal delta-wind exhibit extrema at about the same longitudes. But, while the intensifications of the EIA and the delta-wind are in phase during the evening hours, they are out of phase in the morning. Possible coupling mechanisms are investigated.
 Recently, growing evidence is presented that upper atmospheric parameters are modulated by tidal effects originating in the tropical troposphere. For example, Immel et al.  reported four-peaked patterns of ionospheric emissions in the UV images taken by the IMAGE satellite during April 2002 in the post-sunset sector. They interpreted their observations as longitudinal variations of the equatorial ionization anomaly (EIA). Similarly, England et al.  presented evidence for a nearly in-phase longitudinal variation of the noon-time equatorial electrojet (EEJ) intensity. It was now possible, for the first time, to identify a wave-4 modulation of the zonal wind at about 400 km altitude by means of accelerometer measurements onboard the CHAMP satellite [Häusler et al., 2007]. They present wave-4 patterns observed during equinox seasons. Maximum wave amplitudes are found around 09h and 21h local time (LT). The phase of the wind modulation is shifted by about 180° between these two time sectors. Furthermore, Oberheide et al.  presented wind measurements in the mesosphere, lower thermosphere (MLT) region obtained by the TIDI instrument on board the TIMED satellite. They identified a dominant wave-4 structure in the longitude profile of the zonal wind. These authors related the wave pattern to the eastward propagating nonmigrating diurnal tide with zonal wavenumber-3 (DE3). According to the Global Scale Wave Model (GSWM) this tidal component is primarily excited by latent heat release in the tropical troposphere [Hagan and Forbes, 2002].
 It is expected that the nonmigrating tides dissipate and break well below 150 km altitude. An eastward electric field, together with the geomagnetic field, sets up an ion fountain at the equator. The EIA, comprising electron density peaks (crests) north and south of the dip equator and a density depletion (trough) in between, is produced by the ion fountain effect [Anderson, 1981]. Immel et al.  suggested that modulation of the E-layer dynamo electric field by the DE3 tide is responsible for their observed changing electron density at F region heights.
 A question that remains: Which mechanism is responsible for the wave-4 modulation of the thermospheric zonal wind? For answering the question we make use of various CHAMP satellite datasets sampled at 400 km altitude. Here we take advantage of the local electron density measurements obtained by the Langmuir probe and the wind estimates derived from the triaxial accelerometer. By directly comparing electron density profiles and wind velocities near the magnetic equator we want to find the relation between these two quantities.
2. Data Set and Processing
 The CHAMP satellite carries among other instruments a sensitive accelerometer. The measured accelerations can be used to estimate the thermospheric mass density and cross-track wind [see Liu et al., 2006]. Readings are averaged over 10s corresponding to a sampling distance of 76 km. The wind data presented here are processed in the same way as for the study of Häusler et al. . For obtaining the longitudinal dependence of the zonal wind we subtract the mean zonal velocity. In this case the zonal mean is a moving average over 24 hours. Resulting residual winds are averaged over a band of ±10° latitude centered at the dip equator. We thus obtain one delta-wind value per CHAMP pass.
 The electron density, sampled every 15s, is obtained from the planar Langmuir probe. For our analysis we consider meridional profiles between ±40° magnetic latitude. For the characterization of the individual EIA profiles we defined a crest-to-trough ratio (CTR) equivalent to the index used in total electron content (TEC) studies [Mendillo et al., 2000]
where ncn and ncs are the peak electron densities at the north and south crests and nt is the density at the equatorial trough. Each electron density profile is therefore characterized by a single number. In case no anomaly has formed, CTR is set to 1. The derived CTR index can be regarded as a measure for the strength of the ion fountain.
 The aim of this study is to perform a direct comparison between the longitudinal variations of the EIA and the zonal wind. From the study by Häusler et al.  we know that the wave-4 pattern in the zonal wind data is prominent around equinox seasons. Suitable months in the CHAMP data set are August 2004 for the pre-noon interval and September 2004 for the post-sunset sector.
 The CHAMP data of the two instruments are sorted into 24 overlapping longitude bins of 30° width. To obtain a further smoothing running averages over 5 days are calculated. Figure 1 shows observations for August 2004. During this month the sampled local time runs from 8.9 LT at the top to 11.6 LT at the bottom. In the top panel the distribution of the CTR index is presented in a longitude versus time frame. For emphasizing the wave-4 pattern we performed in a next step a Fourier transform over all longitudes for every day. The middle panel shows the Fourier analyzed fourth harmonic signal of CTR. Throughout the investigated month the wave-4 structure in CTR is always present. The phase of minima and maxima is almost constant over the displayed period.
 The same binning and filter procedure was applied to the delta-wind readings. Results are presented in the bottom panel of Figure 1. Positive values represent eastward winds. Also here we find a persistent wave-4 pattern with amplitudes up to ±30 m/s. The phase is, however, shifted by about 180° with respect to the CTR pattern.
 The post-sunset situation is depicted in Figure 2. During September 2004 CHAMP crossed the equator in the 18.0–20.7 LT sector. The ionosphere is known to be susceptible to density irregularities during these hours [Kelley, 1989, section 4]. Our derived CTR values from this late evening sector are in several cases affected by that phenomenon. Consequently, the distribution of that index, shown in the top frame of Figure 2, is not as well ordered as in Figure 1. The filtered wave-4 signal (middle panel) yields, however, again in a well organized longitude pattern. It is interesting to note that the phasing of CTR minima and maxima is the same as during the morning interval.
 Also during the post-sunset period the zonal wind exhibits a clear wave-4 pattern (Figure 2, bottom). The amplitudes are comparable to that during the morning hours but the phase is switched by about 180°. Generally, it can be stated that the CTR and the wind patterns are about in-phase during this post-sunset interval, but out of phase in the morning. A detailed analysis of the wind and CTR wave fronts reveals that in case of August 2004 (pre-noon interval) CTR peaks on average at a longitude 88.4°E (and multiples, 90° apart), while strongest negative delta-winds are observed at 97.5°E. For the month September 2004 (post-sunset interval) we find an average longitude for the CTR peak at 92.5°E, and the positive delta-wind peak at 82.5°E. In case of the latter interval the early and late days of September were omitted from the average because of vanishing wave-4 signature in CTR.
 In section 3 we presented a direct comparison between the longitudinal variations of the zonal wind and the equatorial ionization anomaly at about 400 km altitude. The CTR index should qualitatively be comparable with the intensity of ionospheric emission presented by Immel et al.  since it represents the electron density contrast between the crest and the equatorial trough. Our Figure 2 may thus be compared with Immel et al.'s Figures 3 and 4. The double peaked density structure (CTR > 1) is best developed during the hours 19 and 20 LT. At that time the CTR index peaks around longitudes of 90° and multiples of it. Immel et al.  found largest emission some 10° to 20° further east. A good part of this phase shift can be attributed to the seasonal phase variation of the DE3 tide (∼8°) between April and September [Oberheide et al., 2006]. This means, we can reproduce their essential findings with quite a different instrumentation.
 Our observations during the pre-noon sector go beyond the so far presented longitude structures of the EIA. The data presented in Figure 1 clearly reveal the 4-peaked longitude structure in the CTR index and show that the EIA intensity peaks also at longitudes of about 90° during the morning interval. From this we may conclude that the 4-peaked longitude structure of the EIA is a stable and persistent feature at least during equinox seasons.
 A distinct difference between the wave-4 patterns of the two quantities is that the delta-wind switches sign between pre-noon and post-sunset hours. Here we have to recall that the zonal mean of the wind velocity was subtracted from the readings before analyzing the presented data. From these studies [e.g., Wharton et al., 1984; Liu et al., 2006] we know that the wind is directed westward in the morning and eastward in the evening. When adding the harmonic variation to the background wind before noon, we find that the negative phase (westward) delta-wind adds to the mean westward wind just at longitudes where the ionization anomaly is strongest. In the evening positive delta-winds add to the mean eastward wind also at longitudes where the EIA peaks.
 For reconciling this apparent inconsistency we may recall that a stronger fountain effect enhances the electron density at anomaly latitudes but reduces it at the dip equator [e.g., Rama Rao et al., 2006]. At the same time the crests are displaced to higher latitudes, as is evident in Figure 4 of Immel et al. . A possible explanation could be that the EIA crests act as an obstacle for the zonal wind and deflect it into the high-speed channels at equatorial latitudes. Here the fast air flow would be confined in latitude by the crests, as depicted in Figure 3.
 From the above scenario we could envisage the following chain of processes. Atmospheric nonmigrating tides (in particular DE3) modulate the ion fountain effect by influencing the E-layer dynamo. In a second step the EIA undulations cause longitudinal variations of the thermospheric zonal wind. A consequence of that scenario should be that the wave-4 patterns at anomaly latitudes (10°–20° MLat) are out of phase with that at the dip equator because of the enhanced ion drag in the EIA peaks. We have tested this prediction and found an overall reduction of the wave-4 amplitude at crest latitudes, but the longitudinal variation of the wind is still in phase with that at equatorial latitudes. This result disqualifies the suggested modulation of the zonal wind by the EIA.
 For assessing the significance of the wave-4 wind amplitude we should compare it with the typical speed of the total zonal wind which is 100–150 m/s [e.g., Liu et al., 2006]. When considering the observed wave-4 velocities of more than ±20 m/s this is equivalent to a ±15% modulation. Zonal winds are the prime driver of the F region dynamo [e.g., Rishbeth, 1971; Lühr and Maus, 2006]. Consequently, the wave-4 pattern should be visible in all phenomena influenced by the F region dynamo.
 In summary, the eastward propagating diurnal tide with wavenumber-3 (DE3) excited by latent heat release in the tropical troposphere is considered to be responsible for a wide range of phenomena at E and F region heights. We have tested the hypothesis whether the wave-4 pattern of the EIA could be responsible for the longitudinal modulation of the thermospheric zonal wind speed. The partly controversial result implies, however, that another mechanism rather then the longitudinal variation of the EIA seems to couple the DE3 tidal signal into the thermospheric zonal wind velocity.
 The operational support of the CHAMP mission by the German Aerospace Center (DLR) and the financial support for the data processing by the Federal Ministry of Education and Research (BMBF), as part of the Geotechnology Programme, are gratefully acknowledged. The DFG supported K. Häusler through the Priority Programme “CAWSES,” SPP 1176.