Ground and satellite observations of nighttime medium-scale traveling ionospheric disturbance at midlatitude

Authors


Abstract

[1] We have investigated a nighttime medium-scale traveling ionospheric disturbance (MSTID) observed by an airglow imager at Shigaraki (34.9°N, 25.4°MLAT), Japan, on 17 May 2001. The structure was identified in the airglow images of OI (630.0 nm and 777.4 nm) as NW-SE band structures (horizontal wavelength: 230 km) moving southwestward with a velocity of 50 m/s. Neutral wind velocity was measured simultaneously from the Doppler shift of the 630.0-nm emission by a Fabry-Perot interferometer at Shigaraki. From these parameters, we performed model calculations of MSTIDs generated by gravity waves and by an oscillating electric field. We found that for the case of gravity waves, the estimated vertical wavelength was too small to explain the observed amplitudes of airglow intensity. For the case of the electric field, we found that an electric field oscillation of ∼1.2 mV/m was sufficient to reproduce the observed airglow amplitudes. This modeled electric field was comparable to that observed by the DMSP F15 satellite as it passed over Shigaraki during our observing period on 17 May 2001. The DMSP ion drift data show that the oscillation of the polarization electric field correlated with the MSTID structure in the airglow image, suggesting that the polarization electric field plays an important role in the generation of MSTIDs.

1. Introduction

[2] Medium-scale traveling ionospheric disturbances (MSTIDs) are a characteristic wave phenomenon in the ionospheric F-layer at midlatitudes. They have periods from a few tens of minutes to a few hours and wavelengths of several hundred kilometers (see Hunsucker [1982] for a review). Previous observations of such mesoscale structures in the ionosphere have used radio wave techniques, such as ionosonde [e.g., Bowman, 1990], HF Doppler [e.g., Waldock and Jones, 1986, 1987; Shibata and Okuzawa, 1983], satellite beacons [e.g., Evans et al., 1983; Ogawa et al., 1987; Jacobson et al., 1995], and ground radars [e.g., Fukao et al., 1991; Ogawa et al., 1994; Oliver et al., 1995]. From these observations, the MSTIDs were considered to be caused by atmospheric gravity waves [Hines, 1960; Hooke, 1968].

[3] Since the latter half of the 1990s, the airglow imaging technique has been widely used to measure ionospheric structures using the OI (630.0 nm) emission, which has an emission layer in the bottomside of the F-layer at 200–300 km [e.g., Mendillo et al., 1997; Taylor et al., 1998; Garcia et al., 2000; Kubota et al., 2000, 2001; Saito et al., 2001; Ogawa et al., 2002; Shiokawa et al., 2002, 2003]. Another weak oxygen emission line, OI (777.4 nm), has an emission layer around the F-layer peak at altitudes of 300–400 km [e.g., Sahai et al., 1981]. Makela et al. [2001] have shown that the height distribution of ionospheric structures can be estimated using these two emission lines (provided that the 777.4-nm emission has a detectable level of intensity). However, such simultaneous measurements of 630.0-nm and 777.4-nm images have not been made yet for midlatitude MSTIDs.

[4] The electrodynamics of mesoscale ionospheric structures have been extensively studied using the airglow imager and the incoherent scatter (IS) radar at the Arecibo Observatory [e.g., Behnke, 1979; Kelley et al., 2000, and references therein; Kelley and Makela, 2001]. Miller et al. [1997] have shown that an intense electric field exists in the mesoscale structure in the nighttime F-layer based on simultaneous measurement using an airglow imager and the IS radar at Arecibo. Saito et al. [1995] have reported similar electric field fluctuations at midlatitudes from the DE 2 satellite. Saito et al. [1998] have shown that such electric field fluctuations observed by the Freja satellite are associated with traveling ionospheric disturbances detected by the middle and upper atmosphere (MU) radar. However conjugate observations of airglow images and satellite electric field have not been made.

[5] The above references show that the electric field fluctuation is associated with the mesoscale ionospheric structures. This fact suggests that the electrodynamic (Perkins) instability in the ionosphere proposed by Perkins [1973], Miller [1997], and Hamza [1999] is a likely source of the mesoscale structures, though the linear growth rate of the Perkins instability is generally very small. It seems that the mesoscale structures seen in the airglow images and discussed in these previous references correspond to the traditional MSTIDs detected by radio wave techniques at midlatitudes. Thus, we will use the term “MSTID” in this paper to express the nighttime mesoscale structures seen in the airglow images at midlatitudes. Shiokawa et al. [2003] have shown that the occurrence frequency of the mesoscale airglow structures has a seasonal variation similar to that of the midlatitude spread-F, though the coincidence of the airglow structures and the spread-F is only 10–15%.

[6] To investigate the generation of MSTIDs by gravity waves and/or by the electrodynamic Perkins instability, information on the background neutral wind is essential. Besides providing two-dimensional images of MSTIDs, the airglow emission can be used to measure the neutral wind velocity from the Doppler shift of the emission line using, for example, the Fabry-Perot interferometer [e.g., Hernandez, 1986]. However, simultaneous measurements of airglow images and neutral wind have not been reported for midlatitude MSTIDs.

[7] In this paper, a comprehensive data set was used to investigate a prominent MSTID event observed at midlatitude at Shigaraki (34.8°N, 136.1°E in the geographic coordinates, 25.4°MLAT, inclination (dip angle) = 49.1°), Japan, on 17 May 2001. We used airglow images at 630.0 and 777.4 nm, neutral wind velocity measured by a Fabry-Perot interferometer for the 630.0-nm emission, and ion drift velocity and density measured by the Defense Meteorological Satellite Program (DMSP) F15 satellite. Almost no phase difference is detected in the MSTID structures at 630.0 nm and 777.4 nm. Model calculations were performed using the observed parameters of the MSTID. The calculations indicate that the vertical wavelength of the gravity waves is too small to account for the observed intensity variations of the airglow. We conclude that the polarization electric field plays an important role in the generation of nighttime MSTIDs.

2. Airglow Observations

[8] The all-sky airglow imager at Shigaraki is part of the Optical Mesosphere Thermosphere Imagers (OMTIs) [Shiokawa et al., 1999, 2000]. The imager uses a thinned and back-illuminated cooled charge coupled device (CCD) with 512 × 512 pixels. It measures airglow emissions of OI (630.0 nm), OI (777.4 nm) and OH (several bands at 720–910 nm) sequentially, with exposure times of 150 s (OI) and 10 s (OH) and a time resolution of 6.5 min (3 min each for the two oxygen lines and 0.5 min for OH). Background emission was monitored at a wavelength of 572.5 nm every 30 min with an exposure time of 105 s. The bandwidths of the band-pass filters for the measurement of the two oxygen lines and the background emission are 1.9–2.0 nm.

[9] Figures 1a and 1b show the MSTID images of 630.0 and 777.4 nm, respectively, obtained at Shigaraki on 17 May 2001. The original all-sky images are converted into geographical coordinates of 1024 km × 1024 km assuming airglow heights of 300 km and 350 km for 630.0 nm and 777.4 nm, respectively. The absolute intensity in units of Rayleighs (106/4π photons cm−2sr−1s−1) was obtained using the sensitivity data of the imager and by subtracting the contamination from background continuum emission. The average intensity of each image was subtracted in Figure 1 to show the structures more clearly.

Figure 1.

MSTIDs in airglow images in geographical coordinates (1024 km × 1024 km) at wavelengths of (a) 630.0 nm and (b) 777.4 nm, observed at Shigaraki at 1200–1530 UT (2100–2430 LT) on 17 May 2001. The altitude assumptions used to convert the all-sky coordinates to the geographical coordinates are 300 km and 350 km for 630.0-nm and 777.4-nm images, respectively. The average intensity for each image was subtracted. The white dots in the images are stars. The semi-circles seen in some of the images (e.g., 1201:19 UT in (a) and 1434:20 UT in (b)) are artifact (joint line of the integrating sphere that was used for the calibration of the imager).

[10] The MSTIDs in the 630.0-nm images are recognized as several band structures elongated from NW to SE in Figure 1a. They have wavelengths of ∼230 km (e.g., upper left part of the image at 1231:19 UT) and move southwestward continuously. On the other hand, the images of 777.4 nm in Figure 1b show small-scale NW-SE structures. From movie images on a computer display, we recognize that the NW-SE structures of 777.4 nm move northeastward continuously, which is opposite to the motion of the MSTID structures in the 630.0-nm images.

[11] Figures 2a and 2b show NE-SW cross sections (keograms) of the airglow images shown in Figures 1a and 1b, respectively. Clear southwestward motion of the structures (toward the bottom of the panel) is seen in the 630.0-nm keogram. The velocity of the motion is estimated to be ∼50 m/s from this keogram. The intensity of 630.0-nm airglow gradually decreased from 1230 UT (2130 LT) to 1500 UT (2400 LT), probably because the nighttime electron density decreases in time due to recombination.

Figure 2.

NE-SW cross sections (keograms) of the airglow structures shown in Figure 1 for (a) 630.0 nm and (b) 777.4 nm observed from Shigaraki at 1230–1630 UT (2130–0130 LT) on 17 May 2001. The airglow structures are shown in color plots in units of Rayleighs.

[12] The keogram of 777.4 nm (Figure 2b) shows clear northeastward motion of small-scale airglow structures. Similar northeastward-moving band structures were simultaneously observed in the OH airglow images by the same imager. The OH airglow has an emission layer around the mesopause region (altitude ∼86 km) and often shows this sort of small-scale gravity wave structure [e.g., Taylor et al., 1995]. The OH airglow emission has many band structures from visible to infrared wavelengths [Broadfoot and Kendall, 1968]. Because the OH(9–4) P2(2) line (wavelength: 778.2 nm) falls into the pass band of our 777.4-nm filter, we conclude that the structures seen in the 777.4-nm images are the contamination of the OH airglow emission from the mesopause region. Note that the airglow intensity at 777.4 nm is weaker than that at 630.0 nm, as indicated by the color scale of Figures 2a and 2b. Looking at the movie images of 777.4 nm on a display, however, we could recognize southwestward-moving structures embedded in the intense northeastward-moving small-scale structures.

[13] In order to remove the contamination of the OH emission, we performed a two-dimensional Fourier analysis on the keograms, similar to that applied to airglow images by Kubota et al. [1999]. By taking a two-dimensional power distribution of the keogram, we can extract only the southwestward-moving component of the airglow structures. The result is shown in Figures 3a and 3b for 630.0 and 777.4 nm, respectively.

Figure 3.

NE-SW cross sections (keograms) of the airglow structures calculated by inverse Fourier transform of the two-dimensional power spectra of Figure 2 for (a) 630.0 nm and (b) 777.4 nm. Only the structures moving southwestward are shown by taking the first and third quadrants of the two-dimensional power spectra obtained from the Fourier transform of the keogram in Figure 2. The airglow structures are shown as a deviation from average airglow intensity for each time. Black dots at both panels indicate the peak locations of the 630.0-nm structures in the keogram.

[14] In Figures 3a (top) and 3b (bottom), the airglow intensity variations are shown in percent from average intensity of each time. By removing the northeastward components using the above two-dimensional Fourier analysis, the southwestward-moving structures are weakly seen in the 777.4-nm keogram. The locations of the peak in intensity in the 630.0-nm keogram (Figure 3a) are indicated by the black dots in both Figures 3a and 3b. Clearly, the southwestward-moving structures at 630.0 nm and 777.4 nm have almost no phase difference, suggesting that the vertical wavelength of the observed MSTIDs is rather large compared with the height difference of the 630.0-nm (300 km) and 777.4-nm (350 km) airglow layers.

[15] Figure 4 shows several cross sections of the keograms from Figures 3a and 3b. From these cross sections, we estimate the peak-to-peak amplitude of the airglow structures at 630.0 and 777.4 nm to be ∼80% and ∼10%, respectively.

Figure 4.

Cross sections of the keograms shown in Figures 3a and 3b. The solid and dashed curves are the intensity variations of 630.0 nm and 777.4 nm, respectively.

[16] Figure 5 shows the neutral wind velocity U in the thermosphere measured by a Fabry-Perot interferometer at Shigaraki on 17 May 2001. The interferometer measures the Doppler shift of the 630.0-nm airglow emissions at a zenith angle of 50° ± 4.75° and at four azimuthal directions of N, S, E, and W every hour. It can measure the wind velocities for the inner (solid curve) and outer (dashed curve) fringes independently. The differences of winds for inner and outer fringes are mostly less than 20 m/s, indicating a random error of the wind measurement. Details of the interferometer are described by Shiokawa et al. [2001; A two-channel Fabry-Perot interferometer with thermoelectric cooled-CCD detectors for neutral wind measurement in the upper atmosphere, submitted to Earth, Planets, and Space, 2003].

Figure 5.

Thermospheric neutral wind velocities measured by a Fabry-Perot interferometer at Shigaraki on 17 May 2001. The interferometer can measure neutral wind velocities for inner and outer interference fringes independently from the Doppler shift of the 630.0-nm airglow emission.

[17] The wind was mostly southward (∼70–100 m/s) and eastward (∼0–50 m/s) for the plotted interval. The eastward wind gradually decreases to zero toward midnight. These variations in the wind velocities are similar to the typical thermospheric tidal variation at mid-low latitudes [e.g., Hernandez and Roble, 1995; Biondi et al., 1999].

3. Model Calculations

[18] Table 1 summarizes the wavelength and apparent phase velocity of the observed MSTIDs, the angle α between the k-vector of the MSTIDs (southwestward) and the geomagnetic east, and neutral wind velocities measured by the Fabry-Perot interferometer for the MSTID event of 17 May 2001. Using these parameters, we have performed model calculations for the two cases when the observed MSTIDs are generated by gravity waves and by the electric field. The airglow intensities were estimated based on the method adopted by Otsuka et al. [2003].

Table 1. Observed and Estimated Parameters of the MSTIDs Observed at Shigaraki on 17 May 2001
 1221 UT1321 UT1421 UT1521 UT
  • a

    Angle between the wave k-vector (southwestward) and the geomagnetic east.

  • b

    Fabry-Perot interferometer.

  • c

    Estimated from dispersion relation of gravity wave.

  • d

    Angle between the direction of E0 + U × B and the geomagnetic east.

Wavelength λh (imager)230 km230 km204 km273 km
Apparent phase velocity c (imager)50 m/s50 m/s50 m/s38 m/s
αa (imager)216°208°201°193°
Northward wind (FPIb)−78 m/s−83 m/s−82 m/s−116 m/s
Eastward wind (FPIb)55 m/s58 m/s35 m/s20 m/s
Southwestward wind U (FPIb)17 m/s18 m/s34 m/s69 m/s
Vertical wavelength λzc23.9 km23.2 km11.5 km22.4 km
θ*d37°37°23°
Growth rate1.4 × 10−6 s−17.3 × 10−6 s−11.9 × 10−6 s−1−2.0 × 10−6 s−1

[19] For the model calculations, the background height profiles of electron and oxygen densities are given by the IRI-95 [Bilitza, 1997] and MSIS-86 [Hedin, 1987] models, respectively, and are shown in Figure 6a. The chemical reactions that generate the 630.0-nm airglow emission are:

equation image
equation image
equation image

Because reaction (1) dominates the whole process, the production of the 630.0-nm emission is proportional to the molecular oxygen density [O2] and the oxygen ion density [O+] (≃Ne in the F-layer). Thus, the peak height of the emission layer becomes the bottomside of the nighttime F-layer, as shown in Figure 6b. The chemical reaction coefficients given by Sobral et al. [1993] are used for the present calculation.

Figure 6.

Altitude profiles of (a) electron density (Ne, solid curve) and molecular oxygen density ([O2], dashed curve) and (b) volume emission rates of 630.0-nm airglow (V630, solid curve) and 777.4-nm airglow (V777, dashed curve), used as an initial condition of the model calculation of the airglow structures of 17 May 2001. The profiles of Ne and [O2] are taken from IRI-95 and MSIS-86, respectively.

[20] The chemical reactions that generate the 777.4-nm airglow emission are:

equation image
equation image

Because the production of the 777.4-nm emission is proportional to both the oxygen ion density [O+] and the electron density Ne, and [O+] ≃ Ne in the F-layer, the peak height of the emission layer is close to the nighttime F-layer peak, as shown in Figure 6b. The chemical reaction coefficients given by Sahai et al. [1981] are used for the present calculation.

3.1. Gravity Waves

[21] The linear dispersion relation of the gravity waves is given as [Hines, 1960]

equation image

where m, N, U, c, k = 2π/λh, and H are the vertical wave number, Brunt-Väisälä frequency, background wind velocity in the direction of wave propagation, observed wave phase velocity, horizontal wave number, and scale height, respectively. Taking typical values of N ∼ 2π/12 min and H ∼ 45 km, we can estimate the vertical wavelength λz = 2π/m from the above dispersion relation of gravity waves. The estimated values of λz are shown in Table 1. Note that the values of λz are fairly small (10–25 km). In the right hand side of equation (6), the main contribution to m2 is the first term. The small values of λz are mainly because of the small value of U-c (20–30 m/s), as listed in Table 1.

[22] For the case in which the above gravity waves generate electron density perturbations in the F-layer, we performed a model calculation to show how much variation in airglow emission can be obtained from the perturbations, by using a method similar to that described by Hooke [1968]. The electron density perturbation Ne is given by

equation image

where u and k are the velocity and wave number of the gravity wave parallel to the ambient magnetic field, respectively, and I is the magnetic inclination (dip angle). The gravity wave in the neutral atmosphere causes field-aligned ion motion u through neutral-ion collisions. The first and second terms of equation (7) indicate the electron density perturbations caused by vertical motions of the layer and by convergence and divergence of ions along the field line, respectively.

[23] Figure 7 shows the result of the model calculation in the north-south cross section. The electron density perturbation shown in the top panel is caused by the gravity wave, which has a horizontal wavelength of 230 km, an intrinsic phase velocity (U-c) of 30 m/s, and an angle α from the geomagnetic east of 225°. These adopted values are based on the observations. The inclination I was assumed to be 45°. The amplitude u of the gravity wave was set to be 1.0 m/s for all altitudes. This amplitude gives an electron density perturbation of ∼1.5 × 1012 m−3, as shown in the top panel, which is ∼50% from the background electron density shown in Figure 6.

Figure 7.

Model calculation of MSTID structures for the case of an atmospheric gravity wave. The color scale, black solid line, black dashed line, and purple line in the top panel indicate the electron density Ne (m−3), peak altitude of Ne, original peak altitude of Ne without perturbation, and contour of the 630.0-nm airglow (every 5 photons cm−3s−1), respectively. The bottom panel indicates height-integrated airglow intensity. Because the vertical wavelength of the gravity wave estimated from the observation and the dispersion relation is very small (21.7 km), the amplitude of the height-integrated airglow variations is very small of less than 0.5%.

[24] For the present case, the convergence and divergence of ions along the field line (second term of equation (7)) mainly cause the electron density perturbation in the top panel. The wave phase surfaces are elongated in the horizontal direction due to the small vertical wavelength of the gravity wave. The bottom panel shows height-integrated airglow intensity variations caused by the electron density perturbations in the top panel. Because the vertical wavelength of the gravity wave is very small, ∼22 km (which is far smaller than the thickness of the F-layer and the airglow layer in Figure 6), the height-integrated amplitudes of the airglow variations are very small, less than 0.3%, for both 630.0 and 777.4 nm. This is not consistent with the observed amplitudes of airglow variations, which were ∼80% and ∼10% for 630.0 and 777.4 nm, respectively.

3.2. Electric Field

[25] We performed a similar model calculation for the case in which the oscillating electric field was imposed on the ionosphere. From the continuity equation of ions, the electron density perturbation Ne is related to the perpendicular velocity v′ by E′ × B drift due to the oscillating electric field E′, as given by

equation image

by assuming that the background electric field is zero. For oscillating perturbations, the perturbed quantities are assumed to be plane waves of Ne, v′ ∝ exp{it - k · x)}. The perturbed velocity v′ is normal to k and the geomagnetic field, considering that the perturbed velocity v′ is caused by the polarization electric field associated with the electron density perturbation Ne. We can assume ∇ · v′ = 0 (noncompressible). From equation (8), we obtain

equation image

where D is the declination of the magnetic field.

[26] Figure 8 shows the result of the model calculation for the case of the oscillating electric field. The background electron and molecular oxygen density profiles and the chemical reaction parameters are the same as those used in Figures 6 and 7. The given electric field perturbation was sinusoidal, perpendicular to the local magnetic field line (dip angle = 45°), and parallel to the k-vector, which was α = 225° from geomagnetic east. The horizontal wavelength, phase velocity, and amplitude of the given electric field were 230 km, 50 m/s, and ∼1.2 mV/m (=25 m/s of ion drift velocity), respectively.

Figure 8.

Model calculation of MSTID structures for the case of oscillation of the electric field, in the same format as that of Figure 7. A sinusoidal perturbation of the electric field with an amplitude of ∼1.2 mV/m (drift velocity of 25 m/s) is imposed on the ionosphere to reproduce the observed airglow amplitude of ∼80% at 630.0 nm.

[27] The given eastward (westward) electric field component causes northward and upward (southward and downward) ion drift. As a result, the electron density forms field-aligned structures associated with the upward and downward motions of the F-layer, as shown in the top panel of Figure 8. The height-integrated airglow intensities shown in the bottom panel are ∼80% and ∼40% for 630.0 and 777.4 nm, respectively. The amplitude of electric field perturbation (∼1.2 mV/m = 25 m/s in drift velocity) was chosen to reproduce the observed amplitude of 630.0-nm airglow of ∼80%. We could successfully reproduce the observed 630.0-nm amplitude by a relatively small perturbation of the electric field. The calculated amplitude of 777.4 nm (∼40%) is larger than that observed (∼10%). This discrepancy is possibly due to the contamination of OH airglow emissions at wavelengths near 777.4 nm, as discussed in section 2.

4. DMSP Observations

[28] During the MSTID event of 17 May 2001, the DMSP F15 satellite traversed the sky near Shigaraki and measured horizontal and vertical ion drift velocities at an altitude of ∼846 km. Figure 9a shows ion drift vectors on the 630.0-nm airglow images at Shigaraki at 1220:49 UT (2120:49 LT) on 17 May 2001. The DMSP F15 satellite traversed the image from south to north at 1221:18–1224:29 UT. It measures horizontal (perpendicular to the satellite track) and vertical ion drift velocities with a time resolution of 1/6 s. The vectors plotted in Figure 9a are ion drift v perpendicular to the local magnetic field line, and are obtained by assuming that the parallel ion drift velocity is zero. Only the perturbation (high-frequency) components of ion drift are plotted by subtracting a running average of 1 min (corresponding to a spatial scale of 450 km) for the DMSP data. The satellite track was mapped from 846 km to 300 km along the local magnetic field line.

Figure 9.

(a) Ion drift velocity and (b) electric field vector measured by the DMSP F15 satellite along the footprint on the 630.0-nm image of the MSTID observed at Shigaraki at 1220:49 UT (2120:49 LT) on 17 May 2001. The images are deviations in percent from each 1-hour average image, in geographical coordinates of 1024 km × 1024 km at an altitude of 300 km. Only the variation component of the drift velocity and the electric field are indicated by subtracting a running average of 1 min for the DMSP data. The vectors of ion drift and electric field are indicated on the frame perpendicular to the local magnetic field. These values are calculated from vertical and horizontal ion drift velocities measured by DMSP F15, assuming that the parallel drift velocity is zero. The satellite track was mapped from 846 km to 300 km along the local magnetic field line.

[29] It is interesting to note that there is one-to-one correspondence between the airglow structures and the structures of ion drift vectors. The ion drift vectors are mostly parallel to the phase surfaces of the MSTIDs. The velocity is most intense and southeastward in the brightest airglow band.

[30] There are two possible causes of the observed ion drift variations, i.e., the neutral wind (which causes ion drift parallel to the magnetic field) and the electric field (which causes ion drift perpendicular to the magnetic field). However, it is not likely that the neutral wind variations (associated with gravity waves) caused the observed ion drift variations, because the structures at two different altitudes show fairly good one-to-one correspondence (airglow structures at ∼300 km and ion drift structures at 846 km) in Figure 9a. Thus, we consider the possibility that the observed ion drift structures were caused by electric field variations imposed on the ionosphere. It is quite conceivable that the electric field variations were similar along the magnetic field line between the airglow altitudes and the DMSP altitudes.

[31] Figure 9b shows electric field vectors E calculated from the ion drift velocities in Figure 9a using the relation v = (E × B)/B2. The electric field vectors tend to be perpendicular to the phase surfaces (parallel to the k-vector) of the MSTIDs. They indicate positive (negative) charge accumulation at the northward (southward) edge of the most intense airglow band. The intensity of the electric field variation (∼1.0–1.5 mV/m) is fairly comparable to that estimated from the model calculation (∼1.2 mV/m) in section 3.2.

[32] Figure 10 shows a comparison of DMSP ion drift and airglow intensity variations. The airglow intensity variation was extracted from the airglow image in Figure 9 along the satellite footprint. The variation in local ion density at an altitude of ∼846 km measured by the DMSP F15 satellite is also shown in the top panel. For the DMSP data, 1-min running averages (corresponding to a horizontal scale of 450 km) were subtracted to show only the variation component, because the absolute values of the ion drift velocity contain instrumental offset. The background ion density in the plotted interval was ∼4–5 × 106 cm−3.

Figure 10.

Vertical and horizontal ion drift and ion density measured by the DMSP F15 satellite near Shigaraki on 17 May 2001, at an altitude of 846 km. The drift velocities are the same as those plotted in Figure 9. Variations in the 630.0-nm airglow intensity observed at Shigaraki along the satellite footprint at an altitude of 300 km are indicated by dashed curves. Only the variation in the drift velocity and the density are plotted by subtracting a running average of 1 min for the DMSP data.

[33] The airglow intensity increases (decreases) for the downward (upward) ion drift region, particularly for the interval of 1222:30–1223:30 UT in the bottom panel of Figure 10. This indicates that when the ionosphere is forced down by the westward electric field, the electrons in the bottomside F-layer interact with rich molecular oxygen at lower altitudes, as indicated in Figure 6a, and thus the 630.0-nm airglow intensity increases.

[34] The ion density in the top panel of Figure 10 is also anti-correlated with the airglow intensity. This can again be explained by the vertical ion drift motion, namely, the region where the ionosphere is pushed down corresponds to the large airglow intensity, and because of the downward motion of the ionosphere, the ion density in the topside ionosphere decreases.

[35] The correlation between the vertical ion drift and airglow intensity is weaker at 1222:00–1222:30 UT, but still seems to hold. For example, the upward drift at 1222:25 UT coincides with slight decrease of airglow intensity, and vice versa at 1222:10 UT. There may be the effect of line-of-sight integration of airglow intensity, which weakens the amplitude of airglow variation in the direction perpendicular to the wave phase surface. The shorter spatial scale of the ion drift variation at 1222:00–1222:30 UT also possibly contributes to the line-of-sight integration effect.

5. Discussion

[36] The comprehensive ground optical observations give the information on the wavelength and phase velocity of the MSTIDs and background neutral wind. As discussed in section 3.1, the vertical wavelength of the gravity wave estimated from these observed parameters using gravity-wave dispersion relation is very small, which is inconsistent with the observed large amplitude of the MSTIDs. As shown in Table 1, the southwestward component of the neutral wind velocity gradually increases and exceeds the apparent southwestward phase velocity of MSTIDs, which were almost constant at 40–50 m/s during the observation interval. Such variations would also be difficult to explain if the MSTIDs were generated by gravity waves. From these facts, we conclude that the observed MSTIDs in the airglow images were not caused by the gravity waves.

[37] On the other hand, the ion drift variations observed by the DMSP satellite indicate a close association between the electric field structures and the MSTID structures in the airglow image. The model calculation indicates that an electric field perturbation of ∼1.2 mV/m, which is comparable to that observed by the DMSP satellite, can explain the observed airglow intensity variations.

[38] The electric field vectors in Figure 9b indicate positive (negative) charge accumulation at the northward (southward) edge of the most intense airglow band. From the Fabry-Perot interferometer measurement, the neutral wind U was southeastward at this time, as shown in Figure 5. Thus, the direction of the ionospheric current J caused by this neutral wind was northeastward (J = Σp (U × B), where Σp is the field-line-integrated Pedersen conductivity). As shown in Figure 10, the region of enhanced 630.0-nm airglow intensity corresponds to the region where the ionosphere was pushed down and the Pedersen conductivity is higher than the surrounding region. Thus, the direction of J from U and the charge accumulation (positive/negative at the northeastward/southwestward edge of high airglow intensity) in Figure 9b are consistent with each other, indicating that the “polarization” electric field develops due to the inhomogeneity of the Pedersen conductivity and the background ionospheric current.

[39] One model of the generation of the polarization electric field in the ionosphere is the Perkins instability [Perkins, 1973]. The basic concept of the Perkins instability can be expressed as follows. If a localized region of the ionosphere moves slightly upward, the Pedersen conductivity Σp in the region would decrease due to a reduction in ion-neutral collision. The decrease in local conductivity would cause an eastward polarization electric field if the background electric current in the ionosphere had an eastward component. The eastward polarization electric field drives further upward motion of plasma by E × B drift. This positive feedback should be relaxed by gravitational diffusion of the plasma along the field line at higher altitudes. However, Perkins [1973] showed that when the direction of the wave k-vector is between the background electric field vector and geomagnetic east, the linear growth rate of the instability becomes positive.

[40] The growth rate of the Perkins instability is expressed as

equation image

where g, 〈νin〉, Hn, and θ* are the acceleration of gravity, height-integrated ion-neutral collision frequency, scale height of the neutral atmosphere, and the angle between geomagnetic east and the direction of E0 + U × B. E0 is the background electric field.

[41] Equation (10) suggests that the maximum growth rate of the Perkins instability is obtained when α = θ*/2. Because the neutral wind vector U gradually shifts from southeastward to southward during the observation (as shown in Figure 5), the direction of the maximum growth rate (θ*/2) is expected to shift from southwestward (or northeastward) to westward (or eastward). As shown in Figure 1a and Table 1, the observed k-vector of MSTIDs gradually changes from southwestward to westward. Thus, the change in the MSTID k-vector is consistent with the shift of the maximum growth-rate directions expected from the neutral wind variation.

[42] From equation (10), we calculated the growth rate γ of the Perkins instability using the observed parameters. For the calculation, we assume that the values of g, I, and Hn are 9.8 m/s, 45°, and 45 km, respectively. The angle α of the phase surface is estimated from the airglow images and indicated in Table 1. To estimate the angle θ* shown in Table 1, we used the neutral wind velocity measured by the Fabry-Perot interferometer and assumed E0 = 0. The height-integrated ion-neutral collision frequency 〈νin〉 is expressed as

equation image

The ion-neutral collision frequency vin is given as [Rishbeth and Garriot, 1969]

equation image

where n is the neutral density and M is the atomic number of oxygen (=16). We used the MSIS-86 and IRI-95 models indicated in Figure 6a for the calculation of neutral and electron densities, respectively.

[43] The calculated values of γ are indicated in Table 1. The growth rates γ are positive except for that at 1521 UT. The value itself, however, is very small on the order of 10−6 s−1 (more than 1 day). Similar small growth rates were reported in previous references [e.g., Miller et al., 1997; Garcia et al., 2000]. We performed a similar calculation using the values of E0 obtained by averaging a 7-year ion-drift measurement (1986–1992) made by the MU radar at Shigaraki [Takami et al., 1996]. The averaged ion drift was mostly eastward and had a value of ∼20–25 m/s, which is smaller than the neutral wind velocities observed during the present event. However, the calculated values of γ become negative for all the times in Table 1.

[44] The linear assumption used to obtain the above growth rate may not work for the present event, because the observed and estimated amplitudes of the electric field fluctuation are ∼1.0–1.5 mV/m (20–30 m/s of ion drift velocity), which is not a small perturbation compared with the observed background wind velocity (∼100 m/s). As discussed by Garcia et al. [2000], investigation of nonlinear development of this type of instability will be needed.

[45] Saito et al. [1995] have shown nighttime electric field fluctuations at midlatitudes based on ionospheric satellite observations. The present event probably corresponds to the electric field fluctuation events in their study. They indicate that the fluctuation is a hemispherically conjugate phenomenon, and suggest electrodynamic coupling of the fluctuations between the northern and southern hemispheres through field-aligned current. In that case, the growth of the polarization electric field should include the processes in the conjugate hemisphere. It may also be necessary to include the effect of the ionospheric E-layer, which is connected to the F-layer by the field line.

[46] The observed wavelength of 230 km cannot be explained by the Perkins instability, because its growth rate does not have a preference for scale size. The wavelength of MSTIDs may be determined by that of gravity waves, which can be a seed of the instability. The k-vector direction (southwestward) of MSTIDs may also be determined by the seed gravity waves. However, as shown by statistical studies of the MSTIDs in airglow images [Garcia et al., 2000; Shiokawa et al., 2003], the direction is always southwestward, while the gravity waves can propagate in all directions in the thermosphere. Kelley and Makela [2001] have proposed introducing an additional polarization electric field in the direction parallel to the wave surfaces (perpendicular to the k-vector), to explain this southwestward preference.

6. Conclusions

[47] Using a comprehensive data set obtained by a multichannel airglow imager, a Fabry-Perot interferometer, the DMSP satellite, and model calculations, we have investigated the nighttime MSTID (NW-SE band structures moving toward SW) of 17 May 2001, observed at midlatitude at Shigaraki. The DMSP F15 satellite observes oscillations of polarization electric field associated with the MSTID structures in the airglow images. The conclusions of the present study can be summarized as follows:

[48] 1. The observed nighttime MSTID is not directly caused by the gravity waves, because the vertical wavelength estimated from gravity-wave dispersion relation is too small to explain the observed amplitudes of the MSTID, and because the phase propagation of the MSTID does not have a direct association with the background neutral wind variation.

[49] 2. The polarization electric field, as measured by the DMSP F15 satellite, plays an important role in the generation of the nighttime MSTIDs. Our model calculation indicates that the observed amplitude of electric field oscillations was sufficient to produce the observed airglow variation in MSTID.

[50] The observed direction of the polarization electric field is consistent with the idea of electric field generation by spatial inhomogeneity of the Pedersen conductivity (inferred from the airglow variation) and the ionospheric current (inferred from the neutral wind), as expected from the Perkins instability. However, the estimated linear growth rate of the Perkins instability was very small. There are several unresolved problems in the generation of the nighttime MSTIDs at midlatitudes, i.e., nonlinear growth of the Perkins instability, coupling with the ionosphere of the conjugate hemisphere and the E-layer, and coupling with the thermospheric gravity waves, which may be a seed of the instability.

Acknowledgments

[51] We thank Y. Katoh, M. Satoh, and T. Katoh of the Solar-Terrestrial Environment Laboratory, Nagoya University, for their kind support of airglow imaging observations. The optical observation at Shigaraki was carried out in collaboration with the Radio Science Center for Space and Atmosphere, Kyoto University. The MU radar at Shigaraki belongs to and is operated by the Radio Science Center for Space and Atmosphere, Kyoto University. K. S. is grateful to A. Saito for his helpful comments on this work. This work was supported by a Grant-in-Aid of the Ministry of Education, Culture, Sports, Science, and Technology of Japan (11440145 and 13573006).

[52] Arthur Richmond thanks Jonathan Makela and another reviewer for their assistance in evaluating this paper.

Ancillary