Simultaneous F-region airglow, E-region coherent-scatter radar, and ionosonde observations were made in Greece during the summer sporadic-E season in 2002. In this paper we report on two case studies during which patchy sporadic-E layers were accompanied by midlatitude spread F, coherent VHF radar echoes (including two-stream echoes), and traveling ionospheric disturbances registered by the airglow instrument. We argue that these events give strong evidence that polarization electric fields are built up in the E region and are mapped upward to the F region, creating rising and falling regions in the bottomside plasma. The resulting structure creates conditions for midlatitude spread F, as detected by the ionosonde. This correlation between patchy sporadic E and midlatitude spread F is further supported in a companion paper. Upward coupling of this sort is particularly efficient in regions of F-region plasma uplift and airglow depletion, since the F-region Pedersen conductivity is low which reduces the electrical load on the E-region generator.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Traveling ionospheric disturbances (TIDs) can be organized into two classes. One class occurs at large scales (>750 km) and typically propagates from high to low latitudes. These TIDs are generally thought to be generated by impulsive heating in the auroral oval. The main ionospheric effects are height changes due to meridional winds and density changes due to heating and compositional changes. Medium-scale or mesoscale TIDs (typically 50–500 km in horizontal scale) are thought to be generated by gravity waves through a combination of wind effects and temperature-dependent recombination. The sources of medium-scale TIDs are thought to be tropospheric weather and orographic effects, but there is good evidence that electrodynamic factors and possibly plasma instabilities play a role in the formation of these types of TIDs [Garcia et al., 2000a, 2000b].
 Medium-scale TIDs are an important source of midlatitude spread F, since the associated corrugation on scales the order of an ionosonde beam size leads to multiple reflection paths at the same frequency and hence to a spread in the ionosonde trace, particularly in range [Bowman, 1990]. Spread in the ionosonde trace can also be due to intermediate-scale (1–50 km) ionospheric structures, particularly height variations of constant density contours, which can lead to both frequency and range spreading. These features are most likely due to plasma processes, since neutral atmospheric disturbances are heavily damped at such scales. At equatorial latitudes an enormous and, in large part, successful effort has been made to understand the plasma physics of equatorial spread F and its relationship to ionospheric and thermospheric conditions, including the role of TIDs. Less progress has been made at midlatitudes until fairly recently, when all-sky imaging became available as a tool to help visualize the ionospheric disturbances.
 The all-sky imaging technique was first extensively combined with incoherent scatter radar (ISR) measurements during a 10-day run at Arecibo in 1993 [Kelley and Miller, 1997; Mendillo et al., 1997; Miller and Kelley, 1997]. They found that the typical TID was oriented from northwest to southeast and moved in the southwest direction at several tens of meters per second. The horizontal scale of the TIDs was several hundred kilometers and, most surprisingly, they were found to have sizeable internal electric fields. The TIDs were associated with earlier reports by Behnke , Fukao et al. , and Kelley and Fukao , who also reported large electric fields and/or high velocities perpendicular to the magnetic field within these features. The direction of motion and scale size are consistent with studies using ionosonde techniques capable of determining frontal motions [Bowman, 1968]. A key feature of this class of TID is periodicity in the horizontal direction. All-sky studies by Taylor et al. , Garcia et al. [2000a, 2000b], and Saito et al.  all report this periodicity. Although the Perkins instability [Perkins, 1973] may play a role in the development of these electrified mesoscale features [Kelley and Fukao, 1991; Kelley and Makela, 2001], this process alone does not seemingly lead to structure in the intermediate scales. For example, Saito et al.  found from in situ data that the spectrum of irregularities is very steep for mesoscale midlatitude structures. The very rare occurrence of VHF coherent scatter from the midlatitude F region [Swartz et al., 2000b, 2000a] implies that unique circumstances and, possibly, a new source of free energy is needed to generate intermediate and small-scale structures.
 Mechanisms to generate intermediate-scale structures at midlatitudes are thus an open problem. One possibility recently discussed by Kelley et al.  is that secondary instabilities occur in the F region. For example, they reported an event in which one side of a medium-scale TID, detected via airglow measurements, steepened and began to develop intermediate-scale features. They suggested that the intermediate-scale structure was due to either a zero-order, wind-driven E × B instability of the density gradient or a poleward, electric field-driven E × B process on the same gradient. Such poleward fields are often found inside TIDs [Kelley et al., 2000]. The event was unique since the Arecibo radar detected density structures on this edge with cross-field scale on the order of a kilometer [Mathews et al., 2001]. The power-law spectrum of these irregularities was studied by Kelley et al.  and was found to be less steep than results reported by Saito et al.  and much more typical of those in ESF and in midlatitude barium cloud studies, as well as in simulations of these phenomena. These secondary processes are all of the generalized E × B instability type, in which either the gradient or a large-scale electric field or both are created by a primary process. The E region and the conjugate F region are passive loads in this sort of F-region process and damp the electrostatic fields.
 Here and in the companion paper [Haldoupis et al., 2003, hereinafter referred to as Paper 1] we report on new measurements in the European sector during which coherent-scatter observations were made in the E region in conjunction with an ionosonde and, in some cases, an all-sky camera. The goal was to explore the coupling between the E and F regions and to study the electrodynamics and plasma instabilities associated with TIDs and midlatitude spread F. We believe that another mechanism has been identified as a source for intermediate F-region structure and midlatitude spread F.
2. Experiment Description
 For this study in July 2002, we fielded the Cornell All-Sky Imager (CASI) on the Greek Island of Milos. Milos was chosen due to its proximity to the E-region, radar-scattering volume of the SESCAT (Sporadic-E Scatter Experiment) radar, a CW bistatic system located on Crete. We explored the possible relationship between mesospheric waves in the neutral atmosphere and the sporadic-E irregularities above. No such connection was found but, to be fair, the camera was not operating quite as well as expected during that period. We did, however, observe some interesting F-region features, reported here, as well as an apparent correlation of these structures with E-region scatter. For such a coupling to exist, the F-region structure must be located to the south of the E-region scattering volume, but the images in the F region extend far enough south from Milos to be useful in this regard.
 CASI uses a thinned and back-illuminated CCD that is liquid-cooled. The system is fronted by telecentric optics, a five-position filterwheel, and an all-sky lens. Due to a software problem, only four of the 2-nm bandwidth filters were properly exposed. We obtained useful 557.7-nm, 630.0-nm, and 777.4-nm images, as well as background exposures for image correction and star removal.
 The radar system is a continuous-wave Doppler radar designed for midlatitude E-region coherent-backscatter studies (for details, see Haldoupis and Schlegel ). SESCAT is located along the northern coastline of Crete and operates at 50.52 MHz with the transmitter and receiver antennas beaming northward to a region perpendicular to the Earth's magnetic field at E-region altitudes near 105 km. The system observes a fixed volume confined at ranges from about 170 km to 210 km, with the most likely range for optimal echo reception at about 185 km. SESCAT observes almost directly along the geomagnetic meridian, thus measuring north-south Doppler motions.
 In the setup used for this experiment, the transmitter antenna consisted of an array of four 11-element Yagis, each separated by 8 m and providing an overall 3-dB beamwidth of 8°. At the receiving site, a phase-coherent, dual-channel receiver was used to perform single-line azimuthal interferometry using two identical 11-element Yagis, each having a beamwidth of 16° and separated by approximately 8 m in the east-west direction. The intersection of the transmitting and receiving antenna array patterns, as well as the severe magnetic aspect sensitivity of backscatter, define an E-region observing area of about 25 km in the meridional direction and about 60 km in the zonal direction. The viewing area is centered at about 36.7°N, 24.6°E geographic, just 15 km to the east of the island of Milos in the central Aegean Sea. This location corresponds to 30.8° invariant geomagnetic latitude, an L-shell value of 1.35, a magnetic dip angle of I = 52.5°, and a magnetic declination of D = 2.5°.
 Both of SESCAT's receiver audio outputs were digitized and processed on site by using two identical digital signal processor (DSP) units, both housed in a personal computer. Fast Fourier transforms (FFTs) and subsequent power-spectrum and cross-spectrum (coherency and cross-phase) calculations were performed in real time for both output signals with high efficiency. Given that the useful Doppler spectral band is between 800 Hz and 1200 Hz with the zero Doppler shift being exactly at 1 kHz, a digitization rate of 2.5 kHz was used, leading to a sampling interval of 0.4 ms and a Nyquist frequency of 1.25 kHz. The Fourier coefficients were computed for each receiver output by performing the 2048-point FFT every 0.8192 s. This yields a frequency resolution of 1.22 Hz or a Doppler velocity resolution of 3.80 m/s. The spectral and cross-spectral estimates were averaged over 4.92 s and stored on disk for further processing.
 The ionosonde data were obtained with the Athens Digisonde (AD), the new ionospheric station of the National Observatory of Athens situated just to the north of Athens at Penteli (38°N, 23.5°E geographic), approximately 150 km northeast of the SESCAT observing region and well within the CASI F-region field of view. The fully automated ionospheric sounder uses two cross-delta antennas on a 30-m tower for its 300-W transmission and an array of four crossed-loop antennas for reception. The AD provides vertical-scanning ionograms and automatically scaled ionospheric parameters on a routine basis. In addition, it serves the international scientific community with real-time data streaming to the World Data Centers. The ionospheric parameters used in this study were obtained from scanning ionograms taken every 5 min and were also manually checked and validated. The ionosonde data reported in the companion paper were obtained with a portable ionosonde operated in Milos during the summer of 1996.
 With the vagaries of weather and the sporadic nature of the radar events, we first determined the times when the camera data were usable and the radar was fully operational. Within these relatively few nights, we searched independently for periods with interesting structures or radar echoes, respectively. This report thus involves case studies, whereas Paper 1 deals with a more extended data set.
 The most striking events and the ones studied in the most detail here occurred on the evenings of 3–4 and 4–5 July 2002. The Kp index never exceeded 2− on the first night and was only as high as 3 on the second. For reference, the event reported by Swartz et al.  was obtained during magnetically quiet conditions. Two series of all-sky camera pictures using the 630.0-nm filter are presented in Figures 1 and 2 for these two nights. The images were processed with a 7 × 7 median filter to reduce the noise. Since we are mainly concerned with large-scale features (greater than 50 pixels), this processing will not affect our results. An emission height of 275 km was assumed to project the images into geographic coordinates. Note that the intensity scales shown in Figures 1 and 2 are in arbitrary units related to CCD pixel counts. The different scales reflect changes that were made to the camera system between nights. As can be seen from the model calculations based on the digisonde data shown in the bottom panel of Figure 3, we expect the emissions to have roughly equivalent intensities on the two nights.
 In both nights, the emissions are modulated by a depleted region with a horizontal wavelength on the order of 500 km that seems to progress towards the southwest. The estimated velocities are 83 m/s for 3–4 July and 93 m/s for 4–5 July. The orientation and velocity are essentially identical to similar images taken over Puerto Rico, Japan, and Hawaii [e.g., Kubota et al., 2000; Kelley and Makela, 2001; Kelley et al., 2002]. The spatial scale can be noted with reference to the map of Greece and Turkey, also shown in these two figures, and by noting that each image covers an area of approximately 1000 km × 1000 km. The rectangular box between Crete and the mainland (above Milos) is our best estimate of the E-region scattering volume observed by the SESCAT radar, and the box just to the north of Crete is an estimate of the F-region area connected to that region via the magnetic flux tubes. Local time is 3 hours later than UT.
 Since the image extends to the F region over Athens, we can compare it with data from the ionosonde located there. This is done for both events and the results are shown in Figure 3, where line plots provide various F and E-region parameters. The panels show f0F2, f0Es, hmF2, and the normalized 630.0-nm emission calculated from the ionosonde data, respectively. The latter is calculated using a Chapman profile based on NmF2 and hmF2, along with neutral parameters from MSIS-86, plasma parameters from IRI-95, and rate and transition coefficients from Link and Cogger  to estimate the volume-emission rate. The intensity profile is then integrated along the ray path to give 630.0-nm intensities in Rayleighs. This method is discussed in further detail in the work of Makela et al. . The light intensity detected by CASI, averaged in a box 100 km × 100 km over Athens and normalized to the ionosonde data, is shown as diamonds in the bottom panel of Figure 3b. This is not done for the night of 3–4 July, as too few images were obtained for a reliable normalization to the ionosonde data to be performed.
 Panel 2 in the ionosonde data of Figure 3 shows that sporadic-E events occurred on both nights. f0Es reached 8 MHz on the first night and 6 MHz on the second. Since F-region data were obtained as well, the E-layer ionization must have been structured, allowing the ionosonde signal to penetrate to the F region. Each such event was also accompanied by strong backscatter, detected by SESCAT. During the night of 3–4 July, SESCAT observed echoes for only about an hour, from 2030 to 2135 UT, at about the same time when the Athens Digisonde recorded the strongest sporadic-E layers of that night, with layer critical frequencies, f0Es, reaching 8 MHz. Note that exact temporal agreement is not expected due to spatial separation and requirements for the instability. The SESCAT spectra and cross-spectral observations are summarized for the 3–4 July event using the three Doppler spectrograms illustrated in Figure 4. The Doppler power spectrogram (Doppler velocity-time-intensity) is shown in the upper panel for only the west receiver since the east receiver is identical. Note that the bright spots and narrow lines across the spectrum are irrelevant to E-region scatter as they relate to meteor echoes.
 The entire backscatter event in Figure 4 is characterized by relatively large, negative-mean Doppler velocities (motions away), ranging from about −50 to −200 m/s. The mean velocities approached −250 m/s for about 10 min from 2235 to 2245 UT when a strong localized region of backscatter becomes unstable to the Farley-Buneman instability [e.g., Haldoupis et al., 1997]. These large, negative velocities are consistent with the notion of an eastward polarization electric field which sets in within a patchy sporadic-E layer, in accordance with the polarization process proposed by Haldoupis et al. . This field, which can be maintained through a field-aligned current-closure system [e.g., Shalimov et al., 1998; Hysell and Burcham, 2000; Tsunoda and Cosgrove, 2001], can drive northward and upward E × B electron drifts, which at times can exceed the ion-acoustic speed threshold of the Farley-Buneman instability.
 Finally, the cross-phase spectrogram in the lower panel shows, at times, a progressive change in color (say, from dark blue to light blue and then to green) which translates to a westward bulk motion of the unstable plasma structures as they drift with the neutral wind because of the high collision frequency. More detailed analysis of the cross-phase changes with time (dΦ/dt) inside various Doppler bands show that in this event the westward bulk motions range from about 20 m/s to 90 m/s, with the larger speeds referring to the most negative Doppler shifts seen from about 2035 to 2045 UT. This supports the postulation that the westward wind is likely responsible for the strong polarization electric fields inside sporadic Es plasma patches, which then become unstable, and it is not necessary to appeal to highly elongated (N-S) patches [Shalimov et al., 1998].
 During the night of 4–5 July, SESCAT observed echoes only during a 40-min time interval from about 2150 to 2230 UT. This scattering event also occurred at about the same time as when the Athens Digisonde observed a burst in sporadic-E activity, with f0Es reaching values somewhat above 5 MHz. The SESCAT observations are presented in Figure 5 in the same format as in Figure 4. The similarities between the two events are rather striking.
 At E-region heights we may assume that the ion clouds move with the neutral wind. Using the interferometry estimate of the cross-beam motion, the temporal duration of the sporadic-E patches in both the ionosonde and radar data suggests a scale size of 50–150 km. The internal structure of the backscatter signal in Figures 4 and 5 suggest an even finer patchiness down to scales of 10 km. If electrically polarized as the data strongly suggests, we argue below that the electric fields should easily map to the local F region.
 The series of Athens ionograms in Figure 6 shows the development of structure in the ionosphere on the night of 3–4 July and the occurrence of both spread Es and spread F at about the same time as SESCAT observed strong echoes. As the sporadic-E layer developed near 100 km, its trace began to spread, indicating an irregular layer. Likewise, we see that the F-region trace also began to spread. As shown in Figure 7, a similar behavior occurred on the following night of 4–5 July, again at about the same time when plasma instabilities and large electric fields were detected by SESCAT. In Paper 1, a more extended data set is used to show that these two events are not unique.
 The two events described above bear a striking resemblance to the one reported by Swartz et al.  over the Caribbean. In that event, a strong and highly structured sporadic-E layer was detected for a period of 30 min using the Arecibo incoherent-scatter radar. Two coherent-scatter radars, also on Puerto Rico, detected strong 50-MHz backscatter from the layer with Doppler shifts exceeding 200 m/s at times. At higher altitudes, the radar detected regions of the F layer which were elevated by 50 km and exhibited highly variable line-of-sight Doppler shifts exceeding 100 m/s. Since these measurements were obtained at a zenith angle of only 15°, much higher velocities are indicated across the magnetic field. Although no all-sky images were obtained on this night, the uplifted regions would almost certainly have exhibited low levels of 630.0-nm airglow such as those found on the nights reported here. Generation of such large electric fields by an internal F-region process acting alone would seemingly be very difficult, since the electric field fluctuation due to the instability cannot exceed UB where U is the driving wind velocity in the thermosphere. Wind speeds in excess of 300 m/s are very unlikely and we conclude that E-region electric fields must be involved.
 Having the Arecibo-measured plasma density allowed Swartz et al.  to calculate the height-integrated Pedersen conductivities in the E and F layers. For most of the night, the ratio of F to E-layer conductivity was 5 or greater, but when the combination of an uplifted F layer and an intense sporadic-E layer were present at the same time, the ratio ranged from 1 to 2. For the events studied here, we have calculated the F-layer conductivity from the airglow intensity using the method described by Makela and Kelley  and find values in the range of 1–2 Siemens in the low-airglow regions. The E-layer contribution is more difficult to estimate but using the peak-density values from the ionosonde data and reasonable values for the layer thickness and altitude, the ratio of F to E-layer Pedersen conductivity estimates range from 1.5 to 10. Thus the F layer is capable of electrically driving Pedersen currents in the E layer most of the time in both events, although at times of strong uplift an F-region driver is marginal.
 But what about the other direction? Can the E layer electrically drive the system? Dagg  was the first to discuss coupling of E-region electric fields to the F region. The first quantitative study of the mapping process was carried out by Farley . He introduced the substitution z′ = z(σp/σo)1/2, where σo is the parallel conductivity and σp is the Pedersen conductivity, for the direction parallel to the magnetic field. This converts the Laplace equation for the potential to an isotropic form, the implication being that a structure with a typical scale across the magnetic field of λ will map along the field lines to a height λ(σo/σp)1/2. Since this ratio is on the order of 50 or greater above 100 km [Kelley, 1989], reasonable mapping should occur for scales as small as 4 km at E-region heights. Heelis et al.  and Heelis and Vickrey  extended this work to include diffusion and recombination effects but concentrated on mapping from the F region to the E region. Their one example of upward coupling indicates a 10-km scale electric-field structure at an altitude of 100 km should map to the base of the F layer with at most a 50% attenuation. In our case, we have the radar evidence for large E-region fields, which thus must exist in the face of diffusion and recombination in the source region. We have already argued that the Swartz et al.  measurements of high plasma-drift velocities in the F region point to an E-region source.
 Although not directly quoted in the publication by Swartz et al. , W. E. Swartz (personal communication, 2003) found a height-integrated E-region Hall conductivity of several Siemens, which is the same order as estimates we have made for the present events. Thus it seems very plausible that the polarization electric fields in patchy sporadic E, thought to generate the Farley-Buneman instability at midlatitudes [Haldoupis et al., 1997] and possibly a primary gradient drift instability as well (C. Haldoupis, personal communication, 2003), can create large electric fields which map up to the F region. These electric fields are quite large by midlatitude standards and could easily create F-region density irregularities by advection of the bottomside F layer by rising and falling plasma regions. For example, a 5 mV/m localized electric field acting on a 20-km gradient scale length yields a 50% relative density irregularity in about two minutes. In Figure 7 of Swartz et al. , 40-km uplifts of density contours were found with the same horizontal scale size as the high Doppler shifts. Such a distorted bottomside ionosphere with localized vertical features would clearly provide an excellent source of spread F. Such a coupling process is more efficient when found underneath a region of plasma uplift and low airglow due to the reduced Pedersen conductivity. A detailed discussion of this mechanism is given in Paper 1.
 The unique feature of these data and that of Swartz et al.  is that a strong, unstable sporadic-E layer existed, and we concentrate here and in Paper 1 on its implications. We must point out, however, that by no means do all medium-scale TIDs and structured F regions have associated sporadic-E layers beneath. A good example of an event with a highly structured F region without sporadic E below was studied by Mathews et al.  and later by Kelley et al. . As noted above, we believe events of this sort are driven by a secondary F-region generalized E × B instability.
 As a final comparison with the Swartz et al.  case, and one which may be quite crucial, note that the strong VHF signals were collocated with a considerable enhancement in the E-region plasma density. These authors pointed out that it is very difficult to imagine an F-region process that would enhance the plasma density in the E region below.
 Further experiments are clearly necessary, but we put forward the following chain of events as our best working hypothesis for these events and those reported in Paper 1. A detailed theoretical study is beyond the scope of this work but we hope these ideas will inspire more work on E and F-region coupling. (1) Midlatitude, classical sporadic-E events occurred over Greece on both nights. (2) Due to the neutral wind in the E region and the patchy nature of the sporadic-E layer, the layer polarized, forming large electric fields at intermediate scales and local plasma instabilities at smaller scales. (3) The intermediate-scale fields map to the F region due to the enhanced E-region conductivity and reduced F-region values, particularly in regions of F-region uplift (low airglow). The F-layer uplift reduces its Pedersen conductivity, which then has less of a shorting effect on the E-region source, enhancing the process further. (4) The mapped electric fields generate F-region structure, leading to midlatitude spread F.
 This hypothesis is expanded and further supported in the companion paper. It seems clear that patchy and highly polarized sporadic E (as evidenced by VHF echoes) can affect the F region above at intermediate scales. We turn now to the relationship, if any, between the existence and motion of the sporadic-E layer and the mesoscale airglow depletion above. One possibility is that they are only accidentally related. In this model, the TIDs travel independently in their usual southwestward manner and, when a sporadic-E patch is below a low-airglow, low-Pedersen conductivity zone, strong coupling occurs, including spread F. The westward E-region wind we deduce from the VHF interferometry results would allow the sporadic-E patch to remain under the F-region TID longer than one might expect. With production of F-region irregularities in only a few minutes, as estimated above, exact velocity matching is not required to produce spread F.
 A more interesting possibility, discussed in detail in Paper 1, is that the westward E-region wind polarized the entire sporadic-E patch and created a small eastward electric field. The Hall drift of the electrons in the E region then, in turn, caused a much larger poleward electric field. The resulting westward F-region drift kept the two regions in contact. A fascinating aspect of this idea is that there is a feedback mechanism which may amplify the effect. The weak eastward polarization electric field causes an uplift in the F layer, which reduces the F-region conductivity and enhances the efficiency of the mapping process from the E region, resulting in the feedback.
 The authors would like to thank Manolis and Stella Mallis, as well as Kαπϵταν Tασoς himself, for their hospitality and assistance in Milos. JJM acknowledges support from a National Research Council Research Associateship Award at the Naval Research Laboratory. Research at Cornell University was supported by the Atmospheric Science Division of the National Science Foundation under grant ATM-0000196. Partial support for this work was provided by the European Office of Aerospace Research and Development (EOARD), Air Force Office of Scientific Research, Air Force Research Laboratory, under contracts F61775-01-WE004 and FA8655-03-1-3028 to C. Haldoupis.
 Arthur Richmond thanks Erhan Kudeki and another reviewer for their assistance in evaluating this paper.