Evidence is presented that supports the suggestion by Tsunoda (2006) that a polarization electric field, if generated by a sporadic-E (Es) layer instability (Cosgrove and Tsunoda, 2002), should map to the base of the F layer and seed equatorial plasma bubbles. Seeding, which leads to bubble development, seems to be a four-step process operating in the bottomside F layer: (1) amplification of seed perturbations in a region of westward-drifting plasma, which is found below the velocity-shear node; (2) upward transference of this modulation through a rotational velocity shear to eastward-drifting plasma; (3) further seed amplification, when the F layer begins to descend; and (4) bubble growth from the seed via the Rayleigh-Taylor instability. The time available for interaction between F-region plasma and elongated channels of enhanced polarization electric field (the latter formed by the mentioned instability) appears to be crucial for seed amplification. The need for a descending F layer to increase interaction time with the eastward-drifting plasma is appealing because plasma bubbles display a propensity to appear after, and not during, the postsunset rise of the F layer. Although plausible, the need for Es presence, perhaps low solar activity, and a multistep process suggest that this mechanism may be a more occasional contributor than the collisional-shear instability (Hysell and Kudeki, 2004).
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.
 Plasma bubbles are geomagnetic flux tubes that are depleted of plasma relative to that of the background nighttime equatorial F layer [e.g., Tsunoda, 1980a]. With radar backscatter “plumes” as tracers of plasma bubbles [e.g., Tsunoda, 1980b], bubbles have been shown experimentally to develop from seed perturbations in the bottomside of the F layer upward into the topside of the F layer [Tsunoda, 1981; Tsunoda and White, 1981; Tsunoda et al., 1982]. Current research interests are driven largely by space-weather concerns, such as the deleterious effects on communication and navigation, which are produced by radio-wave scintillations [e.g., Basu et al., 2002]. The most intense and far-reaching effects are produced by kilometer-scale irregularities that are embedded in bubbles. A high-priority research task, in this regard, is to identify the source mechanism responsible for day-to-day variability in development of bubbles and associated plasma structure. (Below, we refer to all plasma structure in the F layer as equatorial spread F, or ESF).
 The identity of the source mechanism continues to be sought, without success, mostly through examinations of parameters that describe the Rayleigh-Taylor (RT) instability [e.g., Sultan, 1996]. Of these, the postsunset rise (PSSR) of the F layer appears to be most influential [e.g., Fejer et al., 1999]; Mendillo et al., 2001]. Curiously, the PSSR, although correlated climatologically with ESF occurrence, can appear not to have control when evaluated on a day-to-day basis [Fejer et al., 1999; Hysell and Burcham, 2002]. This evaporation of apparent ESF control by PSSR, when examined with a shorter timescale, is curious because ESF, which occurs at times distant from the PSSR, does seem to be causally related to reversals in transport of the equatorial F layer, from downward to upward, at least during periods of low solar activity [Chandra and Rastogi, 1972; Bowman, 1974; Fejer et al., 1976; Rastogi and Woodman, 1978; Niranjan et al., 2003]. Niranjan et al. , for example, found that 80% of the occurrences of ESF in the postmidnight sector (during summer months at Waltair and low solar activity) were related to altitude increases of the F layer. Viewing these findings together, we are led to the possibility that inroads into the identity question could be made by examining reasons why ESF development may be more complex in the evening sector.
 Another role for velocity shear could be through some kind of control over seeds that become bubbles (through amplification by the RT instability). In this regard, likely importance of velocity shear has been underscored by recent discovery of a collisional-shear instability [Hysell and Kudeki, 2004]. By this mechanism, a seed with a preferred horizontal wavelength, determined by the spatial scale of the velocity shear, is amplified in the vicinity of the velocity-shear node. This mechanism is attractive because quasiperiodic seed structures with horizontal wavelengths that appear comparable to those predicted by theory have been both observed and simulated by computer to occur in the vicinity of the velocity shear node [Hysell and Kudeki, 2004; Hysell et al., 2005a, 2005b, 2006].
1.3. Seeding by the Sporadic-E Layer Instability
 Besides the mentioned processes, Tsunoda  has suggested that another role could be played by velocity shear in seeding plasma bubbles via the RT instability. He showed that a polarization (), if generated by a sporadic-E layer (EsL) instability [Cosgrove and Tsunoda, 2002], should map along geomagnetic field () lines to the base of the F layer and modulate the plasma in the form of large-scale wave structure (LSWS), which then spawns plasma bubbles from its crests [Tsunoda, 2005]. To illustrate, the mapping geometry from Tsunoda is shown in Figure 1. The EsL instability produces one or more channels of enhanced , which would be directed northeastward (southwestward) in a raised (lowered) Es band; an example band of raised Es is shown in the sketch as a light gray band. This channel of enhanced will map along the dashed lines to the dip equator, where it appears in the sketch as a dark gray band that is tilted downward toward the east. Plasma within this channel would be transported upward and westward over the dip equator.
 Because large-scale Es patches (from which bands form) tend to drift equatorward, the interactive region between an channel and the base of the F layer (see sketch) should move westward. (We note, in passing, that Es patches could be present in either or both of the conjugate E regions. Both will contribute to increasing the growth rate of the EsL instability, and they will develop mirrored patterns of raised and lowered bands.) Growth time of a seed perturbation in the interactive region can be lengthened, if the F-region plasma drifts westward; infinite interaction time, often referred to as “spatial resonance” [Whitehead, 1971] would occur if plasma drift can be matched to that of the interactive region.
 A sketch of this concept in the vertical equatorial plane is shown in Figure 2. The region between straight dashed lines represents the channel of enhanced at time t1. It is shown to intersect a modulated isodensity contour in the bottomside F layer (wavy dashed curve). The transport of plasma within that channel is indicated by an arrow labeled V. Later, by time t2, the channel would be displaced to the location of the straight solid lines. With westward drift, the modulated isodensity contour has also been displaced, as shown by the wavy solid curve. Because the same perturbation remains within the channel, its amplitude would grow, as sketched.
 In Tsunoda , the velocity-shear node demarcates a region of eastward plasma drift, where field-line-integrated Pedersen conductivity of the F layer () is greater than that of the E layer () from that of westward drift, where the converse is true [e.g., Haerendel et al., 1992]. This interpretation seems consistent with evening model results, which place the condition at an altitude close to 300 km [Stephan et al., 2002]. Because EsL instability operates on lines, whose field-line-integrated Hall conductivity of the Es structure () is comparable to or greater than , the instability growth rate should be largest in the region of westward plasma drift [Tsunoda, 2006]. Hence with largest growth rates and longest interaction times, conditions for seed growth would seem to be best below the velocity-shear node. (Another geometry for extended interaction time is described in section 3.3).
1.4. Experimental Validation
 To validate experimentally the concepts put forth in Tsunoda , we have analyzed 155.5 MHz data obtained with ALTAIR, a fully steerable incoherent-scatter (IS) radar, located on Roi-Namur island in the Kwajalein atoll [Tsunoda et al., 1979]. The approach was to use coherent echoes to characterize behavior near the base of the F layer, where IS measurements of low N are extremely difficult. Key findings for validation were extracted from time sequences of spatial maps that captured the distribution of both IS and coherent scatter. Maps of this kind were constructed from data collected, while keeping the antenna beam perpendicular to during west-to-east scans. Data used for this study were obtained on several summer nights in 1977 and 1978 during previously conducted field campaigns. The pulse width was 30 μs (4.5 km range resolution). The results presented in this paper are shown to be consistent with the notion of seeding plasma bubbles with amplified by the EsL instability.
 Besides validation, several new findings have arisen from this study. We have found that seeding of plasma bubbles appears to be a four-step process: (1) amplification of seed perturbations in a region of westward-drifting plasma found below the velocity-shear node, (2) an upward transference of this modulation through a rotational velocity shear to eastward-drifting plasma, (3) further seed amplification, when the F layer begins to descend, and (4) plasma bubble development from the amplified seeds via the RT instability. The first step appears to be as described in Tsunoda . The remaining three steps had not been considered previously. Perhaps most appealing is the finding that a reversal of F-layer transport from upward to downward, after the PSSR, is important for the generation of plasma bubbles (and ESF). The reason why ESF does not develop more often during the PSSR, when the gradient drift (i.e., -driven rather than gravity-driven interchange) instability should be active, has eluded researchers until now.
 To validate the mechanism described in Tsunoda , we need a description of velocity shear, isodensity contours in the bottomside of the F layer (or ), and plasma bubbles. We have selected ALTAIR data from 1977 and 1978 for this purpose because coherent echoes in the vicinity of the bottomside F layer, referred to as bottom-type (BT) and bottomside (BS) echoes [Woodman, 1994; Hysell and Burcham, 1998], seem to be most prevalent during periods of low solar activity [e.g., Hysell and Burcham, 1998]. (All data presented in this paper, except where indicated, have not been published previously.) Use of data from periods of low solar activity also allows us to compare the results with those from the postmidnight sector and those obtained by other researchers.
 The database consists of eight nights in August 1977 and eight nights from late July and early August 1978. Of these, data useful for this study, that is, those collected prior to 2040 LT [0930 universal time (UT)], were obtained on five nights each in 1977 and 1978 (LT leads UT by 11 hours 10 min). Echo-type activity is summarized in Table 1. We see that BT or BS echoes were observed on all 10 nights, which agrees with the finding over Jicamarca that these echo types are commonplace during nights of low solar activity [Hysell and Burcham, 1998]. In most cases, BT and BS echoes did not appear at the same times. Because these ALTAIR data do not contain Doppler information, BT echoes were identified as such by their location well below (typically, 50 km) the 105 el/cm3 isodensity contour associated with the bottomside of the F layer and, where possible, by the direction of zonal displacement of recognizable features.
Table 1. Summary of Echo Activity
Description of Echo Activity
Examples from these three nights are presented in this paper.
 It should be evident that the success of this analysis depends on the extent to which we can relate the characteristics of BT echoes to those of the background ionosphere. We will attempt to do so by various means. First, we show that the backscatter obtained with ALTAIR at 155.5 MHz is indeed the same type as those obtained with the Jicamarca radar at 50 MHz [Woodman, 1994; Hysell and Burcham, 1998]. We claim equivalence by noting their similar locations in altitude near the base of the F layer and their westward drifts in cases where zonal displacements in recognizable echo features could be discerned. The location of BT echoes in altitude, relative to the background N profile, was first shown by Tsunoda et al.  using ALTAIR data from 18 August 1977 (sunspot number ∼40, ΣKp = 22, Kp = 3). A map from their paper showing the spatial distribution of backscatter is presented in Figure 3. We see that coherent echoes, labeled FAI(F), which we now call BT echoes, came from well below the 105 el/cm3 isodensity contour associated with the bottomside of the F layer. Echolocation between 210 and 240 km altitude is consistent with those of BT echoes detected over Jicamarca, which can be as low as 200 km during solar minimum [Hysell, 2000]. We note that although BT echoes have been inferred to be layer-like from altitude-time-intensity plots, spatial maps such as that in Figure 3 contain valuable information on the spatial distribution of BT echoes. In this example, BT echoes can be described in terms of its tilt, structural detail, and finite zonal extent. We show below how information of this kind can be used to assist with the interpretation.
 If possible, we would like to know whether BT echoes are related to any features in the background N. As an example, the distribution of BT echoes relative to the background N profile is presented in Figure 4, also from Tsunoda et al. . These altitude profiles of echo intensity (after range-squared correction) are from measurements obtained near the center of the scan shown in Figure 3. The scale for echo intensity, along the abscissas, is given in terms of equivalent N (assuming equal electron and ion temperatures, which is reasonable for IS returns from the nighttime F layer). The F layer, clearly visible at altitudes devoid of coherent echoes, is seen to lie well above the BT echoes. That this is the case is seen from the profile in the center panel, where coherent echoes have been suppressed by directing the beam away from perpendicular intersection with . Coherent echoes are again seen in the bottom panel, when the beam was, once again, redirected to achieve perpendicular intersection with . We see that the BT echoes were spread over 50 km in altitude, extending downward into the valley region from the base of the F layer, where N was 2 × 104 el/cm3. The suggestion here is that we could perhaps argue that the top of BT echoes is abutted to the base of the F layer, which would be useful for discerning LSWS at the base.
 However, results from two rocket measurements appear to indicate that the region of field-aligned irregularities (FAI) responsible for BT echoes, while found in the valley region, is not necessarily always abutted to the base of the F layer. Raghavarao et al.  presented N profiles obtained with a rocket launched from Sriharikota, India (5.5° dip latitude) at 1909 LT on 16 February 1982 (solar maximum). During rocket ascent, irregularities were detected at the base of the F layer, 290 km in altitude, where ∇N began to steepen (see their Figure 3). Hysell et al. [2005b], on the other hand, presented radar and rocket measurements from Kwajalein, made on 7 August 2004 (low solar activity), which showed that FAI responsible for BT echoes occurred within the valley region but were not abutted to the base of the F layer. In situ measurements taken at the same time indicated that the FAI appeared below 250 km in altitude, where the drift was small or westward. It, therefore, appears that BT echoes cannot be used, at this time, to locate the base of the F layer.
 On the other hand, the fact that BT echoes can actually occur as a relatively homogeneous layer (e.g., Figure 3) suggests that N associated with the background N profile can play a controlling role in determining the spatial distribution of BT echoes. From other data, including those presented below, we conclude that the BT layer often (if not always) traces out the location of enhanced local ∇N. For example, we see from Figure 3 that the BT layer was tilted slightly, higher on the east side, while the 105 el/cm3 isodensity contour, located 50 km above the top of the BT layer, was more or less horizontal. We thus associate the tilt of the BT layer to a corresponding tilt in isodensity contours. If so, this tilt, which appeared only in lower-N isodensity contours and not in the 105 el/cm3 contour, represents first evidence that can differ at the two altitudes.
2.3. Altitude Modulation of a BT Layer
 The notion that isodensity contours for lower-N values could be altitude modulated, while that for 105 el/cm3 is not, is intriguing because implication is that the responsible for the modulation probably originated in the E region. A perhaps more convincing example, which shows the temporal development of pronounced spatial modulation, occurred on 20 August 1977 (sunspot number ∼30, ΣKp = 16+, Kp = 3−, 2). A sequence of maps, which displays this temporal growth of spatial modulation, is presented in Figure 5. (Times are scan start times.) Two distinct features can be seen to have evolved. One of these appeared near the west edge of the scans and proceeded to develop eastward. The other is located near the east edge of the sector scans.
 The first feature, similar to the tilted BT layer in Figure 3, appeared first in map (b). The leading edge of this feature is seen to extend progressively eastward with time, until this BT layer was overhead of ALTAIR, by the time of map (e). The tilt of this BT layer, higher on the east side, is similar to that in Figure 3. Echo intensity increased by 10 dB (a contour interval) as the BT layer increased in altitude, tilt, and zonal extent. By map (f), the BT layer extended from 230 km in altitude at the west edge of the scan, 100 km west of ALTAIR, to 270 km in altitude at the crest of the modulation, about 50 km east of ALTAIR. (The tilt angle is about 15°.) At the same time, the 105 el/cm3 isodensity contour has remained more or less horizontally stratified.
 The second feature, which is located at the east edge of map (a), is seen to have developed dramatically [maps (b) through (e)] from a relatively thin protrusion not unlike the first feature in map (b) into an altitude-extended S-shaped echo region. Expansion was in both directions, upward and downward about the altitude of the original protrusion in map (a). We see that the S-shaped echo region protrudes westward and slightly upward, appearing to point toward the east end of the above-described tilted BT layer. Finally, we see that the two features become connected in map (f), 30 min later.
 We interpret the observed behavior in terms of two upwellings that were part of LSWS development on this night. Development of one of the upwellings occupied most of the scanned sector and is partially outlined by the tilted BT layer in most maps; it is fully outlined in map (f). The other upwelling is partially described by the S-shaped region with its eastern side falling outside of the scanned sector. The occurrence of an increasingly tilted BT layer is consistent with the growth of an upwelling in the valley region and the development of FAI along the west wall of that upwelling, as described by Tsunoda [1981, 1983]. The source of FAI is believed to be the gradient drift instability driven by an eastward neutral wind () in the moving frame of the plasma. Instability growth rate is larger for plasma that is nearly stationary or drifting westward than for plasma drifting eastward [Kudeki and Bhattacharyya, 1999].
 Our interpretations differ in that we assume the westward component of ∇N is from a relatively smooth distribution of background N, whereas Kudeki and Bhattacharyya  assumed presence of “minivortices” presumably generated near the velocity-shear node. In our view, it is not clear how such vortices would produce preferred occurrences of tilted BT layers that are higher on the east side. On the other hand, it seems conceivable that the occurrence of BT echoes along the east wall in map (f) could be associated with minivortices of some kind, given the spectacular development of S-shaped region.
 To graphically capture the evolution of altitude modulation of the BT layer, we have sketched, in Figure 6, our rendition of how isodensity contours may have become distorted to provide the ∇N necessary to generate FAI and BT echoes. The straight though tilted solid line, labeled a, is intended to represent a smooth visual fit to the 105 el/cm3 isodensity contour in map (a) in Figure 5. The straight dashed line, labeled b,c, which represents the 105 el/cm3 isodensity contours in maps (b) and (c), is seen to be displaced upward from that in map (a), indicating that the F layer was rising at the times associated with those maps. In the same manner, we also conclude that the F layer was no longer rising at times between maps (b) and (c). Given the local times of these measurements (1930 to 2000 LT), upward displacement may have been associated with the PSSR, but the tilt itself may have been associated with an LSWS because a large lift velocity is needed to produce such a noticeable tilt [e.g., Tsunoda and White, 1981].
 The possible presence of a similar tilt in isodensity contours, where BT echoes occurred, is suggested by the gray solid line, which was drawn to connect the starting locations of the BT echoes at the east and west ends of the maps and to provide a reference line from which displacements by a large-scale could be estimated. The similarity of the slope of this gray line with those drawn through the 105 el/cm3 isodensity contours suggests that the tilts could be from a common source. The appearance of the curved isodensity contour in map (d) suggests onset of an upwelling, but from maps (e) and (f), the 105 el/cm3 isodensity contour appears to have returned to being horizontally stratified. Without Doppler information, we cannot tell whether eastward transport of the corresponding plasma was responsible for these changes or not.
 Nonetheless, while the 105 el/cm3 isodensity contour remained well behaved and more or less horizontal, we see from the tracings in Figure 6 that isodensity contours, assumed to be associated with the S-shaped region of BT echoes, could have become substantially modulated in altitude. Using the gray solid line as reference, we see the development of shallow modulation on the west side and dramatic modulation on the east side. In fact, the small upward displacement seen in curve d on the west side is similar to that seen in the 105 el/cm3 isodensity contour. This behavior could be produced by a large-scale, eastward , perhaps with a wavelength of 150 km. The behavior on the east side, however, would have to be produced by transport in the directions shown by the arrows. (The near-vertical growth indicated by the contours could be a consequence of eastward plasma drift above the velocity-shear node.) Growth of modulation was clearly rapid. If we assume the S-shaped region grew nearly ±50 km (in altitude) in 20 min, the average growth velocity would have been about ±40 m/s. The appearance of downward (and eastward) transport is not expected; any eddy cell associated with the velocity shear should exhibit clockwise, not counterclockwise, rotation.
 If we accept the interpretation that BT echoes in Figure 5 were associated with westward components of ∇N, the inferred distortion of underlying isodensity contours leads us to three key conclusions:
 1. Substantial modulation of low-N isodensity contours occurred at altitudes within the valley region near the base of the F layer.
 2. Modulation was confined to altitudes below that of the 105 el/cm3 isodensity contour.
 3. Modulation, in this case, was produced by oppositely directed , that is, southwestward and northeastward.
 These results are consistent with the mechanism proposed in Tsunoda . That is, the responsible for the modulation must have been generated by an Es-driven polarization process rather than one in the F region. The absence of at higher altitudes within the bottomside F layer is consistent with a source mechanism such as the EsL instability, which could function at lower altitudes but not at higher altitudes, because of loading by the F layer [Tsunoda, 2006]. And, the upward and westward (downward and eastward) transport by longitudinally confined and oppositely directed is consistent with the mapping of channels from banded Es structures, as sketched in Figure 1. For the 20 August 1977 event, the oppositely directed could have been associated with a pair of Es bands, one raised and the other lowered relative to the wind-shear node.
2.4. Velocity Shear and Seeding of Plasma Bubbles
 Thus far, the examples have not contained any information about velocity shear or plume development. (The only hint of velocity shear in Figure 5 is the formation of the S-shaped region in which the top portion appears to extend eastward, while the bottom portion appears to grow in westward extent.) We now present evidence for velocity shear, in the form of counterstreaming BT and BS echoes, and for the seeding of backscatter plumes (Plumes have been shown previously to be collocated with plasma bubbles [Tsunoda, 1980a, 1980b]). Using data from 31 July 1978 (sunspot number ∼36, ΣKp = 3−, Kp = 0+), we show how altitude modulation, imposed on plasma located below the velocity-shear node, seems to be transferred to plasma above the velocity-shear node and to lead to the appearance of plumes from the top of the BS layer.
2.4.1. Westward Drift
 The first of two sets of backscatter maps from this data set is presented in Figure 7. The first three maps, with scan start times of 1946:12, 1953:48, and 2001:15 LT, contain a chain of weak BT echoes with peak intensities of 10 dB above IS from N ∼ 106 el/cm3. Weaker, isolated BT echoes were detected in an earlier scan (not shown), at 1938:48 LT. The appearance of BT echoes, starting around 1930 LT, is similar to that in our 20 August 1977 example. Observations made with the 50 MHz Jicamarca radar have indicated first appearances as early as 1900 LT [Hysell and Burcham, 1998, Figure 2]. Westward drift of the BT chain with a speed of about 40 m/s is evident in Figure 7 from displacements of similar features in adjacent maps, shown connected by dashed-line segments. The westward drift of these echoes suggests that they resided below the node of a vertical shear in zonal drift [e.g., Valenzuela et al., 1980; Tsunoda, 1981; Kudeki et al., 1981; Tsunoda et al., 1981]. Westward drift is usually associated with a region where , indicating control by the E-region dynamo [e.g., Kudeki et al., 1981].
2.4.2. Altitude Modulation of BT Echoes
 The small patches of enhanced echoes in the chain were spaced quasiperiodically with a horizontal wavelength of about 75 km; in comparison, the modulation scale in the first example was about 150 km. Shorter wavelengths have been detected with the Jicamarca radar. Hysell et al. [2004, 2005b], for example, found BT echoes to be separated horizontally by about 30 km. They noted that these wavelike features could be indicative of atmospheric gravity waves (AGW) seeding, but because the wavelength appeared also to be comparable to the depth of the shear layer, it could be suggesting a causal relationship between waves and shear, perhaps via a collisional-shear instability [Hysell and Kudeki, 2004]. Although we do not have a measure of shear depth, the observed thickness of the BT echoes, if representative of shear depth [e.g., Hysell et al., 2004], appears to be much less than 75 km.
 It is important to note that as the BT chain became modulated in altitude (left column, maps 3, 4, and 5), the echo enhancements tended to appear on west walls of this oscillation. This behavior, which is similar to the structuring of the west wall of a larger-scale upwelling in Figure 5, lends credence to our earlier interpretation that BT echoes appear to be associated with westward components of ∇N in altitude-modulated isodensity contours in that region. Hence we conclude that isodensity contours developed altitude modulation in a region of westward bulk plasma drift below the velocity-shear node.
2.4.3. Eastward Expansion and Velocity Shear
 The first clear example of BS echoes in this paper is seen in left column, map 4 (2011:32 LT) of Figure 7 as an intrusion from the west side of the scanned sector at altitudes slightly higher than those of BT echoes. These echoes display the typical features of BS echoes [Woodman, 1994; Hysell and Burcham, 1998]. The intensity of BS echoes was 20 dB stronger than that of BT echoes, and IS measurements indicated that they remained within the bottomside of the F layer. Unlike BT echoes, which drifted slowly westward, BS echoes expanded rapidly eastward, as seen in the remaining maps in Figure 7. They were also more dynamic, and their appearance led soon thereafter to the development of plumes.
 There appears to be a difference between eastward expansion of the BS echo region and bulk plasma motion. Displacements of the leading (east) edge of the BS echo region lead to eastward expansion speeds of 203, 343, and 324 m/s. However, displacements of topside features in BS echoes (i.e., plumes, which are discussed below) lead to an eastward drift speed of 90 m/s. In comparison, the average zonal drift over Jicamarca is zero around 1630 LT and increases to 100 m/s by 2100 LT [Fejer et al., 1991]. The drift speed is about 90 m/s at 2030 LT, which corresponds with the times of the maps being discussed. Given the relative stability of zonal plasma drift in the equatorial zone, we have good reason to accept 90 m/s as the eastward drift speed. We can also conclude that the velocity-shear node was located between the BT and BS echoes. Hence the bulk plasma flow seems to have been 40 m/s westward below the node and 90 m/s eastward above the node.
2.4.4. Absence of Initial Modulation in BS Echoes
 Having shown presence of altitude modulation below the velocity-shear node, we wish to establish whether similar modulation was also present above the velocity-shear node. From maps 4 and 5 in left column of Figure 7, we receive the impression that BS echoes expanded eastward without displaying significant initial modulation. The thickness of the BS echo region in map 4, for example, is similar to the unstructured BS echo region in map 5. Only the leading portion of the BS region in map 5 is thicker. Structure, particularly along the topside of this initially smooth BS region in map 5, developed shortly thereafter, in maps 6 through 8 (right column of Figure 7). If we accept this interpretation, we can conclude that quasiperiodic altitude modulation of isodensity contours were imposed only on westward-drifting plasma but not initially on eastward-drifting plasma. This finding is consistent with our earlier finding from Figures 5 and 6 that from the EsL instability could map to the westward-drifting plasma below the velocity-shear node and impress its signature [Tsunoda, 2006].
2.4.5. Plumes From BS Echoes
 The initiation of backscatter plumes from the topside of BS echoes is seen in maps in the right column of Figure 7. Development of these features into altitude-extended plumes is seen in the three maps in Figure 8. Plume development was rapid, from 370 km at 2028:15 LT to 440 km at 2035:58 LT, a growth of 70 km in 8 min, a bubble rise velocity of 146 m/s. In comparison, other bubble rise velocities measured with ALTAIR ranged from 125 to 350 m/s [Tsunoda, 1981]. The westward tilt of developing plumes could be explained in terms of the presence of an eastward in the moving frame of the plasma [e.g., Woodman and La Hoz, 1976]. The development of BS echoes before plumes could be associated with the nature of the seed. The implication is that seed wavelengths comparable to plume spacing were not significant at the time of initial appearance of the BS echoes. This statement is consistent with our earlier observation that, initially, BS echoes appeared to be smooth and unmodulated.
2.4.6. Rotational Shear and Seeding
 If the isodensity contours that underlie BS echoes were initially unmodulated, how was the modulation transferred from BT to BS echoes? We suggest that transference occurred because plasma flow was rotational and not simply a counterstreaming in antiparallel directions. To illustrate, we have traced and superimposed portions of the 10-dB contours from the first six maps in Figure 7 into Figure 9. The contours are labeled with corresponding map numbers. The four enhanced BT regions from map 1 are drawn as solid black regions. The leading BT echo, initially located around 270 km west of ALTAIR, is seen to have drifted westward and upward and disappeared after map 3. The westward displacement of this patch between maps 1 and 2 was larger than between maps 2 and 3, which is consistent with a shear-associated decrease in zonal drift with increasing altitude. The second BT echo, initially located 130 km west of ALTAIR, also drifted westward and upward. It appears that this patch came to a standstill in map 4, which indicates that the shear node was close to 300 km in altitude. The third BT echo, initially located around 80 km east of ALTAIR, drifted westward (through map 5) and then reversed to an eastward drift between maps 5 and 6. This reversal places the node around 280 km.
 Given that quasiperiodically modulated isodensity contours were associated with BT echoes, they, too, must have been transported upward and into the region of eastward plasma flow. Hence in this manner, modulation imposed on westward-drifting plasma becomes modulation in eastward-drifting plasma. The lowering of the velocity-shear node with time is partly due to a continued decay of N in the E region and perhaps also to a reversal of the background (although not measured) from eastward to westward, which should occur at the end of the PSSR.
2.4.7. Downward Protrusion
 Given the formation of the S-shaped region in Figure 5, it does not seem far-fetched to interpret the downward protrusion first seen in map 5 of Figure 7 as arising from a similar process. Both features appear explainable in terms of plasma transport that was initially directed downward and eastward. That is, a bulge seems to have formed by the downward transport of higher-N plasma, most likely from above the base of the F layer. This downward transport process seems to have been short-lived, as suggested by the three curves that appear to be nearly on top of one another in the sketch in Figure 6. In contrast, upward transport continued with a slight clockwise rotation, as also seen from the sketch in Figure 6. The downward protrusion in Figure 7, once formed, rotated counterclockwise, suggesting that the initial thrust, downward and eastward, changed to one that became eastward and upward.
 The pattern of echoes changed significantly between maps 1 and 2 in Figure 8, which were taken about 23 min apart. It appears that the BS region was transported eastward more rapidly than the finger-like region. The connecting region, which created the hook-like shape in map 1, appears to have been ‘dragged’ eastward more rapidly than the echo region at lower altitudes, thus creating what appears to be a more extended connector between the bulge and the last of the BS echoes, as seen in map 2. The connecting region appears to be dissipating in map 3 of Figure 8. We are thus left with the impression that the region between the eastern end of the hook-like region and the remnant of the BT layer (located less than 100 km east of ALTAIR) is a zone where westward drift changes to eastward drift. If a westward drift persists, it must be occurring at altitudes below 200 km.
3.1. Altitude Modulation in the Valley Region
3.1.1. Source of Echo Modulation
 The strength of the evidence presented herein depends on the correctness of our interpretation that the shapes of BT echoes reflect similar distortions in the underlying isodensity contours. The basis for this interpretation depends on the nature of the source mechanism responsible for BT echoes. The consensus mechanism seems to be the gradient drift instability [e.g., Tsunoda, 1981, 1983; Kudeki and Bhattacharyya, 1999; Hysell, 2000]; by its label, the sources of free energy are ∇N and plasma drift relative to the neutrals. In the case of BT echoes, this drift can be large because the plasma can be either stationary (at the velocity-shear node) or in retrograde motion in the valley region. The question becomes what is the source of the westward component of ∇N?
 Two sources for a westward component of N have been proposed: (1) tilted isodensity contours along the west walls of upwellings [Tsunoda, 1981, 1983], and (2) structure produced by “microvortices,” presumably spawned by the background velocity shear [Kudeki and Bhattacharyya, 1999]. Occurrences of tilts with preferred orientation in BT layers or in shapes of echo enhancements in modulated BT layers (Figures 3, 5, and 7) would seem to favor the first interpretation. Severe distortions in a BT layer, such as the S-shaped region in Figure 5, are more easily explained in terms of localized distortions in isodensity contours produced by a corresponding pattern for rather than in terms of localized distortions in the velocity-shear pattern. That is, the S-shaped region would seem to require counterclockwise vortical flow rather than the expected clockwise vortical flow (because eastward drift region occurs above the westward drift region). In the end, however, this discussion may be moot because even distortions in shear flow require a distorted pattern of .
3.1.2. Key Conclusion
 The important conclusion is that structured can occur in the valley region without a comparable occurring in the bottomside of the F layer, at least not to the altitude, where N reaches 105 el/cm3. The implication is that this must be generated by a polarization process in the E region because the valley region is threaded by lines controlled by the E-region dynamo [Tsunoda, 2006]. We can also conclude that an capable of modulating the 105 el/cm3 isodensity contour was not generated by polarization of the F-region-dynamo-driven or the gravitationally driven Pedersen current.
 This behavior is well matched to the scenario proposed by Tsunoda . That is, in the valley region could have been generated by the EsL instability, but above the velocity-shear node, it should be damped because of loading by the large on those lines. Even in cases when is not particularly large, longer interaction times should produce a more substantial altitude modulation, and longer interaction times are expected between equatorward-drifting Es bands and westward-drifting, but not eastward-drifting, F-region plasma, at least not during the PSSR. (This “spatial resonance” condition can be visualized to occur in the “interactive region” in Figure 1).
 The underlying process involves Hall polarization of an Es band [Haldoupis et al., 1996; Tsunoda, 1998] and mapping of the generated to the F region. The importance of the coupling of this was pointed out by Tsunoda and Cosgrove , when considering sources for plasma structure in the nighttime midlatitude F layer. We have also found, during the preparation of this manuscript, that Prakash  had already considered the possibility of seeding ESF by the generation of through Hall polarization and the effects of that when mapped to the bottomside of the equatorial F layer. His treatment, however, differs from that presented in Tsunoda  in the following ways. He considered Hall polarization on lines controlled by the F-region dynamo, that is, above the velocity-shear node, and did not include the EsL instability. Hence his mechanism would not explain the observations described in this paper. He also used an AGW to modulate E-region conductivity, which led to Hall polarization of the E-region dynamo. (The role of conductivity modulation by an AGW is not considered in this paper.) Hall polarization in Tsunoda is produced by local altitude displacements of Es bands about a wind-shear node.
 Two interesting predictions can be made from the scenario described in Tsunoda . First, development of modulation in the valley region must occur around or after E-region sunset because a solar-produced E layer would damp the EsL instability. Moreover, presence of westward drift would not allow daytime seed structure, if any, to be transported into the postsunset sector. Hence precursor signatures of seed structure that lead to ESF and plumes are not likely to be detected before sunset. And second, the largest modulation for a given from the EsL instability would develop when the steepest N, which would be associated with the base of the F layer, enters the region of westward flow. This situation should arise when an intense Es layer is present. In this case, would be large, which would place the velocity-shear node at a higher altitude, where , presumably above the base of the F layer.
3.2. Related Topics
3.2.1. Disruption of Existing Es Layers
 The scenario for seeding plasma bubbles put forth in Tsunoda  requires the presence of an Es layer on the lines that basically thread the BT layer. Some discussion, therefore, is needed about observed disruptions of existing Es layers that occur during active PSSR conditions [Abdu et al., 2003; Carrasco et al., 2005]. These disruptions were found to commence around 1800 LT and last for 3 hours over Fortaleza in association with a large eastward during the PSSR. These results are relevant to our study here because the E region over Fortaleza maps along to near the base of the F layer over the magnetic dip equator. Abdu et al.  noted that the eastward did not appear to be directly related to the disruption. Instead, they found the presence of a westward in the E region and suggested that formation of the Es layer could be from that westward [see Rowe and Gieraltowski, 1974; Abdu and Batista, 1977] with disruption from an upward (and northward) . These findings are important because the observed disruptions suggest that the EsL instability may not be able to operate when the Es layer is disrupted. If so, the findings of this paper may turn out to be applicable mostly under conditions found during solar minimum years when the PSSR is mild.
3.2.2. Fringe Electric Fields in Valley Region
 Several papers have been written to suggest the presence of F-region-generated in the valley region. Zalesak and Ossakow , for example, suggested that the extremely low N found inside of large plasma bubbles could be a consequence of extraction of E-region plasma by the fringe associated with large plasma bubbles. Similar simulations have been reported by Sekar et al. . Kherani et al.  also performed a similar simulation to show that rising equatorial structures in the upper E region could be transported upward by these fringe fields. Patra et al. [2004, 2005] have invoked F-region driven fringe fields to explain the observed weakening or disappearance of FAI in the E region in conjunction with the initial development of radar plumes [Hysell et al., 1994; Rao et al., 1997; Chau et al., 2002; Patra et al., 2004, 2005]. The models used or referred to in all of these cases, however, contain insulated backgrounds, where and parallel currents were set to zero. By doing so, loading effects on the RT instability (and F-region dynamo) are forbidden. On the other hand, it should be clear that plasma drifts westward below the velocity-shear node because is large, and parallel currents flow to close in the E region. By the same token, the fringe from the F-region dynamo should be small in the valley region. Hence interpretations of a fringe extending below the velocity-shear node should be reexamined.
3.3. Modulation Growth in Eastward Flow Region
 The notion that a rotational velocity shear allows the transfer of modulation from a westward- to an eastward-drifting region requires further discussion. Upward transport may be all that is needed to simply plant the seed in the upper region, but a continued interaction with the from the E region may be necessary, given that the growth rate of the RT instability is still quite small at these altitudes. If we assume that the Es bands continue to drift equatorward at the same speed, the interaction time with the westward-drifting F-layer plasma must decrease because westward drift speed will decrease as the modulation moves upward toward the velocity-shear node. Interaction time will continue to decrease as the modulation passes through the node and into regions of increasing eastward drift.
 Modulation growth can only be sustained if the F layer begins to descend, that is, if it develops a downward velocity. For example, if the modulation reaches the velocity-shear node, where zonal drift is zero, the region of intersection with an channel will appear to drift westward at that altitude. If we allow the velocity-shear node to drift downward in altitude, the westward drift of the region of intersection with the channel will be decreased. Spatial resonance for this case is sketched in Figure 10. The channels shown here are identical to those in Figure 2. This time, the isodensity contour is displaced downward (and slightly eastward). As before, we see that continued interaction is also possible when the F layer is descending. This requirement, in fact, seems to be consistent with the development of plumes in Figures 7 and 8, when the BS layer appears to be descending.
 This need for a descending F layer to amplify modulation in eastward-drifting (or stationary) plasma is a key finding because it offers, for the first time, an explanation for the virtually repeatable appearance of the first plume of the night, just after the zonal reverses from eastward to westward. There is a strong tendency for plumes not to appear during the PSSR but to appear around the time of reversal, or shortly thereafter, as seen in numerous altitude-time-intensity plots of equatorial F-region radar backscatter [e.g., Woodman and La Hoz, 1976; Kelley et al., 1981; Hysell and Burcham, 1998, 2002; Yokoyama et al., 2004]. In fact, the onset of ESF in ionograms has been shown to display this same behavior [e.g., Osborne, 1951]. Onset of ESF and plumes appears to coincide with F-layer sunset [Osborne, 1951; Yokoyama et al., 2004], which turns out to coincide with the time when the F layer reaches maximum altitude. We are unaware, however, of any underlying physics that might be indicated by this relationship. On the other hand, this behavior has previously been attributed to a spatial-resonance type of interaction [Whitehead, 1971] between an AGW and F-layer plasma [e.g., Röttger, 1976; Kelley et al., 1981].
 To place the seeding process in perspective, we begin with the finding that zonal plasma drift in the peak of the equatorial F layer reverses from westward to eastward around 1600 LT [e.g., Fejer et al., 1981]. And, as would be expected, velocity shear is observed as early as 1600 LT [Hysell et al., 2005a]. Plasma drift during the day, prior to 1600 LT, is upward and westward over the dip equator. The implication is that the F-region dynamo takes over control of the F-layer peak from the E-region dynamo, around 1600 LT. With passage of the solar terminator, the E-region dynamo should weaken and the velocity-shear node should move to lower altitudes. Velocity shear appears to persist until around 2200 LT [Eccles et al., 1999; Hysell et al., 2005a]. Plasma drift during the remainder of the night is downward and eastward.
 A vortical flow pattern is observed [e.g., Tsunoda et al., 1981] because incompressible flow (i.e., ∇ × = 0) requires closure of streamlines driven by the two dynamo processes. That is, the nighttime downward and eastward flow must close with the upward and westward flow that is still driven by the E-region dynamo as late as 2200 LT. Without disruption by the solar terminator, streamlines would close around 1600 LT, when the F-region dynamo enters as a player. However, there is a westward gradient in conductivity at sunset, which sets up a prereversal enhancement of the zonal . The resultant upward flow is strong enough to cause closure of the vortex shortly after E-region sunset, rather than at 1600 LT.
 The scenario described in this paper comes into play around this time. Seed structure, perhaps ubiquitous but weak, is amplified by the EsL instability. The resultant then maps up to the base of the F layer, where it produces × motion. However, significant modulation develops only where there is sufficient interaction time, that is, where the background plasma is drifting westward (and upward). Upward transport, however, causes the modulation to enter regions of decreasing westward drift and eventually into a region of eastward drift. Modulation growth would stop or even decrease if the modulation is transported into a region where the sign of is reversed. Once the PSSR ends and the F layer proceeds to descend, however, the modulation can continue to be amplified within the region of eastward plasma drift. At some point during this amplification process, the RT instability should begin to contribute to the continued growth of the seeded plasma bubbles.
 In closing, we should mention that the coupling process envisioned by Tsunoda , though plausible and consistent with the presented experimental evidence, should, at this time, be considered as perhaps an occasional contributor to the bubble-generation process. The restrictions that must be waived before activation are clearly more numerous than those for the collisional-shear instability [Hysell and Kudeki, 2004]. The need for an intense Es layer would seem to be most restricting, although little is yet known about such layers in low-latitude regions. Our experience at Kwajalein indicates frequent presence of such layers around 1700 LT during summer months. There is also the question of whether long horizontal wavelengths associated with LSWS could be excited in the E region. This is not a problem for the EsL instability, which is wavelength independent. Instead, we suggest that a long-wavelength modulation of an Es layer is not likely to be detected with sensors, as currently deployed. The reason is that the spatial displacement produced by a given is an order of magnitude larger in the F region than in the E region. For example, a 50-km displacement in the F region would be no more than 5 km in the E region, and it would be easy to overlook a 5-km displacement over a horizontal distance of 200 km. However, rather than viewing this process with skepticism, we should proceed to consider how it may coexist or couple with other processes such as collisional-shear and RT instabilities to produce the day-to-day variability in ESF occurrence.
 This investigation was supported by the National Science Foundation under grant ATM-0318674.
 Amitava Bhattacharjee thanks Miguel Larsen and another reviewer for their assistance in evaluating this paper.