Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 The day-to-day variability in ionospheric irregularity generation giving rise to equatorial scintillation has remained an unresolved issue over many decades. We take a fresh look at the problem utilizing the global imagery provided by the Global Ultraviolet Imager (GUVI) instrument on NASA's Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics satellite. GUVI has been acquiring images of 135.6-nm emission in the Earth's ionosphere-thermosphere system since 2001. These GUVI disk images at dusk have been used to identify cases where the equatorial ionization anomaly (EIA) crests lie near the magnetic equator over a relatively narrow longitude range, so that the anomaly looks collapsed. A 16-month period of GUVI data collected during evening at solar maximum is used to study the morphology of these so-called collapses, since the EIA collapse is shown to be linked to the suppression of equatorial plasma bubbles and scintillations. In particular, we look at the June solstice, during which the Atlantic and Pacific show very different climatology and EIA collapses are most frequent in the GUVI data. On the other hand, EIA collapses are a relatively rare occurrence during the equinox period when scintillations are most prevalent globally. We obtained a few dramatic examples of day-to-day variability in EIA behavior and scintillations over India. The Sami3 is Also a Model of the Ionosphere (SAMI3) model was used to investigate the conditions during the evening collapse of the anomaly in the Indian longitude sector, where measurements of total electron content (TEC) and scintillations and estimates of the daytime vertical drifts and those at dusk were available. Results from SAMI3 show that the observed collapse of the anomaly at dusk can be simulated by a reversal of the upward vertical drift in midafternoon in agreement with the drift estimates from magnetometer observations. Such reversed vertical drifts at this time of the day are generally seen during counterelectrojet events. Introduction of neutral winds into SAMI3 better approximates the dusk behavior of TEC at low-latitude stations in India. This study reveals that the longitudinally confined EIA collapse may explain some of the differences in day-to-day variability of scintillations at different locations around the globe.
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.
 In the equatorial region during the daytime, dynamo electric fields in the off-equatorial E-region locations map along the magnetic field to F-region heights at the magnetic equator and the resulting E × B force lifts the F-region plasma to higher altitudes. Then, by action of forces parallel to B due to gravity and plasma pressure gradients, the uplifted plasma diffuses along the magnetic field toward the north and south of the magnetic equator leaving a trough in the ionization density at the magnetic equator. Two belts of high ionization density are formed nominally around magnetic dip latitudes of 15°N and 15°S. These are the crests of the equatorial ionization anomaly (EIA) [Hanson and Moffett, 1966]. At the time of sunset, the F-region zonal neutral wind and conductivity gradient caused by the sunset terminator interact to develop an enhanced eastward electric field on the dayside of the terminator and a westward electric field on the nightside [Rishbeth, 1981; Farley et al., 1986; Eccles, 1998]. The enhanced eastward electric field, called the prereversal enhancement (PRE), causes the F region to move upward. If this PRE is of sufficient magnitude, the bottomside of the F region steepens and the threshold for rapid development of the Rayleigh-Taylor instability can be crossed. As a result, plasma bubbles are generated [Ossakow, 1981], and irregularities ranging from hundreds of kilometers to tens of centimeters are formed. Thus the PRE of the eastward electric field is of crucial importance for the destabilization of the ionosphere, which in turn creates the kilometer-scale irregularities giving rise to scintillation of satellite signals [Basu et al., 1996; Fejer et al., 1999]. This electric field also leads to a resurgence of the EIA in the postsunset period. As a result, scintillation at frequencies as high as 4 GHz can be encountered near the EIA crests during solar maximum [Basu et al., 1987].
 Scintillations at low latitudes are a serious problem for communication and GPS-based navigation because very large adverse effects are seen on these systems even for magnetically quiet days [Basu et al., 2002]. As quiet days are far more prevalent than disturbed days, it is quite possible to have scintillations at VHF on a large fraction of the days during the “scintillation season” at a given site, except for the unpredictable day-to-day variability [Tsunoda, 2005; McDonald et al., 2006]. To further complicate matters, global scintillation and bubble morphology has a distinct longitudinal-cum-seasonal pattern [Basu and Basu, 1985; Aarons, 1993; Nishioka et al., 2008]. As discussed in these publications, the primary maxima are generally seen during the equinoxes, whereas the solstitial behavior is a function of longitude. Thus in the Atlantic sector, the December solstice shows more activity, while the activity shifts to the Pacific in the June solstice. While the climatology is somewhat better understood [Maruyama and Matuura, 1984; Tsunoda, 1985], the day-to-day variability remains unresolved even after several decades of research.
 Within the past decade, several space-based systems have provided us with the ability to remotely sense the EIA structure at far ultraviolet (FUV) wavelengths. The Low Resolution Airglow and Aurora Spectrograph (LORAAS) on the Advanced Research and Global Observation Satellite (ARGOS) launched in 1999 [McCoy et al., 1992], the FUV Imager on the IMAGE satellite launched in 2000 [Mende et al., 2000], and the Global Ultraviolet Imager (GUVI) on board the Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics (TIMED) satellite launched in 2001 [Christensen et al., 2003] have been extensively used to study the characteristics of the low-latitude ionosphere in general, and the behavior of the EIA peaks in particular [Kil et al., 2004]. Much new information has been obtained on the longitudinal variability of the anomaly peaks and the occurrence of plasma bubbles within them. For instance, longitudinal variations in the anomaly structure are easily seen in the GUVI data [Henderson et al., 2005], as well as depletions in airglow caused by equatorial bubbles [Kelley et al., 2003]. Henderson et al.  have performed a statistical study of several aspects of the EIA region based on GUVI data, including their average brightness and separation. They found that quiet time EIA morphology is well characterized by data taken during 2030–2130 MLT and the equatorial plasma bubble morphology is dependent on longitude, local time, season, and solar flux.
 Recently, the emphasis in the field has been to study the equatorial longitudinal variability driven by atmospheric tides. The tidal effects are synthesized usually by averaging either the 135.6-nm emissions as in the case of using IMAGE [Sagawa et al., 2005; Immel et al., 2006], GUVI [England et al., 2006], and LORAAS [McDonald et al., 2008] or by utilizing the in situ electron densities, such as those from CHAMP [Lühr et al., 2007]. The radio occultation technique has also been used with COSMIC data [Lin et al., 2007] and total electron content (TEC) data from TOPEX to further study the tidal influences [Scherliess et al., 2008]. Whatever the technique, a large amount of averaging is used in studying the longitudinal behavior of the parameters such as plasma densities or TECs.
 The objective of this paper, however, is to study the day-to-day variability of the EIA and scintillations observed using VHF transmissions from geostationary satellites. The 250-MHz scintillation data were obtained from the Scintillation Network Decision Aid (SCINDA) system receivers at several equatorial stations [Groves et al., 1997]. The focus is on the morphology of extreme cases of day-to-day variability in the longitudinal structure of the EIA region and its impact on the generation of equatorial plasma density depletions or bubbles, which give rise to scintillations. Under normal conditions the EIA crests are separated by a fairly wide trough in ionization density around the magnetic equator. In general, most of the days with scintillation show this kind of EIA behavior [Valladares et al., 2001]. However, we focus our attention on cases where the anomaly crests are not well separated, nearly coming together at the magnetic equator over a fairly narrow longitude range, thereby obliterating the ionization trough. In other words, the EIA appear “collapsed” and this terminology is used in the rest of the paper.
 It is important to note that during daytime the low-latitude F-region ionosphere is strongly influenced by the zonal electric field that develops in the E region near the magnetic equator owing to neutral winds. In the E region the Pedersen current due to the eastward electric field and the eastward current due to the vertical Hall voltage combine to cause an equatorial electrojet (EEJ), a band of enhanced current around the magnetic equator. Owing to high parallel conductivity, these dynamo fields in the E region are transmitted along the magnetic field lines to the F region where they are responsible for the E × B force that lifts the F-region plasma to higher altitudes which then diffuse north and south down the magnetic field lines to form the crests of the EIA as mentioned before. Occasionally, the normal daytime eastward directed EEJ appears to reverse into a westward counterelectrojet (CEJ). CEJs are observed as depressions in the diurnal variations of the horizontal intensity of the Earth's magnetic field measured in the equatorial regions. These depressions are assumed to be caused by a reversal of the EEJ current in the ionosphere [Rastogi, 1974; Somayajulu et al., 1993, and references therein]. We shall show that the collapse observed in the GUVI data at dusk could be related to CEJ events in the afternoon hours.
 We present in section 2 the variability of the EIA in the GUVI data at dusk during a 16-month period over a high sunspot cycle phase when the 135.6-nm emissions are high and anomaly dynamics, such as longitudinal variability in its development and collapse are easily discernable on quiet and disturbed days. We are also able to identify instances of dramatic day-to-day variability in the large-scale behavior of the EIA and scintillations. Specifically, in section 3 we identify a CEJ event that was associated with a case of EIA collapse and quenching of equatorial irregularities over India on one day followed by a well-developed and latitudinally separated EIA peak on the next associated with equatorial bubbles and intense GHz scintillations. In section 4, we use SAMI2 [Huba et al., 2000a] and its 3-D counterpart SAMI3 [Huba et al., 2005] to simulate the background conditions using some observed drivers of equatorial electrodynamics to provide a better understanding of the short-term variability in EIA dynamics and its impact on irregularity generation. In section 5 we present our discussion and summary.
2. Morphology of Equatorial Anomaly Collapses Observed by GUVI and Scintillations
 The GUVI instrument is a scanning imaging spectrograph that makes observations in five emission bands [Paxton et al., 1999; Christensen et al., 2003]. These emissions have been extensively used for lower thermospheric and ionospheric studies. In this paper we will concentrate on the nighttime disk images taken of the O I 135.6-nm emission feature. This emission arises from the radiative recombination of atomic oxygen as well as a small contribution from ion-ion mutual neutralization. Neglecting this secondary contribution, its brightness is proportional to the square of the electron density. Thus, variations in brightness of the EIA act as an excellent tracer for the dynamics of the low-latitude nighttime ionosphere on a global scale. As seen in the example composite image of GUVI disk scans in Figure 1, where the emission is mapped to 150 km above the Earth's surface, the EIA crests are clearly visible over a large portion of the globe; they are typically separated by about 15° on either side of the magnetic equator. However, a collapse of the EIA is also clearly visible between 45° and 60°E longitude. On examining consecutive days of GUVI data, it becomes evident that there is a large amount of day-to-day variability in the separation and density of the two crests. This day-to-day variability appears to have a direct influence on the occurrence of scintillations along transionospheric radio wave links.
 To study the morphology of the occurrence of collapsed EIA crests, we have examined GUVI disk data from 1 February 2002 to 8 June 2003. In order to detect collapses, the data were plotted for each day in a format similar to Figure 1. Collapsed crests were noted visually and recorded. The TIMED spacecraft precesses slowly such that each pass on a given day occurs at approximately the same local time. However, the spacecraft cycles through all local times about every 60 days. For this work, we have concentrated on data collected at dusk between 1900 and 2200 LT, when plasma bubbles are generated and the relative brightness of the EIA crests are high. This corresponds to 98 days during the 16-month analysis. The dates provide coverage over all seasons and with the near Sun-synchronous TIMED orbit, all longitudes were uniformly sampled during these time periods. Among the 98 days studied, 43 instances of anomaly collapses were observed on 28 unique days. Thus at least 30 percent of the days had a collapsed EIA feature as in Figure 1 at some longitude. Most of these collapses are narrow longitudinally confined features quite prominent in the global images provided by GUVI.
 The fact that a collapse of the EIA will lead to quenching of instabilities is perhaps not very surprising, since this indicates that the eastward electric field in the F region, which drives the EIA at dusk, is either weak, absent or reversed in direction. What is surprising is its possible linkage to CEJ events earlier in the afternoon. We investigate this question in section 4, where we model such a collapse.
 We now discuss the GUVI daily plot shown in Figure 1. We find that for this magnetically quiet day 7 August 2002 (Ap index is 4), the EIA was well developed near Guam (13.58°N, 144.85°E), bubbles were observed (only the ones near locations being discussed are identified with yellow arrows on this and later diagrams) and intense scintillations were recorded at that station for many hours. On the other hand, at Ancon (11.77°S, 77.15°W) a weak EIA is seen and a short-lived scintillation structure was found which drifted into the field of view from the west. This is consistent with the expected scintillation climatology discussed earlier. No scintillation data were available in the Arabian Sea region where the anomaly collapse was observed.
 In Figure 2, multiple EIA collapses were seen near Guam on 2 August 2002, a magnetically disturbed day (Ap of 42), and this was accompanied by a total inhibition of scintillation. On the other hand, the EIA in the Ancon sector was well developed, and bubbles were observed together with intense scintillations. The Ancon longitude, being in the dusk sector during the development phase of the storm, shows enhanced EIA development, bubble generation and intense scintillation, as opposed to Guam where the dusk sector was in the recovery phase of the storm generally associated with a westward electric field, and no scintillations were seen [Fejer et al., 2008a]. Thus, during magnetic storms, the time history determines which longitude sectors will be conducive to the generation of plasma bubbles and scintillations rather than the normal climatology observed under magnetically quiet conditions. The objective of this section was to show some of the ways in which the day-to-day variability in scintillations can be generated and the GUVI images provide a clear picture of the longitudinally limited region over which such variability is operative.
 Since the SCINDA coverage was longitudinally limited in 2002, a survey was also carried out using the dusk orbits of the Sun-synchronous DMSP F14 and F15 satellites. This database has been extensively used in the past for studying plasma bubble morphology and association with scintillations at sunspot maximum [Huang et al., 2001]. The equatorial crossings of these satellites vary between 2030 and 2130 MLT. On the collapse dates and at longitudes adjacent to the GUVI orbits no bubbles were seen at the 840-km altitude in situ data, indicating that bubbles did not reach these high altitudes. Thus, we determined on a one-to-one basis that collapses were generally associated with a lack of equatorial irregularities. A reasonable conclusion would be that the ionospheric conditions that lead to collapses of the EIA are not conducive to scintillation-causing irregularity growth.
3. Day-to-Day Variability in Equatorial Anomaly at Dusk and Scintillations
 Several instances of fairly dramatic day-to-day variability in anomaly behavior and in the occurrence of scintillation were observed during the 16-month study period. In the remainder of this paper, we focus our attention on two consecutive days in order to better understand the possible drivers of such variations. In this section we discuss the GUVI data and other complementary measurements in some detail. We present a modeling study of the two days in section 4.
3.1. GUVI, DMSP, and Scintillations
Figure 3 shows the GUVI radiances projected to 150 km for 3 February 2002. This is a fairly typical magnetically quiet day, where the anomaly crests, though of varying radiance, are well separated in all longitude sectors and are at their nominal values of 15°N and 15°S dip latitudes. We shall refer to this as our “noncollapsed” day. In contrast, 2 February 2002 shows the occurrence of an EIA collapse over the Indian sector (Figure 5). In the following discussion, this will be referred to as the “collapsed” day. It should be noted that the FUV emissions are much more intense in February (Figures 3 and 5) when compared to the data for August 2002 shown in Figures 1 and 2. This seems to be consistent with the findings of Fejer et al. [2008b] on the importance of conductivity effects on the electrodynamics of the equatorial ionosphere. An enlarged version of the GUVI data over India on the noncollapsed day is shown in Figure 4g. The GUVI scans look more separated as the data are projected to a height of 350 km (rather than the 150-km projection height seen in Figure 3), the nominal height of the F peak. A bubble is seen in the GUVI data at 70°E longitude. Two consecutive orbits from each of DMSP F14 (dashed line) and F15 (solid line) satellites are also shown superimposed on the GUVI data with their magnetic equatorial crossings being marked with white dots. The Ionospheric Penetration Point (IPP) for the scintillation receivers located at Calcutta (22.6°N, 88.4°E; 14°N dip latitude) and Singapore (1.2°N, 102.6° E; 8°S dip latitude) are shown as yellow diamonds for the VHF channel and as a red diamond for the Calcutta GHz channel. The IPP for Calcutta for both channels (almost on top of one another) are at the equatorward edge of the northern anomaly peak whereas that for Singapore is closer to the magnetic equator but in the Southern Hemisphere.
 To the left of the expanded view are diagrams exhibiting electron density data measured at 840 km from the two DMSP satellites showing the presence of bubbles around the magnetic equator (Figures 4a–4d). The local times for these Sun-synchronous satellite orbits are between 2012 and 2118 MLT and agree well with the GUVI scans between 2046 and 2057 LT in Figures 3–6 (in the equatorial region the difference between MLT and LT is small). The data in Figures 4e and 4f show saturated scintillations at 250-MHz (S4 > 1) and significant 1.5-GHz scintillations at Calcutta whereas Singapore had only 250-MHz data available which shows fairly high scintillation levels. Multiple scintillation structures are seen at both sites indicating the drift of several discrete irregularity structures across the satellite raypaths for both stations. The S4 index is a quantitative measure of intensity scintillation and is defined as the ratio of the standard deviation of signal intensity fluctuations and the mean signal intensity. Scintillation strength, and therefore the S4 index, is expected to increase with increasing electron density. Thus, the larger scintillation magnitude seen from Calcutta as compared to Singapore is consistent with the IPP from the former station being located near the high-density EIA peak rather than being near the low-density trough over the magnetic equator.
 To pursue the day-to-day variability we consider the GUVI data of 2 February 2002 at dusk. The anomaly characteristics shown in Figures 5 and 6g are dramatically different with the two anomaly peaks shown to collapse closer to the magnetic equator over India. No equatorial bubbles were seen in the DMSP data (Figures 6a–6d) and no scintillations (Figures 6e and 6f) were measured at either of the two stations. The collapse and its impact on the generation of small-scale irregularities are quite noteworthy. Also of note is the limited nature of the longitude sector over which the collapse takes place: approximately 30–40 degrees.
3.2. Counterelectrojet Investigation
 To investigate this difference in anomaly characteristics on the two successive days described above, we determined the daytime EEJ strength obtained from magnetometer measurements made by the Indian Institute of Geomagnetism, Mumbai, at the two stations Tirunelvelli (8.7°N, 77.7°E; 0.4°N dip latitude) close to the magnetic equator, and Alibag (18.6°N, 72.8°E; 12.8°N dip latitude) outside the EEJ region [Rastogi, 1974]. We used the Anderson et al. [2004a] technique to convert the delta H measurement, the difference between the horizontal magnetic intensity at Tirunelvelli and Alibag, shown in Figure 7a to the vertical E × B drifts shown in Figure 7b. It is very interesting to note that whereas 3 February (Figure 7, dashed line) is a normal day, 2 February (Figure 7, solid line) is a CEJ day of the type described by Rastogi , since the vertical drift changes sign from upward to downward at approximately 1500 LT. Thus it seems that the anomaly collapse at dusk observed in the GUVI data could have happened because of an earlier suppression of the normal daytime upward drift caused by the CEJ in the middle of the afternoon. By studying magnetograms from different equatorial locations around the globe, Rastogi [1973, 1974] came to the conclusion that CEJ events are fairly localized and on some occasions may not occur on the same day at two stations separated by as little as 2 or 3 h (30° or 45°) in longitude. Further, they are most prevalent during the June solstice. Both of these findings are consistent with the collapse morphology seen in the GUVI data. In his study, Rastogi  mentions that the CEJ phenomenon occurs on quiet days as the geomagnetic activity tends to reduce or cancel the afternoon decrease of the magnetic field. In this particular case, 2 February is moderately disturbed (Ap = 18) compared to 3 February, which is a very quiet day (Ap = 5), but even so the afternoon depression in the magnetic field is clearly defined.
3.3. TEC Measurements
 One way of investigating the effect of the CEJ on the F-region plasma distribution is to study total electron content (TEC) measurements. Unfortunately, TEC data were not routinely available from Calcutta in February 2002. However, TEC was available from a dual-frequency GPS receiver located at Bangalore (13.0°N, 77.6°E; 5°N dip latitude) as part of the International GPS Service (IGS) network. The equivalent vertical TEC data plotted in TEC units (1 TECU = 1016 electrons m−2) from this site are shown in Figures 8a and 8b for the two days. The multiple lines are from the different raypaths to the various GPS satellites, which are simultaneously visible from the station. There is an absence of any structuring at dusk (LT = UT + 5 h) on 2 February while the TEC fluctuations observed after 1800 LT on 3 February are indicative of the plasma bubbles seen on that day. It is important to note from the previous diagrams (Figures 4g and 6g) that on 3 February the anomaly peak was near Calcutta and there was less emission at the latitude of Bangalore than on the collapse day when a much weakened anomaly peak was closer to Bangalore. This is consistent with the higher TEC values in the postsunset time period over Bangalore shown in Figure 8a versus Figure 8b. A closer inspection of the GUVI data shows that the collapse was asymmetrically located with respect to the magnetic equator indicating that perhaps a transequatorial neutral wind was active on this day. The secondary peak at 1500 UT (2000 LT) on 2 February could be the result of ionization transport due to an equatorward neutral wind [Anderson and Klobuchar, 1983].
4. SAMI3 Modeling of Day-to-Day Variability of Equatorial Anomaly Structure
 Model simulations using Sami2 is Another Model of the Ionosphere (SAMI2) [Huba et al., 2000a] and Sami3 is Also a Model of the Ionosphere (SAMI3) were performed to assess the role of the afternoon counterelectrojet in the collapse of the equatorial anomaly observed at dusk in the Indian longitude sector.
4.1. SAMI3 Simulations
 SAMI3, a global, three-dimensional, physics-based model of the ionosphere [Huba et al., 2005], is based on the SAMI2 [Huba et al., 2000a] two-dimensional model of the ionosphere. SAMI3 models the plasma and chemical evolution of seven ion species (H+, He+, N+, O+, N + 2, NO2+ and O + 2). These models include ion inertia along the geomagnetic field so that it captures certain essential features of collisionless physics (e.g., ion sound waves) at high altitude [Huba et al., 2000b]. To our knowledge SAMI3 is the only global ionosphere code that models full plasma transport for all of these ion species, including the molecular ions. The complete ion temperature equation is solved for three ion species (H+, He+ and O+) and the electron temperature equation is solved. These codes include 21 chemical reactions and radiative recombination. In the current version of SAMI3, the neutral composition and temperature are specified using the empirical NRLMSISE00 model [Picone et al., 2002], and the neutral wind is specified using Hedin Wind Model (HWM07) [Drob et al., 2008]. The solar EUV flux used in the photoionization calculations is determined from the NRLEUV model [Warren et al., 2001].
 SAMI3 uses a flexible dipole model of Earth's geomagnetic field. In the low-to midlatitude ionosphere it mimics the International Geomagnetic Reference Field (IGRF) by fitting a dipole field to the IGRF at each longitude in the grid while at high latitude it uses a standard tilted dipole. The nonorthogonal, nonuniform, fixed grid is closed so that there are no open field lines in the high-latitude ionosphere that require specified boundary conditions. The plasma is transported transverse to the geomagnetic field using a finite volume method in conjunction with the donor cell method. The present implementation of SAMI3 uses an externally specified E × B drift function, such as a simple sinusoid or the Fejer-Scherliess model [Scherliess and Fejer, 1999]. The maximum latitudinal extent of SAMI3 is ±89° and altitudes as large as 8 Re are allowable [Meier et al., 2009].
4.2. Simulations of Day-to-Day Variability in the Anomaly
 We performed a number of SAMI3 runs corresponding to the collapsed and noncollapsed days. In each case, solar and geomagnetic conditions were set to measured values for the given day. More importantly, the E × B drifts were varied on the basis of data from the collapsed and noncollapsed days. Model inputs include F10.7 of 241, 81-day average F10.7 of 217, Ap of 18 and day-of-year 33 for the collapsed day, and F10.7 of 234, 81-day average F10.7 of 217, Ap of 4 and day-of-year 34 for the noncollapsed day. In these runs, the NRLEUV model was used, but with an enhancement factor of 1.3 to account for the fact that the NRL model is known to underestimate EUV at high F10.7 indices (J. Lean, private communication, 2008).
Figure 9 shows the two E × B drift functions used in the simulations. The solid squares in Figure 9 indicate drift values derived from the magnetometer measurements during 0700–1800 LT (Figure 7). The maximum PRE was estimated from ionosonde measurements at Trivandrum (8.5°N, 76.8°E; 0.3°N dip latitude) of the virtual height (h′F) of bottomside of the F region. The measurements of this parameter at dusk provided approximate rise velocities of 40 m s−1 on 3 February and 20 m s−1 on 2 February and are indicated by open squares. At all other local times, the E × B drift was specified with the Fejer-Scherliess model [Scherliess and Fejer, 1999]. On the collapsed day, the E × B drift function labeled “Feb 2” was used in the Indian sector between 60° to 100°E and the noncollapsed (“Feb 3”) E × B drift function was applied at all other longitudes. Linear interpolation was used to smooth the transition between the Indian sector and the rest of the globe. The “Feb 3” E × B drift function was applied at all longitudes on the noncollapsed day.
 To directly compare the SAMI3 results with the GUVI measurements, the O I 135.6-nm radiance is calculated from the model results. The airglow emission is assumed to be created by radiative recombination of O+ (O+ − O− mutual ionization is ignored) so that the total radiance is given by
where the effective recombination rate coefficient α135.6 is 7.3 × (1160/Te)1/2 × 10−13 cm3 s−1 [Meléndez-Alvira et al., 1999]. The SAMI3 radiances are shown in Figure 10 for a constant local time of 2100 LT on both the noncollapsed day (Figure 10a) and the collapsed day (Figure 10b). On the noncollapsed day, the equatorial anomaly is strong over a large fraction of the globe and well separated at all longitudes; whereas on the collapsed day the anomaly crests are weaker and closer together in the Indian sector where the 2 February E × B drift function was applied. Overall, the SAMI3 radiances are in fair agreement with the GUVI measurements with better agreement at most longitudes on the collapsed day (Figure 5) than on the noncollapsed day (Figure 3). A likely explanation for this is that the winds on the collapsed day are probably better represented by the climatology (HWM07) than on the noncollapsed day. In the Indian sector, there is good agreement with the reduction in the radiance of the anomaly crests, but the anomaly crests do not collapse north of the magnetic equator as indicated by the GUVI data.
 Finally, in Figure 11 we look at SAMI3 TEC versus time at latitude 12°N, longitude 77°E, which is shown for comparison to corresponding measurements taken at Bangalore (Figure 8). The simulated and measured curves are in fairly good agreement, with the exception of the measured increase in TEC at around 1500 UT on 2 February and 1800 UT on 3 February. In the SAMI3 plots, the rate of TEC decrease is arrested, but do not actually increase.
 It is interesting to note that a model run using SAMI2 does show a collapse before the occurrence of the PRE at dusk on 2 February. Figure 12 shows the electron density at 1800 LT as a function of geographic latitude and altitude at the longitude of Bangalore. Indeed a collapsed EIA is seen centered approximately around 10°N, near the location of the magnetic equator at this longitude. By toggling the winds on and off, we find that the collapse at 1800 LT is due to both the suppression of the normal daytime E × B drifts and equatorward winds earlier in the afternoon. This is a rather important result indicating the effectiveness of the CEJ in drastically changing the electrodynamics of the equatorial ionosphere. Additional studies with SAMI2 indicate that further reduction in the separation of the anomaly crests could have been achieved at the time of the GUVI transit shown in Figure 10b if the PRE had not occurred on the collapsed day. However, the ionosonde measurements indicate that a weak PRE was in fact present and was therefore included in the SAMI3 simulation that is shown. We conclude that the CEJ that occurred in the afternoon of 2 February in the Indian sector was responsible for reducing the electron density and thus the radiance of the anomaly crests at those longitudes. The CEJ also contributed to the reduction in the separation of the crests, but the effect of the E × B drifts alone do not explain the observed dramatic collapse of the crests somewhat to the north of the magnetic equator as seen in Figure 5. We conjecture that the F-region wind conditions, particularly the meridional winds contributed in a significant way to the observed collapse of the equatorial anomaly. Clearly, more measurements and modeling studies will be necessary to confirm this.
5. Discussion and Summary
 We have used 16 months of GUVI disk data at dusk, equatorial scintillation measurements from various locations and DMSP data to study large-scale day-to-day variability in the low-latitude evening ionosphere with the purpose of identifying conditions that suppress the occurrence of scintillation activity. Specifically, we have focused our attention on days when the equatorial anomaly crests are collapsed into one feature near the magnetic equator over a fairly narrow longitude region. We have found that scintillation does not occur in the vicinity of the collapsed region.
 We next investigated two consecutive days in order to compare large-scale conditions and their relationship to the occurrence of scintillation (or lack thereof). One day showed a collapse of the anomaly in the Indian sector and the following day was identified as a typical, noncollapsed day. The latter shows well-separated anomaly crests at all longitudes at 2100 LT, as well as strong scintillation and bubbles in the Indian sector. On the day of anomaly collapse in the Indian sector, no scintillation or bubbles were present. Magnetometer data in the vicinity of the anomaly collapse showed a CEJ event on this day. We used SAMI3 to reproduce the observed collapse of the anomaly crests in the postsunset hours. It was found that applying the inferred (from magnetometers) E × B drifts under CEJ conditions in the afternoon and using the measured weak PRE over a narrow longitude sector, it was possible to approximate the collapse observed by GUVI. Prior to the onset of the PRE, the model showed a collapse of the EIA peaks to the magnetic equator in the Indian longitude sector at 1800 LT. It was further found that the model results matched the data best when the PRE was suppressed; though the role of meridional winds merits further investigation.
 This study sheds some light on conditions that may give rise to the day-to-day variability of bubbles and scintillations over limited longitude regions. The dependence of scintillations on the magnitude of the E × B drift corresponding to the PRE and the existence of a threshold value of 20 m s−1, which must be exceeded for scintillations to occur [Basu et al., 1996; Anderson et al., 2004b] are in agreement with the present findings. However, the fact that a CEJ event in the afternoon may contribute to the inhibition of scintillations at dusk is counterintuitive; the former is controlled by the E-region dynamo and the latter by the F-region dynamo. However, the results of the SAMI2 simulation clearly indicate the collapse of the EIA peaks at 1800 LT when a CEJ event was present a few hours earlier. To our knowledge this is the first such simulation of the effect of a CEJ event on the structure of the equatorial F region near sunset. Perhaps this result provides some evidence for the unexplained positive relationship between the strength of the EEJ during the daytime and the occurrence of scintillations at dusk observed by authors in the Indian longitude sector [Dabas et al., 2003].
 While preliminary results from SAMI3 provide some insights into this type of day-to-day variation of scintillations, further work is necessary, particularly on the relative effects of the PRE of the vertical drifts, time of this drift reversal, neutral winds, and the conductivity in the E region on the generation and suppression of instabilities. We hope such modeling efforts will eventually lead to the isolation of a unique set of drivers that control large- and small-scale plasma structuring in the low-latitude ionosphere at dusk.
 This work was supported by NASA under its Living with a Star Program. Sunanda Basu was also supported by ONR grant N00014-07-1-0217. The work at Boston College was partially supported by Air Force Research Laboratory contract F19628-02-C-0087 and AFOSR task 2311AS. Work at the University of Illinois was supported under grant N00173-05-G904 from the Naval Research Laboratory. Various colleagues kindly provided different types of measurements: The GUVI team provided access to their FUV image data, without which this paper would not have been possible. The magnetometer data were from S. Alex, Indian Institute of Geomagnetism, Mumbai. The Trivandrum ionosonde data were provided by S. Ravindran, Vikram Sarabhai Space Center, Trivandrum. Assistance with the vertical drift analysis was provided by A. Anghel of NOAA, Boulder, Colorado. The DMSP in situ data were from F. Rich, now at MIT Lincoln Laboratory, Lexington, Massachusetts. The Bangalore TEC data were obtained from the IGS network.
 Amitava Bhattacharjee thanks R. S. Dabas and another reviewer for their assistance in evaluating this paper.