A comprehensive explanation for the complex climatology of the so-called equatorial spread F (ESF) has eluded researchers for more than 70 years. Recently, however, a seeding hypothesis has been proposed, which appears to provide the final major piece of this puzzle. The hypothesis is based on the discovery that a direct link exists between regions of deep convective activity in the troposphere, where atmospheric gravity waves (GWs) are spawned, and the occurrence frequency of ESF during solstices. The objective here is to answer two questions that may impede the general acceptance of this hypothesis. We first show why seed plasma perturbations should develop from GW-driven neutral-wind perturbations, but only when the GW source region is located very close to the magnetic dip equator. We then reexamine this relationship using a data set on GW source regions that is better matched (in time and longitudinal coverage), than that used previously, to the data set on ESF activity used by Tsunoda (2010a). We conclude that seeding is indeed playing an important role in the development of ESF.
 Plasma structure in the nighttime equatorial F layer, commonly referred to, generically, as equatorial spread F (ESF), displays a complex climatology, which includes variations in both longitude and season. Most attempts to explain this behavior have been in terms of variations in the linear growth rate of the generalized Rayleigh-Taylor (GRT) instability (γGRT) [e.g., Sultan, 1996]. The most successful explanation to date is based on a hypothesis that ESF activity increases, when the solar terminator aligns with the geomagnetic field () [Tsunoda, 1985]. Following Tsunoda [2010a], we refer to it as the solar terminator– alignment (STBA) hypothesis. The underlying notion is that the gradient in field-line-integrated conductivity becomes steeper, when STBA occurs, and that a steeper gradient elicits appearance of a stronger polarization electric field (p). The rest of the explanation is usual; the stronger p drives the postsunset rise (PSSR) of the F layer more vigorously, and the enhanced rapidity of the rise and the loftier altitude attained by the F layer, both act to strengthen γGRT. And, indeed, the strength of the PSSR has been shown to play a major role in ESF development [e.g., Fejer et al., 1999]. Climatologically, this hypothesis works well around the equinoxes because the declination of is typically small at most longitudes [Tsunoda, 1985]; it also works well during years of high solar activity, when PSSR and ESF are generally strong [e.g., Gentile et al., 2006].
1.2. Limitation of STBA Hypothesis
 The STBA hypothesis, however, has a shortcoming; it cannot explain the existence of ESF activity during the solstices, at least not in most longitude sectors. Most conspicuous in this unexplained behavior is the occurrence of enhanced ESF activity in the Peruvian portion of the South American (West Pacific) sector during the December (June) solstice, and very low ESF activity during the June (December) solstice. (It is important to note that the STBA hypothesis does appear to hold in the Brazilian portion of the South American sector, where the magnetic declination is large [e.g., Tsunoda, 1985].) In this regard, Basu and Basu  have stated that “…the problem with the Tsunoda hypothesis is that it does not predict the secondary maximum at many longitudes.” Aarons , in his review, raised similar questions after comparing morphologies of radio wave scintillation in the South American and West Pacific sectors. The STBA hypothesis fails simply because it does not offer any useful information, when STBA does not occur. In passing, we note that although the meridional neutral wind () can affect ESF climatology during the solstices [Maruyama and Matuura, 1984], it cannot enhance ESF activity. It simply acts to reduce γGRT by producing a hemispherical asymmetry in F region plasma density (N), which increases the electrical loading of the GRT instability.
 A promising possibility is that ESF activity could be enhanced without increasing γGRT, if the amplitude of seed plasma perturbations has a seasonal and longitudinal dependence. Usual interpretation is in terms of atmospheric gravity waves (GWs), generated in the troposphere, which propagate up to the thermosphere, where they transfer their wavelike perturbations in to the F region plasma [e.g., Röttger, 1977, 1981; McClure et al., 1998]. Although promising, this notion has not received mainstream acceptance yet, perhaps because its role in ESF development had remained murky, until recently [Tsunoda, 2010a].
 For example, Röttger  showed that almost two-thirds of medium-scale traveling ionospheric disturbances (MSTIDs) originate in equatorial regions and propagate in the meridional direction. He further showed that tropical rainfall activity displays a strong diurnal variation, which appears to be correlated with the diurnal component of MSTID activity; the time lag for maximum correlation was found to be up to about three hours, which agrees with the time GWs take to propagate from troposphere to ionosphere. He considered meridionally propagating MSTIDs to be consistent with having a source region in the intertropical convergence zone (ITCZ), which was located to the north of his observing station. These results, as presented, appear to show that GWs from the troposphere affect F region plasma. On the other hand, Röttger  did not explain how meridionally propagating GWs might cause plasma perturbations in the plane transverse to . Instead, as discussed below, the most effective means for neutral-ion coupling appears to be a polarization response, which involves zonally propagating GWs [e.g., Klostermeyer, 1978; Kelley et al., 1981; Huang and Kelley, 1996; Tsunoda, 2010a, 2010b].
 In another study, Röttger  surmised that thunderstorm electric fields may penetrate directly to the ionosphere. Although effects of this kind have also been reported by others [e.g., Woodman and Kudeki, 1984; Kelley et al., 1984], there is no evidence that this process controls the seasonal seeding of ESF. Besides, the time delay for maximum correlation between lightning and ESF would be much less than the few hours, reported earlier [Röttger, 1977]. Seeding, however, has continued to receive attention as a possible contributor to ESF development [e.g., Abdu et al., 2009; Alam Kherani et al., 2009].
1.4. GWBA Hypothesis
 Recently, a direct link was found between GWs and ESF activity during solstices. That is, ESF activity was shown to be enhanced during solstices, primarily in longitude regions, where the intertropical convergence zone (ITCZ) was located near the magnetic dip equator [Tsunoda, 2010a]. The basis for this relationship was described as follows. A divergent Pedersen current, driven by a GW-induced dynamo, leads to the appearance of a zonally directed p. Local transport of plasma by p, in the presence of a background gradient in N, results in the development of plasma perturbations. If GWs are assumed to be plane waves and is horizontal (as at the dip equator), an p should appear as a polarization response, but only when GW phase fronts are exactly aligned with [Klostermeyer, 1978; Huang and Kelley, 1996; Keskinen and Vadas, 2009; Tsunoda, 2010a]. Hence, seed plasma perturbations should arise only from zonally propagating GWs. (A zonally propagating plane wave is a reasonable approximation for a GW that is in the far field of its source region, which is located near the dip equator.)
 Because most GWs are spawned from mesoscale convective cells (MCCs), which are centers of deep tropospheric convection found in the ITCZ [e.g., Waliser and Gautier, 1993], it seemed to follow that seeding should occur when the ITCZ is located near the dip equator. Hence, observations of a direct link could be interpreted, seemingly reasonably, in terms of the so-called GW- alignment (GWBA) hypothesis [Tsunoda, 2010a]. (Earlier, McClure et al.  surmised that ITCZ migration with season might be affecting the climatology of ESF. They, however, were not able to demonstrate convincingly that this is the case.)
1.5. Need for Further Validation
 Soon after the discovery of a direct link between ITCZ location relative to the dip equator and ESF occurrence during solstices [Tsunoda, 2010a], we pointed out that GWs launched from MCCs are likely to have phase fronts that are more circular than planar, as suggested by Vadas and Fritts , Vadas , and references therein. We were able to show that this added realism does not invalidate the neutral-ion coupling mechanism [Tsunoda, 2010b]. With a circular GW, a reasonably strong, polarization response still appears, where GWBA is satisfied locally; that is, where a circular phase front becomes tangent to . Notice, however, that if the MCC is not on the dip equator, GWBA can still occur in a location away from the dip equator. This finding (for GWs with vertical phase fronts) implies that seeding could occur without requiring that the ITCZ be located close to the magnetic dip equator. Hence, the validity of the GWBA hypothesis as the explanation for a direct link between ITCZ and ESF activity was threatened by the use of vertically aligned, cylindrical phase fronts to describe the GWs that are involved with seeding.
 To place the seeding hypothesis on firmer ground, we address two questions in this paper:
 1. Why should seeding be affected by the proximity of the ITCZ to the dip equator?
 2. Could the use of disparate data sets in Tsunoda [2010a] lead to less accurate (or erroneous) conclusions regarding the link between GWs and ESF?
 We have already discussed the first question at some length. In order for the GWBA hypothesis to remain valid, we must be able to show why GWs cannot elicit a polarization response at latitudes away from the dip equator (see section 2.3). The second question arises because the climatology for ESF activity, which was used to infer existence of a direct link with GWs, comes from AE-E measurements taken between January 1978 and September 1980, a period of high solar activity [McClure et al., 1998]. Estimates of the mean ITCZ location, on the other hand, was made using averages of highly reflective cloud data, obtained over a 17 yr period [Waliser and Gautier, 1993]. Because the latter spans nearly two solar cycles, the existence of a dependence on solar activity, if any, would be lost in the averaged behavior. Hence, a demonstration of the existence of a direct link would be more convincing, if both data sets were from the same phase of the solar cycle.
 The results are organized as follows. We begin with a brief review of ESF and what is known about seed plasma perturbations. We distinguish between GWs, which produce perturbations in the neutral gas, from plasma perturbations, which seed the development of ESF. They are not identical, and the development of plasma perturbations from neutral perturbations depends on the effectiveness of the neutral-ion coupling process. As stated above, we show why the development of plasma perturbations must be confined to the vicinity of the dip equator. We then describe the geographic maps of outgoing longwave radiation (OLR), which provide a convenient means for identifying regions of deep convective activity. The OLR measurements used herein were made with a scanning radiometer on board a polar-orbiting, sun-synchronous satellite [e.g., Gruber and Winston, 1978]. The emissions covered by the radiometer are in the infrared band, from 10.5 to 12.5 μm. Measurements of OLR have been shown to be related to convective activity [e.g., Waliser et al., 1993], and have been used previously to show that a correlation sometimes exists between tropospheric and ESF activity [Ogawa et al., 2006]. We then proceed with a comparison of OLR distribution and ESF activity, and conclude that the GWBA hypothesis appears to be well founded.
2.2. Nature of Plasma Perturbations
 There are two types of ESF, range spread and frequency spread. We are concerned with the range type, which appears after the PSSR and is most intense during premidnight hours. This type is associated with equatorial plasma bubbles (EPBs) and strong radio wave scintillation. All indications are that the range type is produced largely by oblique reflections, not by underdense scatter from small-scale irregularities [e.g., King, 1970]. The spread in range that appears in ionograms is basically the presence of multiple, often unresolved, replicas of the main F trace [e.g., Reinisch et al., 2004]. When resolvable, these replicas are referred to as satellite traces. It should be evident from this description that range spread occurs, when the observing station is located under concave isodensity contours (i.e., upwellings) in the bottomside of the F layer. Hence, not surprisingly, the most reliable precursor of ESF is the appearance of a single, isolated satellite trace, which has a slightly longer group delay than the main F trace [e.g., Lyon et al., 1961]. Such traces have been shown, on a case-by-case basis and statistically, to always precede the development of ESF [Lyon et al., 1961; Abdu et al., 1981].
 Using ALTAIR, a fully steerable incoherent scatter (IS) radar, Tsunoda and White  showed that upwellings are indeed present, and that they are associated with the crests in large-scale wave structure (LSWS) found in the bottomside F layer. That a satellite trace is associated with an upwelling was demonstrated, by comparing ionogram and IS radar data [Tsunoda, 2008]. Another signature of an upwelling has also been found [Tsunoda, 2009]. Again, not surprisingly, radio signal strength become enhanced, when reflected from concave isodensity contours, which leads to the appearance of multireflected echoes (MREs) in ionograms; these, too, have been shown to be associated with an upwelling [Tsunoda, 2009]. Because MREs appear to occur prior to satellite traces, the focusing responsible for MREs is likely associated with earlier LSWS development than that associated with satellite traces.
 The finding that upwellings are seats of ESF activity [Tsunoda and White, 1981; Tsunoda, 2005] is consistent with findings that ESF occurs in patches, which are distributed quasiperiodically in longitude [e.g., Röttger, 1973; Singh et al., 1997; Eccles, 2004; Thampi et al., 2009]. Up to 10 patches have been detected, distributed spatially in longitude, on a given evening. Using results from Röttger , we infer that the wavelength of LSWS can vary from less than 100 km to more than 1500 km, with a median value of 380 km. Hence, we conclude that LSWS is the seed plasma perturbation that is responsible for ESF development. Because the properties of LSWS are similar to those of GWs that reach the thermosphere [e.g., Vadas, 2007; Makela et al., 2010; Tsunoda, 2010b], we further infer that LSWS develops from GWs through neutral-ion coupling. If coupling is inefficient, simple presence of GWs does not imply a corresponding presence of LSWS.
2.3. Confinement of Neutral-Ion Coupling to the Dip Equator
 Thus far, GWs have been assumed to have phase fronts that are either planar or circular. Circular phase fronts were introduced because they appear to provide a more accurate representation of GWs that reach the F layer after being launched from an MCC in the troposphere [Tsunoda, 2010b]. But, this representation seemed to negate the need for GW source regions (i.e., MCCs) to be located near the dip equator, and without this geometrical need, we no longer had a physical basis for arguing that a direct link exists between GW source regions in the troposphere, and ESF activity.
 The source of this quandary is that we had overlooked a GW property, which, when included, reestablishes the need to have the ITCZ close to the dip equator, in order for enhanced neutral-ion coupling to occur. That property is a downward tilt (from the vertical) in the phase front of the GW. That is, all GWs that have an upward group velocity (such as those that propagate upward from the troposphere to the thermosphere) will have phase fronts that are tilted downward [e.g., Hines, 1960]. Thus far, none of the GW representations used to date has included a downward tilt. For example, circular phase fronts that were used are actually vertically aligned cylinders, when drawn in three dimensions.
 Once we allow for a tilted phase front, we can easily show that GWBA occurs only at the magnetic dip equator. To keep it simple, we use a plane phase front for the GW. When the phase front is vertical, GWBA is satisfied when it is tangent to the magnetic meridian. In this case, the entire line is contained in the plane of a GW phase front. If we rotate the plane phase front in azimuth, we immediately lose GWBA, which leads to the requirement for zonally propagating GWs. If we add curvature to the phase front, GWBA is achieved only locally, where the phase front remains tangent to the magnetic meridian. Notice that the dip latitude, where local GWBA occurs, is identical to that of the MCC. Notice further that the magnetic inclination (I) angle of is finite here, but it does not matter because remains tangent to the vertical phase front. But if the phase front is tilted, will be in the plane of the phase front only where I = 0°, that is, over the dip equator.
 To illustrate, the interaction geometry for a finite I is sketched in Figure 1. Cartesian coordinates are used with the x axis along the magnetic east direction, y axis along magnetic north, and z axis is directed upward. A portion of the GW phase front is represented by a tilted rectangle (bordered by a thick gray line). The magnetic meridian is shown by a rectangle that is in the yz plane. The intersection of the two planes is indicated by a dashed line, which is horizontal and in the magnetic meridian plane.
 We see immediately that the interaction zone (marked by a small ellipse) is essentially a point, except when is straight and horizontal, and aligned with the dashed line. Elsewhere, where makes an angle I with the horizontal plane, the intersection is just a point. (A more rigorous treatment, which is beyond the scope of this paper, should include a determination of the polarization response, using a dipole line and a GW with spherical phase fronts.) Hence, we conclude that only MCCs that are located along the magnetic dip equator will launch GWs with phase fronts (planar or spherical) that are aligned with , which can elicit a polarization response. (Notice that equatorward propagating GWs in the magnetic meridian could have phase fronts that are aligned with , but only at latitudes removed from the dip equator, where I is larger. This alignment is irrelevant because it does not elicit a polarization response; GW-induced perturbations in are along , which do not drive spatially divergent Pedersen currents.) This explanation is satisfying because we are able to restore the GWBA hypothesis as the physical basis for why the ITCZ must be located near the dip equator, by simply using a more realistic description of GWs.
2.4. Outgoing Longwave Radiation
 Having restored the GWBA hypothesis as a reasonable physical basis for claiming a direct link between GW activity in the troposphere (i.e., ITCZ) and ESF activity, we have good reason to examine further the findings of Tsunoda [2010a]. This time, we use another data set, which is more closely matched in observation period (i.e., high solar activity) to that of the AE-E data set, which was used in the earlier comparison. The data set consists of measurements of outgoing longwave radiation (OLR) [e.g., Waliser et al., 1993], which are used here for identifying regions of deep convective activity, as were highly reflective cloud data in the work of Tsunoda [2010a]. Maps representing convective activity were constructed (1) to match the seasonal groupings used by McClure et al.  and (2) to display the regions of deep convection as a function of latitude and longitude. The format of the maps is well suited for evaluating the need for MCC alignment with the dip equator, which should be controlling the effectiveness of the neutral-ion coupling process.
2.4.1. OLR Data Description
 Maps of interpolated OLR data are available from the NOAA Earth System Research Laboratory, Physical Sciences Division Web site. Data have been collected by various satellites in sun-synchronous orbits [Gruber and Krueger, 1984], which include the Tiros N satellite (1 January 1979 to 31 January 1980; equatorial crossings were at 1530 and 0330 LT), and the NOAA 6 (1 February 1980 to July 1981; equatorial crossings at 1930 and 0730 LT). (Data from 17 March to 31 December 1978 are unavailable.) The interpolated OLR data set is an average of OLR measurements made during the day and night satellite passes at a given longitude. The original spatial resolution of OLR data appears to be ∼8 km in the subsatellite direction, but the data have been averaged and binned into an array with a grid size of 2.5° in latitude and longitude [e.g., Gruber and Winston, 1978]. Interpolation, both spatial and temporal, was performed to fill the array, as described by Liebermann and Smith . In comparison, the highly reflective cloud (HRC) data used by Waliser and Gautier  include only daytime measurements. Nevertheless, Waliser et al.  found, after comparing the HRC and OLR data sets, that both reproduced all major features in the equatorial zone, with some differences at higher latitudes; the latter is of no concern here.
2.4.2. Basic Features in OLR Maps
 The OLR maps used for the comparison with ESF activity are presented in Figure 2. The strength of OLR (in W/m2) is indicated with a color scale, which is shown at the top of Figure 2. Waliser et al.  found that 230 W/m2 is a reasonable choice for a threshold value that could be used to distinguish regions of convective and nonconvective activity. For discussion purposes, we will arbitrarily assume that deep convection was associated with blue and bluish-green regions, which appear to correspond with OLR strengths less than 200 W/m2. The OLR data used for this study are from 1979 and 1980, two of the three years used in the AE-E data set by McClure et al. . The intensity of OLR, which is plotted as a function of latitude and longitude, was averaged over the same 3 month intervals used by McClure et al. . They are MJJ (May, June, July), NDJ (November, December, January), ASO (August, September, October), and FMA (February, March, April). Features of interest found during these 3 month intervals are described below.
2.4.3. MJJ Months
 Maps for this 3 month interval are presented in Figures 2a and 2b, 1979 and 1980, respectively. Given that GW travel time, from troposphere to thermosphere, is several hours [e.g., Vadas, 2007], most relevant should be measurements made in the afternoon, that is, 1530 LT in 1979 (i.e., Figure 2a) rather than 1930 LT in 1980 (i.e., Figure 2b). Comparing the two maps, we find them to be virtually identical, especially features that are likely to be associated with regions of deep convection. The implication is that the regions of convective activity were stable and persistent throughout the course of a day. This finding is consistent with Waliser et al. , who noted that “synoptic conditions favorable for convective activity can persist for days, but the individual mesoscale features are much shorter lived (∼1 day).” Typical spatial scales are 1 to 102 km2 for a convective cell, 102 to 104 km2 for mesoscale precipitating features, and 104 to 106 km2 for cloud clusters [Waliser et al., 1993].
 The ITCZ is the band consisting of blue colored and dark green colored regions. Its latitudinal width seems narrower over the oceans and wider over landmasses (e.g., Africa, South America), although there are exceptions, especially in the Indonesian and West Pacific sectors. (Gray lines, if visible to the reader, outline the landmasses. Otherwise, the sector labels in Figure 2c can be referred to as a guide.) For the most part, the mean center of the ITCZ is located, in the northern hemisphere, in the vicinity of 10° geographic latitude, during these months. For comparison, the location of the dip equator can be estimated by interpolating between the various observing stations (shown with white circles), which are located very nearly on the dip equator. The southward penetration, which occurred around 170°E longitude, is a persistent feature, referred to as the South Pacific Convergence Zone (SPCZ) [Waliser et al., 1993], but it is not of interest here because it is a permanent fixture located away from the dip equator.
2.4.4. NDJ Months
 The OLR maps for NDJ months are presented in Figures 2d and 2e, January 1979 and NDJ 1979–1980, respectively. (Both data sets are from Tiros N with a daytime satellite crossing of 1530 LT.) The two patterns for deep convection are somewhat different, but the differences are not significant for our evaluation of the GWBA hypothesis. The most important difference in features with the OLR maps from MJJ months is the shift in ITCZ location from the northern hemisphere to the southern hemisphere. That is, virtually all regions of deep convection (blue areas), certainly over landmasses, are located below the geographic equator (dotted horizontal line). We also notice that OLR over Africa is stronger during NDJ months than during MJJ months. The seasonal behavior over oceans is different from that over land. We find that the ITCZ weakens in strength, rather than migrating in latitude. For example, blue regions can be seen in the northern hemisphere, in the East Pacific sector, during MJJ months. During NDJ 1979–80, we see a weak remnant (green region) of the ITCZ, which is still in the northern hemisphere. The weakly convecting (green) region in the Atlantic sector also remains in the northern hemisphere during MJJ and NDJ months.
2.4.5. FMA and ASO Months
 Maps for FMA months are presented in Figures 2f and 2g, and those for ASO months are presented in Figures 2h and 2i. These maps confirm that the ITCZ migrates in latitude with season, as shown previously [Tsunoda, 2010a]. The migration pattern during equinoxes can differ, as can be seen by comparing the patterns for FMA months with those for ASO months. We also see a bifurcation of the ITCZ, which can be seen during ASO months in the West Pacific sector.
2.5. Comparison of OLR Maps With ESF Activity
 The equatorial F region irregularity (EFI) probability curves, redrawn from McClure et al. , are presented in Figure 2c. (We refer to these as ESF activity curves.) The variations displayed in these curves are discussed below in terms of features associated with five longitude sectors, African, Indonesian, West Pacific, East Pacific, and American. The features of interest with regard to the seeding hypothesis are (1) maxima in ESF during a solstice, (2) minima in ESF during a solstice, (3) minima in ESF during both solstices, (4) seeding effects during equinoxes, and (5) exception to GWBA hypothesis. Below, we discuss why the first four features provide strong evidence for the validity of the seeding hypothesis. At present, only the fifth appears to escape explanation in these terms.
2.5.1. Maxima in ESF Occurrence During a Solstice
 If seeding controls ESF activity during solstices, ESF activity in Figure 2c should maximize in longitude sectors, where regions of deep convective activity in Figures 2a, 2b, 2d, and 2e, are located near the magnetic dip equator. There are three such longitude sectors in Figure 2c: (1) the African sector during MJJ months, (2) the Peruvian portion of the South American sector during NDJ months, and (3) the West Pacific sector during MJJ months. And, indeed, from Figures 2a and 2b, we find deep convective activity along the dip equator over the African continent (between Ibadan (Ib) and Addis Ababa (AA)). In fact, the peak in ESF activity is nicely centered on the region of deep convective activity; the tapering of ESF activity also agrees with weaker convective activity to the west, in the Atlantic sector, and absence of convective activity near the east coast of Africa. From Figures 2d and 2e, we find deep convective activity along the dip equator in the vicinity of Jicamarca (J), but less so in the vicinity of São Luis (SL), over the South American continent. The high ESF activity in the Peruvian subsector is consistent with seeding, whereas that in the Brazilian subsector is consistent with the STBA, which occurs during NDJ months. Steep tapering of ESF activity to the west agrees with the complete absence of convective activity in the East Pacific sector. There is almost no reduction in ESF activity to the east of Brazil, which agrees with the STBA hypothesis, because the magnetic declination remains large as the dip equator enters the northern hemisphere east of São Luis.
 The West Pacific sector differs from the African and South American sectors in that deep convective activity is not controlled by the presence of a continent. Nevertheless, this smallest of the three maxima in ESF activity is associated with deep convective activity, at least according to Figure 2b, although not obviously so, according to Figure 2a. The gradual taper in ESF activity to the east, seems to agree with the weakening of convection. (Whether convective activity to the east should be strong enough to seed ESF activity remains to be seen.) The taper to the west cannot be explained in terms of seeding, as discussed below, under the paragraph heading, “Exceptions to the Seeding Hypothesis.” We can conclude, however, that the three prominent maxima in ESF activity are consistent with the existence of a direct link between seeding and ESF activity.
2.5.2. Minima in ESF Occurrence During a Solstice
 If a direct link does exist, the converse statement should also hold. That is, if seeding controls ESF activity during solstices, ESF activity in Figure 2c should minimize in longitude sectors, where regions of deep convective activity in Figures 2a, 2b, 2d, and 2e, are not located near the magnetic dip equator. From Figure 2c, there are three longitude sectors, where ESF activity during a solstice is much weaker than that during the other solstice: (1) the South American sector during MJJ months, (2) the West Pacific sector during NDJ months, and (3) the African sector during NDJ months. The difference in ESF activity between solstices is most dramatic in the Peruvian portion of the South American sector. This finding is consistent with the complete absence of convective activity along the dip equator in that sector, as clearly seen from Figures 2a and 2b. This interpretation for the West Pacific sector, however, is not as conclusive as it is for the South American sector. From Figures 2d and 2e, deep convective (blue) regions were situated mostly in the southern hemisphere, well removed from the dip equator, which agrees with the seeding hypothesis. The slightly higher level of ESF activity in the West Pacific sector than in the Peruvian portion of the South American sector could be related to the exceptional behavior found in the Indonesian sector (see below).
 In the African sector, the minimum in ESF activity was higher than in either of the other two sectors. Referring to Figures 2d and 2e, we find that convective activity was to the south and well removed from the dip equator, similar to the South American sector during MJJ months, and clearly more removed than found in the West Pacific sector. At present, we have no explanation for the low but significant level of ESF activity during NDJ months in the African sector. But, we can conclude that there is a general tendency for ESF activity to be weak during a solstice, when the ITCZ is not located near the dip equator. Hence, the seeding hypothesis appears capable of explaining occurrences of maxima and minima in ESF activity during solstices.
2.5.3. Minima in ESF Occurrence During Both Solstices
 Perhaps the most convincing example for the existence of a direct link between seeding and ESF activity is the behavior found in the East Pacific sector, as already mentioned in Tsunoda [2010a]. The evidence is compelling because we find that the electrodynamical response (in the form of ESF activity) to a different circulation pattern in the lower atmosphere is consistent with the seeding hypothesis. The East Pacific sector is unique because of the absence of a large landmass and the location of the dip equator in the southern hemisphere. Here, the ITCZ does not migrate in latitude with season, as it does in other longitude sectors. Instead, it remains in the northern hemisphere; the only variation is in the intensity of convective activity in the ITCZ, which is stronger during MJJ months than during NDJ months. Consequently, the ITCZ is always distant from the dip equator, regardless of season. And, indeed, from Figure 2c, we see that ESF activity is equally very low, during both MJJ and NDJ months.
2.5.4. Seeding Effects During Equinoxes
 Although the focus of this paper is on establishing the role of the seeding hypothesis during solstices, we should also be interested in whether seeding effects are discernible during equinoxes. During equinoxes, STBA effects appear to dominate the ESF climatology [Tsunoda, 1985], which suggest that support from even the weakest level of seeding may be sufficient. On the other hand, the asymmetry in ESF activity during equinoxes in the East Pacific sector, for example, where seeding should be the weakest anywhere, may be another indication that even small differences in seed amplitude may be detectable in relative levels of observed ESF activity.
 We suggest that, in the East Pacific sector, the finding that ESF activity is enhanced during ASO (but not FMA) months could be explained in terms of the GWBA hypothesis, and the large magnetic declination angle of in the East Pacific sector. When the declination angle is large and eastward, MCCs that are located north of the dip equator should be able to contribute GWs that satisfy GWBA at locations on the dip equator. Comparing Figures 2h and 2i with Figures 2f and 2g, we find that convective activity to the north of the dip equator was stronger during ASO months than during FMA months. We surmise that although convective activity may not be strong enough to produce ESF through seeding alone, it may be able to do so with the assistance of a strong PSSR, which should be present during this time. Hence, we conclude that the complete climatology of ESF activity in the East Pacific sector could be explained in terms of the combined effects of GWBA and STBA.
2.5.5. Exception to GWBA Hypothesis
 Although the evidence presented is convincing, there are still exceptions to the notion that seeding depends solely on GWBA. The most conspicuous exception is found in the Indonesian sector. (We have referred to this region as the Vietnam-Philippines sector in the work of Tsunoda [2010a].) In this sector, ESF activity is very low throughout the year, which is not unlike that found in the East Pacific sector. According to Figures 2d, 2e, 2f, and 2g, seeding should have been weak during NDJ and FMA months, because the ITCZ was located to the south and remote from the dip equator. But, according to Figures 2a, 2b, 2h, and 2i, seeding should have been strong, because strong convective (blue) regions were located along the dip equator in this longitude sector. Interestingly, this exceptional behavior in the Indonesian sector may actually extend eastward into the West Pacific sector. For example, that there was significant convective activity in this sector during ASO months, perhaps more than in MJJ months; yet ESF activity is less intense in ASO than MJJ months.
 To summarize, three papers have now been written on a seeding hypothesis for the development of ESF; the other two are Tsunoda [2010a, 2010b]. In our view, the accumulated evidence, both theoretical and experimental, is undeniable. The fundamental theory is a neutral-ion coupling process, which requires GWBA [Klostermeyer, 1978; Huang and Kelley, 1996; Keskinen and Vadas, 2009]. But no one, until now, had shown why GWBA should occur only at the magnetic dip equator and not at finite dip latitudes. The key finding, presented herein, is that GWBA does not occur where I is finite because the GWs of interest are likely to have phase fronts with a downward tilt angle. This finding holds for phase fronts that are planar, as well as locally for phase fronts that have curvature. Concern also arose as to whether curved phase fronts necessarily lead to a shorting out of the polarization response. After all, neutral-ion coupling would not occur if an p does not appear. Hence, another key finding was that there is a substantial polarization response, even when GW phase fronts are curved [Tsunoda, 2010b]. These findings, taken together, provide a sound theoretical basis for believing that the seeding of plasma perturbations should occur, when GWs are launched from MCCs that are located near the dip equator.
 Experimentally, the evidence that seeding via the above described mechanism is playing an important role in the development of ESF is overwhelming. We have shown that the amplitude of the plasma perturbations varies with season and longitude, and that this behavior is controlled by the migration of the ITCZ in latitude with season. The migration of the ITCZ can be seen in data from any given year (shown herein), or even in data that have been averaged over as much as 17 yrs [Tsunoda, 2010a].
 Most importantly, we have been able to show that ESF activity, which varies with season and longitude [McClure et al., 1998], is basically consistent with seeding during solstices, as prescribed by the GWBA hypothesis. We have shown that both maxima and minima in ESF activity, which occur during solstices, can be explained to large extent by the seeding hypothesis. We have even uncovered an unusual situation, which occurs in the East Pacific sector, where the magnetic dip equator is located in the southern hemisphere, and the ITCZ remains in the northern hemisphere. Here, we find minima in ESF activity during both solstices, exactly as predicted by the seeding hypothesis. We have even found that ESF activity contains minor features, which appear explainable in terms of seeding effects during equinoxes. In fact, virtually all of the major features in ESF activity can be explained in terms of the STBA and GWBA hypotheses. The only exception is the behavior found in the Indonesian sector with eastward extension into the West Pacific sector.
 The notion that GWs may be more realistically represented with circular (or spherical) phase fronts is interesting, because the polarization response appears to favor longer wavelengths, which are comparable with those of GWs that reach the thermosphere [Tsunoda, 2010b]. This finding is consistent with the notion of seeding, as described, because the wavelengths of LSWS that have been observed are also comparable in length.
 An important finding that should follow from the GWBA theory is that not all GWs or MSTIDs that are observed at low latitudes are necessarily involved in the seeding of LSWS. For example, zonally propagating GWs that originate from an MCC not located on the dip equator are not expected to produce plasma perturbations. On the other hand, meridionally propagating GWs could produce MSTIDs through ion drag effects along , but they are not likely involved in the seeding process that leads to ESF. It is possible that, when a circular GW is launched at the dip equator, its zonal component could produce an LSWS, while its meridional component could excite an MSTID, if the GW propagates far enough poleward in latitude to where I is significant.
 The finding, that two distinctly different processes (STBA and GWBA) are responsible for ESF activity, is also satisfying because it is consistent with the finding that UHF scintillation tends to favor equinoxes, whereas VHF scintillation tends to favor solstices [Aarons, 1993]. Differences in ESF (or scintillation) properties should be expected, if the source mechanisms are different, and it seems reasonable that a PSSR-related (instability) process may be more capable of producing smaller-scale irregularities, hence, scintillations at UHF but not VHF, than a seeding-related process.
 We continue to be puzzled by the persistently low ESF activity in the Indonesian sector (with possible extension into the West Pacific sector). The OLR maps indicate broad regions of deep convective activity, similar to those found in HRC data presented by Waliser and Gautier . According to Figure 2, ESF activity should be high during ASO and MJJ months, but not during NDJ and FMA months. The notion that proximity to the MCC may act to depress ESF activity, which is based on expected properties of circular GWs [Tsunoda, 2010b], needs to be revisited, using more realistic models. The behavior of Fresnel zone effects should be considered by using GWs with spherical phase fronts and a dipole field.
 Much more needs to be done, but significant inroads have been made toward solving the enigma of the day-to-day variability in ESF occurrence [e.g., Tsunoda, 2005]. Clearly, we must understand ESF climatology before we can understand day-to-day variability. To make further progress, there is crucial need to measure the properties of LSWS, which reflect the effectiveness of the neutral-ion coupling process. At present, only the Communications/Navigations Outage Forecasting System (C/NOFS) satellite provides a convenient means for doing so. With C/NOFS, it is possible to measure total electron content as a function of longitude, receiving the beacon signals from C/NOFS at ground stations that are distributed in longitude [e.g., Thampi et al., 2009; Tsunoda et al., 2010].
 Interpolated OLR data provided by the NOAA/OAR/ESRL/PSD, Boulder, Colorado, USA, from their Web site at http://esrl.noaa.gov/psd/. This research was supported by the National Science Foundation under grant ATM-0720396 and by Air Force Research Laboratory/Air Force Office of Scientific Research under contract FA9550-10-C-0004.
 Robert Lysak thanks the reviewers for their assistance in evaluating this paper.