Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 The relation of equatorial bubbles to the equatorial anomaly is important because scintillation that is most disruptive to transionospheric RF propagation occurs when it passes through the intersection of the two. However, measurement of the relation between the two and of the electric field from which both arise is difficult because of large separations in space and time. This first attempt to perform these measurements employs a latitudinal array of ionospheric sounders spanning 0° to 40° dip latitude (DLAT) in the Western American sector. Measured on each day of a solar maximum year are the following: (1) the maximum electron density of the postsunset equatorial anomaly, Ne, at 16° and at 20.3° DLAT at 2100 LT, the time when the anomaly crest is at its maximum latitude; (2) equatorial spread F (ESF), detected by the occurrence of macroscopic bubbles and of bottomside spread F (BSSF), the latter recorded at levels of none, weak and strong; (3) Kp averaged over the 6 hours before sunset. Ne and ESF are considered functions of the maximum prereversal F layer drift E × B drift velocity measured by the Jicamarca incoherent scatter radar also during solar maximum and at the same longitude. Parameters are averaged over two levels of Kp for the three seasons, the E months (March, April, September, and October), D months (November–February), and J months (May–August) to yield the following results: (1) Ne measured at 16°, at 20.3° DLAT or at the anomaly crest are linearly dependent on maximum E × B drift velocity. (2) Occurrence of each level of ESF increases with Ne approximately linearly during the E and J months but not during the D months. (3) ESF occurrence is dependent on and increases approximately linearly with maximum E × B drift velocity during the E and J months. During the D months this dependence is absent. Except for the D months, these results indicate that scintillation increases with maximum prereversal E × B drift velocity: at L-band at the bubble-anomaly intersection because bubble occurrence increases, Ne increases, and the latitudinal extent of the anomaly increases; and at VHF/UHF near the equator because the occurrence of strong BSSF increases.
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.
 Scintillation is most disruptive to transionospheric RF propagation when it transits the irregularities produced at the intersection of an equatorial bubble with the high electron density of the postsunset equatorial or Appleton anomaly. The bubble and anomaly result from the postsunset prereversal enhanced eastward electric field which is measured by the upward E × B drift velocity of the equatorial F layer [see for example Kelley, 1989; Fejer et al., 1999]. However, the two arise from different mechanisms: bubbles and all levels of equatorial spread F (ESF) arise from the Rayleigh-Taylor instability produced by the upward vertical E × B drift plasma drift velocity and the resulting high altitude of the equatorial F layer, and the anomaly arises from plasma transport processes known as the enhanced or postsunset fountain effect.
 The times of occurrence emphasize the differences. At ∼1830 LT the prereversal drift velocity of the equatorial F layer plasma reaches its maximum. At ∼1930 LT ESF onset occurs near the maximum altitude of the F layer which is the time of reversal of drift velocity from positive to negative. ESF then can persist for several hours and extend over thousands of km in latitude and longitude. At ∼2100 LT the anomaly crest reaches its maximum latitude at about ±15° dip latitude (DLAT), nearly 2.5 hours after E × B drift maximum and ∼1.5 hours after reversal, and persists for many hours. [Argo and Kelley, 1986; Balan and Bailey, 1995; Whalen, 1998]. A particular consequence of the extent of and differences in space and time make measurement of the interrelation of all three very difficult.
 The most extensive measurements relating maximum prereversal E × B drift velocity (hereafter described as E × B drift) to ESF are by Fejer et al. , who used the Jicamarca incoherent scatter radar between 1968 and 1992 to examine irregularity level and its relation to E × B drift as a function of solar flux, season, and magnetic activity. They resolved bottomside spread F (BSSF) into three levels, none, weak, and strong, and found that weak and strong levels increased with increasing E × B drift. Bubbles, although not recorded separately, were accompanied by strong spread F, the threshold for which increased with increasing solar flux. During high solar flux levels they found the seasonal average of E × B drift to be dependent on season and on magnetic activity, quantitative results that will be incorporated into the present study.
 An extensive study of the postsunset anomaly and its relation to ESF has been by described by Valladares et al.  who recorded the anomaly profile in TEC at 2000 LT daily over the course of a year using a chain of GPS receivers located in the Western American sector. They found in general that the anomaly was greatest in latitude and TEC level near the equinoxes, least near June solstice, and intermediate near December solstice. ESF was indicated by the presence of plumes recorded by Jicamarca radar or by scintillation observed near the dip equator. They found that the presence of ESF as compared with its absence was associated with increased magnitude of TEC at the anomaly crest and decreased magnitude in the equatorial trough region. Their paper together with the above references give quite complete discussions of and references to previous work.
 The only joint measurements of E × B drift as cause together with Ne and with ESF as effects are by Whalen , measurements made during an equinox month at solar maximum in the Western American sector. Bubbles were observed to form when the four coinciding thresholds were exceeded: E × B drift of 50 m/s; anomaly crest maximum latitude occurring at 2100 LT of ∼15.4° DLAT; anomaly crest NmF2 at 2100 LT of ∼3.8 × 106 el/cm3; appearance of strong bottomside spread F.
 The anomaly crest latitude and magnitude were in a linear relation to one another above this threshold, a dependence that was judged to be a linear function of E × B drift. The highest-latitude bubbles coincided with the highest E × B drift and with the anomaly crests that were highest in electron density and latitude. Because the latitudinal extent of the profile also was observed to increase with the latitude of the crest, the indication was that the scintillation associated with these bubbles was the greatest because both Ne and its extent were greatest. In a related study of the entire year, ESF at each level decreased with increasing geomagnetic activity [Whalen, 2002], relations consistent with those found by Fejer et al.  and Huang et al. .
 This paper extends the equinox study to the entire year. Its purpose is to measure the dependence of the anomaly on E × B drift, the relation of ESF to the anomaly, and the dependence of ESF on E × B drift, measurements not previously undertaken. Also the implication of the observations to scintillation will be described.
 In outline, section 2 describes the experiment. Section 3 describes the measurements used to describe the anomaly. Section 4 describes the compilation of the daily measurements for the year. Section 5 describes the dependence of the anomaly on E × B drift. Section 6 describes the variation of the anomaly by month and its interpretation in terms of E × B drift. Section 7 describes the relation of ESF to the anomaly and their joint dependence on E × B drift. Section 8 is the discussion and section 9 is the summary and conclusions.
 Observations are by the array of ionospheric sounders shown in GLAT/GLONG in Figure 1a and in DLAT/GMLONG in Figure 1b. The period is the International Geophysical Year (IGY) of 1958 during the highest recorded solar maximum. These are part of the approximately 150 sounders that were in place worldwide during this period, producing the most comprehensive ionospheric database ever acquired.
 The sounders, even of this archaic vintage, can detect all the phenomena studied here. Equatorial bubbles are detected as range spread F (RSF) as demonstrated by the AFRL Airborne Ionospheric Observatory in a series of experiments in which RSF was found to coincide with bubbles that were identified separately by other means: the arcs of depleted 6300 Å airglow; the plumes of the Jicamarca incoherent radar; and scintillation on air-to-satellite RF transmission links [e.g., Weber et al., 1980].
 In addition to detecting bubbles, the array of sounders used here has been shown to measure bottomside spread F (BSSF) at levels of none, weak, and strong, the equatorial anomaly NmF2 in latitude and LT, and the prereversal E × B drift velocity by the rate of increase of h′F, the virtual height of the postsunset F layer [e.g., Whalen, 2000, 2001]. Altogether, this array of sounders is the only system of measurement extensive enough to permit the comprehensive study undertaken here. As such in spite of its antiquity, this database is in the forefront of observation of the complex equatorial region.
 Using this data set has the advantage that levels of ESF and NmF2 can be adequately measured by hourly tabulations of foF2 supplied by National Geophysical Data Center, as was shown in case studies that compared them with foF2 read directly from ionograms recorded at 5 and 15 min intervals [Whalen, 2002]. On the other hand, the measurement of E × B drift requires the difficult task of evaluating h′F from ionograms recorded on 35 mm film at 5 to 15 min intervals for several hours on each day of the year. Instead, this work will use E × B drift measured by the Jicamarca incoherent scatter radar reported by Fejer et al. . In addition to being the only such quantitative data set available, it was acquired also during solar maximum and in the same longitude sector.
3. Anomaly Represented at 16° Dip Latitude
 A principal difficulty in extending the above study to the entire year is that only during September was the sounder array complete enough to measure the latitudinal profile of NmF2 and so to determine the latitude and magnitude of the anomaly crest. The alternative is to use NmF2 recorded by the sounder at Bogota located at 16° DLAT because during September the anomaly crest parameters at 2100 LT, the time of its maximum latitude, could be inferred from Bogota alone. The relation of NmF2 at the anomaly crest measured on each day of September is shown as a function of NmF2 at Bogota in Figure 2a and of latitude of the anomaly crest in Figure 2b. Both relations are well described by linear functions over much of the observed ranges, to the extent that both crest parameters could be adequately determined from Bogota alone.
 Based on these results the anomaly will be represented by NmF2 at Bogota and described as Ne(16) to indicate its origin. Its validity as a measurement of the anomaly will be examined further by measurements at Panama and at the anomaly crest as described in sections 5 and 6.
4. Relation of ESF, the Anomaly, and Magnetic Activity
 The daily measurements are organized by season as the E months (March, April, September, and October), the D months (November–February) and the J months (May–August), hence near the equinoxes, the December local summer solstice, and the June local winter solstice. Shown in Figure 3 are the measurements on each day of the year:
 1. The level of ESF of no BSSF (N), weak BSSF (W) and strong BSSF (S) observed near the dip equator, and macroscopic bubbles (B) observed at the anomaly. Because B is found to occur only in the presence of S, all S is recorded as solid points, and those days on which B are observed are identified by the addition of an asterisk [Whalen, 2002]. The inset shows the designations that will be further described in Section 7.
 2. The anomaly at 2100 LT represented by Ne(16) recorded at Bogota.
 3. Geomagnetic activity represented by Kp averaged in the 6 h before the measurements, 1800–2100 and 2100–2400 UT, hence spanning 1300–1900 LT.
 In the E months, high levels of ESF indicated by B and S appear principally at high Ne(16) and at low Kp. With increasing Kp, Ne(16) decreases, occurrences of B and of S decrease and occurrence of N increases.
 In the D months, Ne(16) decreases more rapidly with Kp, than in the E months but without the decrease of S and B and increase in N seen in E months. As a result B and S appear at much lower values of Ne(16) than in the E months. In the J months, Ne(16) is low and ESF is mostly N, nearly the opposite of the E months.
 These measurements will be used in section 5 to describe Ne(16) as a quantitative function of E × B drift and in section 7 to describe the joint dependence of ESF and Ne(16) on E × B drift. They previously were used in the study of ESF as a function of Kp [Whalen, 2002].
5. Dependence of Anomaly on E × B Drift
5.1. Bogota Ne(16)
 Ne(16) shown in Figure 3 will be considered as a function of E × B drift measured by Fejer et al.  using the Jicamarca incoherent scatter radar under the same conditions of solar maximum and longitude. The average drift velocity for the same E, D, and J months determined for two levels of average Kp, ≤2.3 and ≥3.0 their (Figure 8) provide six values of maximum prereversal drift velocity, described here as E × B drift. They found this maximum to determine the level of ESF, a relation also found by Whalen  who in addition found E × B drift to determine the anomaly maximum. To correspond to these E × B drift values, Ne(16) in Figure 3 has been binned into the two Kp populations for each of the three seasons.
 The resulting six median values of Ne(16) are plotted versus E × B drift in Figure 4. Kp < 2 are solid circles and Kp > 3 are open circles, and the flags are the upper and lower quartiles. In each set, E month data are highest, D months are intermediate, and J months are lowest. The least squares straight line fit to the points is given by Ne(16) = 0.0820 * E × B drift + 0.733, where Ne(16) is in 106 el/cm3 and E × B drift is in m/s.
 Linear fits to the upper and lower quartiles of Ne(16) (shown dashed) have nearly the same dependence on E × B drift as the median and provide a measure of the fit in that the quartile range representing half of the days of the year is approximately ±0.7 × 106 el/cm3. Nearly the same dependence appears in the upper and lower deciles so that the linear dependence on E × B drift of the anomaly measured at Ne(16) appears at percentile levels of 10, 25, 50, 75, and 90.
 The result is the first measured dependence of the anomaly on E × B drift. Under the conditions described, the measurement of E × B drift determines Ne(16) but also the measurement of Ne(16) can be used to determine E × B drift. This correspondence between the two will be invoked in section 6 to describe the variation of E × B drift by month and in section 7 to determine ESF as a function of E × B drift.
 The use of Ne(16) as a metric for the anomaly will be examined further by the comparison with two other measurements, to the anomaly crest, and to Ne over Panama.
 The only direct measurements with which to compare Ne(16) are those of the anomaly crest measured in September by Whalen  (shown there in Figure 4b). These are shown as the crosses in Figure 4. Although only 3, the linear relation is supported by their being members of the set of 17 days in which the crests were linear in NmF2 versus anomaly DLAT. The crests have nearly the same slope as Ne(16), but are about 10 m/s higher in E × B drift. The difference in E × B drift is significant because it is the same as that of the thresholds for strong BSSF and therefore bubbles, observed to be 40 m/s by Fejer et al.  but 50 m/s by Whalen .
 The above-mentioned difference may result from the average solar flux during IGY being higher than that during the measurements of Fejer et al , based on their finding that the threshold for strong BSSF increased with solar flux. If this accounts for the difference of 10 m/s, Ne(16) is nearly the same as NmF2 measured at the crests so that the interpretation that Ne(16) is a measurement of the crest seen for the September measurements in Figure 2a is expanded to the entire year.
 The difference may due to the different measurement of velocity, Doppler by Jicamarca radar and dh′F/dt by the Huancayo sounder. The fact that both are similar in linear dependence suggests that the separate velocity measurements are internally consistent in determining the relative dependence but vary in magnitude by a constant factor [e.g., Bittencourt and Abdu, 1981, Rastogi et al., 1991, Fejer et al., 1996].
 At Panama at 20.3° DLAT the only other sounder located in the anomaly, NmF2 has also been measured at ∼2100 LT on each day and evaluated at the two levels of Kp in the same way as Ne(16). The resulting six median values described as Ne(20.3) are shown as the square points in Figure 4. The least squares straight line fit to the points is given by Ne(20.3) = 0.0220 * E × B drift + 1.106. As with Ne(16), the dependence of the quartiles is the same as that of the median.
 Although the dependence of Ne(20.3) on E × B drift is much less than that of Ne(16), the fact that the anomaly at even such a high latitude is also a linear function of E × B drift, reinforces the validity Ne(16) as a measure of the anomaly. Section 6 will show that the association of Ne(20.3) to Ne(16) is much more evident when observed by month.
6. Monthly Dependence
6.1. Ne(16) and Ne(20.3)
 The observations in Figure 3 displayed in the same format by month in Figure 5, show Ne(16) to increase from January to March, decrease from April to June, increase from July to September, and decrease from November to December. This variation is seen to be continuous in the median and quartiles of Ne(16) plotted in Figure 6. The variation is quite similar to that seen by Valladares et al.  in the overplots of TEC profiles during 1998 in this same longitude sector when solar flux was much lower and was increasing throughout the period.
Figure 6 also shows the electron density over Panama Ne(20.3) (shown dashed) to vary by month in nearly the same manner as Ne(16). This agreement is evidence that the anomaly can be observed throughout a range of latitude and can be measured throughout that range. This dependence on month at both locations and their interrelation are shown here without reference to E × B drift, but the variation can be used to further relate E × B drift to bubbles and to BSSF.
6.2. E × B Drift
 The relation of Ne(16) to E × B drift that has been derived in Figure 4 is used to construct an approximate scale of E × B drift that is shown along the right-hand y-axis in Figure 6a. Although there is the additional uncertainty in this scale because of the uncertainties in the fit of the linear relation in Figure 4, it is evidence that the monthly variation of the anomaly seen at Bogota and at Panama is the result of the monthly variation of maximum E × B drift. Furthermore, because of the one-to-one correspondence between Ne(16) and E × B drift, it is possible to infer E × B drift for each month by means of this scale and that E × B drift varies continuously by month.
 This result is consistent with E × B drift measured directly as Δ(h′F)/Δt at Huancayo during 1979–1980 by Batista et al.  which is shown in Figure 6b. The latter is lower in magnitude perhaps because of the lower solar flux and different technique of measurement. In any case the consistency suggests that E × B can be derived at least approximately via Ne(16).
 Bubbles have been recorded during solar maximum at two extremes of magnitude and duration: macroscopic bubbles detected in ionospheric soundings during 1958 [Whalen, 1997, 2002] and essentially all bubbles recorded in patches of scintillation at 137 MHz in this same longitude sector during the solar maximum of 1979–1980 [DasGupta et al., 1983]. In addition bubbles detected in electron content as depletions ≥1 TEC unit by DasGupta et al.  had very similar monthly dependence. The two have the same general seasonal characteristics as shown together in Figure 6c. Moreover these characteristics are nearly the same as those of the anomaly and E × B in Figure 3a, maximum in the equinox months, minimum near June solstice, and intermediate near December solstice. Based on the similarities of monthly dependence, both the anomaly and bubbles result from the maximum prereversal E × B drift velocity.
 The two levels of BSSF, total and strong BSSF as measured on each day of the year, are shown in relative daily occurrence in Figure 6d. Both of these independently measured levels are maximum in the D months, intermediate in the E months, and minimum in the J months, therefore departing from the behavior of bubbles and from the anomaly in the D months. Furthermore, it differs from E × B drift in the D months and is therefore not dependent at least solely on E × B drift. Section 7 will discuss this dependence together with quantitative descriptions of the relations of all the phenomena.
7. ESF as a Function of the Anomaly and of E × B Drift
 The daily occurrence of each ESF level shown for each season in Figure 3 has been binned at integral levels of Ne(16) and plotted as a function of Ne(16) in Figure 7. The format is described in the key, showing the ESF levels individually that are plotted together in the Ne(16) bin of 4 × 106 el/cm3. Included is total BSSF (T), all RSF recorded above threshold by Cohen and Bowles .
 The independently measured levels are found to be in the hierarchy that T is inclusive of S which is inclusive of B. As a result all three can occupy the same bin and thereby display each level individually as well as its relation to the others. Because T records all measurements above detection threshold and N, all below, the sum of the two records all the days within the bin.
 A more quantitative ordering of these relations is in the relative occurrence, the number of days of each ESF level within each Ne(16) bin divided by the total number of days within that bin as shown in Figure 8. Bins with fewer than 10 days have been ignored in this display.
7.1. E Months
 With increasing Ne(16), relative occurrence of ESF increases nearly linearly at each level B, S, and T and decreases nearly linearly for N. The fact that the levels increase together separated by 0.3 to 0.4 indicates the close relation of B to both S and of T.
 In particular, bubbles are most likely to occur with a large anomaly. In addition the crest latitude also increases with Ne(16) as shown in Figure 2b and the latitudinal width also increases with latitude [Whalen, 2001]. As a result scintillation increases in severity with this increase, as described further in section 8. Note that these relations have been found without reference to E × B drift velocity. However, the following will show the dependence of both Ne(16) and ESF on E × B drift.
 The dependence of ESF on E × B drift can be derived from Figure 8 using the linear relation of Ne(16) to E × B drift described in Figure 4 as was done in Figure 6. This dependence is represented by the linear scale of E × B drift on the upper axis. As before, there is additional uncertainty in this scale shown in the fit in Figure 4, in which the quartile range of about ±0.7 × 106 el/cm3 is somewhat larger than the binning range of ±0.5 × 106 el/cm3. Even so, it is possible to conclude that each level of ESF increases with E × B drift, approximately linearly and jointly with Ne(16).
7.2. J Months
 ESF levels are the lowest of the seasons and are lowest overall in Ne(16) and lowest in E × B drift. However very similar to the E months, with increasing Ne(16) and increasing E × B drift, occurrence increases also nearly linearly for S and T and perhaps B but decreases for N. Note that Ne(16) = 3 × 106 el/cm3 and E × B drift = 28 m/s ESF levels are very similar to the E months. Here it is evident that scintillation is minimal because bubble occurrence is minimal and the anomaly is small.
7.3. D Months
 In contrast to both the E and J months, the D months display no consistent relation of ESF occurrence to Ne(16) and hence to E × B drift, although Ne(16) has the same linear dependence on E × B drift as the other months. Thus ESF and in particular BSSF is not a function of prereversal E × B drift alone. The relation between bubbles and the anomaly is very different from E months and as was seen in Figure 3, bubbles occur not only at high Ne(16) but also at low Ne(16); the latter implying lower scintillation in magnitude and in extent.
Fejer et al.  observed that BSSF was not dependent on E × B drift and attributed it to the fact that reversal occurred later and that the postreversal drift velocity was lower in magnitude that is, (less negative) in the D months than in the E or J months. As a result, in the D months the F layer remained longer near its maximum altitude, thereby prolonging production of irregularities which in the E and J months were more rapidly quenched by the earlier and more rapid lowering of the F layer. It appears then that the success of E × B in describing ESF is in large part a function of the time of reversal and the magnitude of the post reversal drift velocity.
 The intersection of the macroscopic bubble with the anomaly described here is of particular interest because it is the locus of fading depths that can exceed 20 db at L-band for durations of ∼1 hour during solar maximum [Basu and Groves, 2001].
 In the E months, bubble occurrence increases with increasing anomaly Ne(16) and E × B drift velocity. In addition the crest latitude also increases with Ne(16) as seen in September in Figure 2b. Furthermore the latitudinal extent as of the anomaly increases as crest latitude and magnitude increase [Whalen, 2001]. As a result, the scintillation formed at the intersection of the bubble with the anomaly is expected to increase with E × B drift because the likelihood of bubbles increases, the electron density of the anomaly increases, and its latitudinal extent increases.
 In the J months the dependence on E × B drift and Ne(16) is the same, but E × B drift is much lower than in the E months. Accordingly, scintillation is low because bubble occurrence is low, Ne is low, and its latitudinal and extent is low.
 In the D months bubbles are most likely at the highest E × B drift which is at the highest Ne(16), conditions that produce the highest scintillation effects as in the E months. However bubbles also occur at low E × B drift, but those that do produce lower scintillation because the anomaly is lower in electron density and in latitudinal extent.
 However scintillation is not limited to the bubble-anomaly intersection. Strong BSSF can produce scintillation at UHF and VHF hence in a band nominally 12° in latitudinal width centered at the dip equator that can persist for many hours [Costa and Kelley, 1976; Aarons, 1993; Whalen, 2002].
8.2. Local Times
 The dependence of the anomaly seen as a linear function of E × B drift (Figures 4 and 5) is evidence that the effect observed at the anomaly near 2100 LT is the result of E × B recorded near 1830 LT. Furthermore the relation of all three is shown in Figure 8: ESF at each level with onset near 1930 LT and Ne(16) recorded near 2100 LT are both continuous functions of E × B recorded at near 1830 LT. The exception to this is in the D months because ESF is not primarily a function of E × B drift.
 As summarized in Figure 8, in the E months the measurement of E × B drift determines Ne(16) and vice versa so that the measurement of either yields the relative occurrence of ESF. For example at E × B drift = 55 m/s, Ne(16) = 5 × 106 el/cm3, relative occurrence interpreted as probability of occurrence of B is ∼26%, S ∼60%, T ∼95%. Because S is the necessary condition for all bubbles, the probability of having bubbles of any magnitude is 60%, and of having no bubbles is 40%.
8.4. Dependence on DLAT
 The dependence of Ne(16) on E × B drift for Bogota at 16° was seen above to be Ne(16) = 0.0820 * E × B drift + 0.733, and Panama at 20.3° was Ne(20.3) = 0.0220 * E × B drift + 1.063. At E × B drift = 0, Ne(16) = 0.733, and Ne(20.3) = 1.063 × 106 el/cm3, indicating constant levels above which Ne(16)/Ne(20.3) = 3.7. This is approximately the same factor seen independent of the relation to E × B drift in Figure 6. At latitudes between the two the dependence on E × B drift is probably intermediate.
8.5. Other Measurements
 The monthly dependence of the anomaly seen at Bogota and at Panama in Figure 6 is consistent with that of the anomaly profile seen by Valladares et al. , though at a lower and increasing level of solar flux. As a result, it is possible to conclude that their seasonal variation seen by month is the result of the variation of E × B drift. Also the relation of ESF to the anomaly seen there at two levels is seen here in Figure 8 to be continuously related for the E and J months, a relation that exists because each is a continuous function of E × B drift.
 The results here apply strictly to the Western American sector during solar maximum and therefore are expected to differ at other longitudes and lower levels of solar activity. However there may be wider application in longitude based on observations during IGY in the east Asia sector ∼180° in longitude away and therefore where the dip equator is in the Northern Hemisphere [Lyon et al., 1960]. There ESF occurrence was minimum in the D months hence local winter which is consistent with the minimum seen here in the J months which is local winter in the Western American sector. However the scintillation measurements compiled by Aarons  show decided variation in longitude implying variation in E × B drift.
 The global model of Scherleiss and Fejer  shows E × B drift to be greatest and most nearly constant in longitude in the E months so that the results found here are most likely to apply worldwide during this season. This condition appears to apply also during solar minimum conditions where E × B drift is much lower even though the threshold for strong BSSF/bubbles is decreased so that the occurrence of bubbles may not decrease. Even so, scintillation would be lower because the anomaly is expected to be lower in electron density and in latitudinal extent.
9. Summary and Conclusions
 This paper reports the first measurements of the dependence of the anomaly on E × B drift, the relation of ESF to the anomaly, and the dependence of ESF on E × B drift
 ESF is measured at each level of BSSF on each day of the year by an array of ionospheric sounders extending from the dip equator to the anomaly. The anomaly at ∼2100 LT is measured on each day by Ne(16), NmF2 at Bogota at 16° DLAT, based on its quantitative relation to the magnitude and latitude of the anomaly crest observed during September, and to NmF2 at Panama at 20.3° DLAT. E × B drift is measured in E, D, and J months by Jicamarca ISR [Fejer et al., 1999]. Kp is the average over the 6 hours prior to measurement.
 Ne(16) is a linear function of E × B drift where Ne(16) is averaged at high and low Kp in the E, D, and J months and plotted versus E × B drift averaged in those same categories. This linearity is consistent with that of individual crests observed during September indicating that this linearity also extends to the crests throughout the year.
 Panama Ne(20.3) is also linear with E × B drift, having a dependence that is reduced from that of Bogota Ne(16) by a factor of about 3.7. Agreement between the two is also evident when seen by month, indicating that the dependence of the anomaly on E × B drift is not confined to the crest, but is a property of a wide range of DLAT.
 The median value of Ne(16), together with that of Ne(20.3), varies continuously over the year, increasing from January to a maximum in March, decreasing from April to the minimum in June, increasing from July to the maximum in September, and decreasing from November to an intermediate level at December which is close to that seen at January. As a result of the one-to-one correspondence between Ne(16) and E × B, the monthly variation Ne(16) records the monthly variation of E × B drift velocity, and therefore of the maximum prereversal eastward electric field.
 Based on the similarity of monthly dependence, bubbles are a function of maximum prereversal E × B drift velocity. Based on the dissimilarity in the D months, BSSF is not solely a function of prereversal E × B drift velocity in those months.
 Relative occurrence of ESF at each level increases with Ne(16) nearly linearly in the E and J months but not in the D months. As a consequence of the linear dependence of Ne(16) on E × B drift, ESF is dependent on and increases approximately linearly with E × B drift during the E and J months. In the D months, the fact that ESF does not show this dependence is consistent with its dependence on postreversal E × B drift velocity.
 Under the uncertainties indicated, measuring E × B drift implies Ne(16), measuring Ne(16) implies E × B drift, and measuring either yields the probability of occurrence of each level of ESF during E and J months but not during the D months. These findings supplement those of Whalen , in which measuring average Kp yields the probability of occurrence of ESF in E and D months but not in the J months.
 At the bubble-anomaly interface, scintillation at L-band increases with E × B drift because bubble occurrence increases, and because the anomaly also increases both in electron density and in latitudinal extent. Near the equator scintillation at VHF/UHF increases with E × B drift because occurrence of strong BSSF increases.
 This work was sponsored in part by the Air Force Scientific Research Task 2311AS. Data were provided by the National Geophysics Data Center, Ionospheric Services Branch.
 Lou-Chuang Lee thanks Michael Kelley and another reviewer for their assistance in evaluating this paper.