Multitechnique observations may considerably improve our understanding of factors responsible for the generation, growth, and dynamics of the destabilized nighttime equatorial F region plasma irregularities. In order to investigate the dynamics of plasma density irregularities of different scale sizes, a campaign of observations was conducted during 11–20 November 2001 at the Brazilian magnetic equatorial station São Luís (2.57°S, 44.21°W, dip latitude 1.73°S). We carried out observations using VHF coherent backscatter radar, two spaced GPS-based scintillation monitors, and one digisonde. Range type spread F on ionograms and radar plume signatures on range-time-intensity maps of the VHF radar started at similar times. In order to compare GPS L1 (1.575 GHz) scintillations and radar plumes we used the scintillation S4 index computed for the signal transmitted by the highest elevation satellite. GPS scintillations were not observed during the initial bottom-type layer shown by the radar; however, stronger scintillations (higher S4 values) were observed concurrently to stronger radar echoes. Although the time duration of GPS scintillation is longer than the duration of the plumes observed by the radar, ionosonde spread F is still much longer than scintillation occurrence, confirming that smaller scale-size irregularities decay faster. Zonal and vertical velocities of 5-m irregularities measured by the radar were analyzed jointly with the apparent zonal velocity of ∼400-m irregularities measured by the spaced-receiver scintillation method. Larger values of the zonal velocity measured by the scintillation technique were found during the explosive growth phase of radar plumes associated with large values of vertical drifts measured by the radar.
 Many aspects of the generation, growth, and dynamics of the equatorial F region plasma irregularities, also referred to as equatorial spread F (ESF), especially their day-to-day variability, are still not completely understood. Multitechnique investigations may provide valuable information for their better understanding. Ionospheric electron density irregularities produce amplitude and phase scintillations on transionospheric satellite signals from VHF to L band mainly near the geomagnetic equator [Aarons, 1982; Basu et al., 1988]. The strongest L band scintillations, with signal fades of about 20 dB, occur during solar maximum years, at ±15° dip latitude, i.e., in the equatorial anomaly regions during postsunset period [Aarons, 1982; Basu et al., 1988]. The impact of scintillation on Global Positioning System (GPS) navigation has generated a new impetus in view of the increasing reliance on satellite-based positioning systems in critical applications such as air traffic control and precision landing [Kintner et al., 2001]. Earlier multitechnique investigations of ESF, such as the CONDOR project [Basu et al., 1986], addressed the comparison of vertical wave number spectrum obtained by rocket and UHF scintillation spectrum. More recent Multi-instrumented Studies of Equatorial Thermosphere Aeronomy (MISETA) campaigns in South America [Basu et al., 1996; Valladares et al., 1996; Mendillo et al., 2001] focused on neutral and plasma dynamics in relation to the day-to-day variability of ESF. In this paper, coordinated multitechnique measurements using a recently installed VHF (30 MHz) coherent backscatter radar, GPS L1 band (1.575 GHz) amplitude scintillation, and digisonde spread F observations are used to study the evolution of the ESF irregularities and their dynamics. A campaign of observations was carried out at São Luís, Brazil, in the magnetic equator during 11–20 November 2001. The radar probes 5-m scale-size irregularities (half the radar wavelength), whereas scintillation occurs due to radio wave scattering by decameter-scale irregularities predominantly of the Fresnel scale (∼400 m in this case) [Kintner et al., 2001]. Spread F observed on ionograms is caused by irregularities of kilometric scale sizes [Basu and Basu, 1993]. The zonal drift of the irregularities causing scintillation is derived employing the spaced-receiver technique [Kil et al., 2000; de Paula et al., 2002], and it is compared with that of 5-m scale irregularities obtained from radar interferometer measurements.
2. Motivation for the Present Study
 Recently, several authors have addressed the effects of scintillations over satellite-based navigation and communications systems [see, e.g., Bandyopadhayay et al., 1997; Skone et al., 2001; Kintner et al., 2001]. Skone et al.  have performed GPS L1 and L2 cycle slip measurements in order to compare the effects of high- and low-latitude scintillations over GPS receiver tracking performance. They found that low-latitude scintillations are much stronger than high-latitude scintillations, confirming the results of Basu et al. . They also showed scintillation effects on receiver tracking performance. Bandyopadhayay et al.  reported some examples of degradation in the position accuracy during periods of scintillation activity. More recently, Kintner et al.  pointed out that when the ionospheric puncture point velocity of a moving receiver and the scintillation pattern velocity match, the probability of loss of lock increases due to the longer duration of the amplitude fades. This case of a moving GPS receiver would be applicable to receivers on board airplanes.
 It is evident from the cited works that a better knowledge of ionospheric plasma density irregularities and their dynamics is important to understanding their effects on practical systems. The generation, evolution, and decay of these irregularities, especially in the low-latitude regions where stronger scintillation is observed, should be well known in order to estimate these effects.
 Also, the large availability of spread F measurements carried out during the last decades using different observational techniques motivates this study. It would be valuable to determine the relation between the intensity or duration of spread F events observed by any other technique and by GPS scintillation since other data sources could then be used to evaluate scintillation activity where or when scintillation measurements are not available.
3. Description of Instrumentation and Observations
 In order to investigate the ionospheric plasma irregularities of different scale sizes we have performed an observation campaign using one VHF coherent backscatter radar, two GPS scintillation monitors, and one digisonde. The observation campaign was conducted at São Luís (2.57°S, 44.21°W, dip latitude 1.73°S), located near the magnetic equator in Brazil during 11–20 November 2001. The VHF (30 MHz) coherent backscatter radar generates real-time range-time-intensity (RTI) maps of 5-m scale size irregularities. Postprocessing data analysis, using Doppler information and interferometer techniques, gives the vertical and zonal drift of the irregularities, respectively. The radar maps cover an altitude range from about 200 km up to 1300 km during nighttime.
 Measurements of 1.575 GHz amplitude scintillations were carried out using two GPS-based scintillation monitors. These monitors were developed by Cornell University using a GEC-Plessey GPS development kit, and they are capable of logging the signal intensity at 50 samples/s from up to 11 visible satellites simultaneously. The S4 scintillation index, which is defined as the standard deviation of the signal power divided by its average, is calculated every 1 min (3000 data points) for all satellites tracked during the observation night (1800 LT to 0600 LT in the morning). Beach and Kintner  give more details about this instrument. According to scintillation theory, L1 GPS signals are sensitive to ∼400 m scale size irregularities situated at 350 km altitude in the zenith [Yeh and Liu, 1982; Kintner et al., 2001].
 One digisonde was also operated simultaneously with the radar and GPS scintillation monitors. Inograms are available at intervals of 15 min. The occurrence of range and frequency type spread F on these ionograms was analyzed.
4. Results and Discussion
4.1. Occurrence of Plumes, L Band Scintillation, and Spread F on Ionograms
Figure 1 shows an example of an RTI map obtained by the VHF radar during the campaign period and simultaneous GPS scintillation and ionogram spread F observations for 16 November 2001. Echo intensities shown in the RTI map are divided into two ranges (signal-to-noise ratio (SNR) > 5 dB and SNR > 15 dB) for a better data visualization. The RTI map shows a low-altitude thin layer of 5-m irregularities developing from about 400 km at 1845 LT. Suddenly, around 1915 LT the layer explodes into a high-altitude plume. Radar plumes are interpreted as a manifestation of large plasma depletions known as ionospheric plasma “bubbles” that originate in the bottomside F region and may extend over several hundred kilometers in altitude. Sahai et al.  inferred from airglow data that plasma bubbles can attain equatorial heights larger than 1500 km. The RTI map also shows that initially the bottom portion of the plume is situated around 600 km altitude. However, 5-dB echoes may also be seen as high as 1200 km. The strongest echoes (SNR > 15 dB), which correspond to the presence of the most intense 5-m ESF plasma density irregularities, are observed mainly around 1930 LT. From 1900 to 2000 LT the plume reached high altitudes and then moved down, confining into a thin layer of about 200-km thickness.
 Initial efforts to compare radar plumes and L band scintillations were done by Basu et al.  using signals transmitted by geostationary satellites, which provide a fixed subionospheric point. In the present work we use data from GPS satellites that are constantly moving along different azimuth-elevation paths. Nowadays, GPS-based scintillation monitors have been extensively used and a better understanding of the relation between the data output by them and other techniques for spread F observations is important. Comparison would show whether data from other sources could be used to assess scintillation occurrence and/or intensity.
 In order to compare the dynamics of the radar plumes and the occurrence of GPS scintillations caused by irregularities over the observation site, we analyzed only the S4 index computed for the signals transmitted from the higher elevation angle satellites (>50°). Data gaps in the S4 data shown in Figure 1b are due to the unavailability of satellites with elevation higher than 50°. Figure 1b also shows the elevation of the satellite (dashed line) from which the signal was used to compute the S4 index. Figure 1 is a typical example (16 November data) showing that the maximum GPS scintillation activity seems to occur when stronger echoes are observed in the RTI map. At the time of maximum S4, the elevation and azimuth of the GPS satellite were 64° and 324°, respectively. Assuming an ionospheric penetration point at 600 km, the signal intercepts the ionosphere at about 76 km away from the observation site (in the magnetic west direction). Looking at the RTI map, the portion of the plume with stronger echoes lasted about 30 min. Assuming an average eastward velocity of 100 m/s, it gives us a first estimate of a structure 180 km large in the magnetic E-W axis. It confirms that the signal is intercepting the region of the plume with stronger echoes when the S4 reaches its maximum. We also assumed that the perpendicularity of the antenna beam with geomagnetic field is over the radar site since it is located in the magnetic equator, but it may be a few kilometers southeast with no further implications for our considerations.
Figure 1c shows the occurrence of spread F on ionograms for 16 November. The columns over the time indicate the intensity of the spread F, and the labels over the columns indicate the spread F type. R indicates range, F indicates frequency, and M indicates that both types of spread F were observed. The intensity is estimated only for range and range- and frequency-mixed spread F types and is indicative of the intensity of spread observed in the ionogram traces. Weak spread (spread layer with thickness <100 km) is defined as intensity 1. Moderate spread (spread layer with thickness between 100 and 200 km) is defined as intensity 2, and strong spread (spread layer with thickness >300 km) is defined as intensity 3. The range type spread F starts almost at the same time of the initial low-altitude thin layer (bottom-type) in the radar map. It is important to observe that the ionograms were taken at 15-min intervals. The concurrent beginning of the radar plumes and range spread F on ionograms was typical during the campaign period, indicating the sensibility of both instruments to irregularities in the F region bottomside. Figure 2 shows the satellite path in azimuth-elevation coordinates for pseudo-random number (PRN) 11 satellite from which we used the signal to compute the S4 index during strongest radar echoes. For the campaign period, the maximum intensity of GPS scintillation coincides well with the strongest echoes observed by the VHF radar and weaker scintillations associated with posterior weak thin layers that continue thereafter.
 Data analysis for other campaign days showed the occurrence of GPS scintillations even after the complete decaying of the radar plumes, similar to observations of Basu et al. . We observed also the absence of GPS scintillation during the occurrence of initial low-altitude thin layers (bottom-type layers) that precede the topside radar plumes [Hysell and Burcham, 2002]. Figure 3 shows the RTI map of the longest duration bottom-type layer preceding a topside plume we observed during the campaign period. During the occurrence of the bottom-type layer as shown in Figure 3 (∼1840–2000 LT), we did not observe any effect in the scintillation activity. Scintillation S4 index starts to increase only at the rising of the layer and development of the topside plume, when regions of stronger echoes are observed. The absence of scintillation during the bottom-type layer shows the good correspondence between S4 values computed for the highest elevation satellite and the radar echoes. We emphasize at this point that even using S4 data from high-elevation satellites we do not probe exactly the same ionospheric region probed by the radar, but we still obtain results similar to those pointed out by geostationary satellites. Once again, the range type spread F on ionograms starts almost simultaneously with the echoes shown in the radar RTI map.
4.2. Ionospheric Irregularities Zonal Velocities
 In order to analyze the zonal velocities of GPS scintillation irregularities jointly with zonal and vertical velocities of the 5-m irregularities probed by the VHF radar, two GPS-based scintillation monitors were installed 70 m from each other in the magnetic east-west direction. From the time delay of the maximum cross correlation of the signal power measured by the two scintillation monitors, we computed the apparent velocity of the irregularities. The apparent velocity υ′ is given by Briggs's full correlation method [Kil et al., 2000]:
where υo is the mean true velocity and υc is the characteristic random velocity. While υ′ values were directly obtained from cross-correlation analysis, further analyses were not performed to determine υc values since υo do not differ very much from υ′ [Kil et al., 2000]. However, proper correction for the ionospheric puncture point velocity in the case of orbiting GPS satellite was applied to obtain the plasma irregularity zonal velocity.
Figure 4 shows the vertical and zonal drifts measured by the radar as well as the zonal drift computed using the spaced GPS receivers for the same night of the RTI map shown in Figure 1 (16 November). The same data set used to compute the S4 values (highest elevation satellites) in Figure 1 was used to compute the zonal velocities using the spaced-receivers technique. The radar data show a spatial distribution of eastward velocities varying from about 100 to 250 m/s. During the explosive growing phase, radar data show maximum values of eastward zonal velocities (1900–2000 LT), and then the values start to decrease. Scintillation-derived zonal velocities also are abnormally large during the plume growth phase and attain steady values of about 200 m/s. Scintillation zonal velocity values are possibly caused by the large vertical velocity during the growing phase of the plumes, as we will discuss later. Zonal velocities of the irregularities sounded by the radar reach higher values; however, zonal velocity values measured by scintillation technique can be seen as averaged and weighted values by all irregularities crossing the signal line of sight. Also, the zonal velocities of radar irregularities decrease faster with respect to time.
 According to the statistical study of de Paula et al. , the zonal velocities of irregularities causing GPS scintillation do not vary very much before 2400 LT, when they then start to decrease. De Paula et al.  found average zonal velocities of about 150 m/s with standard deviation of about 25 m/s at the premidnight sector for December 1998 to February 1999. These values are smaller than the values we obtained during the campaign. This may be related to the lower solar activity period analyzed by them. Our zonal velocity values are also considerably larger than Jicamarca zonal drifts for high-solar-activity years [Fejer et al., 1991]. This is consistent with the works of Kil et al.  and Valladares et al. , who found larger values of scintillation irregularity zonal velocities than the zonal velocities of ambient plasma measured by incoherent scatter techniques at Jicamarca.
 The growth phase of the 16 November radar plume is characterized by a strong vertical drift as indicated by the vertical velocity radar map in Figure 4a. Apparently, the strong vertical drift is affecting the zonal velocity derived by the spaced scintillation method, which gives abnormally large and highly variable zonal velocities at this time. Bhattacharyya et al.  have shown that the observed irregularity velocity υo has a contribution from the eastward velocity (υe) and from the vertical velocity (υz) such that
where θ and ϕ are the zenith and azimuth angles, respectively, of the signal path. According to the elevation-azimuth plot shown in Figure 2, the GPS signal used to compute the zonal drift by the spaced-receivers method during the growth phase of the plume was transmitted by a satellite located at θ ≈ 23° and ϕ ≈ 318°. For this specific geometry, the observed irregularity velocity is given by υo = υE + 0.28υz. A positive contribution, proportional to υz, is then added to the zonal velocity values giving origin to the observed large values. After about 1945 LT the contribution by υz is negligible and the scintillation zonal drift velocity shows steady values.
 To confirm the vertical drift effect in the zonal velocity estimation, Figure 5 shows the same type of data set shown in Figure 4 but for 13 November. In this case the large values of radar plume vertical drift were not observed and we did not observe the large values on the scintillation zonal velocities.
 We did not observe GPS (L band) scintillation during low-altitude irregularity layers (bottom-type layers) shown in the RTI maps in the initial development phase of the plumes. If we assume that ∼400-m scale size irregularities are collocated with the 5-m irregularities, it suggests that such irregularity layers do not cause detectable GPS scintillation magnitudes. When the plume develops to higher altitudes, stronger radar echoes and simultaneous maximum S4 scintillation index values are observed.
 According to scintillation theory [Yeh and Liu, 1982], S4 depends on the electron density deviation (ΔN) of the ionospheric irregularities and also on the thickness (L) and height (z) of the irregularity layer. Bottom-type layers shown by the radar are situated in an ionospheric region of low plasma density and ΔN is expected to be very small. The radar also shows that bottom-type layers we observed are characteristically thin (<50 km). However, when the plume develops, the irregularity layer reaches the F region peak where the plasma density is much higher, giving rise to larger density deviations. During this time the irregularity layers as shown by the radar may be much thicker (>500 km), and this may explain higher levels of scintillation observed. It should be mentioned that even in the F region peak, the electron density is not very high at the equator due to the fountain effect that takes the plasma to higher magnetic latitudes (∼±20°), and this explains why we observed only weak scintillation levels during the observation campaign.
 GPS scintillation continues even after complete decaying of the radar plumes as first observed by Basu et al. . This observation may be explained in terms of the faster decaying of smaller irregularities. During the observation campaign, range type spread F started almost simultaneously with the beginning of echoes on the radar RTI maps, indicating the sensibility of both techniques to irregularities in the bottomside F region. As expected, because of the slower decaying of the large scale size irregularities, spread F on ionograms continues after the decaying of the radar plume and after the end of the scintillation activity.
 Using the vertical velocity maps of the radar, we could observe the strong effect of the vertical velocity over the spaced-receivers method to compute the zonal velocity of irregularities. Simultaneous radar and scintillation measurements confirm the statement of Bhattacharyya et al.  that attributed large zonal velocities (also computed by the spaced-receiver method) to the E × B drift arising from perturbation electric fields in the early phase of development of equatorial plasma bubbles.
 It was difficult to compare the zonal velocities given by the radar and the zonal velocity of the irregularities causing GPS scintillation. Zonal drift radar maps showed both spatial and temporal distribution of velocities varying in the range of about 100–250 m/s eastward (values for 16 November, for instance), while spaced-receiver scintillation techniques showed velocities of about 200 m/s.
 Scintillation-derived zonal drift values we measured during the campaign are larger than the values measured by Kil et al.  during November 1998 and by de Paula et al.  during November 1998 to February 1999 at Cachoeira Paulista, Brazil, located under the equatorial anomaly. Larger values at the magnetic equator were expected to be due to vertical shear of the zonal drift [Basu et al., 1996]. Our larger zonal velocity values may also be related to the increase of the solar activity.
 We acknowledge helpful comments given by B. G. Fejer. F. S. Rodrigues especially acknowledges T. Pedersen and K. Groves from the U.S. Air Force Research Laboratory and the Windows on Science Program for supporting his participation in the IES 2002. This work is part of the FAPESP project 00/13325-5. São Luís radar was funded by FAPESP under project 99/00026-0. K. N. Iyer acknowledges CNPq for visiting researcher grant under process 301213/00-3. We are grateful to A. Cunha for equipment maintenance and operation at São Luís and to F. T. Martins and M. G. S. Aquino for ionosonde spread F data reduction.