HF Doppler radar observations of low-latitude spread F

Authors


Abstract

[1] HF (5.5 MHz) Doppler radar observations of nonspread F and spread F echoes over Visakhapatnam (17.7°N, 83.3°E; dip 20°) are presented. The echoes appearing suddenly and nearly simultaneously in 16 successive range bins at 7.5 km intervals in association with spread F have been investigated. Two to five episodes of spread F activity were found to appear at intervals of 1–2 hours during individual nights. At the time of onset of spread F conditions, the Doppler velocity for each range bin changed rapidly from a negative maximum to a positive maximum followed by a gradual decrease to a steady ±10–15 m/s or to a large negative velocity and then again to a large positive. At the time of small constant velocity or velocity change from negative to positive, the spread F echoes were weak or even below the detection level of the radar. This disappearance in the higher ranges causes the decrease in range extent of spread F echoes. The positive and negative maximum velocities of spread F were in the range of +70 to −60 m/s. The maximum upward and downward velocity is not the same in all events of spread F activity. The width of the Doppler velocity spectrum for spread F echoes was found to vary with velocity. For zero velocity the width was a minimum of 50 m/s in contrast to 25 m/s for nonspread F events. These features were consistently observed for all spread F incidences. The observed results are compared with already reported HF/VHF observations and are discussed in the light of equatorial plasma dynamics during the growth phase of Rayleigh Taylor instability leading the incidence of spread F.

1. Introduction

[2] The phenomenon of equatorial spread F (ESF), manifested as diffuse echoes in the equatorial and low-latitude ionograms has been a topic of active scientific research since its first observations by Berkner and Wells [1934]. It is now understood that the large-scale (10–100 km) plasma depletions, or “bubbles,” are generated in the bottom side of the nighttime F layer via Rayleigh-Taylor (RT) instability or the gradient drift instability which evolve through nonlinear processes into the topside of the F region. This process produces a spectrum of electron density irregularities of scale sizes ranging from a few centimeters to a few hundred kilometers. Theoretical models of rising plasma bubbles have been developed by several researchers [e.g., Haerendel, 1973; Scannapieco and Ossakow, 1976; Ossakow and Chaturvedi, 1978; Ott, 1978; Anderson and Haerendel, 1979; Cakir et al. 1992; Huang and Kelley, 1996a, 1996b, 1996c] and these models were generally consistent with the available experimental evidence. From the in situ measurements using satellites and rockets [Kelley et al., 1976; McClure et al., 1977; Morse et al., 1977] information on the size and percentage depletion of plasma bubbles is reasonably known. The AE-E satellite traversing through the plasma bubble showed that the electron density inside the bubble is much smaller than the density out side [Tsunoda et al., 1982] and is highly structured.

[3] The plasma irregularities associated with the ESF bubbles were studied using many instruments, including coherent and incoherent scatter radars [Woodman and Lahoz, 1976; Tsunoda, 1980, 1983; Tsunoda et al., 1982; Hysell et al., 1994; Hysell and Woodman, 1997; Patra et al., 1995; Rao et al., 1997; Hysell and Burcham, 1998] and optical instruments such as photometers and all sky imagers [Weber et al., 1978; Rohrbaugh et al., 1989; Tinsley et al., 1997; Kelley et al., 2002], each experiment being sensitive to a particular region of the irregularity scale-size spectrum. The coherent radar echoes are produced by scattering from Field Aligned irregularities (FAI) of scale size equal to one half the radar operating wavelengths for radar beam incident normal to the geomagnetic field lines. Clemesha [1964] and Kelleher and Skinner [1971] made ESF observations using 18 and 27.8 MHz radars. More recently Tiwari et al. [2004] reported 18 MHz radar observations of ESF from Thumba, an Indian station on the magnetic equator. Cecile et al. [1996] have reported near simultaneous observations of ESF using ionosonde and 2.8 and 3.5 MHz fixed frequency radar from West Africa. They showed that the range-time-intensity (RTI) maps of the 2.8 and 3.5 MHz echoes depicted several nearly identical successive arches extending over more than 100 km.

[4] Unlike VHF radars which provide information on the meter scale irregularities that are formed by the disintegration of the large-scale structures, the HF radars are sensitive to the decameter-scale size irregularities and therefore provide information on the spread F irregularities during the formation times. Also, the HF radar provides information on the ambient background ionization dynamics before the onset of spread F and during nonspread F events. The simultaneous observations of spread F using HF radar operated at 5.5 MHz situated at Visakhapatnam and ionograms from three stations Visakhapatnam (VSP), Sri Harikota Range (SHAR), and Trivandrum (TRV) facilitated to study dynamical character of the ESF events. It is well known that large-scale (10–100 km) irregularities originate in the bottom side and then evolve through nonlinear instability process into topside, and degenerate into small-scale structures. Therefore, investigating the bottom side of the F region prior to and after the triggering of ESF using HF radar is of interest since the HF radars are believed to be better instruments for the investigation of decameter irregularities which are important during the ESF growth phase. An attempt is also made to compare the 5.5 MHz radar observations with the earlier HF/VHF radar observations from other locations.

2. Experiment and Data Analysis

[5] The observational data on ESF used in this study were acquired using 5.5 MHz pulsed HF Doppler (HFD) radar at Visakhapatnam (17.7°N, 83.3°E, geomagnetic dip 20°) and ionograms from Trivandrum (8.5°N, 76.9°E, geomagnetic dip 0.6°), SHAR (13.7°N, 80.2°E, geomagnetic dip 5.5°) and Visakhapatnam. The specifications of the radar are given in Table 1. The data of three nights, 30 September 2004 (day 274 had AP and F10.7 values are 3 and 88, respectively), 19 and 20 October 2004 (day 293 and 294) on which the ionograms had shown strong spread were presented in this study as representative of the time histories of ESF plumes observed with 47 nights of spread events studied with HF radar. The 5.5 MHz frequency usually penetrates the ionosphere over Visakhapatnam during the postmidnight. But during the spread nights the echoes were observed even up to 2130 UT (0300 IST) of the next day.

Table 1. HF Doppler Radar System Specifications
ParameterSpecifications
Radar typeCoherent detection type
Operating frequencyAlternate pulses of 5.50 MHz
Pulse width100 μs
Pulse repetition frequency50 Hz
Transmitter peak power4 kW plus
Transmitting antenna3 wire folded dipole
Direction of the beamVertical
ReceiverCoherent detection type
Receiving antennaHalf-wave dipole
Receiver outputsI and Q channels and amplitude
Receiver outputs range±5 V for I and Q and 0–5 V for amplitude
Number of range bins16
Sampling pulse interval50 μs
Sampling pulse width2.5 μs
Number of data inputs (maximum possible)16 single ended
Number of FFT points used for spectrum512

[6] The HFD radar pulse width was 100 μs and PRF was 50 Hz. Both the transmitting and receiving antennae were dipoles with a beam width of about 60° in the east west plane. The “I and Q” outputs of the received radar echoes, were sampled and held simultaneously for 16 successive range bins at intervals of 50 μs, were sequentially scanned, digitized and transferred to the computer hard disk. The split buffer mode of acquiring and transferring data to the hard disk ensured no loss of data between data files. The data recorded in successive files were arranged sequentially to form a continuous time series of I and Q values for each of the 16 range bins. For each range bin, the I and Q time series were analyzed by 512 point FFT routine to compute the complex Doppler spectra at intervals of 10.24 s. The Nyquist frequency of the spectra is 25 Hz giving an unambiguous maximum Doppler velocity of ±682 m/s and the resolution was 2.66 m/s. It was noted that the spectral amplitude outside −6.25 Hz through 6.25 Hz frequency bins is only noise and the spectral power within this band contained all the echo power. For each Doppler spectrum, the signal-to-noise ratio (SNR), line of sight velocity (V) and the spectral width (SW) were computed from the three lower-order, zeroth, first, and second, moments (denoted as M0, M1, and M2, respectively) representing the total signal power (PS), weighted mean Doppler shift (equation image) and variance (fw2) which is a measure of the dispersion from the mean Doppler frequency (equation image), respectively, using the expressions given by Woodman [1985]

equation image
equation image
equation image

where m and n are the lower and upper limits of the Doppler bin of the spectral window, Pi and fi are the powers and frequency of the Doppler bins within the spectral window. The Doppler velocity (V) is −M1 × λ/2. The Doppler frequency shift being positive (negative) for downward (upward) velocities the definition of “V” in this study implies upward (downward) velocities are positive (negative). Signal-to-noise ratio in dB is calculated as

equation image

where N and L are the number of Doppler bins (FFT points) and mean noise level per frequency bin, respectively. From a survey of all the Doppler spectra of F region echoes, the Doppler velocity V was always found to be within ±100 m/s except during the presunrise and postsunset times in which the Doppler velocity some times could be as large as ±120 m/s. We take the spectrum within ±6.25Hz, corresponding to Doppler velocity of ±170 m/s, contained the entire echo power. L is calculated as the mean spectral power for the frequency bins −25 to −6.25 Hz and 6.25–25 Hz and M0 is calculated as the total received power in frequency bins −6.25–+6.25 Hz. Thus N = 128 and N × L gives the total noise over the signal bandwidth. Doppler width is taken to be the full width of the Doppler spectrum and is calculated as

equation image

[7] The I and Q received from 16 successive range bins separated by 7.5 km were processed offline to give SNR, Doppler velocity and spectral width using the expressions given by Woodman [1985].

[8] According to the studies of Namboothiri et al. [1988], Doppler velocities measured with HF radar represent electrodynamic (E × B) drift, but the contribution due to layer decay is significant at lower heights. Bittencourt and Abdu [1981] theoretically computed (E × B) drift velocities and showed that the correction due to layer decay is insignificant above 300 km and is less than 10 percent at about 280 km. It must also be noted that the small contribution due to layer decay to the measured velocities is always upward thus making the measured upward (downward) Doppler velocities to be overestimates (under estimates) of the electrodynamic drift velocities.

[9] It may be noted that the echo from 5.5 MHz Radar is basically a specular reflection from the isodensity layer corresponding to the transmitting frequency during nonspread F condition. However, during spread F condition, the radar beam penetrates through the irregularities due to density variations and gets reflected (though there could be coherent scattering also) from the structures of the same electron density giving rise to the altitude structure of the equidensity layer. In addition to simple reflection, there could be scattering also from decameter-scale size irregularities which is indicated by the Spread in the Doppler spectrum reported in section 3.2.6 of this article.

3. Results

3.1. HFD Observations of Nocturnal F Region During Nonspread F Condition

[10] Figures 1a and 1b show the time variations of the echo signal strength (representing group height until penetration around midnight) and the cumulative phase height on 8 October 2004, a nonspread F night, taking an arbitrary initial phase height of 256.5 km, equal to the group height at 1145 UT (1715 IST). Even though, the variations of phase height and group height have similar trend the magnitudes of variations are different. The difference between the maximum and minimum group heights during the night is more than 100 km whereas the difference for phase heights is <15 km. Noting that the group height is particularly sensitive to the electron density gradients at the reflection height, observations imply very significant changes of both reflection height and integrated electron density below the reflection height during this night. The magnitudes and time variations of the group and phase heights are not the same every day but a treatment of these aspects is beyond the scope of this paper.

Figure 1.

(a) RTS plot for the day 8 October 2004 which shows typical variations of no-spread day. (b) Phase path variations for the same day. (c and d) The time variation of power (V2) and velocity averaged for all heights, respectively.

[11] The range-time-SNR (RTS) map shown in Figure 1a indicates the nocturnal variation of the F region vertical drift during a typical nonspread F night. The first peak occurs around 1230 UT (1800 IST) similar to the E × B drift reversal caused by the F region dynamo which was confirmed by several workers [Rishbeth, 1971; Anderson, 1981]. The vertical drift velocities at subtropical latitudes like Visakhapatnam are a result of the combined effect of the electro dynamic drift and neutral winds in the F region. Considering the results of Anderson [1981] on E × B drifts at the equator and those of Heelis et al. [1974] on the predicted vertical drifts due to tidal and polarization electric fields, Rao et al. [1984] showed that the vertical drifts at Visakhapatnam are predominantly due to electric fields during 1230–1530 UT (1800–2100 IST), those between 1530 UT (2100 IST) to local midnight are dominated by neutral winds and those during postmidnight are again due to electric fields. The bottom side of the F region as seen from the RTS maps shows a smooth variation reflecting the vertical dynamics with out any significant undulations or prominent quasi sinusoidal variations during a nonspread F night.

[12] Figure 1c shows the time variations of the average echo power in successive (usually three and occasionally two or four) ranges in which the SNR is >40 dB (the Doppler velocity is nearly the same for the three range bins). Figure 1d shows the time variations of Doppler velocity computed as the weighted mean for the successive range bins, weighted by the echo power. The times of maximum echo power shown in Figure 1c do not match with the maximum velocity plotted in Figure 1d. This is expected because the maximum echo power depends on the radius of curvature at the reflecting surface and is largest when the radius of curvature is equal to the reflection height. The small-scale variations of the order of 10–20 min are clearly observed in Figure 1c which may be because of the medium-scale TIDs in the F region heights. The vertical drift velocity at 15 min interval obtained from the time derivative of ionosonde measured virtual height (d h′F/dt) are shown as solid symbols on Figure 1d for comparison. The trend of the drift velocity variation matches with the HF Doppler drift velocities, though the magnitudes are slightly different. HF Doppler velocity gives the height variation of the density layer sensitive to 5.5 MHz, while the ionosonde derived drift is contaminated by the variation of the F region bottom side electron density, and hence the small deviation is expected.

[13] Figure 2a shows a typical example of the amplitude spectra of a specular F region echoes in our database. Shown here are the Doppler data corresponding to 16 ranges stacked vertically. The echo is seen in three to four successive range bins (22.5 km to 30 km range extent) because of the 50 μs sampling interval, the echo pulse broadening due to ionospheric dispersion and the receiver band width, to more than 100 μs. For discrete specular echo by Fresnel reflection, the Doppler velocity is to be the same in the successive range bins as observed. If there is more than one echo at the same range, they would be seen resolved in Doppler spectra. Figure 2b shows successive Doppler spectra of the specular echo from 264 km range which shows a smooth change in velocity from 6 m/s to about 20 m/s in 7 min.

Figure 2.

(a) Doppler velocity spectra of discrete F region echo, for 30 September 2004 on which AP and F10.7 values are given as 3 and 88, respectively, seen in three successive range bins 256.5, 264, and 271.5 km. (b) Time variations of spectra of the echo from 264 km during 1745–1752 at intervals of 10.24 s for the same day.

3.2. HFD Observations of Spread F Echoes

3.2.1. Range-Time-SNR Maps

[14] Figures 3a–3c are the RTS (for SNR > 40 dB) maps of nighttime F region showing ESF echoes on 30 September, 19 October, and 20 October 2004, respectively. The signals penetrated the ionosphere during the postmidnight hours. The bottom F region echo on 30 September and 19 October shows the usually observed postsunset enhancement during 1206–1300 UT (1736–1830 IST) followed by a decrease till 1400 UT (1930 IST) which is conducive for the generation of Spread F that was observed simultaneously in all height ranges soon after the postsunset height maximum. The duration of these spread F plumes was about 15 min on 30 September and more than 45 min on 19 October. On 20 October the postsunset enhancement is absent which may be the reason for the delayed appearance of spread F echoes after 1530 UT (2100 IST). Each spread F event lasted for about an hour. It may be noted that the SNR of spread F echoes is about 10 dB less than that of specular echoes from the bottom of the F region since the echoes from ionosphere irregularities are weak compared to specular echoes. This is due to the fact that the 5.5 MHz echo is due to reflection and during nonspread F conditions a strong specular reflection is expected unlike the VHF radar, where the scatter intensity shown in the RTI maps is a function of the scale size of the irregularities.

Figure 3.

RTS maps of nighttime F region echoes on (a) 30 September, (b) 19 October, and (c) 20 October 2004 showing ESF plumes for SNR > 40 dB.

[15] During a Spread F night as seen from Figures 3a–3c, the F region shows a large upward drift during the postsunset hours with peak around 1300 UT (1830 IST). Following this, a significant quasi sinusoidal variation is seen at the bottom side of the F region after which a sudden onset of spread F [Woodman and Lahoz, 1976] was observed. Mendillo et al. [1992] suggested that the nightly requirements for the RT instability growth leading to the onset of spread F are postsunset raise of F region, availability of seed perturbation to launch the RT mechanism and an absence of strong transequatorial thermospheric wind. This quasi sinusoidal variation at the bottom side probably indicates the seed perturbation for the generation of irregularities causing spread F.

[16] It may also be of interest to note that spread F did occur on 20 October 2004 (Figure 3c) even in the absence of postsunset upward drift at Visakhapatnam as measured by the Doppler Radar. However, at the equator, a significant postsunset upward motion of the F layer is observed on ionograms derived heights (Figure 6) which is conducive for the generation of spread F.

[17] As expected, the onset times of the spread F episode and the range extent of all spread F echoes agree well with that of the ionograms scaled at 5.5 MHz. The significant features as observed from the present HF Doppler observations are (1) echoes appear and disappear in a broad range of altitudes; (2) the interval between the observation of successive spread F plumes on a night is same and ranges between an hour to two, which probably ties to the phase of the large-scale bottom side F region modulation for favoring the growth of RT instability; and (3) the SNR strength shows a 10–30 min variation in echo strength during spread F event also as in the case of bottom side HF signal in the absence of spread F, which implies that the echo strength is modulated by the focusing and de focusing of the HF signal due to bottom side curvature caused by gravity wave modulation. This also falls in line with the fact that the medium-scale F region structure indicates a smaller-scale F region modulation present during quiet and spread F days.

3.2.2. Doppler Spectra for Different Ranges

[18] The Doppler spectra of the radar data, for different ranges, before the onset of spread F and at different times of the spread F episode are shown in Figures 4a–4d, respectively. It is observed that the echoes before the onset of spread F show same low value of mean Doppler velocity at both 256.5 km and 264 km while they show a higher value of −36 m/s at 271.5 and 279 km. Under spread F conditions, strong spectral peaks appear for successive lower ranges, for example, below 279 km as can be seen in Figure 4b. Strong spectral peaks appear in two or three successive ranges (15–22.5 km range extent) implying nearly specular echoes from the lower F layer. The echoes were observed simultaneously from all the range bins when the spread F event is observed (Figure 4b) with large spectral width when compared with nonspread condition as shown in Figure 4a. The spectra for each range, in general, show many peaks of both positive and negative Doppler shift and this spreading of spectra implies Doppler resolved echoing region.

Figure 4.

Doppler spectra of the ESF echoes on 30 September 2004 in the 16 range bins at the times noted. The vertical separation between successive range bins is 1.5 V.

[19] Although the beam widths of both transmitting and receiving antennae are very large (3 dB bandwidth is approximately 60° for a simple 3 wire dipole antenna), the location of the echoing regions for each range is limited by the range resolution of the radar. At the lowest range of 300 km, the zenith angular spread is ≈20° for a 15 km transmitter pulse length (corresponding to a pulse width of 100 μs). During spread F conditions, the observed echoes in successive higher ranges may be from increasingly larger zenith angle. From the geometry, it can be calculated that the zenith angular spread for an echo observed from 400 km range with 300 km altitude is 42°. The contribution of the horizontal velocity (VH) of the irregularities in the measured Doppler velocities for the lowest and highest ranges would be 2.5 and 33%, respectively. It is known that the electron density profile under spread F conditions will be highly structured and in such a case the zenith angle of the echoes is limited. Under such condition, the large range extent of echoes is due to increasing distance along the raypath. The highly unrealistic tilts of electron density surface may not lead to reflection of the radio wave and the observed echoes may be because of coherent scattering from ionospheric irregularities close to the reflection level. It is also evident with findings of Cornelius et al. [1975] that approximately 20–30 m irregularities are present when spread F occurs as indicated by coherent scattered echoes on HF sounders. If the scattering electron density irregularities (27.27 m scale size for our radar) are highly aspect sensitive as in the case for ESF echoes in VHF and UHF radars, they should lie in a plane tilted at the dip angle (20° toward north in our case), the range of spread F echoes appearing suddenly should be larger by 20 km and should be discernable. In our observations, the incidence of spread F conditions was never seen separated from the lower F region echoes which implies that the irregularities are not very aspect sensitive unlike the highly aspect sensitive irregularities as seen by VHF and UHF radars. These irregularities are confined to all azimuth directions and a zenith angle of about 20° (set by the radar pulse width).

3.2.3. Time Variations of Spread F Spectra

[20] The Doppler spectra of echoes under spread F conditions, observed on 30 September at 10.24 s intervals are shown stacked vertically for selected ranges are shown in Figure 5. The spectra show considerable spreading of Doppler velocity often with both positive and negative Doppler but, usually, not symmetric about the zero Doppler. The mean Doppler will be weighted more toward either large positive or negative values as the contribution to sum of the product “Pifi” (where Pi and fi are the power and frequency of the Doppler bins) from small Doppler on either side of zero is significantly reduced.

Figure 5.

Time variations of the Doppler spectra on 30 September 30 2004 for the range and time durations noted. The vertical shift of consecutive spectra is equal to the spectral amplitude of 1.5 V.

3.2.4. Spread F in Ionograms Across the Equatorial Region

[21] We have attempted an assessment of latitudinal pattern of spread F across the equatorial region by consulting ionosonde data of Visakhapatnam (VSP), SHAR (located around the northern fringe of the equatorial electrojet belt in India) and Trivandrum, TVM (located closer to the magnetic equator). Figure 6 shows the virtual height extent of 5.5 MHz echo scaled from the ionograms at the three stations. In Figure 6, the upper thick line with circles and the lower thin line represent the maximum and minimum virtual heights of spread F echoes, respectively. Spread F conditions were observed at all the three stations with some similarities and some differences on individual dates. The maximum range observed at around 1330 UT (1900 IST) over Trivandrum was 500–550 km on all the three days. It was about 425 km over SHAR on 30 September and 19 October 2004.

Figure 6.

Time variations of virtual height extent of 5.5 MHz echo scaled from the ionograms at Visakhapatnam, SHAR, and Visakhapatnam. The line with circles represents the top group height and the line without circles is for the bottom height of the ESF echoes in the ionograms.

[22] Range spread in ionograms was observed immediately after the postsunset height rise for the two days (i.e., 30 September and 19 October) which was not observed on 20 October. Over Visakhapatnam there were two plumes on 30 September, one at 1315 UT (1845 IST) and another at 1445 UT (2015 IST) with a minimum range difference in between. The postsunset onset of spread F is usually earlier at TRV and SHAR compared to VSP and so also is the virtual height extent of the spread F echoes which assumes higher values at TRV and SHAR compared to VSP. The data for the night of 20–21 October 2004 illustrate these trends.

[23] Figure 7 shows the virtual height extent of spread F at 3.5, 4.5, 5.5, 6.5, and 7.5 MHz over VSP for the three nights. The height extent is larger for 3.5 MHz. The variations of the minimum group height at all the frequencies are strikingly similar and the maximum group height is varying with sounding frequency. It is also of interest to compare Figures 3 and 6 to see how the range extent as observed by ionograms at 5.5 MHz matches with that observed by the HF Doppler Radar. We see a very good correspondence on the range extent in both the cases.

Figure 7.

Same as Figure 6 but for different sounding frequencies at Visakhapatnam.

3.2.5. Doppler Velocity Variations in Each Range Bin

[24] Figures 8a–8c show the time variations of the Doppler velocity (for SNR > 40 dB) on the three nights. The velocity variations in three successive ranges are shown below. The correspondence of the ESF echoes in the signal-to-noise ratio and velocity are seen clearly. Although, the spectra some times have power in a wide band of positive and negative Doppler frequencies (see Figure 2c), the weighted mean Doppler variations with time are smooth and continuous. At 1430 UT (2000 IST) on 30 September (Figure 8a), ESF appeared in the lower ranges with 20 m/s downward velocity which changed to an upward maximum of about 20 m/s followed by a slow decrease to 20 m/s downward. For ranges above 324 km, the occurrence of spread F was a bit earlier than successive lower ranges and only the change from large positive to large negative are seen. The maximum positive and negative velocities are larger for spread echoes from higher ranges than the specular echoes and are as large as +80 m/s and −80 m/s in the largest range. At 1510 UT (2040 IST) the velocity started changing rapidly from negative to positive. From 1545 UT (2115 IST) onward, the velocity started decreasing slowly to ±15 m/s. Also, the rate of change from negative to positive velocities is larger than that from positive to negative. At the time of the velocity change from negative to positive for ranges above 301 km, the echo is weak and below radar detection limit. This is seen in the RTS maps as appearance and disappearance of ESF echoes.

Figure 8.

Velocity variations plume echoes at different heights for the three spread nights (a) 30 September, (b) 19 October, and (c) 20 October 2004.

[25] If the echoes for each increasing range were from increasingly large zenith angles, then the horizontal velocity component along the raypath should continuously increase with range, which is not seen in the weighted mean Doppler velocity for successive range bins. There are also instances when the velocity in the higher range is less than that in the lower range bins. Therefore, we are considering that the irregularities producing spread F echoes in the increasing range bins reflect the vertical variations and that the large field of view is not contaminating the interpretation of the Doppler velocity as from the vertical component.

[26] Also, there are quasiperiodic velocity changes of about 12 min period superposed on the above velocity variations with large amplitude when the velocity changes from negative to positive. The oscillation is clearer in the lower ranges on 30 September and 19 October. Similar features were observed during the incidence of spread F on the other two nights presented in Figures 8b and 8c. The short-period oscillatory velocities appear at the higher ranges first and then successively in lower ranges after a small but finite time delay, which can be seen clearly on 19 October at 1550 UT (2120 IST) for 256 km to 294 km ranges. For the other episodes of spread F also, when the echo is above threshold, the maximum positive and negative velocities occurred earlier in higher ranges and then at successively lower ranges after a finite delay. Further, the cyclically similar velocity variations at nearly constant intervals of one to a few hours each night imply a propagating wave of the same period. It must be noted that the velocity in the lowest range was within ±15 m/s irrespective of ESF in the higher ranges. It is observed that the larger was the Doppler velocity; the Doppler width of the spectrum also was larger indicating a diffuse echo. It is also interesting to note that the minimum spectral width for zero Doppler velocity is almost constant for the nonspread F days whereas it is larger for higher ranges during spread F event.

3.2.6. Spectral Widths During Nonspread F and Spread F Conditions

[27] Figure 9 shows the spectral widths as a function of Doppler velocity during nonspread F (Figure 9 (bottom)) and spread F (Figure 9 (top)) conditions. It may be noted that the spectral widths are low during nonspread F conditions. The variance in the spectral widths are quite low (60 m/s) when the velocities are downward while they are relatively higher (up to 80 m/s) during upward motion of the layer. However, during spread F conditions, the spectral widths are quite high (up to 80 m/s) during downward motion and up to 110 m/s during upward motion. This indicates the dispersive nature of the spread F echoes. The higher spectral widths and the larger variance during the upward motion of the irregularity also indicate the rapid topside expansion of the plume. The spread of Doppler spectra observed during spread F than those during nonspread F is a clear indication that the echo during spread F cannot be considered as a simple reflection. Also 80 m/s velocity is much higher than the ambient plasma motion and hence it may be true that the HF Radar is also measuring the polarization field present at that time.

Figure 9.

Spectral widths as a function of Doppler velocity for the (top) spread F and (bottom) nonspread F events.

4. Discussion

[28] The HF Doppler radar observations at Visakhapatnam indicate several interesting features that compare and contrast the observations of VHF radars at other locations. The appearance and disappearance of plume structures with a time interval of 1 to few hours compares with those observed by VHF radars. The characteristic feature of the plumes is the local altitude modulation in the bottom side F layer associated with the development of plasma bubbles as suggested by Woodman and Lahoz [1976]. During such event, the most probable bubble size matches the scale size of the initial electron density perturbation, particularly during the generation phase. Hence, the interval between appearance and disappearance of the plume like structures in the HF radar is constant in a day and varies between 1 to a few hours on a day to day basis, making the large-scale perturbing phenomena like the interaction of the propagating gravity wave with the background ionosphere as a leading candidate behind the ESF plume development [Tsunoda, 2005].

[29] HF Doppler radar observations indicate near simultaneous occurrence of ESF in many range bins while the observations from VHF radars indicate that the plume grows upward from bottom side to topside of the F layer as a function of altitude [Woodman and Lahoz, 1976]. The observation of high-altitude echoes almost simultaneously with the lower altitude during spread F growth must be due to coherent scattering from F region irregularities due to plasma depletion in the bottom side. Observation of echoes from large range bins than the normal F layer suggests that the 5.5 MHz signal penetrated the F layer which is possible when the region under sounding is an electron density depleted region. During the growth phase of the plasma bubble it is expected that the bubble walls will have an electron density suitable for producing coherent backscatter and hence may be the reason for near simultaneous observation of echoes in all the range bins during the growth phase unlike the VHF radars.

[30] The velocities in the plume as observed by the HF Radar are in the range of ±80 m/s whereas the velocities reported from VHF radar observations are much higher, above 150 m/s [Patra et al., 2005; Tiwari et al., 2004]. Scannapieco and Ossakow [1976], Ossakow and Chaturvedi [1978], and Ott [1978] from their theoretical simulation reported that the bubble rise velocities can reach 300 m/s if the bubble rise above 400 km and if the percentage depletion is very high. However, these velocities are due to the combined effect of the E × B velocity due to the background Eastward electric field and the rising plasma bubble. But the HF Doppler radar observations reflect the rise velocities during the growth phase and as the radar frequency is 5.5 MHz, they reflect the velocities of medium-scale irregularities. It is of interest to note that the velocities reported by Patra et al. [2005] using a 18 M Hz radar are in the range of ±100 m/s. Second, some difference in velocities could be expected for different radars with different operating frequencies. It is also known that the velocities are mainly controlled by electric field up to 350 km and by buoyancy force above [Ossakow and Chaturvedi, 1978; Anderson and Haerendel, 1979]. It may be noted that the velocities measured by the HF Doppler radar show a smooth variation and highly organized. At higher altitudes, the velocities are higher as reported from VHF radar observations due to the nonlinear growth of the RT instability and also due to the dominant role of the buoyancy force resulting in diminishing collision as a function of altitude. As the present observations are limited in altitude extent to below 400 km, the observed velocities could be lower due to comparatively lower buoyancy. It may be noted that the VHF radars are sensitive to meter-scale irregularities while the HF Radars reflect the characteristics of decameter size irregularities which appear earlier and at lower altitudes compared to meter-scale irregularities that are subsequently generated by the upward diffusion of the plume and subsequent degeneration.

[31] It is also of interest to note that there are rapid variations inside or during the early development of the plasma plume, with higher ranges showing this feature earlier than the lower range altitudes. Also there is negative Doppler velocity at lower altitudes prior to plume development, which could possibly be due to a descent outside the rising plume as a part of the plasma interchange process. Tsunoda et al. [1995] published range-Doppler-intensity (RDI) maps of evolving ESF bubbles observed with their frequency-agile radar operated at 5.51 MHz as part of NICE (Neutral-Ion Coupling at Equator) program at Kwajalein Atoll (9.4°N and 167.5°E) and interpreted the Doppler signatures to be due to wedge shaped plasma depletions which are open for ground based HF soundings. They observed that the ESF plasma bubble development is characterized by distinct updrafting and the pattern of velocity variations can be interpreted as oblique echoes from a tilt or irregularities on a tilted electron density surface located west and moving eastward in the bottom side F layer. Fukao and Kelley [1991] reported Doppler velocity variations of FAI associated with midlatitude spread F using MU radar data. All spread F incidences showed updrafting FAI on a wedge shaped electron density bubble. These bubbles appeared and disappeared at intervals of a few hours. Just as we observed, they also reported weak/no echoes when the velocity was downward and changing to upward.

Acknowledgments

[32] This work is supported by the Council of Scientific and Industrial Research and the Indian Space Research Organization under CAWSES-India Programme.