By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 An analysis is presented of high-frequency (HF) signals from the European Incoherent Scatter HF ionospheric modification transmitter received during 26–30 October 2002 at three sites, two in Europe and one in Antarctica. Two components with different characteristics (“mirror-reflected” and “scattered”) were observed in the signal spectra. The mirror-reflected component can be associated with radiation through the side lobes of the transmitting antenna thus bypassing the modified volume on its way to the receiving sites. In contrast, the scattered component was radiated through the main antenna beam and then scattered by pump-induced ionospheric irregularities above the heater. As a result, variations in the scattered component signal intensity and Doppler frequency shifts (DFS) recorded at the greatly separated sites showed a high level of correlation. It is shown that the Doppler frequency variations can be associated with variations in the plasma density and/or physical motion velocities of stimulated inhomogeneities within the volume common to all propagation paths. Analysis of Doppler frequency shifts at greatly separated sites would allow identifying the mechanism responsible for the self-scattering effect. In the case of DFSs due to motion of the pump-induced scatterers it would be also possible to reconstruct the full velocity vector of the inhomogeneities.
 Powerful high-frequency (HF) radio wave transmitters (commonly about 2 to 10 MHz), often referred to as ionospheric heating facilities, are actively used in the USA, Europe, and Russia to produce controlled modifications in the Earth's ionosphere. The main scientific goal of such experiments consists in the use of the near space environment as a natural plasma laboratory [Gurevich and Shvartsburg, 1973], where a variety of microphysical as well as geophysical phenomena can be stimulated at a known time and in a given location. Four heating facilities, namely Sura (Nizhny Novgorod, Russia), EISCAT (Tromsø, Norway), SPEAR (Svalbard, Norway), and HAARP (Gakona, Alaska, USA), are actively operating in the world today, and one, Arecibo (Arecibo, Puerto Rico, USA), is currently under construction.
 One of the principal effects of the interaction between powerful HF transmissions and plasma is the development of artificial ionospheric turbulence (AIT) and artificial ionospheric irregularities (AII), which, in a variety of radar observations, may play the role of a convenient “target” with distributed parameters. Studies of the dynamics of AIT and AII, including the spatial distribution of inhomogeneities and their interaction with other ionospheric and magnetospheric processes, can yield unique information both on the turbulence itself and on the “undisturbed” background properties of the Earth's environment [Erukhimov et al., 1987; Beley et al., 1997; Sinitsin et al., 1999; Ponomarenko et al., 2000]. Most papers on radio probing of AIT and AII are based on local diagnostics of turbulence either in close vicinity or within a moderate distance from a heating facility [Fejer et al., 1991; Blagoveshchenskaya, 2001; Stubbe et al., 1992; Yampolski et al., 1997] (see also the bibliography in the work of Stubbe ). In this paper we consider radio diagnostics of the pump-plasma interaction volume using the HF pump wave itself as a probe signal propagating over long and superlong radio paths.
 In October 2002 an international experimental campaign was conducted to investigate the long-range propagation of HF radio signals and the effect of natural and HF radio wave pump-induced inhomogeneities on those signals. One of the principal goals was to determine the effect of transmitter, or pump wave, spectrum modification under different geophysical conditions. The pump signal was emitted by the HF transmitting facility located near Tromsø, Norway, and consisted of two simultaneously transmitted waves with a difference in frequency of either 19 or 22 Hz. Signals at both transmitted HF frequencies, f1 and f2, and some of their combinations (specifically, mf1, mf2, and m(f1 + f2), with m = 1, 2, 3) were recorded at three widely separated receiving sites deployed in Antarctica, in the Kharkov region of Ukraine, and near St. Petersburg in Russia (see Figure 1). In addition, extra-low-frequency (ELF, 3–30 Hz) signals in the vicinity of the difference frequency ∣f1 − f2∣ were observed in Antarctica and Ukraine. The latter might be due to radiation from a current system in the lower ionosphere modulated by the high-power transmissions [Stubbe et al., 1982; McCarrick et al., 1990]. This paper however is devoted to analyzing and interpreting spectral features of the fundamental pump frequencies received at the different sites and to determining the connection between these characteristics and the physical processes occurring within the region illuminated by the heating facility.
2. Experimental Technique
 For the experiments reported here the pump wave was emitted by the European Incoherent Scatter (EISCAT) HF transmitter located near Tromsø, Norway, at coordinates 69°35′N, 19°14′E (site EIS, see Figure 1). The maximum of the transmitting antenna pattern was directed at 78° elevation, along the direction of the Earth's magnetic field. The pump wave frequencies were selected to be below the critical frequency of the F2 layer. The value of the critical frequency was monitored by the EISCAT 929-MHz incoherent scatter radar and ionosonde, both colocated with the HF transmitter facility. The HF operating frequencies varied from 4040.000 kHz to 7953.717 kHz depending on the ionospheric conditions. The observational campaign was conducted on 5 days, during the hours when the solar terminator simultaneously crossed the interaction region over Tromsø and the most remote receiving site in Antarctica. The experiments were carried out from 0200 to 0800 UT (Universal Time) between 26 and 29 October and from 0400 to 1200 UT on 30 October 2002. HF transmissions started on the exact hour and lasted for 55 min. The transmitter power during the remaining 5 min of every hour was decreased either to 10% or to 0% level (transmitter off). Local standard time in Tromsø equals UT plus 1 h.
 Transmissions from the EISCAT HF facility were designed to generate 22 Hz and sometimes 19 Hz ELF waves in the ionosphere in the three ways detailed below. However, it is the transmitted fundamental frequencies which are important for the results presented here.
 1. Sine-like modulation of the HF pump wave power at the 22-Hz ELF frequency, with less time spent at high power than at low power. The radiated power was varied between 0 and 100% of full power during the first 55 min of each hour and between 0 and 10% of full power in the last 5 min. This mode was used on 26 October.
 2. Equal-length on-off pulsing of the HF power at the 22-Hz ELF frequency for 55 min followed by 5 min of no transmission. This was done on 28 October.
 3. Splitting of the HF antenna array in two equal parts, with the two halves transmitting at frequencies offset from each other by the desired ELF frequency. This was used on 27 and 29 October (22 Hz) and on 30 October (19 Hz). Between 0300 and 0600 UT on 29 October and between 0400 and 1200 UT on 30 October the transmitter was off during the last 5 min of each hour. The transmitter was also off on 29 October between 0655 and 0657 UT.
 In all modes the peak transmitted power equaled the maximum HF transmitter power. On 29 October X mode was used between 0300 and 0400 and between 0500 and 0600 UT; all other transmissions during the experiment were O mode.
 Mode 3 is of particular interest as it was used most often, and as it brings into play spatial as well as temporal interference effects between the waves transmitted by the two halves of the HF array. In this mode the waves combine to produce an apparent sawtooth scan of the constructive interference maximum of the resultant HF beam across the main HF lobe at the difference frequency. More precisely, the ELF difference frequency between the two half-arrays can be regarded as a constant linear change in the phase difference ϕ between the two halves, causing the lobes of the interference pattern to move steadily at some dϕ/dt through the main lobe of the half-array antenna pattern, where dϕ/dt equals the difference frequency. For a commanded (i.e., programmed) antenna pointing at 12° south of zenith, near field-aligned, and at transmitter frequencies of 4.040 and 7.953 MHz on the 3.9 to 5.4 MHz and 5.4 to 8.0 MHz wide-beam arrays (arrays 2 and 3), respectively, the parameters of the HF beam pattern have been computed using an antenna modeling program (EZNEC+ version 4.0, http://www.eznec.com/). At 4.040 MHz, the angular difference between consecutive interference lobes at the point in the rotational pattern where the strengths of the two largest lobes are equal, each at about 19.8 dB, is 23.2°, with lobe positions of 0.7°N and 22.5°S. At 7.953 MHz, the corresponding figures are 22.3 dB and 17.2°, with lobe positions of 1.5°S and 18.7°S. In the case when there is only one main lobe at 4.040 MHz pump frequency, for a commanded pointing of 12°S the main lobe maximum gain is 22.4 dB at 11.0°S, with a half-power width of 13.9° (−3 dB points at 4.1°S and 18.0°S). For the same case at 7.953 MHz the corresponding figures are 25.0 dB gain at 10.5°S, with a half-power width of 10.2° (−3 dB points at 5.5°S and 15.7°S). Thus the HF beam is always on, although at any fixed point in space the electric field and incident power will oscillate at the chosen ELF difference frequency.
 During the observations the EISCAT 929-MHz incoherent scatter radar antenna was scanned in the east-west plane for 15 min and then held stationary for 15 min. The radar measured backscatter power from ion-acoustic waves versus range on six and sometimes five channels at frequencies spaced by 300 kHz from one another. The sixth channel was on occasion tuned such as to observe electron Langmuir waves, that is to near 929 MHz ± fpump, with fpump being the pump frequency. The radar transmission cycle repeated every 5 ms, meaning that one raw range-power profile was recorded on each channel every 5 ms. Thus 5-ms time resolution can be retained even when averaging the five ion-acoustic wave channels. This in turn made it possible to search for 19- or 22-Hz variations in the backscatter power. In this study however the ion-acoustic data was used to produce electron density profiles for the HF propagation analysis discussed below.
 The pump signal was recorded at all three remote receiving sites (see Figure 1): (1) UAS, the Ukrainian Antarctic Station Akademik Vernadsky, at coordinates 65°15′S, 64°16′W (path 1, ground distance 16300 km); (2) RAO, the Radio Astronomical Observatory of the Institute of Radio Astronomy, National Academy of Sciences of Ukraine, Kharkov region, Ukraine, at coordinates 49°40′N, 36°50′E (path 2, ground distance 2400 km); and (3) AARI, set up by the Arctic and Antarctic Research Institute, St. Petersburg, Russia, and located near St. Petersburg at coordinates 59°57′N, 30°42′E (path 3, ground distance 1200 km). Local standard time at UAS is UT minus 4 h, and at both RAO and AARI it is UT plus 2 and 3 h, respectively.
 Signal reception at UAS was performed using a horizontal dipole with the reception pattern directed toward the transmission facility (EIS). The signal then was converted to an intermediate frequency and filtered within a 300-Hz band by a single-channel coherent receiver with highly stable (Δf/f ≤ 10−9) local oscillators. Finally the signal was digitized by an analog-to-digital converter with a 72-dB dynamic range at the 700-Hz sampling frequency. Further processing was performed using specially designed software. To check both the equipment serviceability and propagation conditions, during the last 5 min of every hour the receiver was retuned to either 4996 or 9996 kHz reference frequencies transmitted by station RWM of the Moscow Time and Frequency Service (coordinates 55°44′N, 38°12′E; path 4, ground distance 15900 km). Local standard time at RWM is UT plus 3 h.
 At RAO a vertical dipole was used as the receiving antenna which was connected to a receiver of the same type as at UAS. The receiver was tuned exclusively to the HF (EIS) pump frequencies, and the received signal was digitized at a sampling frequency of 682 Hz. At AARI the data were recorded using a multichannel HF Doppler receiving system designed by the Arctic and Antarctic Research Institute [Blagoveshchenskaya, 2001], with the equipment characteristics being very similar to those of the equipment at RAO and UAS.
 At the UAS and RAO stations the data were recorded continuously throughout the experiment, while at AARI site the records are available for 29 October only. The EIS (EISCAT) HF transmission was reliably detected at each site during approximately half of the observation time, although the detection periods did not always overlap between the three receiving sites (see Table 1).
Table 1. Dates in October 2002 When the Receivers Were Operating and the EIS HF Signals Were Distinctly Observed at the Receiver Sitesa
Pluses indicate a detection. N/A means that no data are available.
St. Petersburg (AARI)
3. Observational Results
 Analysis of the received EIS HF signals shows that the spectra of both transmitted frequencies in a number of cases contained two components of different characters. One of these was a relatively stable in frequency, narrowband “mirror-reflected” component typical of HF hop propagation in the midlatitude ionosphere. The other, or “scattered,” component showed a wider spectrum similar to that of a signal scattered by ionospheric inhomogeneities at frequencies higher than the maximum usable frequency [Bezrodny et al., 1997]. An example of the amplitude spectrum of the signal received at RAO on 30 October 2002 is shown in Figure 2. The spectrum was averaged over the time interval 1103:30 to 1104:00 UT. On that day two frequencies were simultaneously transmitted, specifically f1 = 7953.717 kHz and f2 = 7953.698 kHz for the time of Figure 2. The mirror-reflected components are distinctly seen in the signal spectrum at 0 and −19 Hz. Two scattered signal components centered near −12 and −31 Hz are also present. Unlike signals scattered by natural midlatitude ionospheric inhomogeneities [Bezrodny et al., 1997], the signatures of the scattered components showed significant frequency variations. The Doppler frequency shift sometimes exceeded 10 Hz.
 The spectral width of the scattered components received in Antarctica from 1000 to 1200 UT on 30 October 2002 reached 20 and more Hz (see Figure 3a), which made it impossible to individually distinguish them during that particular interval. Meanwhile the reference signal spectra from the RWM station at 9996 kHz received during the 5-min pause in the EIS transmission were practically undistorted (see Figure 3b). The station transmitted pulses with a 100-ms repetition period, and distinct spectral components spaced by 10 Hz can unambiguously be detected in the spectrum. Hence, the character of the spectral distortions of the EIS HF signal received in Antarctica was fundamentally different from that of the RWM signal, despite the fact that both paths were similar in length and in sunlit conditions.
 Well-correlated power variations were observed in a number of cases in signals received at the greatly differing distances of RAO and UAS from the transmitting facility, and a considerable amount of correlation between Doppler frequency shifts at the two sites was shown during 70 to 80% of the simultaneous observation time. The phenomenon is illustrated in Figure 4, which presents data from 29 October. At that time two frequencies, 4040.717 and 4040.695 kHz, were transmitted simultaneously. Figure 4 presents power variations (Figure 4a) and the cross-correlation function (Figure 4b) for signals received at RAO and UAS.
 It was noticed during the experiments that the scatter component power depends, possibly, on the transmission polarization, as shown in Figure 5, where the intervals of O and X mode transmission are indicated. The O mode signal was received more reliably and, when detected, the O mode signal strength was sometimes higher than that of the X mode signal.
 The scattered components of both transmitted frequencies quite often showed quasiperiodic frequency variations with periods from 10 s to a couple of minutes and amplitudes of up to 10 Hz. For instance, synchronous Doppler frequency variations with 20- to 40-s periods are clearly seen in the dynamic spectra of signals from EIS received simultaneously at the UAS, RAO, and AARI sites on 29 October 2002 between 0434 and 0437 UT (see Figure 6). It is worth noting that frequency variations of the kind were never seen in the RWM station signals recorded in Antarctica. This fact suggests that only a local ionospheric region above the heater is responsible for the variations, rather than the long-range propagation effects.
 In most cases the quasiperiodic Doppler frequency shift (DFS) variations recorded at the different stations were nearly synchronous and almost equal in amplitude, like those shown in Figure 6. However, a different situation was observed on 29 October after 0400 UT. The periods of the variations, time lags obtained from the shift of the maximum in the cross-correlation function, and the amplitudes of the variations for this day are listed in Table 2. Figure 7a presents an example of DFS variations recorded on 29 October between 0510:00 and 0513:30 UT. The respective cross-correlation function is shown in Figure 7b. As can be seen, its maximum is shifted by 5 s relative to the origin.
Table 2. Parameters of Quasiperiodic Doppler Frequency Shift Variations in the Signals Received at RAO and UAS on 29 October 2002a
Here T is the period of the variations, τ is the time lag, and δfR and δfV are the variation spans at RAO and UAS, respectively. A positive time lag means that variations at RAO advanced those at UAS. The transmission frequencies were 4040.717 and 4040.695 kHz.
 The greatest difference between the spectral characteristics of the signals received at different sites was observed between 1000 and 1200 UT on 30 October. Within this interval the Doppler frequency shift of the scattered signal component at RAO was nearly always negative, with the average magnitude equal to about 10 Hz. Shown in Figure 8 are time variations of the Doppler frequency shift of the signal received at RAO on 30 October between 1025 and 1032 UT. Simultaneously, a broadband signal with no frequency offset and spectral width of no less than 20 Hz was observed in Antarctica (see an example in Figure 3a).
 Now, let us summarize the basic features shown by the spectral characteristics of the EIS signals recorded during the campaign.
 1. The signal spectra typically consisted of “mirror-reflected” and “scattered” components.
 2. Unlike the mirror-reflected component, the scattered component demonstrated great variations in the Doppler frequency shift (up to 10 Hz), which in most cases showed a considerable amount of correlation between different receiving sites. Also in contrast to the mirror-reflected component, a great amount of correlation was observed between scattered component intensities recorded at the different sites.
 3. The scattered component power is apparently dependent on the transmission polarization, with the O mode being detected more reliably and sometimes at greater a strength as compared with the X mode.
 4. The scattered component markedly demonstrated quasiperiodic variations in the Doppler frequency shift with typical periods from 10 s to a couple of minutes and amplitudes up to 10 Hz.
 The behavior shown by the mirror-reflected component is consistent with that of an HF signal propagating by the multihop mechanism [Gurevich and Tsedilina, 1979; Zalizovskii et al., 2007]. Most likely, it can be associated with the side lobe radiation of the transmitting HF antenna, which bypasses the modified volume and propagates through significantly different regions of the ionosphere on its way from EIS to the receivers, thus showing minor deviations of the Doppler frequency shift.
 The behavior demonstrated by the scattered component is of greater interest and we focus mainly on analyzing its spectral characteristics in what follows. The similarity of the variations in the Doppler frequency shifts and intensities of the scattered component of the EIS signal received at the different sites suggests that they are causally related, which in turn implies a shared propagation path. The only possible section of the signal propagation trajectories common for all the radio paths is the way from the EIS transmitting facility to the modified ionospheric region above the heater, which is responsible for the scattering.
 The probable dependence of the scattered signal power on the polarization of the HF transmission can be explained by the existence of the region of artificial ionospheric turbulence and irregularities stimulated by the EIS pump wave which scattered a portion of the EIS signal. According to the single scattering theory [Booker and Gordon, 1950], the power of the scattered signal is proportional to the intensity of the inhomogeneities. As is well known [Gurevich and Shvartsburg, 1973], O mode polarized pump waves produce field-aligned plasma irregularities whereas X mode waves do not such that O mode waves are much more effective at generating ionospheric turbulence than X mode waves. Hence the scattered power should be greater in the first case. This mechanism can also be responsible for the high level of correlation between the signal intensity variations observed at the RAO and UAS receiving sites (see Figure 4; limited data prevent a similar conclusion for AARI, see Table 1). In view of the above consideration it is natural to suppose that the synchronous fluctuations shown by the signal powers at the receiving sites occurred due to variations in the intensity of the ionospheric inhomogeneities within the plasma-pump wave interaction volume in the F region.
 Further evidence of the influential effect of the AIT/AII region on the behavior of the scattered component of the EIS signals is the high spatial correlation between the quasiperiodic variations in the Doppler frequency shift with magnitudes up to 10 Hz. It should be noted that similar frequency variations were not observed in the reference signals from the RWM station received in Antarctica, and that the RWM spectrum received in Antarctica during the experiment was virtually undistorted (see Figure 3b). This allows the suggestion that the main contribution to Doppler frequency shift variations is provided by the ionospheric dynamics above the heater, while propagation from the scattering volume to the receivers affects signal parameters to a much lesser extent. Thus, it is natural to assume that the behavior of the scattered component is almost completely determined by the processes occurring within the region containing AIT and AII.
 Let us consider the details of the Doppler frequency variations in the EIS signal. There are two possible mechanisms of Doppler frequency shifts fD [Bennett, 1968]. One of these is physical motion of the signal source (or scattering inhomogeneities in our case), which yields [Rytov et al., 1978]
Here is the velocity vector of the scattering inhomogeneities and s = r − i is the scattering vector, with r and i being wave vectors of the reflected (scattered) and incident waves. Note that ∣r∣ = ∣i∣ = nk0, where n is the refractive index at the reflection point and k0 = 2πf/c is the free space wave number, with f being the signal frequency and c the speed of light.
 The other mechanism is associated with phase path variations of a ray between a source and a receiver due to fluctuations of the refractive index along the propagation path. As was shown by Bennett , in this case we have
where P is the length of the phase path and l is a ray coordinate along the propagation trajectory L.
 In view of the above considerations it can be expected that Doppler frequency variations at all the receiving sites should always appear coincident in time and equal in magnitude if the main contribution is provided by phase path fluctuations along the upgoing section from the heater to the scattering volume which is common for all the trajectories. However, if they are due to motion of inhomogeneities within the region of AIT, then, in the general case of an arbitrary motion direction of the scatterers, their magnitudes and phases can differ at the different receiving sites.
 To find out which of these two mechanisms might have happened in one or another specific case, let us analyze the Doppler frequency shifts of the EIS HF signals received at different sites. The pump wave vector i was directed along the magnetic field, i.e., almost vertically (see Figure 9). We estimated the elevation of the scattered wave vector r to be under 10°, i.e., nearly horizontal, in order to reach the receiving sites. Hence the scattering vector s has approximately 45° elevation angle in the vertical plane. It should be noted that the AARI and RAO sites are seen from the EIS position at the same azimuth. Therefore, any motions of the irregularities within the vertical plane containing the bisector of the angle between the directions from EIS toward UAS and RAO (or AARI) will contribute equally to the Doppler frequency shifts of the signals observed at these receiving sites. (For the given experimental layout the bisecting plane is nearly meridional, being however rotated by about 8° and 4° clockwise relative to the geographic and magnetic meridians, respectively.) In this case it seems impossible to determine from the measurements whether the effect occurred due to scatterer motions or was produced by plasma density fluctuations. It can only be asserted that the transport of inhomogeneities transverse to the meridional plane is absent. On the other hand, dissimilarities between the Doppler drift amplitudes and phases at UAS and RAO may be evidence that the scatterer velocity vector is oriented at an angle relative to the meridional plane; the more the angle between the velocity vector and the meridional plane the more the difference. Both situations of similar and different behavior of the Doppler frequency shifts at the receiving sites were observed in the experiments.
 Consider the case where Doppler frequency variations showed a different behavior at RAO and UAS. As noted above, some differences occurred on 29 October after 0400 UT, while the greatest dissimilarities were observed at UAS in the interval from 1000 to 1200 UT on 30 October. It should be noted that 30 October is also unique in that 1000 to 1200 UT is also the period when the spectral width reached 20 and more Hz, making it impossible to distinguish between the two transmitted components in the received signals. This was the only day on which observations were made during this time period. One distinction is the higher transmission frequencies used, about 7953 kHz as compared to approximately 4040 kHz and 4544 kHz between 0150 and 0800 UT on 29 October, for example, and between roughly 4040 and 4913 kHz on the other three observation days, for which transmissions also began about 0200 UT and ended at 0800 UT on each day. The much higher frequency used is due to the difference in local times, midday versus early morning, and the correspondingly higher ionospheric electron densities. It is not known whether the timing and the unique observations might be related.
 During the 1000 to 1200 UT period on 30 October, EIS was transmitting at frequencies f1 = 7953.717 kHz and f2 = 7953.698 kHz, and a negative (on average) Doppler frequency variation was observed at RAO, while the signals at UAS were seen to have an average DFS of zero. A long-term constant-sign frequency shift (it lasted for about 20 min from 1015 to 1035 UT, a fragment of the record is shown in Figure 8) cannot easily be explained by the signal phase length variation. The reason is that for an average fD = −10 Hz equation (2) shows that the rate of growth in the path length must be approximately 400 m s−1. Hence, during the 20-min interval the phase length will have changed by nearly 500 km which, when compared to the ground distance of 2400 km, appears to be too much. During this period incoherent scatter radar data show electron density over Tromsø to be roughly constant, resulting in no effect from electron density changes on the phase length.
 Thus in this case it must be the velocities of the scatterers that determined the frequency variations at RAO. The zero value of the Doppler frequency shift in Antarctica signifies the absence of a velocity component along the Tromsø-Vernadsky direction. Hence, the transport was toward the northwest direction, from the illuminated region of the ionosphere to the shadowed one. The magnitude of the velocity component of plasma inhomogeneity drift along the scatter vector can be estimated from equation (1), given that the refractive index at the point of scattering is known. As was shown by Gurevich and Shvartsburg , AIT/AII are generated chiefly at the altitudes where the condition for upper hybrid resonance of the pump wave is satisfied,
Here f is the pump frequency, fpe is the electron plasma frequency, and fce is the electron gyrofrequency. The refractive index at the upper hybrid resonance level can be estimated using the collisionless Appleton-Hartree-Lassen equation [Budden, 1985],
where X = fpe2/f2, Y = fce/f, YL = Y cos θ, and YT = Y sin θ, with θ being the angle between the wave vector and magnetic field vector.
 By combining equations (3) and (4) we can obtain that the refractive index at the upper hybrid resonance level for the ordinary-polarized pump wave at f = 7953.717 kHz and fce = 1.4 MHz is nO ≈ 0.42. If the scattering occurred on inhomogeneities moving horizontally at the upper hybrid resonance height or lower (nO varies from 1 to 0.42), then the motion velocity along the scattering vector estimated after equation (1) for fD = −10 Hz lies between 270 m/s and 650 m/s, which are large but not unreasonable for the high-latitude ionosphere. Velocities measured during this same time period with the EISCAT Dynasonde (an HF radar) [Wright and Pitteway, 1994], which have earlier been shown to compare well with F region velocities measured by the tristatic EISCAT 930-MHz incoherent scatter radar system [Sedgemore et al., 1998], show horizontal velocities under 100 m/s. However, the motion measured by the Dynasonde and 930-MHz radars, which is equivalent to a mirror-reflected bulk plasma motion, may be different than the component scattered by pump-induced ionospheric irregularities reported here and which is here referred to as the scattered component.
 As was mentioned above, in most cases Doppler frequency shift variations showed similar behavior at the different receiving sites. Of course, this might be because the direction of motion of the scatterers lies within the bisecting plane of the angle made by the EIS-RAO and EIS-UAS lines. However, an alternative cause could be electron density variations along the upgoing EIS to scattering volume path which is common for all the trajectories. Consider the possible contribution of this mechanism to the Doppler frequency shift in more detail. Let the pump wave be propagating upward and then scattered at the upper hybrid resonance height, zUH. Then the Doppler frequency shift arising along this path can be obtained by taking the integral in equation (2) over z from z = 0 (ground level) to z = zUH, namely,
Also assume that the ionospheric electron density below z = zUH is rather low such that fpe/f ≪ 1 holds for most of the region z < zUH. In this case the region near zUH where this approximation breaks down makes only a small contribution to the integral along the trajectory to the scattering level. Of course this is a strong assumption, but it has been tested through computing the electron density derivative numerically (equation (6) below) and the difference between the rough estimate and computed results is reasonable. Let Y ≪ 1 in equation (4), which is the nonmagnetized plasma approximation. Then, making use of the expression for the electron plasma frequency fpe2 = αN, where N is the electron plasma density and α = e2/4π2ɛ0me, with e the elementary charge, ɛ0 the admittance of free space, and me the electron mass, we obtain a Doppler frequency proportional to the time derivative of the integrated electron content IT(t) = N(z, t)dz below the scattering level,
Electron density data available from simultaneous EISCAT incoherent scatter radar observations were used to compare the temporal variations of the Doppler frequency shift recorded at RAO and the component of the Doppler frequency shift computed from EISCAT radar-derived electron content measurements in the altitude range from 90 to 260 km using equation (6). The results of the comparison are shown in Figure 10 for a 1.5-min-long interval on 29 October. The time interval corresponds to a stationary tilted position of the radar beam in order to avoid possible effects of the HF heating. The high correlation between the electron content variation rate and observed Doppler frequency shift demonstrates that the phase length deviations played a significant role in the nature of the Doppler frequency measured at the receiving sites.
 We have detected HF signals radiated in the northern Scandinavia and propagated over moderate and superlong range paths to receivers in Russia, Ukraine, and Antarctica. The signals propagated via two different modes. The first one is the regular multihop ionospheric propagation of HF waves radiated by the side lobes of the EISCAT HF antenna. The second mode is associated with scattering of radio waves transmitted in the main beam of the EISCAT HF antenna on artificial inhomogeneities stimulated by the same transmission, and further propagation through the ionosphere to the observation sites. These two modes, which we have called the mirror-reflected and scattered components, respectively, can usually be distinguished in the received signal spectra. The parameters of the scattered component are determined by the properties of the region of artificially induced plasma irregularities and are virtually independent of the ionospheric conditions along the propagation path.
 A distinctive feature of the scattered component is strong variations in the Doppler frequency shift, which in most cases were synchronous at the different receiving sites. Two mechanisms have been analyzed to explain these Doppler frequency variations: (1) velocity changes of the plasma inhomogeneities in the pump-plasma interaction region which are responsible for the scattering and (2) variations of the signal phase path length due to deviations in the plasma density along the propagation trajectory. Time intervals were identified within which one or another of these mechanisms played the principal role.
 To reliably identify the mechanism of the variations in Doppler frequency shift of the signals from the HF facility, data from at least three receiving sites greatly separated in azimuth are needed, with the sites being equipped with normal nondirectional antennas and standard receivers. The results obtained can be used for solving the inverse problem of recovering the parameters of artificially induced irregularities and natural dynamic processes in the pump-plasma interaction region.
 This work has been performed with the support of STCU projects P-330, P-072, and 827-C, INTAS grant 03-51-5583, and in collaboration with the Ukrainian Antarctic Center. The authors would like to thank the staffs of the Ukrainian Antarctic Station Akademik Vernadsky; the Radio Astronomical Observatory of the Institute of Radio Astronomy, National Academy of Sciences of Ukraine; the European Incoherent Scatter Scientific Association; and the personnel of the University of Tromsø. We are grateful to V. G. Sinitsin, V. G. Bezrodny, A. V. Koloskov, and A. S. Kascheev, and to engineers I. I. Pikulik and P. V. Silin, for their assistance in conducting the experiment in Antarctica and at RAO, and for useful comments and discussions.