1 Introduction

The experimental methods using controlled sources of low-frequency radio waves are of particular interest for studies of ionosphere and magnetosphere physics. One can generate the ULF/VLF waves directly in the ionospheric plasma by radio wave heating for creating such sources. The ionospheric generation of artificial VLF signals was first discovered using the medium power transmitter (\(P=150\) kW and array gain \(G=100\)) located near Gorky, Russia (Getmantsev et al. 1974). The first experiments on the generation of artificial ULF signals by powerful RF radiation were conducted in the 1980’s at more powerful facilities: SURA (\(P = 750\) kW, Nizhny Novgorod region, Russia) (Kotik et al. 1983; Belyaev et al. 1987) and EISCAT (\(P = 1\) MW, Tromso, Norway) (Stubbe et al. 1981). It was believed that the mechanism of ULF signals generation is the same as in the VLF range. Namely, it was assumed that a secondary source is formed in the lower ionosphere due to the modulation of the ionospheric conductivity in the region with quasi-stationary currents due to ohmic heating of electrons by RF waves (Kotik and Trakhtengerts 1975). These currents are the dynamo currents at mid-latitudes and the auroral electro jet at high latitudes. In these experiments the signal intensity was correlated with the level of geomagnetic activity. The first experiment on the generation of artificial signals at frequencies of 3–6 Hz in the nighttime ionosphere was carried out at the Arecibo facility (Puerto Rico) in 1985 (Ganguly 1986). Results of this experiment could not be explained by the conventional theories of ionospheric current modulation. From 2009–2012 a series of experiments were carried out at the new HAARP facility (Alaska, USA). The authors of those experiments on generation of artificial micropulsations in the absence of the electrojet proposed a new mechanism associated with low-frequency signal drift currents driven by the RF pumping wave in the Earth’s magnetic field. The important circumstance for their interpretation is the presence of a pressure gradient due to the ohmic heating of the upper ionosphere plasma (Papadopoulos et al. 2011; Eliasson et al. 2012).

The ULF signals at the modulation frequency were also found in the first 2010 experiment at SURA facility at night (Kotik and Ryabov 2011). The comprehensive studies of artificial ionospheric signal generation in the ULF/VLF ranges were conducted at SURA during the last 3 years. The main features of artificial ULF/VLF signals observed in the vicinity of the SURA facility were summarized in the paper Kotik et al. (2013). Here we will present some characteristics of the artificial ULF/VLF signals observed at mid-latitudes during the magnetic disturbances.

2 Equipment and Data Processing

The experiments have been carried out using the SURA facility consisting of three 250 kW transmitters. Each of them feeds a section of the \(12\times 4\) crossed dipoles (144 total). The array can radiate RF-waves of ordinary (O) or extraordinary (X) polarization in the range of 4.5–9 MHz. The angular size of the beam is \(10^\circ \) at the lower frequency range and \(5^\circ \) on top. It is possible to change the direction of the beam in the N–S plane with \(4^\circ \) step from the zenith. During the experiments, the total power was 500–600 kW in an unmodulated mode, and half of this value when square modulation form was in use. Operating frequencies were chosen based on the ionospheric conditions, which were observed by the ionosonde CADI.

Induction magnetometers were used for receiving artificial ULF signals in the range of 2–20 Hz. Data were digitized by the National Instrument ADC and stored on the hard drive of the computer. Two components of the magnetic field were measured in north–south and in the east–west directions. Receiving points were located at 2.6 km distance from the heater (Barkovka village).

Two crossed loop antennas (delta shape) with preamplifiers were used for receiving the VLF signals. The above mentioned ADC was also used for digitizing the data. Data processing was carried out by a standard FFT. An example of 20 Hz signal received in one of our SURA experiments and processed by the spectral analysis is shown in Fig. 1 (left panel). Such method gives us the average value of the signal amplitude over the all receiving time. The right panel of Fig. 1 illustrates the signal processing by the synchronous detection method which allows measuring the signal phase and indicating the amplitude variations in time.

Fig. 1
figure 1

Examples of data acquisitions, FFT method (left) and synchronous detector (right) at night time September 25, 2010

Full size image

The right and left-polarized components of the signal were introduced for studying the polarization of artificial signals. The north–south and east–west components of artificial ULF signals could be written as \(H_{NS} = A\) and \(H_{EW}=nAe^{i\varphi }\) where \(\varphi \) is the phase shift between the components. The right and left polarizations are defined as \(H_R=(H_{NS}+iH_{EW})/\sqrt{2}\) and \(H_L=(H_{NS}-iH_{EW})/\sqrt{2}\). A nonzero phase shift indicates an elliptical polarisation. In the case of \(\varphi =0\), the polarization of the radiation is linear. Experiments were carried out under different geomagnetic conditions. The magnetic disturbances were monitored by the IZMIRAN magnetic variation station \((\hbox {N}55.47^\circ , \hbox {E}37.29^\circ )\), whose latitude is almost the same as the SURA’s one (http://www.izmiran.ru/?LANG=en); for the planetary activity the Kioto site was used (http://wdc.kugi.kyoto-u.ac.jp).

3 Results of the Experiments

During the autumn 2011 campaign, the measurements in the VLF range were carried out at the receiving point ‘New Life’ (30 km south-west of SURA). The ULF measurements were carried out in the automatic mode at the additional receiving point 10 km west of SURA. A small magnetic storm occurred in October 05, 2011 with maximal index \(Ki = 4\) (according to IZMIRAN station). The VLF and ULF signal variations simultaneously were traced at that day and the results are shown in Fig. 2. Variations of VLF and ULF signal amplitudes during the storm differed drastically. The signals at frequencies of 2 and 2.6 kHz are maximized at the beginning of the storm, and then they disappear in the next few hours due to the reduction of electron density in the lower ionosphere. Note that the typical variation of these signals has a maximum at midday. At the same time, in the ULF range the signals are not correlated with the level of geomagnetic activity, showing a small maximum at midday and a small decreased at midnight. From this experiment it is clear that the mechanisms of generation in the VLF and ULF ranges are quite different.

The experiments on the ionospheric VLF generation at middle latitudes are fundamentally differ from the similar experiments in high latitudes. First of all the VLF signal amplitudes observed nearby SURA heater are up to two orders smaller then at HAARP or Tromce heaters. Hence we need more time to detect the VLF: 5–10 min due to low signal/noise ratio instead several seconds for high latitudes heaters. That is why we cant use a number of frequencies or perhaps a ramp techniques which have been successfully employed at mentioned above heaters. The second consequence of the mid-latitude experiments is the lack of strong changes in the lower ionosphere during small magnetic storms because the ionosphere was not affected by energetic particles like in auroral latitudes. That is why we do not expect changes in the lower ionosphere and especially in the height of the upper walls of Earthionosphere waveguide. Magnetic disturbance at mid-latitudes is primarily manifests in enhancement of ionospheric currents which are spreading from the substorm current system. As one can see from Fig. 6 in paper Belyaev et al. (1987) the shift is observed in the first VLF signal maximum from 2.2 kHz at the noon to 1.6 kHz in the evening at magnetically quiet time. But the signals at frequencies around 2 kHz gradually decrease from noon to evening at the magnetic quite time and never were detected at night even during magnetic storm. That is why the frequency 2 and 2.6 kHz was selected as a marker for observing the magnetic disturbance influence on ionospheric signal generated by heater.

So while it is possible that ionospheric variations other than the small storm are present in our observations, our conclusions are based on the typical characteristics of the lower ionosphere and documented diurnal variation of artificial ULF/VLF signals at the SURA facility.

Fig. 2
figure 2

Artificial VLF signals (middle panel, the frequency of 2 and 2.6 kHz) and VLF signals (lower panel, the frequency in Hz are shown in the histogram) during a small magnetic storms (K-index on the top panel), October 5, 2011

Full size image

The observations on the background of strong magnetic storms with \(Kp = 6{+}\) were performed at the first time on June 16–18, 2012 (see Fig. 3, upper panel). Figure 3 also shows the amplitudes of ULF signals at frequencies of 2–17 Hz and VLF signals at frequencies of 2 and 2.6 kHz (middle panels) for all days of observation. As one can see from this figure, there was no correlation between the signal amplitudes and the Kp-index at night as well as at day. There was only a slight decrease in the ULF/VLF signals during the main phase of the storm. Also it should be noted that the time variation of ULF signals quite differs from its pattern during a small magnetic storm, while the amplitude of VLF signals well correlates with the magnetic activity (see Fig. 2).

Let us now overview the ionospheric parameters obtained during the storm time (see Fig. 3, low panel). The main feature is the presence of an extremely intense sporadic E-layer at day and night, whose critical frequency \((foEs)\) achieved values up to 8.5 MHz. Sometimes the sporadic E-layer became so intense that the higher ionospheric layers were even screened. Also during the same period a sufficiently dense E layer was observed. The critical frequency \(foEs\) achieved values of 3.5–3.8 MHz at the noon instead regular 2–2.5 MHz at non disturbed days.

Fig. 3
figure 3

Amplitudes of artificial ULF/VLF signals (middle panels: white daytime values, black nighttime ones), Kp-index and ionospheric data, June, 2012

Full size image

It was detected the modulation of the artificial signal at frequencies of 11.1 and 17.24 Hz with a period of about 30 s during the same storm on the decay phase in the daytime June 18, 2012 (see Fig. 4, left). The modulation was so strong that it was clearly demonstrated at the synchronous detector output (see Fig. 4, right).

In 2013, at the peak of solar activity, the measurements of artificial ULF signals were performed during three magnetic storms: March 20–23 \((Kp = 5-)\), May 23–25 \((Kp = 5+)\) and August 14–17 \((Kp = 5+)\). The decreasing of the VLF signal amplitude almost in all cases was observed when the magnetic perturbation increased. A typical example is shown in Fig. 5. The modulation of artificial ULF signals was also observed in the August, 2013 experimental campaigns. However, at this time the signals were modulated at the frequency of 11 Hz with a period 14.3 s and at the 17.24 Hz frequency with a period 15.8 s. In this case the modulation was observed a day before the magnetic storm also at daytime.

Fig. 4
figure 4

The spectrum of modulated artificial signal induced by SURA on June 18, 2012 and the output of digital synchronous heterodyne filter for the same signal

Full size image
Fig. 5
figure 5

Decreasing of the ULF signal amplitudes during magnetic disturbance and Kp-index, August 13–17, 2013

Full size image

3.1 Polarization Properties of ULF Signals

Figure 6 displays the examples of spectra of right- and left-polarized signals at a frequency of 17.24 Hz for different dates. The night signal polarization is compared in the quiet period 2012-June-15 and during geomagnetic storms June 17, 2012. As it can be seen from Fig. 6, the right polarization became predominant compared to the same periods in quiet geomagnetic conditions. This is typical for all observed frequencies and may be associated with a change in ionospheric plasma parameters during a magnetic storm.

The artificial signals occur both at the fundamental frequency of modulation of the RF wave and at all harmonics falling within the reception band when transmitters were square shape modulated (for example, at a signal frequency \(f = 4.55\) Hz, five harmonics can be received). As a rule, the polarization of ULF signals is the same for the fundamental frequency and harmonics in quiet geomagnetic conditions. Elliptical polarization is typical for the emission at all frequencies with predominance of the left-polarization. Sometimes the linear polarization was observed.

Fig. 6
figure 6

Spectra of right- and left-polarized artificial ULF signals: left quiet geomagnetic field, right during magnetic storm

Full size image

4 Conclusions and Discussion

The features of artificial ULF/VLF signals characteristics under magnetic disturbed conditions can be formulated as follows:

  • No correlation of artificial ULF signals with variations of the Earth magnetic field was observed for weak geomagnetic disturbances \((Kp \le 3)\) while the VLF signals increased in growth phase of the geomagnetic perturbation.

  • At Kp-indexes greater than five there was observed a tendency of decreasing the amplitude of the ULF/VLF signals with increasing the magnetic disturbance.

  • Sometimes it was detected the modulation of artificial ULF signals with a period of 15–30 s before or after the magnetic storm.

  • During a storm time it was detected a change in the polarization of artificial ULF emission. The right polarization becomes predominant during the storm time.

According to the mechanisms proposed in Papadopoulos et al. (2011), Kotik and Ryabov (2011) the ponderomotive force caused by the pressure gradient in the region of the ohmic heating of the plasma by a powerful RF beam plays the major role in the formation of artificial ionospheric source of ULF signals. As a result, a diamagnetic ring current appears in the ionosphere which is given by Eq. (1).

$$\begin{aligned} \overrightarrow{j_\phi } = -cn_e\frac{[\nabla _{\perp }T_e \times \overrightarrow{B_0}]}{{B_0}^2}, \end{aligned}$$
(1)

here \(n_e\) and \(T_e\) are the electron density and temperature, \(B_0\) is Earth magnetic field induction, \(c\) is the speed of light.

Calculation of the electron temperature in the field of a powerful pump wave is based on the theory of the thermal nonlinearity (Gurevich 1978). Since the field of the pump wave has a strong inhomogeneity across the beam axis, the temperature gradient should be parallel to the gradient of the pump wave field. As it follows from Eq. (1) a ring diamagnetic current around the RF waves beam will be induced in the ionosphere. Figure 7 shows the calculated structure of the source formed by a narrow RF beam in the F-layer at nighttime (left panel) and in the E-layer at daytime conditions (right panel), when such a solenoidal structure concentrated in the E-layer the current density in it is ten times lower (Ryabov and Kotik 2012; Kotik et al. 2013).

Fig. 7
figure 7

Calculated current density of the ionospheric source formed by a narrow RF beam in the F-layer at nighttime (left panel) and in the E-layer at daytime conditions (right panel)

Full size image

The main observed peculiarity of artificial ULF/VLF signals is the decreasing of their amplitudes during the storm time. There are two possible reasons for the observed dependence: (a) increasing the absorption of HF and ULF/VLF waves in the lower ionosphere and (b) decreasing of the F-layer critical frequency and Earth magnetic field usually accompanied by a magnetic storm.

As one can see from Fig. 3, very intensive sporadic layers Es and E-layer have been observed. The critical frequency foE achieved the value of 3.8 MHz instead of regular values 2.5–3 MHz during magnetically quiet days. Obviously such conditions in the lower ionosphere should cause additional absorption of the RF pump wave and consequently decrease the heating of the F-layer as well as direct decay of the low frequency waves passing to the observation point.

The variations of foF2 could also be one of the main factors in decreasing the ULF/VLF signals because the ionospheric source current (1) directly depends on the electron density. This dependence was traced more convincingly on May 24–25 when during a storm time SURA had operated from the evening until 6:00 MST in the morning. The signal amplitude explicitly followed the F-layer critical frequency variation as it is shown on Fig. 8. By itself this dependence is a further evidence of source localization in the upper ionosphere at night time.

Fig. 8
figure 8

Measured signal amplitudes and F-layer critical frequency variation, May 24,25, 2013

Full size image

Sometimes it was detected the modulation of artificial ULF signals with a period of 15–30 s. Note that the periods of 15–30 s is the main period for the torsional and toroidal oscillation modes of the geomagnetic field lines at latitudes around the SURA facility (Westphal and Jacobs 1962; Nishida 1978). It is possible to transfer the oscillation of Earth’s magnetic field with Pc-3 periods on the current source (1) as it directly proportional to the magnetic field induction. Unfortunately we have not a fluxgate magnetometer at SURA site therefore this conclusion cannot be confirmed by direct measurements.