1 Introduction

Waves from ground based powerful VLF transmitters can penetrate into the ionosphere and even heat the ionosphere, which has been verified in the theoretical simulation and actual observations (Galejs 1972; Inan 1990; Rodriguez et al. 1994). After the launch of DEMETER satellite in June of 2004, more evidences have been revealed to demonstrate the interaction between the ground-based powerful VLF transmitter and the ionosphere. Parrot et al. (2007) first reported the ionospheric heating phenomena from a powerful VLF transmitter –NWC in Australia, with perturbations observed in broad band plasma waves, lower hybrid waves, electron densities, electron temperatures, ion temperatures, etc. Sauvaud et al. (2008) found an enhancement in 100–600 keV electron fluxes at L = 1.4 ~ 1.7 linked to NWC transmitter, and Graf et al. (2009) showed electron precipitation from the inner radiation belt at \({\text{L}}\sim 1.9\) induced by NPM transmitter in Hawaii. Bell et al. (2011) compared the significant effects in the topside ionosphere from two transmitters of NWC and NAA, and the magnetically conjugate region of NWC, in which the quasi-electrostatic VLF plasma waves and ELF turbulence below 300 Hz were studied. As for the interaction mechanism between VLF transmitter and ionospheric perturbations, Starks et al. (2009) simulated the Doppler-shifted ground-based VLF transmitter signals by linear mode coupling as VLF transmitters scatter from small-scale plasma density irregularities. Mishin et al. (2010) explained the broadband lower- and upper- hybrid electrostatic waves and plasma perturbations by VLF-driven nonlinear effects. Němec et al. (2020) studied the nightside effects of NWC transmitter on the upper ionosphere, with (1) significant perturbations of plasma density and electron temperature within a distance of 200 km from the transmitter; (2) considerably enhanced wave intensity above 14 kHz; and (3) clear induced electron precipitation east of transmitter at somewhat larger L-shells. Xia et al. (2020) studied the spectral broadening of NWC transmitter, and they observed that the width and intensity of broadening power are proportional to the NWC wave amplitude, and the broadening spectral intensity shows negative correlation with electron density which drives a lower threshold for spectral broadening.

In the present paper we summarize the plasma parameter effects measured by the DEMETER satellite induced by the powerful NWC VLF transmitter during the extreme solar minimum years, and compare with HF heating effects from SURA and HAARP. We then investigate the plasma structure at the topside ionosphere related to VLF transmitter, especially on the ion drift ducts, and finally discuss their interaction mechanism.

2 Background information

DEMETER satellite had operated for about 6.5 years with a sun-synchronous orbit at the altitude of 660–710 km since the end of June in 2004 to the early December in 2010 (Cussac et al. 2006), which covered the extreme solar minimum in the 23rd/24th solar cycles and will provide a good opportunity for studying the topside ionospheric heating effects from VLF transmitters during this time period. Onboard this satellite, topside electron density (Ne) and temperature (Te) were continuously detected by ISL (Instrument Sonde de Langmuir), and the ion parameters such as density (Ni), temperatures (Ti) and drift velocity were provided by IAP (Instrument d’Analyse du Plasma) instruments (Lebreton et al. 2006; Berthelier et al. 2006a). Instrument Champ Electrique (ICE) observed the three-component electric field with four electric probes at ULF, ELF, VLF and HF frequency bands from DC to 3.5 MHz (Berthelier et al. 2006b). The data was stored as half orbits, with up-going half-orbits from south to north corresponding to local nighttime 22:30 and down-going ones from north to south at daytime (10:30LT) respectively.

NWC transmitter is located at North West Cape of Australia (21.816° S, 114.166° E). It provides very low frequency radio transmission for communication and navigation. Its center frequency is 19.8 kHz, with bandwidth of 200 Hz. The transmission power is 1 MW, which makes it the most powerful VLF transmitter in the whole world.

3 Data analysis

3.1 Statistical analysis on topside ionospheric perturbations

According to the results of Parrot et al. (2007), the typical feature in VLF heating of NWC is the significant enhancement in electric field, with the large electrostatic noise between 10 and 20 kHz. The orbits in a distance of 2000 km over NWC were selected, and those with large perturbations in electric field spectrum were chosen out during the solar minimum from March 2008 to December 2009. Since the last half year of 2007, NWC was closed temporarily till February of 2008, so the first 2 months in 2008 were not included in the research. Finally, 78 up-orbits in local nighttime were summarized with typical heating feature during 2008 and 2009. Compared with the 49 events during 2005 and 2006 reported by Parrot et al. (2007), the heating events significantly increased about 59% ((78–49)/49 × 100). 2008 and 2009 are the extremely solar minimum years, and Ne at satellite altitude decreased more than 100% relative to 2005 (Zhang et al. 2014a). It demonstrates that during solar minimum, the plasma irregularities due to VLF heating are easier to be formed. Figure 1 showed the event number in each month of 2008 and 2009. It can be found that in Nov. to Feb. during the summer season at Southern Hemisphere, the events occurred within 0–3, but in other three seasons, VLF heating happened with an averaged level of 4.3 times/month. The seasonal difference in heating events also illustrates the instability of ionosphere with low electron content, and its strong effects on VLF radio wave propagation.

Fig. 1
figure 1

The monthly number of heating events induced by NWC transmitter during the solar minimum years of 2008–2009

Full size image

3.2 Spatial distribution of perturbations in VLF electromagnetic field

The level-1 data of electric field spectrum in VLF electric field onboard DEMETER was selected of all 78 up-orbits over NWC region, and plotted in Fig. 2. The power spectrum density (PSD) of electric field at 19.8 kHz and 19.0 kHz were abstracted respectively for comparison, because there is no transmitter with 19.0 kHz in globe. The results showed that, the heating effects mainly occurred at the north of NWC, and coincided with the normal peak intensity center of NWC at DEMETER altitude (Fig. 2a–c). From Fig. 2d–f, the enhancement in VLF electric field was much clear although the background of 19.0 kHz signals was distributed in a large range of 10–3–102 \({\mu V}^{2} {\text{m}}^{ - 2} {\text{Hz}}^{ - 1}\). The vertical lines in Fig. 2f revealed a blank region at latitude of 17–20° S, showing totally different perturbations in 19.0 kHz with other latitudinal region, with the intensity increasing at least one order even compared with the highest background intensity. Combined the results of 19.8 kHz and 19.0 kHz, the heating region was located at 111–116° E, 16–21° S, within a rectangle region of \(5^{^\circ } \times 5^{^\circ }\). Therefore, the heating region was quite near to the south foot of NWC. And due to the geomagnetic field controlling on VLF electromagnetic waves propagation along magnetic field lines, the disturbed region inclined to north direction of 2–3°.

Fig. 2
figure 2

Spatial distribution of power spectrum density in electric field (logE means log10E with unit of \(\upmu {\text{V}}^{2} /{\text{m}}^{2} /{\text{Hz}},\)) at 19.8 kHz (a–c) and 19.0 kHz around VLF heating region (d–f the solid lines outline the heating region in f, ▲ is the position of NWC in the panels of a and d)

Full size image

3.3 Perturbations in plasma parameters

The previous papers have detected the simultaneous disturbances in Ne, Te and Ti induced by VLF transmitter (Parrot et al. 2007; Bell et al. 2011). Here two examples were selected randomly and provided in Fig. 3 recorded on March 6 and 9 in 2008. It can be seen that the topside ionospheric disturbances occurred in all the plasma parameters of Ne, Te, Ni with three compositions (H+, He+, O+), and Ti, where it should be noticed that along 19699_1 orbit on 9 March 2008 Ni (H+ and He+) cannot be distinguished due to their small values less than 10 around UT14:39. The typical feature in plasma parameters was the general rise in Te and Ti, also in Ni(O+), but decrease in Ne, with the simultaneous perturbations in VLF electric field over NWC at 21.8° S latitude. The maximal variational amplitude was, on March 6 2008, Ne ~ − 39% while Te ~ + 1327 K, Ni(O+) ~ + 284% while Ti ~ + 257 K; and on March 9 2008, Ne ~ − 36% while Te ~ + 791 K, Ni(O+) ~ + 44% while Ti ~ 275 K. The total 78 up-orbits were searched again to find the perturbations in plasma parameters. Final results showed that electrons were more sensitive to VLF heating effects, only 4 orbits without disturbances in Ne and Te in total 78. Compared Ne and Te, Te perturbations occurred more frequently, with quite weak or no Ne variations at the same time. As for the ion parameters, Ti perturbations occurred in 35 orbits, with a half percent in total orbits. While Ni disturbed 50 times, with higher occurrence probability than Ti.

Fig. 3
figure 3figure 3

VLF heating perturbations recorded over NWC transmitter on March 6 (orbit: 19655_1) and 9 (orbit: 19699_1) in 2008. From the top to the bottom the plots represent, the VLF spectrum of one electric component up to 20 kHz with the white line of fLHR overlapped on it, the electron density, the electron temperature, the ion composition and ion temperature

Full size image

In order to understand the relationship between plasma parameters in the time period of VLF heating, the ISL and IAP data along the orbit of 19699_1 on March 9 2008 was processed, in which the background is quiet, and no apparent increasing and decreasing trend variations with latitude increase shown in all plasma parameters. Due to difference in the sampling rate between IAP and ISL, the correlations of Ni(O+) ~ Ti and Ne ~ Te were calculated respectively as presented in Fig. 4. It showed that Ni(O+) ~ Ti presented high positive linear correlation with R (correlation coefficient) ~ 0.74 (Fig. 4a), while Ne ~ Te gave weak negative relation with R ~ 0.19 (Fig. 4b). The sampling rate is 1 s in ISL and 4.429 s in IAP under survey mode, so more detailed changes can be detected in ISL, which may cause the reduce of linear correlation between Ne ~ Te.

Fig. 4
figure 4

The correlation between plasma density and temperature on March 9 2008 (a Ni(O+) ~ Ti; b Ne ~ Te)

Full size image

Kakinami et al. (2011) investigated the correlation of Te and Ne winthin |MLat| ≤ 40° measured by the Hinotori satellite, and illustrated the well-known negative correlation between daytime Te and Ne when Ne is low, while when Ne is significantly high (> 106 cm−3), the correlation turns positive, especially during LT11-15 and around |MLat| < 10°, in which Te in the low-latitude topside ionosphere is suggested to be controlled by photoelectron heating and cooling through Coulomb collisions with ions that are related to the integrated Ne. On the basis of DEMETER observations, Zhang et al. (2014b) studied the correlation of Te and Ne around daytime of LT10:30 and nighttime of LT22:30, and the significant negative correlation of Te and Ne is revealed in local daytime at equator and low latitudes, while in local nighttime Te and Ne presented relatively weak positive correlation at low latitudes and almost no correlation occurred at |MLat| > 25°. In this paper, all heating events occurred in local nighttime, which means if there is no heating source in this region, Te and Ne have weak positive correlation or no correlation in normal condition, so the heating process from NWC is definitely the additional source to contribute to the negative correlation of Te and Ne as shown in Fig. 4. The increase of Te and Ti over NWC during heating time may accelerate the ion movement, and the recombination rate of electrons and ions, and finally to induce the decrease of Ne at the satellite altitude.

3.4 Comparison of VLF heating-induced perturbations in different solar years

Combined with the research results in Parrot et al. (2007) and Bell et al. (2011), typical VLF-induced ionospheric heating events had occurred in September over NWC during 2005 and 2006. Ne over NWC is with typical annual period with peak values occurring in summer season (Zhang et al. 2014a), and September is at its middle level of each year, but heating events are frequently detected in this month as those in winter season (Fig. 1), which illustrate that the heating process can also take place at higher plasma environment like in equinox or in 2005 and 2006 with a higher solar activity years relative to 2008–2009. Figure 5 presented the significant events in near day in September during 2005 to 2010, in which 2007 was lost because NWC was turned off at its last half year. So five events were collected respectively on 21 Sep. 2005, 22 Sep. 2006, 22 Sep. 2008, 22 Sep. 2009, and 18 Sep. 2010, and the main variations in plasma parameters and electric field were plotted accordingly with Ne, Te, Ni(O+), Ti, and EE (Fig. 5a–e). The critical frequency in F2 layer foF2 (Fig. 5f) was also showed in those days from the ionograms published on the website of Digital Ionogram DataBase (http://ulcar.uml.edu/DIDBase/) at Learmonth station (21.8° S, 114.10° E). Last two panels presented the daily ∑Kp and F10.7 (Fig. 5g, h), which represented the geomagneticand solar activity condition. F10.7 decreased since 2005 to the minimum in 2008, and recovered in 2009 and 2010. The low ∑Kp values meant that there was no geomagnetic storms taking place in those heated days (Fig. 5g), so all the variations were directly related to the NWC heating process. It can be seen that the biggest PSD values in electric field at 19.8 kHz occurred in 2006 and 2010 (Fig. 5e) with quite low foF2 in 2006 (no foF2 data available in 2010). The largest δNe was detected in 2009 (Fig. 5a), while the smallest one in 2005. δNi(O+) changed intensely due to the small values of normal background in local nighttime (Fig. 5c), but it still demonstrated much stronger perturbations in 2006, 2008 and 2009 than those in 2005 and 2010, and also ∆Ti gave similar feature (Fig. 5d). So the variations in plasma density all illustrated that the larger variations might be related to the lower ionospheric environment in solar minimum or even in high solar activity years but with low electron contents. ∆Te did not exhibit clear characteristics related to the year (Fig. 5b).

Fig. 5
figure 5

The parameter variations during typical events on 21 Sep. 2005, 22 Sep. 2006, 22 Sep. 2008, 22 Sep. 2009, and 18 Sep. 2010 respectively (a δNe; b ∆Te; c δNi(O+); d ∆Ti; e PSD of E; f foF2 at; g daily ∑Kp; h F10.7 index)

Full size image

4 Discussion

The quasi-electrostatic (QE) waves above the lower hybrid frequency have been detected during all the heating periods. The comparison in electric field PSD as shown in Fig. 2 illustrate that, these QE waves are lower of 2 orders in PSD amplitude than that at 19.8 kHz. From the model of Mishin et al. (2010), these QE waves may be generated through nonlinear parametric interaction driven by the input VLF transmitter signals, and the parametric instability could eventually produce QE waves between lower hybrid frequency and the transmitter frequency. This model indicates the abundance of O+ at the satellite altitude (Mishin et al. 2010). By the observations shown in the panels for Ni in Fig. 3, O+ increased during the ionospheric heating time although O+ is not always the dominant species in local nighttime at DEMETER altitude, while H+ decreased even it was the dominant one. This increase in O+ coincides with the expectation of Mishin et al. (2010). To exhibit the movement of ions, Figs. 6 and 7 respectively give the ion velocity and ULF electric field observations along two revisited orbits on March 6 and 22 in 2008, in which O+ all increased within the heating period. About the satellite coordinate, Parrot et al. (2006) has defined it as + Xs Nadir (Earth direction), + Ys along the normal to the orbit, and − Zs along the velocity vector (Fig. 6c). As shown in Fig. 6, the ion velocities along the downward direction (Vx) and Vz opposite to satellite-flying direction presented strong variations around UT14:45. On March 6 (Fig. 6a), the background ions moved upward, and during heating time their upward movement slowed down from − 35 to − 5 m/s, while they turned their movement from north to south in Vz from − 200 to 60 m/s. On March 22 (Fig. 6b), Vz still showed north to south direction turning during the heating time from − 100 to 200 m/s, but Vx accelerated to upward movement from − 5 to − 35 m/s. These two orbits are revisited orbits, and similar ionospheric heating due to NWC took place at the same local time and the same region within the same month of 2008. O+ all increased at 660 km altitude, and the ions exhibited same southward and upward movement during heating time. Their vertical movement is complicated with one accelerated and one slowed down respectively, being not simply uplifted. From Fig. 7, the ULF electric field in Ex and Ez all decreased during the heating time, with the similar amplitude of − 2 mV/m in Ex and − 4 mV/m in Ez. The movement of ions is contributed to the motional electric field variations by the effects of \(\mathop{V}\limits^{\rightharpoonup} \times \mathop{B}\limits^{\rightharpoonup}\) in Lorentz force equation (F = qE + qv × B), in which \(\mathop{B}\limits^{\rightharpoonup}\) at three directions are calculated at NWC by IGRF model with northward X = 30,206.9 nT, eastward Y = 329.7 nT, downward Z = − 43,482.0 nT, and inclination I = − 55.211°. These ULF disturbances indicate the existence of plasma irregularities at this region. It also can be seen that ULF disturbances modulated many times during the heating period, which illustrates the thermal plasma instability (Lee and Kuo 1984) occurring in both cases.

Fig. 6
figure 6

Ion velocity along two revisited orbits of 19655_1 (a) and 19890_1 (b) on March 6 and 22 in 2008 and the satellite coordinate of DEMETER

Full size image
Fig. 7
figure 7

The three components of electric field on March 6 (a) and 22 (b) in 2008

Full size image

About the formation of QE waves above lower hybrid frequency, Galinsky et al. (2011) further improved the nonlinear interaction into a novel 3D code with vector non-linearity included to illustrate the intense broadband lower hybrid electrostatic waves generated by whistler-mode waves from NWC and to explain missing 20 dB absorption of powerful VLF waves in the topside ionosphere. According to their results, with the input signals of 2.2 mV/m at 20 kHz and lower hybrid frequency of 10.7 kHz, the electric field at 19.7, 18.8, 17.7 kHz all showed equal strength with the driven VLF waves, while the strengths at 15.3 kHz and 10.4 kHz decreased quickly to a few orders lower. This result has some differences with our observations. As shown in Fig. 2, during the heating time, the PSD in electric field at 19.8 kHz increased to 105.4 μV2 m−2 Hz−1, while at 19.0 kHz being about 103.2 μV2 m−2 Hz−1, exhibiting the obvious differences above 2 orders of magnitude. Figure 8 gives the electric field PSD at 10–20 kHz and 18–20 kHz along 19655_1 on 6 March 2008, and it presents that broaden frequency band at heating time reached ± 300 Hz to 19.5 kHz. Below 19.5 kHz to the fLHR over 10 kHz, the PSD in electric field maintained the similar amplitude, but reduced more than two orders in magnitude relative to that at 19.8 kHz. The observations illustrate that the QE waves do not maintain the same total wave energy with the input NWC signals as those simulated by Galinsky et al. (2011). In the research of Bell et al. (1983), the spectral broadening may be resulted from a Doppler shift effect where the input signals scatter from plasma irregularities in the F region and couple into quasi-electrostatic modes of short wave length. By using the full-wave method, Lehtinen and Inan (2009) simulated the electromagnetic field in the stratified ionosphere from NWC, and their results presented overestimated maximum field amplitude by ~ 8 dB for Ew and ~ 14 dB for Bw, but for NWC case the observed Ew exceeded the calculated values at the regions far from the maximum with a relatively small-scale structures and a significant Doppler broadening there. The passive linear conversion mode from VLF transmitter to QE waves was studied by Bell and Ngo (1990), and their model predicts that the electromagnetic whistler mode waves can experience significant effective damping as the lower hybrid waves are excited. From Fig. 8, the input whistler waves have been enhanced with amplitude increasing of 1–2 orders with the occurrence of plasma irregularity during the heating process, while the broadened frequency band around 19.8 kHz expands from ± 100 Hz to ± 300 Hz with amplitude decreasing to about 1 order to the peak intensity at 19.8 kHz, and the PSD in electric field between 19.5 kHz to the LHR (Low Hybrid Resonance, which is plotted on the panels of electric field in Figs. 3 and 8) frequency further reduces at least 2 orders to the input transmitter signals but being higher 1–2 orders to their normal background (Figs. 2e, f, and 8). The boundary of broadened frequency band with QE band is quite clear with large decreasing amplitude in electric field. Whatever the linear or nonlinear conversion mode from whistler wave to QE, the intensive energy loss on the input signals considered in the model did not observed at the DEMETER altitude, and the enhancement in QE frequency band stronger than the input signals predicted by them were also not consistent with the actual observations. So this QE generation induced by VLF transmitter whistler waves is still need verified.

Fig. 8
figure 8

The electric field PSD along 19655_1 at 10–20 kHz (left) and 18–20 kHz (right)

Full size image

In HF heating research (Zhang et al. 2016), three HF-induced heating events were summarized from SURA heating facility in Russia. Figure 9 gives the spectrum in VLF and HF electric field on May 12 2008 and April 18 2010. It can be seen that with the occurrence of HF heating, the enhanced QE waves above 10 kHz were also detected. In addition, on April 18 2010, a few VLF transmitter signals were enhanced over the heating region, especially at 18.1 kHz where the frequency band was broadened obviously. Compared the QE waves above the lower hybrid frequency induced by HF heating, they did not show clear lower hybrid boundary over HF heating region as those over VLF transmitter, but it was similar that the VLF transmitter signals were enhanced with frequency broadened. Vartanyan et al. (2016) studied the generation of whistler wave by HF heating, and they thought that the VLF waves at LH (Low Hybrid) frequency (7–10 kHz) were due to the interaction of LH waves with meter-scale field-aligned striations and the VLF waves at twice the LH frequency (15–19 kHz) were due to the interaction of two counter-propagating LH waves. But for the VLF heating, QE waves are not much enhanced at the LH frequency, therefore they may not be generated from LH waves. From Fig. 3 along the orbit of 19655_1, many vertical lines covering the whole frequency band were detected in VLF electric field spectrum, which may illustrates intensive activity of lightning over this region. The existence of plasma cavity during VLF heating may provide the artificial duct for these lightning-whistler waves propagation, and also the transmitted signals from NWC were enhanced with broadened frequency band when they pass through this plasma cavity, which is consistent with the occurrence of MF (Medium Frequency) signals over VLF transmitters (Parrot et al. 2009).

Fig. 9
figure 9

The electric field spectrum at VLF and HF bands along 20641_1 (left) on May 12 2008 and 31014_1 (right) on April 18 2010 during SURA HF heating time period (red lines in HF spectrum) recorded from DEMETER

Full size image

Compared with the artificial plasma ducts caused by HF heating (Vartanyan et al. 2012), Te and Ni(O+) all increased over the VLF and HF heating region, but the perturbations in Ne was totally reversed with increasing in HF heating while decreasing in VLF-induced. In our simulation results about HF-induced ionospheric heating through ohmic process (Zhang et al. 2016), the electron density will increase firstly around 110–180 km altitude, and then a depletion is formed at the reflection altitude of the bumping HF waves at 215–260 km, finally electron density increases again around 530–710 km. According the previous studies on VLF-induced ionospheric heating (Parrot et al. 2007; Bell et al. 2011), the plasma depletion most likely takes place between the E-region and 700 km altitude due to the high decreasing amplitude of Ne about 50% at the DEMETER altitude as shown in Fig. 3. From Fig. 1, the VLF heating from NWC mostly occurred from March to October with less plasma density during these seasons. How to produce the ionospheric heating from the VLF transmitter is still in question, such as ohmic process as HF heating, wave-particle interaction, or other mechanisms? And also the actual altitude scale of VLF-induced plasma cavity (ducts as shown in Ne in Fig. 3 and enhanced electromagnetic waves in Fig. 2) still needs further observations and model verification in future.

5 Conclusions

Aiming at the VLF-induced ionospheric heating from NWC at extremely low solar activity years of 2008–2009 (Fig. 1), the VLF heating events are summarized and studied in spatial and temporal distribution features. Besides the obvious perturbations in Ne and Te during the heating time, the ion density and temperature are studied, including the ion movement. Finally the QE waves above lower hybrid frequency are compared with those induced by HF heating, and the QE generation mechanism is discussed. The main conclusions can be summarized as follows.

  1. (1)

    During VLF-induced ionospheric heating, Te and Ne show opposite variations at the altitude of 670 km in the topside ionosphere, and Ne always decreases above 50%. As for ions, the apparent increasing variations always occur in Ni(O+) and Ti simultaneously with good positive correlation (Fig. 4).

  2. (2)

    The comparison of PSD in electric field at transmitted frequency of 19.8 kHz and neighboring frequency at 19.0 kHz presents that the heating regions all move to the north of NWC to 2–3°toward the equator (Fig. 2a, d), and the longitudinal and latitudinal scales are about 200–300 km radius (Fig. 2).

  3. (3)

    The relative variation amplitudes in Ne, Ni(O+) and Ti all show larger values in low solar activity years (Figs. 3 and 5), which illustrate that the low plasma density will play an important role in VLF-induced ionospheric heating.

  4. (4)

    Two revisited orbits with VLF-induced ionospheric heating in March of 2008 are selected to investigate the features in ion movement as shown in Fig. 6, in which the ions along both orbits all present upward movement, and the simultaneous perturbations are detected in ULF electric field (Fig. 7) due to the effects of \(\mathop{V}\limits^{\rightharpoonup} \times \mathop{B}\limits^{\rightharpoonup}\). But along both orbits, one gives accelerated upward movement in ions, one decelerated, reflecting the complex state in ions during VLF heating time.

  5. (5)

    The PSD in VLF electric field is compared at different frequency band. During heating time, the transmitted signals at 19.8 kHz are enhanced to increase one order of amplitude, and the broadened frequency extends to 300 Hz (Fig. 8). While below 19.5 kHz to the lower hybrid frequency, the electric field maintains similar strength, with at least 2 orders of amplitude decreasing relative to those at 19.8 kHz. It is suggested to be generated by lightning whistlers over this region, not from the VLF whistler conversion.

  6. (6)

    Compared the QE waves due to HF heating, the enhanced and broadened VLF transmitter signals are also detected in HF heating (Fig. 9), while QE waves do not have the clear boundary feature in lower hybrid frequency here. The mechanism of VLF heating and the spatial features of plasma cavity formed due to VLF heating are still in question.