Results of Ionospheric Heating Experiments Involving an Enhancement in Electron Density in the High Latitude Ionosphere

Results of Ionospheric Heating Experiments Involving an Enhancement in Electron Density in the High Latitude Ionosphere WU Jun ( ) 1,2, WU Jian ( ) 1,2, XU Zhengwen ( ) 1,2 1 Key Lab for Electromagnetic Environment, Beijing 102206, China 2 China Research Institute of Radio Wave Propagation, Beijing 102206, China Abstract Observations are presented of the phenomenon of the enhancement in electron density and temperature that is caused by a powerful pump wave at a frequency near the fifth gyrofrequency. The observations show that the apparent enhancement in electron density extending over a wide altitude range and the enhancement in electron temperature around the reflection altitude occur as a function of pump frequency. Additionally, the plasma line spectra show unusual behavior as a function of pump frequency. In conclusion, the upper hybrid wave resonance excited by the pump wave plays a dominating role and leads to the enhancement in electron temperature at the upper hybrid altitude. The phenomenon of apparent enhancement in electron density does not correspond to the true enhancement in electron density, this may be due to some mechanism that preferentially involves the plasma transport process and leads to the strong backscatter of radar wave along the magnetic line, which remains to be determined. Keywords: ionospheric heating, electron density enhancement, electron temperature enhancement PACS: 52.50.Qt, 52.40.Db, 94.20.Tt DOI: 10.1088/1009-0630/18/9/03 (Some figures may appear in colour only in the online journal) 1 Introduction The impact of powerful high frequency (HF) radio waves on the ionosphere leads to the development of a wide range of phenomena, the most outstanding of which are electron density perturbations, in the form of either enhancement or reduction, at an altitude close to the reflection level of the pump wave. Birkmayer [1] observed the small-scale plasma depletions ranged from 3% to 55% induced by ionospheric heating, which involve Langmuir waves trapped in density cavities. Using the Arecibo HF facility and incoherent scatter radar (ISR), Duncan et al. [2] reported the observations of density depletions exceeding 50% extending hundreds of kilometers along the geomagnetic field. In June 1992, a National Aeronautics and Space Administration (NASA) sponsored sounding rocket flew through the Arecibo heater beam both above and below the reflection level of the pump to study the structure of the heated volume. Over 180 deep filamentary density depletions were detected and they had a mean depletion depth of 6% [3]. Stocker et al. [4] found both density enhancements and depletions on the order of 10% in the vicinity of the pump wave reflection altitude using European Incoherent Scatter Scientific Association (EISCAT) heater and ultra high frequency (UHF) radar located near Tromso, Norway. These effects were thought to be thermally driven. Using the facility of The High Frequency Active Auroral Research Program (HAARP), Pedersen et al. [5,6] observed large enhancements in electron density that created the artificial ionospheric layers, which were attributed to heat-driven ionization. In addition to the above observations of the effects of polarization in the O-mode, Blagoveshchenskaya et al. [7,8] reported EISCAT UHF radar observations of an increase up to 30% in electron density induced by an X-mode pump wave with frequency f H f x F2, where f x F2 is the critical frequency of the extraordinary wave. Senior et al. [9] also presented EISCAT UHF observations of apparent large electron density enhancements with wide altitude extent during X-mode pump waves, which show that these apparent density enhancements appear to exhibit aspect-sensitive backscattering and are not associated with corresponding changes in the frequency of the incoherent scatter plasma line. Senior et al. [9] hypothesized that the large electron density enhancements with wide altitude extent do not in fact correspond to true enhancements in electron density and may be due to an unknown physics that does supported by National Natural Science Foundation of China (No. 40831062) 890

WU Jun et al.: Results of Ionospheric Heating Experiments Involving an Enhancement in Electron Density not involve a change in plasma density but rather the enhancement in the incoherent scatter ion line spectrum and the backscatters of the radar wave along the magnetic field line. The above results were obtained by using a pump in X-mode polarization and at a single fixed frequency. This paper will present further experimental results of the apparent large electron density enhancements with wide altitude extent, which was deduced from EISCAT UHF observations during ionospheric heating experiments using pump operated in O-mode polarization and at frequencies sweeping near the fifth gyrofrequency, and especially shows the pump frequency dependency. with a slope of 1 12 nt/10 s from 12:30UT to 13:30UT, and 1 30 nt/10 s from 13:30UT to 14:19UT, while decreases linearly and monotonically with a slope of 10 33 nt/10 s from 14:19UT to 14:30UT. The total magnetic intensity at altitude of 200 km shows same configuration as that on ground, and varies in the interval of [49210 nt, 49240 nt], thus the corresponding fifth electron gyrofrequency should be in the interval of [6.8922 MHz, 6.8964 MHz]) and are located in the interval of pump frequency [6.7 MHz, 7 MHz] exactly. 2 Experiment and observations The EISCAT heater [10] is located at Ramfjordmoen near Tromsø, Norway (69.58 N, 19.21 E, magnetic dip angle I=78 ). The 12 transmitters can generate up to 1.2 MW of continuous wave power in a frequency range from 3.85 MHz to 8 MHz. There are three antenna arrays that cover the frequency ranges of 3.85 5.65 MHz and 5.5 8 MHz, with a gain of 24 db (dependent on frequency), and can produce a beam width of 14.5 and a maximum effective radiated power (ERP) of 360 MW. The EISCAT UHF radar [11] located approximately 500 m from the EISCAT heater is an ISR operating at 930 MHz. The antenna is a 32 m parabolic dish with a beam width of 0.5 at half-maximum power. It is fully steerable in azimuth and elevation. Using an EISCAT heater and UHF radar, the experiment was conducted at 12:30UT 14:30UT (universal time) on 11 March 2014. The EISCAT heater was operated at a frequency sweeping in steps of 2.804 khz from 6.7 MHz to 7 MHz near 5Ω ce, where Ω ce is the local electron cyclotron frequency at the altitude of 200 km with a value of 1.36 MHz in Tromsø, Norway, and in O-mode with a modulation cycle of 18 min, followed by 12 min off. The period of each step frequency sweep was 10 s, that is, 108 frequencies over a period of 18 min of heating. The heating beam was field-aligned (12.5 zenith, 186.2 azimuth), and ERP was estimated to be 100 MW. The UHF radar started observation at 12:32:30UT and its beam remained field-aligned. It should be emphasized that the local geomagnetic condition was relatively inactive during the experiment. Fig. 1 shows the total magnetic intensity at 12:30UT 14:30UT on 11 March 2014 in Tromsø, where the data on ground were recorded by Tromsø Geophysical Observatory, University of Tromsø, and the data at altitude of 200 km were derived from the International Geomagnetic Reference Field model (IGRF) by extrapolating. The total magnetic intensity on ground varies in the interval of [53452.5 nt, 53485 nt] and increases linearly and monotonically Fig.1 The total magnetic intensity in Tromsø during experiment, where the solid curve and dashed curve represent the values on ground and at the altitude of 200 km, respectively Fig. 2 is an altitude profile of the ratios of electron density N e (top panel) and electron temperature T e (middle panel) to the undisturbed values of N e14:30 and T e14:30, respectively, as a function of heating cycle, where N e14:30 and T e14:30 were taken from the final 30 s of the last cycle before completion of the UHF radar observations. In the bottom panel of Fig. 2, the schematic of the heating cycles and the pump frequency are shown, with the pump frequency decreasing from 7 MHz to 6.7 MHz in the intervals of 12:30UT 12:48UT and 13:30UT 13:48UT and increasing from 6.7 MHz to 7 MHz in the intervals of 13:00UT 13:18UT and 14:00UT 14:18UT, and EISCAT heater off in the intervals of 12:48UT 13:00UT, 13:18UT 13:30UT, 13:48UT 14:00UT and 14:18UT 14:30UT, respectively. Fig.2 The ratios of N e and T e to N e14:30 and T e14:30, respectively, versus heating cycles, where the black dashed lines corresponds to heating on and off When the heating is on, there is a strong enhancement in (T e /T e14:30 ) extending in the vicinity of the reflection altitude that varies with pump frequency (f HF ) in particular, and which especially almost disappears when the heating is off. When f HF 891

approaches 6.7 MHz, it sweeps in the low band (LB, 6.7 6.812 MHz) of the pump frequency, (T e /T e14:30 ) enhances strongly up to the order of 1.6, whereas when f HF is in the high band (HB, 6.851 7 MHz) of the pump frequency, there is slightly less enhancement in (T e /T e14:30 ), approximately up to the order of 1.4. When f HF is in the gyrofrequency band (GB, 6.815 6.849 MHz) that is, very close to 5Ω ce (T e /T e14:30 ) is approximately on the order of 1.2 and is less than in both the LB and HB. Consequently, it is apparent that (T e /T e14:30 ) LB > (T e /T e14:30 ) HB > (T e /T e14:30 ) GB, (1) where (T e /T e14:30 ) LB, (T e /T e14:30 ) HB and (T e /T e14:30 ) GB indicate the increase in electron temperature when the pump wave operates in LB, HB and GB, respectively. Moreover, Fig. 2 also shows the synchronization of the enhancement in electron temperature to the heating cycles, which is consistent with an electron temperature response time on the order of µs. Although the enhancement in electron temperature by a pump wave at a single fixed frequency is commonly observed with O-mode heating, the magnitude of enhancement presented here is a function of the pump frequency. This means that multiple physical processes are invoked and play a dominating role at different band of the pump frequency sweeping; that is, in LB, GB and HB, respectively. As is well known, an O- mode pump wave can couple through striations into an upper hybrid wave at an upper hybrid resonance altitude [10 14], where pump frequency yields the other hand, when the pump frequency sweeps up, such as in the heating cycles at 13:00UT 13:18UT and 14:00UT 14:18UT, the enhancement is absent in LB and appears in HB, and it does not disappear immediately after the heating is switched off, but decays to an undisturbed level within approximately 60 s. The top panel of Fig. 2 shows an enhancement in N e /N e14:30 extending from 100 km to 225 km, especially at 12:30UT 13:00UT, which does not depend on heating cycles, but due to the larger ratios of N e to the undisturbed values N e14:30. In addition, when sweeping in GB, an enhancement of 1.3 in N e /N e14:30 occurs around reflection altitude. The top panel of Fig. 3 is the altitude profile of the plasma critical frequency corresponding to the altitude profile of the electron density derived from the ion line analysis, which depends on f pe 9 N e, (3) where f pe (Hz) denotes the plasma frequency and N e (m 3 ) the electron density. When pump frequency is sweeping in HB, the increase in the plasma frequency is up to 11 MHz around an altitude of 300 km and extends from approximately the pump wave reflection altitude to the limit of the radar measurement. There is no evidence of a change in plasma frequency when the pump frequency sweeps in LB. f 2 HF = f 2 UH = f 2 pe + Ω 2 ce, (2) where f UH denotes the upper hybrid frequency. Therefore, the upper hybrid wave propagates in a direction perpendicular to the magnetic field and dissipates energy through Ohmic heating and wave trapping. This leads to an enhancement in electron temperature at the upper hybrid resonance altitude. In addition, the evidence that the magnitude of the enhancement in electron temperature reduces when pump frequency is near 5Ω ce is consistent with the idea that the trapping of the upper hybrid wave excited by heating is poor when the pump frequency is slightly below the harmonic electron gyrofrequency [5], and the cyclotron damping of the pump wave and Ohmic heating of the upper hybrid wave may play a dominating role. In addition, during heating at a HB frequency, there is strong enhancement up to the order of 1.8 in N e /N e14:30, which extends from approximately the pump wave reflection altitude to 600 km and is apparently altitude-independent. The enhancement does not immediately occur when heating was turned on at 13:30UT, but it has a delay of 30 s and develops in HB and disappears with the falling pump frequency when the pump frequency sweeps down in the heating cycles at 12:30UT 12:48UT and 13:30UT 13:48UT. On Fig.3 The altitude profile of critical frequency and plasma line spectra at an altitude of 351 km as a function of heating cycles, where the black dashed lines correspond to the HF on and off As an example, the second and third panels of Fig. 3 illustrate the downshifted plasma line spectra versus the heating cycles at altitude of 351 km. The downshifted plasma line spectra come from two separated plasma line channels covering the Doppler frequency offsets of 9.65 MHz to 7.15 MHz and 7.25 MHz to 4.75 MHz, respectively. In the channel of 9.65 MHz to 7.15 MHz, one can see an obvious natural plasma line covering 8.1 MHz to 8.6 MHz, which does not show the expected rise in plasma line frequency corresponding to the increase in electron density. Moreover, there exists an unusual phenomenon in that the natural plasma line becomes much weaker when heating in HB than that in LB or heating off condition. The line almost disappears in GB, which for convenience is termed breaking down here In the channel of 7.25 MHz to 4.75 MHz, no natural 892

WU Jun et al.: Results of Ionospheric Heating Experiments Involving an Enhancement in Electron Density plasma line was observed, but as the case of the channel of 9.65 MHz to 7.15 MHz, there is also breaking down of the plasma line spectra when pump frequency sweeps into GB and HB. Furthermore, in the period of pump frequency sweeping down, the breaking down of plasma line spectra in the channel of 7.25 MHz to 4.75 MHz occurs 60 s ahead of that in the channel of 9.65 MHz to 7.15 MHz. Its reappearance lags by 60 s behind that in the channel of 9.65 MHz to 7.15 MHz. When the pump frequency sweeps up, the breaking down behavior of the plasma line spectra is similar to that when sweeping down. The period of breaking down of plasma line spectra in the channel of 7.25 MHz to 4.75 MHz is longer than that in the channel of 9.65 MHz to 7.15 MH. Fig.4 The altitude profile of critical frequency at 14:30UT As an example, the top panel of Fig. 4 is the altitude profile of critical frequency at 14:30, which is extracted from the top panel of Fig. 3 and shows the background condition of ionosphere when heating off. To showing this in more detail, a subset of the altitude profile of critical frequency with an altitude range of 190 230 km and a critical frequency range of 5 8 MHz is given as the bottom panel of Fig. 4. Considering the magnetic intensity at 14:30UT shown by Fig. 1, the electron gyrofrequency should be 1.3792 MHz at the altitude of 200 km, then the reflection altitude of pump at 7 MHz and 6.7 MHz should be 218 km and 216 km, respectively, and furthermore the upper hybrid resonance altitude 204 km and 200 km, respectively. 3 Discussion In theory, the plasma line is the backscatter from Langmuir waves, and the behavior of the electron density in the enhanced region should reconcile with that of the plasma line, whose frequency ω satisfies the following dispersion relation: ω 2 = ω 2 pe + γ e k 2 v 2 th. (4) Where k denotes the wave number, v th the thermal speed and γ e = 3 for the one-dimensional case. With Eq. (4), we attempt to seek the evidence of the enhancement in electron density by taking insight to the dependence of the behavior of the plasma line on electron density. The top panel of Fig. 3 shows that the plasma frequency obtained from the altitude profile of electron density rises up to 11 MHz at an altitude of 300 km and 10 MHz at an altitude of 351 km when heating in HB, during which, however, the plasma line spectra at an altitude of 351 km become weak, as shown in the second and third panels of Fig. 3. The interpretation may be that the increase up to 10 MHz of the plasma frequency leads to a plasma line spectra exceeding the channels of 9.65 MHz to 7.15 MHz and 7.25 MHz to 4.75 MHz, which can no longer be detected by EISCAT UHF radar. When pump frequency sweeps near the fifth gyrofrequency (GB), however, the enhancement in electron density cannot be observed but the plasma line spectrum becomes weaker than in HB and almost disappears. Therefore, the above interpretation of the enhancement in electron density is difficult to reconcile with the behavior of the plasma line spectra. On the other hand, by comparing the middle panel of Fig.2 with the second and third panels of Fig. 3, one can see that the breaking down of the plasma line spectra at the altitude of 351 km corresponds in the heating cycles (or pump frequency) to the weaker disturbance of electron temperature around the reflection altitude of 210 km. As is well known, however, the enhancement in electron temperature at the altitude of 210 km does not contribute to the behavior of the plasma line spectra at the altitude of 351 km. Additionally, there is no evidence showing the enhancement in electron temperature at an altitude of 351 km. Thus, it can be concluded that the enhancement in electron density shown in Fig. 2 does not correspond to the behavior of the plasma line and the true enhancement in electron density. Senior et al. [9] hypothesized that the enhancement in electron density with wide altitude extent is due to the backscatter of the UHF radar wave along the magnetic field line. Therefore, we should ask what process leads to that scattering body of the radar wave? It is obvious that the apparent electron density enhancements presented here and those previously reported by Blagoveshchenskaya et al. [7,8] and Senior A, et al. [9] are produced by ionospheric heating; that is, they are not natural. There should be some mechanism to be responsible to transport the heating energy upward from the pump wave reflection altitude of 200 km and produce the scattering body of radar wave. Here, we write the pump energy around the reflection altitude as P H = P r + P T + P u + P o. (5) Where P r is the reflected energy, P T the energy used to enhance electron temperature, P u is the energy transported upward and P o the energy used to excite other such processes as the stimulated electromagnetic emission (SEE). To simplify the problem, P o will be neglected. Thus, the ionospheric self-absorbed energy 893

should be expressed as P a = P T + P u. (6) We assume that the level of self-absorption above (HB) 5Ω ce is equal essentially to that below (LB) 5Ω ce. In fact, the assumption is supported by experiment conducted with a vertical pump beam and in O-mode at 14:05UT 15:53UT on 15 May 1991 [15,16]. Fig. 5 is the self-absorption level of pump as a function of pump frequency near the fifth gyrofrequency, and shows that the minimum ( 3 db) in self-absorption occurs in the interval of [ 6.65 MHz, 6.7 MHz]. Since the experiment that was conducted on 15 May 1991 was performed with a vertical pump beam, here we should examine the self-absorption level of pump by considering the vertical component of magnetic. Fig. 6 shows the vertical component of magnetic intensity at 14:00UT 16:00UT on 15 May 1991 in Tromsø, where the data on ground were recorded by Tromsø Geophysical Observatory, University of Tromsø, and the data at altitude of 200 km were derived from the IGRF by extrapolating. From Fig. 6, one can see the vertical component of magnetic intensity at the altitude of 200 km varies in the interval of [47598 nt, 47627 nt], the corresponding fifth electron grofrequency at the altitude of 200 km should then be in the interval of [6.6664 MHz, 6.6705 MHz], which is consistent obviously with the band of the minimum self-absorption in Fig. 5. Fig.5 Self-absorption of pump as a function of pump frequency close to the fifth gyrofrequency on 15 May 1991 [15] Fig.6 The Vertical component of magnetic intensity at 14:00UT 16:00UT in Tromsø, where the solid curve and dashed curve represent the values on ground and at the altitude of 200 km, respectively It can be concluded that the above assumption is feasible; that is, the symmetry of self-absorption level with respect to 5Ω ce in Fig. 5 shows that the level of self-absorption above 5Ω ce is essentially equal to that below 5Ω ce, and the minimum level of self-absorption occurs near 5Ω ce. It should be noted for Fig. 5 and Fig. 6 that it is difficult due to the errors from IGRF model and measurements to determine whether the minimum in self-absorption occurs exactly at 5Ω ce, but the reasonable agreement between experiment and IGRF model suggests that the minimum in selfabsorption is indeed related to the fifth electron gyrofrequency harmonics. In the above principle, the self-absorption of pump shown in Fig. 5 shows that the absorbed energy by ionosphere in HB is equal essentially to that in LB. In the middle panel of Fig. 2, it is obvious that the enhancement in electron temperature when heating in LB is stronger than that in HB, whereas the enhancement in electron density above the reflection altitude takes place only in HB. Based on the energy budget, this may be due to energy distribution at the upper hybrid altitude and around the reflection altitude. When heating in LB, the whole absorbed energy can be contributed to the enhancement in electron temperature and no energy transport occurs; that is, P u = 0 and P a = P T. When heating in HB, a fraction of absorbed heating energy (P T ) is contributed to the enhancement in electron temperature, and the other fraction (P u ) is transported upward and produces the scattering body above the reflection altitude. In fact, the above phenomena also appear in the experimental results presented by Senior et al [9]. There should be two possible ways to be responsible for transporting energy upward when heating. One possible way is that the pump wave is coupled to Z mode near the critical angel and propagates upward. The pump wave cannot usually penetrate further than the critical altitude of X = 1 in such a magnetized plasma of increasing density as the ionosphere, where X = fpe f HF. In high latitudes, however, it is possible for the pump wave to be coupled to Z mode and continue upward in Z mode as it reaches the altitude of X = 1+Y when the pump wave vector is parallel to the magnetic field, where Y = Ωce f HF and Ω ce = eb 2πm, B is the magnetic field intensity, and e and m are the electron charge and mass, respectively. With regard to the experiment conducted at 12:30UT 14:30UT on 11 March 2014, however, X = 1 + Y varies from 1.2 for f HF = 6.7 MHz to 1.19 for f HF = 7 MHz, of which the critical plasma frequencies correspond to 8.05 MHz and 8.35 MHz respectively. Considering the ionospheric background shown by the top panel of Fig. 4, it is not possible that the pump wave in Z mode can reach an altitude of 300 km and above. In addition, Miolhus and Fla [17] have calculated the resonance spatial region (radio window) of Z mode for an EISCAT heater case, which is very localized and displaced horizontally 15 20 km southward from the point that the pump wave undergoes when it arrives at the altitude of X = 1. Thus, a UHF radar with a beam 894

WU Jun et al.: Results of Ionospheric Heating Experiments Involving an Enhancement in Electron Density width of 0.5 cannot probe the Z mode resonance region. Moreover, the enhancement in electron density in Fig. 2 occurs only when heating is in HB. In addition, a linear mode conversion of ordinary and extraordinary waves into an upper hybrid wave and Bernstein wave can take place at an upper hybrid altitude when heating the ionosphere. However, upper hybrid waves and Bernstein waves propagate in directions perpendicular to the magnetic field and cannot transport heating energy upward. Therefore, the way to transport energy upward in wave mode should be ruled out. The other possible way to transport energy upward is a plasma transport process, such as diffusion along the magnetic field due to thermal pressure and density gradients. Neglecting the natural electric field and the natural wind field and considering the thermal pressure gradient and density gradient caused by the ionosphere heating as well as gravity, the plasma diffusion velocity yields [18] 2 10 19 n [ 1 dn e W d = D N e dh + 1 dt p T p dh + 1 H p ], (7) here ( D is ) plasma diffusion coefficient and given by D cm 2 /s for the reasonable daytime condition, the neutral densitynis given by n[o] = ( 5 10 9) cm 3, plasma temperature T p and plasma scale height H p are defined, respectively, by T p = (T e + T i ), (8) 2 H p = 2kT p m i g, (9) where T i denotes ion temperature, k the Boltzmann constant, m i ion mass and g the acceleration due D dn to gravity. Thus, e N e dh, D dt e T e dh and D H p should be the contribution of electron density, electron temperature and gravity respectively to the plasma diffusion velocity. It should be emphasized here that the data used in this paper are measured along the geomagnetic direction, so the diffusion process does not take into account the effect of the magnetic field as well as horizontal diffusion. Moreover, the electron temperature response time is on the order of µs and the characteristic time of the heat conducting process on the order of s, then the heat conducting process excited by ionospheric heating may be neglected. The data obtained during the experiment conducted at 12:30UT 14:30UT on 11 March 2014 show D dn e 0.0025 km/s. (10) N e dh D dt e 0.0012 km/s, (11) T e dh D H p 0.0024 km/s, (12) then W d = 0.0037 km/s at the reflection altitude of pump, where negative sign indicates the downward direction of diffusion along the magnetic field. As shown in the top panel of Fig. 2, however, the pump energy is transported upward. Besides, the enhancement in electron density does not immediately occur with heating at 13:30UT, but a delay of 30 s. Obviously, the measured characteristic time of electron density enhancement is much less than the diffusion time. So the way to transport energy upward in diffusion should also be ruled out. 4 Conclusion In this paper, the experimental phenomenon of the apparent enhancement in electron density with wide altitude extent was presented. The results show the enhancement in electron temperature and the apparent large electron density enhancement with wide altitude extent are a function of pump frequency, and the plasma line spectra are absent and become weak when the pump frequency is very close to and above the fifth gyrofrequency, respectively. The enhancement in electron temperature should involve multiple physical processes, mainly the upper hybrid wave resonance excited by the pump wave in O-mode at the upper hybrid resonance altitude. Although the physics of the phenomenon of apparent enhancement in electron density and the breaking down of the plasma line spectra remain to be determined, it may be due to some unknown mechanism that preferentially involves plasma transport process and leads to the backscatter of the radar wave along the magnetic line, which needs more means of observation. Acknowledgments We would like to thank Dr. M. T. Rietveld for a valuable discussion on the plotting of plasma line spectra and for his diligent operation of the heating facility, and engineers of EISCAT in Tromsø for keeping the facility in excellent working condition. In addition, we thank Tromsø Geophysical Observatory, UiT The Arctic University of Norway, for providing the magnetic data of Tromsø recorded on 15 May 1991 and 11 March 2014, respectively. The EISCAT Scientific Association is supported by China (China Research Institute of Radiowave Propagation), Finland (Suomen Akatemia of Finland), Japan (the National Institute of Polar Research of Japan), Norway (Norges Forkningsrad of Norway), Sweden (the Swedish Research Council) and the UK (the Particle Physics and Astronomy Research Council of the United Kingdom). References 1 Birkmayer W, Hagfors T, Kofman W. 1986, Phys. Rev. Lett., 57: 1008 895

2 Duncan L M, Sheerin J P, Behnke R A. 1988, Phys. Rev. Lett., 61: 239 3 Kelley M C, Arce T L, Salowey J, et al. 1995, J. geophys. Res., 100: 17367 4 Stocker A J, Honary F, Robinson T R, et al. 1992, J. Atmos. Terr. Phys., 54: 1555 5 Pedersen T, Gustavsson B, Mishin E, et al. 2010, Geophys. Res. Lett., 37: 106 6 Pedersen T, McCarrick M, Reinisch B, et al. 2011, Ann. Geophys., 29: 47 7 Blagoveshchenskaya N F, Borisova T D, Rietveld M T, et al. 2011, Geomagn. Aeron., 51: 1109 8 Blagoveshchenskaya N F, Borisova T D, Yeoman T K, et al. 2011, Geophys. Res. Lett., 38: 302 9 Senior A, Rietveld M T, Haggstrom I, et al. 2013, Geophys. Res. Lett., 40: 1669 10 Rietveld M T, Kohl H, Kopka H, et al. 1993, J. Atmos. Terr. Phys., 55: 577 11 Rishbeth H, Van Eyken A P. 1993, J. Atmos. Terr. Phys., 55: 525 12 Dysthe K B, Miolhus E, Pecseli H, et al. 1982, Phys. Scr., T2/2: 548 13 Miolhus E. 1990, Radio Sci., 25: 1321 14 Miolhus E. 1993, J. Atmos. Terr. Phys., 55: 907 15 Stocker A J, Honary F, Robinson T R, et al. 1993, J. Geophys. Res., 98: 13627 16 Stubbe P, Stocker A J, Honary F, et al. 1994, J. Geophys. Res., 99: 6233 17 Miolhus E, Fla T. 1984, J. Geophys. Res., 89: 3921 18 Rithbeth Henry, Garriott Owen K. 1969, Introduction to Ionospheric Physics. Academic Press, New York (Manuscript received 27 August 2015) (Manuscript accepted 22 March 2016) E-mail address of WU Jun: Junwu629@hotmail.com 896