PMSE dependence on frequency observed simultaneously with VHF and UHF radars in the presence of precipitation

Transcription

1 Plasma Science and Technology PAPER PMSE dependence on frequency observed simultaneously with VHF and UHF radars in the presence of precipitation To cite this article: Safi ULLAH et al 2018 Plasma Sci. Technol View the article online for updates and enhancements. This content was downloaded from IP address on 06/09/2018 at 03:39

2 2018 Hefei Institutes of Physical Science, Chinese Academy of Sciences and IOP Publishing Printed in China and the UK Plasma Science and Technology Plasma Sci. Technol. 20 (2018) (9pp) PMSE dependence on frequency observed simultaneously with VHF and UHF radars in the presence of precipitation Safi ULLAH, Hailong LI ( 李海龙 ), Abdur RAUF, Lin MENG ( 蒙林 ), Bin WANG ( 王彬 ) and Maoyan WANG ( 王茂琰 ) School of Electronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu , People s Republic of China hailong703@163.com Received 14 March 2018, revised 21 May 2018 Accepted for publication 30 May 2018 Published 4 September 2018 Abstract Using PMSE (polar mesosphere summer echoes) observations in combination with particle flux measurements obtained with detectors onboard the Geostationary Operational Environmental Satellite (GOES) a special condition is shown for the occurrence of rare observed UHF PMSE. When electron flux observed from GOES satellites show a decrease, then after being in the presence of precipitation UHF PMSE occurs. The heating effect on PMSE is small when the UHF electron density is enhanced at 90 km due to particle precipitation. We analyzed and compared the frequency dependence of PMSE under the condition of high energy particle precipitation in July of 2004 and 2007 at well separated frequencies (224 and 930 MHz) at the same site, height, and time. The frequency index varies with height and time. At different heights, the maximum as well as the minimum value of volume reflectivity at VHF is greater than that at UHF with 2 to 3 orders of magnitude. A new qualitative method for the analysis of dust distribution is used by analyzing the relationship between volume reflectivity and frequency index. In agreement with the results of the model it is shown that dust particles of smaller size generally did not occur at the edges, instead they occurred in the middle PMSE regions. Keywords: polar mesosphere summer echoes (PMSE), volume reflectivity, ionospheric disturbances (Some figures may appear in colour only in the online journal) 1. Introduction One intriguing phenomenon of the polar ionosphere under summer conditions is the strong radar echoes called polar mesosphere summer echoes (PMSE). Apparently PMSE consists of nanometer size charged dust particles whereas NLC (noctilucent clouds), which is found at the lower edge of the PMSE [1], consists of submicron size charged dust particles and can be seen at sunset with the naked eye. PMSE has a connection with NLC [1]. Spectral separation between electrons and neutrals was proposed by Kelley et al [2]. Dueto electron attachments on dust particles, irregularities in the electron density are produced, which cause the PMSE to be observed at half of the radar wavelength [3, 4]. The irregularities of the electron number density should be rapidly destroyed by the molecular diffusion. At PMSE altitudes the electron diffusivity should decrease when more than 50% of the background electrons attach on aerosol particles [5] because they gain the low diffusivity of the aerosol due to ambipolar electric fields. An enhanced Schmidt number results in the decrease of the diffusivity required for the existence of Bragg-scale irregularities. Using the EISCAT heating facility Chilson et al [6] first documented artificial modulation of PMSE backscatter for 10 and 20 s heater-on and heater-off times, respectively. When ionosphere is heated with a highfrequency (HF) radio wave, then due to electron temperature enhancement the electron diffusivity increases causing smoothing of electron density irregularities and hence the /18/ $

3 Table 1. Radar parameters and experimental configuration. EISCAT VHF EISCAT UHF Site (Tromsø) 69.6 N, E 69.6 N, E Frequency (MHz) Wavelength λ (m) Bragg wavelength λ/2 (m) Bragg wavenumber 4π/λ (m 1 ) Beam width ( ) Peak power (MW) Range resolution (km) Time resolution (s) 2 2 Spectral resolution (Hz) Velocity resolution (ms 1 ) Radar program arc_dlayer_ht arc_dlayer_ht System temperature (K) PMSE structure [7]. For a comparatively long heating cycle in which the heater was switched on for 20 s and switched off for 160 s, Havnes et al [8] showed that the dusty plasma had enough time to return back to its state before the heating was switched on. This causes greater electron density gradients than those before the heating was switched on. Consequently PMSE overshoot is produced in which the radar scattering intensity sharply increased, than its equilibrium state (before the heating was switched on). To use heating experiments as a diagnostic tool, the first PMSE heating model was developed by Havnes [8, 9] and later by Biebricher et al [10]. These models are not perfect for all frequencies and work only for high radar frequencies. However, the models of Chen and Scales [3] and Mahmoudian et al [4] simulate well the temporal evolution of PMSE for all radar frequencies. To probe the characteristics of dusty plasma, recent techniques and basic models were presented by Scales and Mahmoudian [11]. One of the most important properties of PMSE is the dependence of volume reflectivity on radar frequency. So far PMSE have been observed at different radar frequencies in the 2.78 MHz 1.29 GHz range. Measurements at different radar frequencies were performed at different times and sites. Due to different geophysical situations their comparison must be done with great caution, but it is obvious that volume reflectivity shows a great increase with decreasing frequency or increasing Bragg scale [12]. A theoretical expression between the frequency and volume reflectivity in view of PMSE is derived which shows an agreement of theoretical and statistical results [13]. It was realized that if observations are carried out at different radar frequencies, then it would be helpful to understand the mechanism of radar echoes. Cho and Kelley [14] compared calibrated observations at 50, 224, and 930 MHz but these measurements were done at different atmospheric volume, time, and height. Cho and Kelley [14] also documented the comparison of volume reflectivities with models based on neutral air turbulence and dressed aerosol scatter. A study of the literature on PMSE demonstrates that volume reflectivity shows an enormous frequency dependence, e.g., for studying the validity of theoretical expectations a common volume and simultaneous observations should be used as ideal tools. Belova et al [15] and Rapp et al [16] presented the case studies of PMSE observations with the ALWIN VHF (53.5 MHz), EISCAT VHF (224 MHz), and UHF (930 MHz) radars. For derivation of the microphysical parameters and the statistical properties of PMSE, Li et al [17] documented simultaneous and common volume observations with the collocated SOUSY Svalbard Radar at 53.5 MHz and the EISCAT Svalbard Radar at 500 MHz. In addition, a statistical study of the EISCAT VHF and UHF radars is presented in [18] but without an analysis of each observation under specific conditions. In this article we analyze and compare the frequency dependence of PMSE for each observation at the same site, height, and time in the presence of high energy particle precipitation at well separated frequencies (224 and 930 MHz) in July of 2004 and UHF radar PMSE observation is difficult and also affected by heat [19]. So we carried out this work in the presence of high energy particle precipitation and also show one special condition which is common in July of 2004 and The purpose of the present work is to consider PMSE studies at higher frequencies than 50 MHz and to further understand the frequency dependence of PMSE. 2. Experimental details and data analysis The data used for the measurements of this paper were taken from three different sources. The PMSE observations were carried out simultaneously by the EISCAT VHF and UHF radars on 13 July 2004 and 12 July As at 930 MHz PMSE is difficult to observe so first the threshold for UHF PMSE was defined as volume reflectivity m 1 [20] and then for the same time and height the corresponding data at 224 MHz were selected. In this study GUISDAP (Grand Unified Incoherent Scatter Design and Analysis Program) was used for the analysis of incoherent scatter measurements with the EISCAT radars [21]. Volume reflectivities can be obtained from apparent electron number densities by using the following relation: h = s N, ( 1) where η is the volume reflectivity, σ equals half the scattering cross section of an electron also equal to m 2, and N e is the apparent electron number density [20]. The technical and physical parameters of the EISCAT VHF and UHF radars are given in table 1. The VHF and UHF datasets were integrated over 20 s intervals. At both frequencies the heating has identical effects [19] but here we still considered the 20 s heating on parts because under the given condition in this study due to particle precipitation the radio waves do not produce clear heating effects. On 13 July 2004 and 12 July 2007 measurements of electron proton and x-rays fluxes in the magnetosphere were obtained by energetic particle detectors onboard the GOES- 10, GOES-11, and GOES-12. The data were obtained from e 2

4 the NOAA Space Environment Center website ( noaa.gov/). To examine the effect of geomagnetic activity on PMSE the K-indices from 13 July 2004 and 12 July 2007 were used from the Tromsø Geophysical Observatory (TGO) at the University of Tromsø (66.66 N, E) Norway. 3. Energetic particle effects on PMSE In the upper mesosphere ionization may increases to high levels [22] due to intensive particle precipitation called solar proton events. Middle atmosphere and subsequently PMSE might be strongly affected by SPE [23]. In the presence of a solar proton event a modeled background electron number density effect on PMSE is presented in [24]. Rapp et al [24] found that a limited amount of aerosol particles cause a significant anti-correlation between the electron number density and the signal-to-noise ratio. An additional ionization in the ionospheric D-region results from precipitating electrons having a kev energy range [25]. In 2007 at the Poker Flat Incoherent Scatter Radar PMSE was observed at nighttime accompanied by particle precipitation [26]. The observations presented in [5] at 1295 MHz and in [16] at 933 MHz also show a positive correlation between the strengthening of particle precipitation and the appearance of PMSE. Due to a greater ability of penetration hard x-rays (1 10 Å) can produce an increase in D-region ionization [27]. Often a positive correlation is found between geomagnetic activity (used as a proxy of high energetic particle precipitation) and PMSE [28]. Kirkwood et al [29] proposed a strong effect of electron density and its gradient on PMSE reflectivities. Electron density at polar latitudes depends on ionization produced either from sporadic high energetic protons and electron precipitation or from solar extreme-ultraviolet radiation [29]. The K-index is the quasi-logarithmic measure of the degree of disturbance of ionosphere derived for the fixed 3 h time intervals. The information about the variation in the horizontal component of the Earth s magnetic field with respect to the geomagnetic quiet conditions can be obtained from the local geomagnetic K-index values from 0 9 where 0 and 9 represent very quiet and extremely disturbed condition, respectively Particle effects on PMSE on 13 July 2004 Figure 1 shows the particle fluxes and K-indices on 13 July The electron fluxes of energy >= 2 MeV and >= 0.6 MeV are shown in the first and second panels of figure 1, respectively. From the first panel it is clear that initially the electron flux of energy >= 2 MeV is high and then decreases gradually. After 3 UT to 9 UT it remains small. After 9 UT the flux again gradually increases with no noticeable variation up to 24 UT. In the second panel a comparatively low energy electron flux (E >= 0.6 MeV) is initially high and then decreases before 3 UT and remains small at 6 UT. Just after 6 UT the flux increases again and remains high and shows no noticeable variation up to 24 UT. The proton flux of energy >= 10 MeV is given in panel 3 of Figure 1. Electron, proton, and x-ray fluxes, and K-indices on 13 July The text in panels 1 4 represent the corresponding GOES satellite, particle flux, and their energy. Figure 2. Electron, proton, and x-ray fluxes, and K-indices on 12 July The text in panels 1 4 represent the corresponding GOES satellite, particle flux, and their energy. figure 1 and is small throughout on 13 July Panel 4 of figure 1 shows the solar x-ray flux of energy 1 8 Å. It is clear that the x-ray flux is also very small throughout. We do not believe that x-rays played any role in the ionization of the D-region on 13 July. To check the effect of the geomagnetic activity on UHF PMSE we focus on the third K-index because it corresponds to the UHF PMSE time period (8 9UT). Panel 5offigure 1 shows that the value of the third K-index (K = 3) is small so in this case geomagnetic disturbance has no significant effect on UHF PMSE Particle effects on PMSE on 12 July 2007 Figure 2 shows the particle fluxes and K-indices on 12 July From the first panel of figure 2 it is obvious that the electron flux of energy >= 2 MeV initially is high, then before 3 UT it decreases and from 3 6 UT it shows a greater 3

5 Figure 3. Subplots 1 2: UHF PMSE observation starts from 07:00 UT on 13 July Subplots 3 4: UHF PMSE observation starts from 08:00 UT on 12 July The color bar shows the log 10 (reflectivity (m 1 )). variation. After 6 UT the flux again increases and shows no noticeable variation during 9 15 UT. After 15 UT the flux further increases and remains high up to 24 UT. The second panel of figure 2 shows the electron flux of comparatively low energy >= 0.6 MeV. From panel 2 it is obvious that from the start to 8 UT the electron flux is comparatively small whereas from 8 24 UT the flux is comparatively greater. From panels 3 and 4 it is obvious that proton and x-ray fluxes are small throughout on 12 July On 12 July 2007 the fourth K-index corresponds to the UHF PMSE time (9 10 UT) and also small (K = 1) indicates no noticeable geomagnetic disturbance. From figures 1 and 2 it is obvious that before 8 9UT on 13 July 2004 and 9 10 UT on 12 July 2007 the electron flux at different energies is comparatively small while at 8 9 UT on 13 July 2004 and 9 10 UT on 12 July 2007 the electron flux is comparatively high. Figures 1 and 2 show that when proton and x-ray fluxes and geomagnetic K-indices are very small and non-significant, then the relatively greater electron fluxes might be important for the D-region ionization to produce PMSE Observations of UHF PMSE and particle precipitation Figure 3 shows the rare observed UHF PMSE on July 2004 and Subplots 2 and 1 of figure 3 show the observation of UHF PMSE on 13 July 2004 at the height ranges of and km, respectively. Subplots 1 and 2 shown that during 7 8 UT no PMSE coincide with particle precipitation but relatively strong UHF PMSE centered around 85.5 km and precipitation, coincide during 08:02:00 08:20:00 UT. In figure 3 subplots 4 and 3 show the UHF PMSE observation on 12 July 2007 at height ranges of and km, respectively. In subplots 3 and 4 the comparatively strong precipitation during 8 9 UT with no PMSE is obvious. On the other hand, comparatively strong PMSE centered around 83.5 km during the time period of 09:15:00 09:24:00 UT coincides with particle precipitation. Figure 3 shows one Figure 4. Subplots 1 2: VHF PMSE observation starts from 07:00 UT on 13 July Subplots 3 4: VHF PMSE observation starts from 08:00 UT on 12 July The color bar shows the log 10 (reflectivity (m 1 )). similarity on 13 July 2004 and 12 July In both cases when comparatively strong precipitation reaches 86 km then in the next hour in the presence of particle precipitation clear UHF PMSE occurs for a short time interval Observations of VHF PMSE and particle precipitation Figure 4 shows the PMSE observation carried out with VHF radar for the corresponding time and height range of figure 3. It is now known that VHF PMSE have an extended structure with altitude. Figure 4 shows the strong VHF PMSE with extended structures, at the time corresponding to UHF PMSE (shown in figure 3) Electron density changes due to particle precipitation A survey of the literature on PMSE heating [9, 10, 19] revealed that in the low electron density case heating increases whereas when below and above, the PMSE region electron densities are high, then the heating effects on PMSE are small or even absent. The effect of particle precipitation on UHF electron density and on VHF, and UHF volume reflectivities on July 2004 are shown in figures 5 and 6, respectively. Subplots 1 and 2 of figure 5 show the observed UHF electron density on 13 July 2004 where UHF PMSE and precipitation are present (as shown in the subplot 2 of figure 3) and as a comparison also shows the UHF electron density profile on 14 July 2004 where no UHF PMSE and precipitation are observed (not shown here). Subplot 1 of figure 5 shows that at 80 km the UHF electron densities show no greater difference on 13 and 14 July However, the case is quite different at 90 km (see subplot 2) where the UHF electron density on 13 July (with precipitation and UHF PMSE) is greater than that on 14 July (without precipitation and UHF PMSE). Figure 6 shows the normalized and mean VHF and UHF volume reflectivities from km on 13 July The normalization of VHF and UHF reflectivities were done by 4

6 Figure 5. At two different altitudes the UHF electron density profiles in the presence (13 July 2004) and absence of particle precipitation (14 July 2004). In both subplots the solid and dash lines represent the UHF electron density on July 13 and 14, respectively. Figure 7. At two different altitudes UHF electron density profiles in the presence (12 July 2007) and absence of particle precipitation (13 July 2007). In both subplots solid and dashed lines represent the UHF electron density on July 12 and 13, respectively. Figure 6. Normalized VHF and UHF volume reflectivities in the presence of particle precipitation (on 13 July 2004). The vertical green bars show the heater-on period. Dashed and solid lines represent the normalized volume reflectivity (η) at VHF and UHF, respectively. The vertical black line at cycle 10 indicates the time at 08:30:00 UT. dividing each data on the maximum value of η VHF and η UHF, respectively. Figure 6 shows that reflectivity at both frequencies from 08:00 08:30:00 UT are comparatively greater than that from 08:30:00 09:00 UT. Just after 08:30:00 UT the reflectivity at VHF again shows some increase. UHF PMSE and precipitation occur during the time period of 08:00 08:30:00 UT (as is clear in figure 3) and also during the same time period (i.e., during 08:00 08:30:00 UT) radar echoes show a comparatively greater increase (as is clear from figure 6). This indicates that radar echoes are not much affected by HF heating during the time period of 08:00 08:30:00 UT. Similarly, subplots 1 and 2 of figure 7 show the effect of particle precipitation on UHF electron density at two different altitudes on July Here 12 July 2007 has UHF PMSE for Figure 8. Normalized VHF and UHF volume reflectivities in the presence of particle precipitation (on 12 July 2007). The vertical green bars show the heater-on period. Dashed and solid lines represent the normalized volume reflectivity (η) at VHF and UHF, respectively. a short interval in the presence of particle precipitation (as shown in subplot 4 of figure 3) whereas 13 July 2007 has neither UHF PMSE nor precipitation (not shown here). From figure 7 it is clear that at both altitudes (see subplots 1 and 2) the UHF electron density in the presence of UHF PMSE and precipitation (on 12 July 2007) is greater than that in the absence of UHF PMSE and precipitation (on 13 July 2007). Figure 8 shows the normalized and mean VHF and UHF volume reflectivities from km on 12 July From figure 8 it is clear that for the first 8 heating cycles reflectivity at both radars are comparatively greater than that during the next 12 heating cycles. As can be seen from figure 7 it is obvious that at both altitudes (80 and 90 km) UHF electron density in the presence of particle precipitation (on 12 July 2007) is greater than that in the absence of particle 5

7 Figure 9. Comparison of volume reflectivities at different altitudes obtained simultaneously with the EISCAT VHF and UHF radars on 13 July In each panel asterisks represent the reflectivities at VHF while the triangles represent the reflectivities at UHF. Figure 10. Comparison of volume reflectivities at different altitudes obtained simultaneously with the EISCAT VHF and UHF radars on 12 July In each panel the asterisks represent the reflectivities at VHF while the triangles represent the reflectivities at UHF. precipitation (on 13 July 2007). Furthermore figure 8 shows that from 09:00 09:24:00 UT the comparatively greater reflectivity at both radars show that radar echoes during the time period of 09:00 09:24:00 UT are not much affected from HF heating. From figures 5 and 7 it is important to note that in both cases in the presence of particle precipitation at 90 km the UHF electron densities are greater than those in the absence of particle precipitation. Therefore, we are confident to say that under our given condition in the presence of high energy particle precipitation electron density increases in the PMSE region so no obvious significant heating effect is produced as is clear from figures 6 and 8 that volume reflectivity at both frequencies are not much affected during the time period containing UHF PMSE and precipitation. 4. Volume reflectivities and frequency index Figure 9 shows a comparison of volume reflectivities simultaneously observed at VHF and UHF frequencies at three different heights on 13 July In panel (a) at 86.5 km η VHF the values fall within the range of to m 1 whereas at the same time and height UHF observations fall within the range of to m 1. At the height of 85.3 km (see panel (b) of figure 9) η values obtained at the 224 MHz frequency fall within the range of to m 1 whereas at 930 MHz observations fall within the range of to m 1. As can be seen in panel (c) of figure 9 at 84.5 km the η values obtained from EISCAT VHF radar fall within the range of to m 1 whereas at the same time and height UHF observations fall within the range of to m 1. Next, the volume reflectivities obtained from the simultaneous observations at VHF and UHF frequencies on 12 July 2007 are compared and shown in figure 10. On 12 July 2007 UHF PMSE occurs at comparatively lower altitudes than that on 13 July At 83.5 km (see panel (a) of figure 10) the volume reflectivity obtained at VHF radar fall within the range of to m 1 whereas at the same time and height the UHF observations fall within the range of to m 1. At 82.5 km (see panel (b) of figure 10) the η values at VHF radar fall within the range of to m 1 while at the same time and height the η UHF values fall within the range of to m 1. A comparison of volume reflectivities at VHF and UHF frequencies show some similarities on 13 July 2004 and 12 July From figures 9 and 10 it is obvious that the maximum value of volume reflectivity for each frequency is quite different for the observations at different heights. The comparison shows that the maximum value of volume reflectivity at VHF is greater than that at UHF with different orders of magnitude at different heights. For example, from figure 9 it is clear that on 13 July 2004 at 84.5, 85.3, and 86.5 km the maximum values of volume reflectivity at VHF are 2, 2, and 3 orders of magnitude larger than that at UHF, respectively. Similarly, in figure 10 on 12 July 2007 the comparison of the maximum values of volume reflectivities at VHF and UHF show that at 82.5 and 83.5 km the maximum values of volume reflectivities at 224 MHz are 3 and 2 orders of magnitude larger than that at 930 MHz, respectively. Similarly, comparing the minimum values of η VHF and η UHF it is obvious that the minimum value of η VHF is greater than the minimum value of η UHF with different orders of magnitude at different altitudes. On 13 July 2004 (figure 9) at VHF the minimum values of volume reflectivities are 2, 3, and 2 orders of magnitude larger than that at UHF at 84.5, 85.3, and 86.5 km, respectively. Furthermore, on 12 July 2007 (figure 10) the comparison of minimum values of η at VHF and UHF show that at 82.5 and 83.5 km the minimum values of volume reflectivities at 224 MHz are 2 and 3 orders of magnitude larger than that at 930 MHz, respectively. From figures 9 and 10 it is also obvious that at each height the η values at both frequencies are not stable and show great variations. However, in panel (b) of figure 9 at 6

8 Figure 11. The circles represent the indices based on equation (2) for the corresponding data shown in figure km from 08:02:40 08:08:00 UT and in panel (a) of figure 10 at 83.5 km from 09:16:20 09:19:40 UT the η values at VHF show stability. But this saturation on the horizontal straight line is only due to the fact that we used the GUISDAP software package where m 3 is the maximum electron density recognized by this software package. So here all the η values greater than m 1 are still considered as equal to m 1 which caused the observed saturation. To analyze the index of radar frequency at each height we use the following equation: log10 ( hvhf) - log10 ( huhf) n =. ( 2) log ( f ) - log ( f ) 10 VHF 10 UHF In equation (2) n is the index of the wavenumber, η VHF and η UHF are the volume reflectivities for the corresponding radars whereas f VHF and f UHF are 224 and 930 MHz, respectively. Figure 11 shows the indices for the corresponding data shown in figure 9. From figure 11 it is clear that at each height the index of the wavenumber shows a fluctuation with time. By using the simultaneous measurements of VHF and UHF we can arrive at the following order-of-magnitude functionality: h = k, ( 3) h = k, ( 4) h = k. ( 5) In equations (3) (5) η is the volume reflectivity in m 1 and k is the wavenumber in radians/m. Equations (3) (5) are derived at 84.5, 85.3, and 86.5 km, respectively for the corresponding data shown in figures 9 and 11 on 13 July From equations (3) (5) it is clear that at altitudes 84.5, 85.3, and 86.5 km reflectivity is inversely proportional to radar frequency to the powers of 4, 5.1, and 4.3, respectively. Similarly, figure 12 shows the corresponding indices of the data shown in figure 10. From figure 12 it is clear that on 12 July 2007 again at each height the index of the wavenumber shows a fluctuation with time. On 12 July 2007 at Figure 12. The circles represent the indices based on equation (2) for the corresponding data shown in figure 10. Figure 13. Relation between volume reflectivity and wavenumber. In both panels the circle symbols correspond to the measurements at VHF and the cross-hatches correspond to the measurements at UHF. different heights the relation between volume reflectivity and frequency is calculated by using the simultaneous measurements of 224 and 930 MHz and we arrived at the following order-of-magnitude functionality: h = k -4.1, ( 6) h = k ( 7) Equations (6) and (7) are derived at 82.5 and 83.5 km, respectively, for the corresponding data shown in figures 10 and 12 on 12 July From equations (6) and (7) it is clear that at heights 82.5 and 83.5 km reflectivity is inversely proportional to frequency to the powers of 4.1 and 5.1, respectively. The relation between volume reflectivity and frequency for the combined data on 13 July 2004 and 12 July 2007 are also calculated. On 13 July 2004 and also on 12 July 2007 the data were combined by removing all the data which did not satisfy the given condition (η m 1 ) for the UHF PMSE. For the combined data we arrived at the following order-of-magnitude functionality: h = k -4.8, ( 8) 7

9 Table 2. Frequency indices for the same time and different heights. 13 July 2004 Height (km) Index 12 July 2007 Height (km) Index 08:02:40 UT :16:00 UT :21:00 UT :07:00 UT h = k ( 9) Equations (8) and (9) correspond to 13 July 2004 and 12 July 2007, respectively. Equations (8) and (9) show that for the combined data the relation between volume reflectivity and frequency is identical on 13 July 2004 and 12 July 2007, that is in both cases reflectivity is inversely proportional to radar frequency to the power of 4.8. For the combined data the relation between volume reflectivity and frequency is shown in figure 13. In both panels the straight line is the result of a first order polynomial fit and the circle symbols and cross-hatches stand for the volume reflectivity at the corresponding frequency. Figure 13 shows that for increasing frequency or the Bragg wavenumber the volume reflectivity decreases. 5. Radius distribution of dust particles in the PMSE region In PMSE region the distribution of dust particles is difficult to obtain. Simulation results show that the mean ice radius changes with time and height [30]. Here based on the model results shown by Li et al [17], one new way to analyze the relation between dust particle radius and frequency index are obtained to carry out a qualitative analysis. From figure 10 of Li et al [17] the dashed and dashed-dotted lines are considered. For the dashed line (with a radius of 20 nm) and the dash-dotted line (with a radius of 80 nm) the indices of the wavenumbers are calculated as 5.5 and 3.1, respectively. Derivation of index n from the model calculation of Li et al [17] shows that the change in the absolute value of index n indicates the change in particle radius, i.e., greater absolute value of the index n represents a smaller particle radius. In figures 9 12 the time at which the data is missing means that at these particular time intervals the corresponding data are not satisfying the UHF PMSE threshold (η m 1 ).Infigures 11 and 12 at different altitudes there are several data points which have the same time. To analyze the dust particle radius at the same time and different altitudes, some data points on 13 July 2004 and 12 July 2007 from figures 11 and 12 are selected and are shown in table 2. In table 2 we can see that when there are two obvious layers (on 12 July 2007) then the absolute value of the index is greater than in the upper PMSE region. This indicates that the dust particle radius is greater in the lower PMSE region. On the other hand, when there are three layers (on 13 July 2004), then the largest absolute value of the index is found at the middle height of the PMSE region. This indicates that the smallest dust particles occur in the middle PMSE region. This result is also supported by our statistical results as shown in equations (3) (7). For the three layers the greater absolute value of the index is obvious at the middle height from equation (4) and for the two layers at upper height from equation (7). By comparing our results with the simulation results given by Rapp et al [30] the dust distribution show a similar character, that is the smaller dust radius particles generally may be occurred in the middle PMSE region and not at the edges. 6. Conclusions In the current study we have carried out the analysis and comparison of simultaneous and common volume measurements of PMSE observed by EISCAT 224 and 930 MHz radars in July of 2004 and 2007 with energetic particles fluxes from GOES satellites. We can draw the following conclusions: One new condition of higher probability to observe rare UHF PMSE is presented in the presence of precipitation and a decrease in electron flux. PMSE affected by HF heating is small when UHF electron density is enhanced at 90 km caused by energetic particles. Volume reflectivity is inversely proportional to frequency, where the frequency index varies with height and time, and consequently varies volume reflectivity. At different heights, the maximum and minimum value of volume reflectivity at VHF is greater than that at UHF with 2 to 3 orders of magnitude. A new qualitative analysis method in agreement with the model results is used to analyze the dust distribution by analyzing the relation between volume reflectivity and frequency index. It is found that generally the smallest dust particles may have occurred in the middle of the PMSE region instead of at the lower edges. Acknowledgments This work is supported by National Natural Science Foundation of China (Nos and ) and the 8

10 Fundamental Research Funds for the Central Universities (Nos. ZYGX2015J039, ZYGX2015J037, and ZYGX2015J041). The EISCAT Scientific Association is supported by the research councils of China, Finland, France, Germany, Japan, Norway, Sweden, and the UK. We are grateful to the NOAA Space Environment Services Center, USA for the data from the Geostationary Operational Environmental Satellites (GOES). We also thank the Tromsø Geophysical Observatory (TGO) at the University of Tromsø (66.66 N, E) Norway. References [1] Von Zahn U and Bremer J 1999 Geophys. Res. Lett [2] Kelley M C, Farley D T and Röttger J 1987 Geophys. Res. Lett [3] Chen C and Scales W A 2005 J. Geophys. Res. 110 A12313 [4] Mahmoudian A et al 2011 Ann. Geophys [5] Cho J Y N, Hall T M and Kelley M C 1992 J. Geophys. Res [6] Chilson P B et al 2000 Geophys. Res. Lett [7] Rapp M and Lübken F J 2000 Geophys. Res. Lett [8] Havnes O et al 2003 Geophys. Res. Lett [9] Havnes O et al 2004 Phys. Scr [10] Biebricher A et al 2006 Adv. Space Res [11] Scales W A and Mahmoudian A 2016 Rep. Prog. Phys [12] Rapp M and Lübken F J 2004 Atmos. Chem. Phys [13] Li H L et al 2009 Plasma Sci. Technol [14] Cho J Y and Kelley M C 1993 Rev. Geophys [15] Belova E, Dalin P and Kirkwood S 2007 Ann. Geophys [16] Rapp M et al 2008 J. Atmos. Solar Terr. Phys [17] Li Q et al 2010 J. Geophys. Res. 115 D00I13 [18] Li Q and Rapp M 2011 J. Atmos. Solar Terr. Phys [19] Næsheim L I, Havnes O and La Hoz C 2008 J. Geophys. Res. 113 D08205 [20] Röttger J and La Hoz C 1990 J. Atmos. Terr. Phys [21] Lehtinen M S and Huuskonen A 1996 J. Atmos. Terr. Phys [22] Roble R et al 1987 J. Geophys. Res [23] Cho J Y N and Röttger J 1997 J. Geophys. Res [24] Rapp M 2002 J. Geophys. Res [25] Zeller O and Bremer J 2009 Ann. Geophys [26] Varney R H et al 2011 J. Atmos. Solar Terr. Phys [27] Sahai Y et al 2008 Ann. Geophys [28] Bremer J et al 2009 J. Atmos. Solar Terr. Phys [29] Kirkwood S et al 2013 Ann. Geophys [30] Rapp M et al 2002 J. Geophys. Res