Transient ionization of the mesosphere during auroral breakup: Arase satellite and ground-based conjugate observations at Syowa Station

Transient mesospheric echo in the VHF range was detected at an altitude of 65–70 km during the auroral breakup that occurred from 2220 to 2226 UT on June 30, 2017. During this event, the footprint of the Arase satellite was located within the field of view of the all-sky imagers at Syowa Station in the Antarctic. Auroral observations at Syowa Station revealed the dominant precipitation of relatively soft electrons during the auroral breakup. A corresponding spike in cosmic noise absorption was also observed at Syowa Station, while the Arase satellite observed a flux enhancement of > 100 keV electrons and a broadband noise without detecting chorus waves or electromagnetic ion cyclotron waves. A general-purpose Monte Carlo particle transport simulation code was used to quantitatively evaluate the ionization in the middle atmosphere. Results of this study indicate that the precipitation of energetic electrons of > 100 keV, rather than X-rays from the auroral electrons, played a dominant role in the transient and deep (65–70 km) mesospheric ionization during the observed auroral breakup.


Introduction
Significant ionization of the mesosphere has long been documented during substorms. One of the most prominent phenomena is found near the onset and is referred to as "absorption spikes, " which are observed by groundbased riometers (Parthasarathy and Berkey 1965;Nielsen and Axford 1977). The absorption spike events are found in a vast majority of substorms (Hargreaves et al. 1997;Aminaei et al. 2006;Spanswick et al. 2006) and are often associated with energetic electron bursts in the inner magnetosphere (Baker et al. 1981;Kremser et al. 1982;1988). In pulsating auroras, the mesospheric ionization is due to the precipitation of energetic electrons of a few hundred keV (Brown et al. 1976;Sandahl et al. 1980;Miyoshi et al. 2015;Oyama et al. 2017). Atmospheric ionization that results from this energetic particle precipitation in the mesosphere and lower thermosphere plays an essential role in linking the space and terrestrial environments via the generation of odd nitrogen and odd hydrogen (Codrescu et al. 1997;Daae et al. 2012;Isono et al. 2014;Seppala et al. 2015;Tomikawa 2017), and, thus, it is important to quantitatively evaluate the fundamental factors that lead to such deep ionization scenarios to understand its mechanisms and characteristics better.

Open Access
*Correspondence: kataoka.ryuho@nipr.ac.jp 1 National Institute of Polar Research, 10-3 Midori-cho, Tachikawa, Tokyo 190-8518, Japan Full list of author information is available at the end of the article In 2017, the Arase satellite began full in situ observation of energetic particles and electromagnetic waves in the inner magnetosphere in an effort to understand the dynamics of radiation belts and the inner magnetosphere better . The energetic particles precipitating from the inner magnetosphere deposit their energy in the atmosphere, and this is improving our understanding of the associated atmospheric response, which has also been an important objective of the Arase satellite project.
The Program of the Antarctic Syowa Mesosphere Stratosphere Troposphere Incoherent Scatter Radar (PANSY) at Syowa Station, in the Antarctic (− 69.00°S, 39.58°E; − 70.46° magnetic latitude) has been in operation continuously since September 2015 at an observation frequency of 47 MHz (Sato et al. 2014). It is accompanied by a suite of important ground-based instruments, including a high-speed auroral imager, a meridian scanning photometer, and imaging riometer. The PANSY radar system detects polar mesosphere winter echoes, which are caused by neutral turbulence in the weakly ionized atmosphere at an altitude of 55-85 km (Czechowsky et al. 1989;Brattli et al. 2006;Lübken et al. 2007). Recent observations revealed that the sudden intensification of mesospheric echoes around 65 km was caused by electron density enhancements in the D region during a solar proton event (Nishiyama et al. 2018), implying that intensified mesospheric echoes may serve as a proxy of electron density variations at mesospheric altitudes.
There have been several occasions when the ionospheric footprint of the Arase satellite has passed through the field of view of the all-sky auroral imagers in operation at Syowa Station. As expected, the Arase satellite, PANSY radar, and other ground-based conjugate instruments simultaneously observed an isolated substorm at Syowa Station on June 30, 2017. The purpose of this study is to report on a rare comprehensive data set from these cutting-edge conjugate observation instruments and discuss the transient mesospheric ionization associated with auroral breakup, combined with a general-purpose Monte Carlo particle transport simulation code particle and heavy ion transport code system (PHITS). In this study, the EGS5 mode (Hirayama et al. 2005) in PHITS version 3.02 ) was employed. Figure 1 shows the data from the meridian scanning photometer at Syowa Station for the 2-h time interval from 2200 to 2400 UT on June 30, 2017. Soft electrons of a few keV dominantly precipitated during the auroral breakup from 2220 to 2226 UT, which was identified from the relatively high auroral intensity of 845 nm compared to 670 nm (Ono 1993;Ono and Morishima 1994). After 2230 UT, the overall auroras became diffuse and began pulsating, and the characteristic energy of the electrons increased, which was also identified from the 845/670 ratio. High-speed auroral imagers clearly captured the rapid complex motion of the bright discrete auroras with a tall curtain-like shape at the auroral breakup until 2226 UT, and faint pulsating auroras were observed thereafter (Additional file 1: Movie A1). Figure 2 shows the backscattered echo power over time and height sections for the 2-h time interval from 2200 to 2400 UT on June 30, 2017, based on the PANSY radar observations. Atmospheric ionization can be sensitively detected based on the relative intensification of echo power, although a quantitative estimation of the ionization rate or electron density from the echo power itself is rather difficult. As expected, a layer at an altitude of ~ 80 km was ionized during the post-breakup pulsating aurora activity from 2230 to 2330 UT. The ionization source can be attributed to the precipitating electrons of > 10 keV (see Miyoshi et al. 2015), which is also supported by the relatively low 845/670 ratio depicted in Fig. 1. The echo altitude gradually descended, which was likely due to the downward propagation of the neutral turbulence at that time. It is also important to note that intermittent and transient echoes above 80 km during the 2-h interval, which did not coincide in five beam directions, were largely caused by meteors (Nishiyama et al. 2015).

Results
Interestingly, the echo was also enhanced at an altitude of 65-70 km as the auroral breakup began from 2220 to 2226 UT, regardless of the dominant presence of lowenergy auroral electrons. It is not possible to rule out the possible role of the direct injection of > 100 keV energetic electrons, even if the auroral electrons were predominantly soft, because the flux of such energetic electrons would be significantly lower than the auroral electrons, and their impact on the auroral emission is negligible.
More specifically, the echo enhancements at altitudes < 70 km were confirmed at only two successive time intervals, specifically 22:21:26-22:22:47 UT and 22:24:44-22:26:05 UT. In the next time frame of 22:28:02-22:29:23 UT, the echo enhancement disappeared at altitudes < 70 km. The signals were clearer at the zenith, and in south/east beams (i.e., poleward beams) compared with those associated with the north/west beams, where auroral emission is also intensified, which was observed in all-sky images (Additional file 1: Movie A1). The altitude range of the transient echo enhancement was found to be broader than the narrow echo band noted during the pulsating auroras.
The imaging riometer at Syowa Station observed cosmic noise absorption during the auroral breakup, thus providing additional evidence to support the PANSY radar observations associated with the mesospheric North-south intensity profiles of the hard-electron aurora at 670 nm and the soft-electron aurora at 845 nm as observed by the meridian scanning photometer at Syowa Station (top two panels). A single north (0°) to south (180°, poleward) scan takes 10 s. The middle panel shows the ratio between the intensities of 670 nm and 845 nm. The bottom two panels show the estimated characteristic energy and total energy flux of auroral electrons as estimated from the two wavelengths (Ono 1993;Ono and Morishima 1994) ionization (Fig. 3). Relatively weak ionizations were also detected during the 1.5-h time period of pulsating auroras. The bottom two panels of Fig. 3 show the in situ observation of energetic particles made by the highenergy electron experiments (HEP)  and extremely high-energy electron experiment (XEP) (Higashio et al. 2018) onboard the Arase satellite. There is a clear signature of rapid flux enhancement of several hundred keV electrons observed by the HEP and XEP instruments that match the timing of the spike identified in the riometer data. Therefore, the flux enhancement of energetic electrons is likely to be one of major causes of the transient mesospheric ionization if these electrons precipitate into the atmosphere. The possibility of the flux enhancement of energetic electrons is quantitatively discussed later in this paper. Figure 4 presents the in situ observations of the plasma waves detected by the high-frequency analyzer (HFA) and onboard frequency analyzer (OFA) of the plasma wave experiment (PWE) onboard the Arase satellite (Kasaba et al. 2017;Kumamoto et al. 2018;Matsuda et al. 2018;Ozaki et al. 2018) for the same 2-h time interval from 2200 to 2400 UT on June 30, 2017. The yellow and magenta lines represent the electron-gyro frequency and half-gyro frequency, respectively. For the 2-h period, the footprint of the Arase satellite passed through the eastern side of the field of view of the all-sky cameras from low latitude to high latitude. It is important to note that possible errors associated with the footprint mapping by the IGRF model can be as much as a few degrees, as evaluated from the T96 (Tsyganenko 1996) and T04s (Tsyganenko and Sitnov 2005) models during this event (Additional file 2: Figure A1).
Broadband noise was observed during the auroral breakup from ~ 2220 to 2230 UT, while continuum emissions from the plasmapause were observed during the pulsating auroras from ~ 2230 to 2400 UT. Note that the Arase satellite traversed across the plasmapause at L = 6.6 at ~ 2150 UT, which was also detected by HFA (not shown), implying that this event was observed just outside the plasmapause. The Arase satellite did not observe electromagnetic ion cyclotron waves nor chorus waves during the 2-h event.

Discussions
The mesospheric ionization is quantitatively evaluated using PHITS ). The following hypothesis is proposed as an explanation for the transient ionization of the mesosphere. Note that X-rays are emitted from the soft auroral electrons of < 10 keV by bremsstrahlung emission at an altitude of ~ 100 km, and these X-rays can penetrate deep into the mesosphere and deposit their energy. Altitude profiles of the local auroral energy deposition due to electrons and X-rays can be calculated by well-established methods (Rees 1964;Berger and Seltzer 1972;Luhmann 1977). In this paper, we use PHITS  to simulate three ionization models, including the auroral breakup model A (soft electrons and high flux), the pulsating aurora model B (hard electrons and low flux), and the energetic electron model C. In this study, the MSISE-90 model (Hedin 1991) was employed for simulating the neutral atmosphere over Syowa Station at 2200 UT on June 30, 2017.
The ratio of 845/670, apparent tall curtain shape, rapid motion, and timing near the substorm onset led to the assumption of relatively broadband electron spectra, which is represented by the wave aurora (Mende et al. 2003) or broadband electrons often associated with broadband noise (Newell et al. 2009). Therefore, for model A, we assumed a Gaussian energy spectrum with characteristic energy E c of 2.5 keV (Strickland et al. 1993) and a relatively  Fig. 2 Altitude profiles of the echo power enhancement as observed by the PANSY radar, Syowa Station. From top to bottom, the directions of the five beams are the zenith, north, east, south, and west, respectively, at a zenith angle of 10°. The beam width is ~ 1°. The altitude range resolution is 0.6 km, and the temporal resolution is ~ 4 min wide Gaussian width W of 0.75 E c . The pitch-angle distribution was assumed to be a beam with a 30° width from the center, where the differential flux can be expressed by Eq. (1). (1) where Q denotes the total energy flux.
On the other hand, pulsating auroras typically have a harder spectrum, which is also supported by the 845/670 ratio (Fig. 2). For model B, an energy spectrum of a kappa distribution (Frahm et al. 1997) was assumed with a peak energy E p of 1 keV and a kappa parameter κ = 8, considering a high-energy tail in addition to Maxwell distribution (Strickland et al. 1993). The pitch angle distribution was assumed to be isotropic in the downward hemisphere, and the differential flux can be expressed by Eq. (2).
(  Figure 5 shows the results obtained from PHITS simulation for models A and B. The ionization levels for a typical total energy flux of 1 mW m −2 are shown for model B, and those for 10 mW m −2 are shown for model A. Direct ionization by electrons with ~ 100 keV energy can be seen at 80 km altitude for model B. Note that X-ray ionization of model A is comparable to or even higher than electron ionization of model B at an altitude of 65-80 km. As a useful reference, some other simulation results are also presented in the Additional file 3 ( Figure A2) with different parameters.
Then, the resultant riometer response was calculated to quantitatively test the above hypothesis. We calculated the cosmic noise absorption by the formulation of 4.6 × 10 −5 νω −2 q 0.5 a −0.5 dB m −1 , where q is the ion pair production rate. The recombination coefficient a was obtained from Gledhill (1986), and the collision frequency ν was obtained from Aggarwal et al. (1979). The observation frequency ω of the riometer was 38.2 MHz. The height integrated intensities of cosmic noise absorption were estimated to be only 0.016 dB and 0.005 dB for models A and B, respectively, which are an order of magnitude smaller than the observed values, as shown in Fig. 3. From the above results, we expect an order of magnitude larger ionization than those of the soft and hard auroral electrons to take place in the mesosphere during the event investigated in this study.
Here we introduce model C to evaluate the possible role of precipitating energetic electrons. The differential flux of the power law distribution can be expressed using Eq. (3), where Q denotes the approximate downward total energy flux. For simplicity, we assumed a lower energy cutoff E 0 of 100 keV, and set the maximum energy cutoff at 1000 keV. To roughly fit the omnidirectional electron flux of HEP and XEP, the possible maximum total energy flux Q of 0.3 mW m −2 was assumed for three different power law indices γ of 3.5, 4.0, and 4.5.
The result of the PHITS simulation for model C is represented by the black lines in Fig. 5b. Comparing model C to models A and B, ionization that was 2-3 orders of magnitude larger was obtained at an altitude of 50-80 km. The integrated cosmic noise absorption intensity of model C was 0.053 dB, which represents a more reasonable value to explain the riometer observations. Additional contributions from the omitted populations of less energetic electrons < 100 keV, which cannot reach to a 65 km altitude, can fill the gap of simulated and observed cosmic noise absorption intensities. Based on the above PHITS simulations, we concluded that the precipitation of energetic electrons of > 100 keV, rather than X-rays from the auroral electrons, played a dominant role in the transient and deep mesospheric ionization during the auroral breakup. The possible cause of the precipitation of energetic electrons is discussed as follows. The possible pitch angle scattering of energetic electrons > 100 keV from the inner magnetosphere is generally attributed to either electromagnetic ion cyclotron waves or chorus waves (e.g., Miyoshi et al. 2008Miyoshi et al. , 2015. However, the Arase satellite did not observe electromagnetic ion cyclotron waves or chorus waves. In general, electromagnetic ion cyclotron waves should be found in the equatorial region. By contrast, the resonance energy of chorus waves depends on the off-equatorial magnetic latitude in the magnetosphere, and even MeV electrons can precipitate into the atmosphere when the chorus waves propagate at higher latitudes (Horne and Thorne 2003; Miyoshi et al. 2015). It is also possible that during this particular event the Arase satellite was not in the correct position to observe the chorus wave activity. Meredith et al. (2012) conducted a statistical survey of chorus waves using multiple satellites and found that between magnetic latitudes of 15°-30° on the nightside, chorus intensities drop off significantly compared to those in the equatorial region. Chorus waves that reach higher latitudes were found to be restricted to dayside. Thus, it is possible that the Arase satellite was outside of the region where one would expect to see the chorus waves or electromagnetic ion cyclotron waves.
Finally, we briefly discuss the transient timescale of the appearance and disappearance of the echo, as shown in Fig. 2. Note that both the horizontal and vertical appearance and disappearance of radar echoes in the wintertime mesosphere have been known to depend on a complex interplay of both ionizing sources and possibly turbulence, in addition to nonionizing radiation, such as daylight, ion-chemistry and the charging behavior of dust or smoke layers (Kirkwood 2007;Lübken et al. 2006;Kirkwood et al. 2015;Nishiyama et al. 2018). It is important to note that during the present spike event the deep ionization disappeared rather quickly. At a nighttime altitude of 65 km, the recombination rate was 10 −9 m 3 s −1 (Friedrich et al. 2004), and the electron density from the results of model C was on the order of 10 7 m −3 . Therefore, the timescale is ~ 10 2 s, which is consistent with the spike appearance. However, the estimated electron density at 65 km was not large enough by itself to generate detectable mesospheric echoes (Lübken et al. 2006;Nishiyama et al. 2018). Therefore, the observed transient echo seems to have been caused by ionization due to electron precipitations and strong neutral turbulence and/or meteor smoke particles. Unfortunately, we have no mesospheric wind information during this period, and a vertical wind profile behind the echo layer cannot be examined. Ion-chemistry, including charging behavior of meteor smoke particles, also plays an important role in forming mesospheric echo layers in the VHF range. Electron density variations related to dust and smoke particles, especially during geomagnetically disturbed periods, should be taken into account in future research efforts.

Conclusions
The transient mesospheric ionization was identified at an altitude of around 65-70 km over Syowa Station during the auroral breakup that occurred from 2220 to 2226 UT on June 30, 2017, when the footprint of the Arase satellite was located close to Syowa Station. Instruments onboard the Arase satellite observed the rapid flux enhancement of several hundred keV electrons at the same time. Such energetic electrons, which represent a different population from auroral electrons, precipitated to transiently ionize the deep mesosphere during the auroral breakup.