Long-lasting injection of solar energetic electrons into the heliosphere
Key Words.:solar energetic particle event, shock acceleration, electron acceleration
- coronal mass ejection
- interplanetary coronal mass ejection
- ground level enhancement
- solar energetic particle
- stream interaction region
- corotating interaction region
- graduated cylindrical shell
- STEREO A
- STEREO B
- active region
Context:The main sources of solar energetic particle (SEP) events are solar flares and shocks driven by coronal mass ejections. While it is generally accepted that energetic protons can be accelerated by shocks, whether or not these shocks can also efficiently accelerate solar energetic electrons is still debated.
In this study we present observations of the extremely widespread SEP event of 26 Dec 2013. To the knowledge of the authors, this is the widest longitudinal SEP distribution ever observed together with unusually long-lasting energetic electron anisotropies at all observer positions. Further striking features of the event are long-lasting SEP intensity increases, two distinct SEP components with the second component mainly consisting of high-energy particles, a complex associated coronal activity including a pronounced signature of a shock in radio type-II observations, and the interaction of two CMEs early in the event.
Aims:The observations require a prolonged injection scenario not only for protons but also for electrons. We therefore analyze the data comprehensively to characterize the possible role of the shock for the electron event.
Methods:Remote-sensing observations of the complex solar activity are combined with in-situ measurements of the particle event. We also apply a Graduated Cylindrical Shell (GCS) model to the coronagraph observations of the two associated CMEs to analyze their interaction.
Results:We find that the shock alone is likely not responsible for this extremely wide SEP event. Therefore we propose a scenario of trapped energetic particles inside the CME-CME interaction region which undergo further acceleration due to the shock propagating through this region, stochastic acceleration, or ongoing reconnection processes inside the interaction region. The origin of the second component of the SEP event is likely caused by a sudden opening of the particle trap.
SEP events mainly consist of electrons and protons with small amounts of heavier ions.
Two phenomena are considered to be the main accelerators of these particles: Solar flares and shocks driven by CMEs.
The widely-used classification by Reames (1999) distinguishes between impulsive, that is, flare accelerated events, and gradual, that is, shock-associated events.
Impulsive events show enrichments of He, heavy elements such as Fe, and electrons, while gradual events are proton- and ion-rich showing compositions more similar to coronal and solar wind material.
Naturally, the extended shock front builds a larger acceleration region producing a larger SEP spread in the inner heliosphere.
However, a significant number of electron events with longitudinal spreads much larger than the expected widths of impulsive events ( degrees, Reames, 1999) have been observed (Wibberenz & Cane, 2006).
Thanks to the two STEREO spacecraft it was possible to detect so-called widespread events (Dresing et al., 2012; Lario et al., 2013; Dresing et al., 2014; Gómez-Herrero et al., 2015) where SEPs including electrons were distributed all around the Sun.
Also, unexpectedly wide He events were observed with the STEREO spacecraft (Wiedenbeck et al., 2013).
The important processes for these extraordinarily wide particle spreads are a matter of debate.
Some authors have suggested that strong perpendicular diffusion might explain the wide particle spreads (Dresing et al., 2012; Dröge et al., 2014, 2016; Strauss et al., 2017). Others suspect a large shock to be the driver (Richardson et al., 2014; Lario et al., 2014). Especially in the case of electrons it is not clear if a shock is able to efficiently accelerate electrons during SEP events providing an extended source region. No in-situ measurements close to the acceleration sites near the Sun exist and the likely mixing of flare accelerated and possibly shock accelerated electrons makes it very difficult to distinguish the populations when measured at spacecraft. While interplanetary shocks, which can be measured in-situ at a spacecraft, have shown to be very inefficient in accelerating electrons ( for keV electrons (Tsurutani & Lin, 1985; Dresing et al., 2016), this might be different close to the Sun where the Alfvén speed is larger (Mann et al., 1999; Gopalswamy et al., 2001) and a greater turbulence and seed-particle population might be present. Especially preceding CMEs can have a preconditioning effect on the ambient medium and have been found to be connected with higher fluxes of SEP events (Lugaz et al., 2017).
The existence of galactic cosmic ray electrons suggests that shocks are capable of accelerating electrons to high energies (Aharonian et al., 2005). Furthermore, the observation of ”herringbone” structures in type-II radio bursts during solar events indicates the presence of shock-accelerated electrons of at least a few tens of keV (Mann & Klassen, 2005). However, whether or not coronal or CME-driven shocks can accelerate electrons up to hundreds of keV or a few MeV, such as what is observed during SEP events, is still uncertain. On the one hand, Haggerty & Roelof (2002) explain the onset delays between the inferred injection time of near-relativistic electrons and the associated flare/type III times with the shock being the electron source. On the other hand, Maia & Pick (2004) suggest that an ongoing reconnection process in the corona driven by the uplifting CME accelerates the electrons and may account for the delayed onsets. Although some authors argue that coronal shocks (Klassen et al., 2002) or standing fast mode shocks involved in flares (Mann et al., 2009), are capable of accelerating electrons up to relativistic energies, an unambiguous distinction between shock acceleration or particle release due to the shock has not yet been made. These delays could, however, also be caused by turbulence and its resulting transport effects, such as scattering or an effective increase of the connecting field line length.
Detailed case studies of extreme events, like widespread events, may help to disentangle the main physical processes involved. The event presented in this study is a widespread electron (and proton) event which was observed on 26 Dec 2013 by the two STEREO spacecraft, separated by 59 degrees at that time, and by the L1-spacecraft SOHO, ACE, and Wind, separated by 150 and 151 degrees with respect to STEREO A (STA) STEREO B (STB), respectively. The event is associated with complex solar activity and a prominent type-II radio burst extending far into the interplanetary (IP) medium. The electron intensities show long-lasting rising phases and unusually long-lasting anisotropies observed at all three positions. The onset delays between the spacecraft are small, even at L1, being poorly connected to the solar event. However, all energetic particle onsets at the three observers show a delay of at least 30 minutes with respect to the assumed flare injection time. A long-lasting and spatially extended injection is therefore required to explain the observations. Although the CME-driven shock seems to be a good candidate, we carefully analyze the event and find that the shock alone does not explain all the characteristics of the event; a complex scenario also involving particle trapping is more likely.
In Sect. 3 we first discuss the remote-sensing observations of the associated complex solar event including the interaction of two CMEs (Sect. 3.1.1). Then we present the energetic particle observations at all three observers (Sect. 3.2.1) followed by the interplanetary context (Sect. 3.2.2). Afterwards we determine the longitudinal spread of the analyzed event in Sect. 3.3 and finally we discuss the possible source of the SEP event in Sect. 4.
The SEP observations at the STEREO spacecraft are provided by the instruments of the IMPACT suite (Luhmann et al., 2007): the High Energy Telescope (HET, von Rosenvinge et al., 2008), the Low Energy Telescope (LET, Mewaldt et al., 2007), and the Solar Electron and Proton Telescope (SEPT, Müller-Mellin et al., 2008). Directional SEP measurements can be obtained only from LET (protons from 1.8-15 MeV) and SEPT (electrons from 30-400 keV, protons from 60-7000 keV). Solar wind plasma and magnetic field observations are provided by the PLASTIC (Galvin et al., 2008) and MAG (Acuña et al., 2007) instruments, respectively. The SECCHI instrument suite (Howard et al., 2008) contains the remote-sensing instrumentation of the STEREO spacecraft. Observations of the coronagraphs COR1 and COR2, as well as images of the EUV cameras (EUVI, Wuelser, 2004) were used in this study. Radio observations are provided by the SWAVES instruments (Bougeret et al., 2008). The STEREO observations of this study are complemented by observations from the Earth’s point of view: We use energetic particle measurements taken by the EPHIN instrument (Müller-Mellin et al., 1995) aboard SOHO, by EPAM (Gold et al., 1998) aboard ACE and by the 3DP detector (Lin et al., 1995) aboard Wind. Solar wind plasma data at the Lagrangian point L1 were provided by ACE/SWEPAM (McComas et al., 1998). The WAVES instrument aboard WIND (Bougeret et al., 1995) provides radio measurements at this position. EUV imaging from SDO’s AIA instrument (Lemen et al., 2012) and the coronagraph LASCO (Brueckner et al., 1995) aboard SOHO complete the 360 degree remote-sensing set.
3.1 Remote-sensing observations
|2:10||filament eruption (region #1) associated to CME1||STA, STB /EUVI|
|2:30||CME1 at 1 R||STA, STB|
|2:35||CME1 first appearance||STA/COR1|
|2:41 - 4:45||main flare (region #2)||STA, STB|
|2:47 - 2:55||1st (faint) type II burst||STB/SWAVES|
|2:47||filament eruption (region #3) and interaction with northern CH||STB/EUVI|
|2:50||impulsive phase of main flare||STA, STB /EUVI|
|2:53||partly occulted type III burst||STA, STB, Wind|
|2:55||partly occulted type III burst||STA, STB, Wind|
|3:02 - 3:12||2nd (main) type II burst||STB, (STA)|
|3:02 - 3:09||main type III bursts||STA, STB, (Wind/WAVES)|
|3:05||CME2 first appearance at 2.7 R||STA/COR1|
|3:15 - 4:17||drifting type II-like structure (see Fig. 4)||STB/SWAVES|
|4:00 - 4:30||flare in region #4||STB/EUVI|
|3:25||latest possible 55-105 keV electron injection (according to|
|onset time assuming a nominal travel time along the Parker spiral)|
|3:45 1min||55-105 keV electron onset||STA/SEPT|
|3:53 5min||55-105 keV electron onset||STB/SEPT|
|4:10 15min||62-103 keV electron onset||ACE/EPAM|
|4:07 15min||0.7-2.8 MeV electron onset||STA/HET|
|4:22 15min||0.7-2.8 MeV electron onset||STB/HET|
|4:30 20min||0.7-3.0 MeV electron onset||SOHO/EPHIN|
|6:45||type III burst||Wind/WAVES|
|6:53||C2.2 flare at W28||GOES, SDO|
|7:05||CME3 first appearance||STA/COR1|
|7:15 - 8:30||SEP onsets of the second component||STA, STB /HET, SOHO/EPHIN,ERNE|
The solar event on 26 Dec 2013 is a very complex event involving four active regions of distinct activity at the Sun (numbered from one to four in Fig. 1 (a)).
In total it has a spatial extent of more than 60 degrees in latitude and longitude centered at the central meridian as seen from STB.
This extent makes it difficult to provide a single coordinate which can be assumed to be the injection site of the flare-accelerated SEPs.
Figure 2 shows the longitudinal constellation of the two STEREO spacecraft (STB in blue and STA in red) and the Earth (green) on 26 Dec 2013.
The gray sector indicates the 60-degree-wide longitudinal range magnetically connected to the suspected solar source region longitudes according to the region of activity involved at the Sun.
While STA is magnetically connected to that sector, the magnetic footpoint of STB lies outside towards western longitudes.
The Earth is situated and connected to the backside of this sector of activity at the Sun.
Figure 1 (a) shows an extreme ultraviolet (EUV) base difference image at 195 Å observed by STB at 4:27 UT. The different numbers indicate the ARs ordered by the sequence of activity. A time-line of the coronal events can be found in Table 1 which is also illustrated in Fig. 3. At 2:10 UT STB (and STA) observe a filament eruption in region #1 towards south/west. This filament eruption is associated to a CME listed in the LASCO catalog
The main event is a large two-ribbon flare in region #2, located at E17S12 as seen from STB point of view (cf. Fig. 1 (a)) and starting at 2:41 UT, which likely triggers the later activity in regions #3 and #4. According to the statistical relation found by Nitta et al. (2013), we estimate the GOES class of this flare as M7 utilizing the disk-integrated emission change in 195 Å observed by STB (not shown here). The flare is accompanied by an EIT wave propagating mainly towards west and south (not shown here). The propagation towards north is blocked by a coronal hole (CH) and towards east by an AR.
This event is also accompanied by another large CME listed in the LASCO catalog as a halo CME with a speed of 1336 km/s. Based on the propagation direction of the EIT wave the CME is likely deflected towards south/west making an interaction with CME1 likely (see Sect. 3.1.1). Figure 1 (c) shows the observations of this CME (henceforth CME2) from the Earth’s point of view. CME1 (marked by arrows in Fig. 1 (c)) clearly posed an obstacle for CME2 which is reflected in the disturbed front of CME2, suggesting interaction between both CMEs. A discussion on that interaction follows in Sect. 3.1.1.
The activity in region #2 triggers region #3, close to the northern polar CH, where a plasma loop moving towards the northern CH leads to an enlargement of the CH at its southern boundary. A filament eruption is then observed at 2:47 UT in region #3. The loop connecting regions #2 and #3 is transequatorial. Finally, a small flare in region #4 (cf. Fig. 1 (a)) is observed from 4:00 to 4:30 UT. The 195 Å observations at STB reveal that there is a connection between the main event in region #2 and the flare in region #4.
The solar event was associated with type-II and type-III radio bursts best observed by STB/SWAVES. Two short-lived type-II bursts, signatures of shock waves in the corona, appeared at frequencies above 10 MHz. We note that no ground-based radio signals at higher frequencies were recorded due to the backside location of the event as seen from Earth. The first (faint) type-II burst occurred between 02:47 and 2:55 UT followed by partly occulted type-III radio bursts (see Fig. 4 showing the radio observations by STB during the early phase of the event). The second main type-II burst appeared between 3:02 and 3:12 UT, a few minutes after the first one ceased, and simultaneously with the main group of type-III bursts, which were observed by STA and STB but not at Wind.
Figure 5 shows the dynamic radio spectra of the whole day of 26 Dec as detected on board STA (top), STB (middle, y-axis reversed), and Wind (bottom). While STB observed the high-frequency part coming from deeper regions in the corona, this part is slightly occulted for STA and all of the radio type-III bursts are only barely visible at Wind.
Immediately after the disappearance of the second type-II burst at 03:12 UT, a broadband structure appears between 3:13 and 4:15 UT with clearly drifting high- and low-frequency cutoffs (indicated by white dashed lines) as shown in the dynamic spectrum in Fig. 4. This whole structure, filled with type-III-like radio bursts, slowly drifts to lower frequencies with a drift rate comparable to the second type-II burst. The drifting cutoffs are not harmonically related, suggesting that the low-frequency border could be a new propagating disturbance starting essentially higher in the corona than the second type-II burst and its driver. As discussed in the following section this drifting type-II-like structure could be related to the interaction of the two associated CMEs. Later (see Fig. 5), after 4:15 UT, the type-II burst extends far into the interplanetary medium until 14 UT and frequencies 0.2 MHz and shows a clumpy broadband frequency structure implying an extended shock front occupying a broad range of densities during its propagation.
The Cme interaction
Figure 6 shows the result of the Graduated Cylindrical Shell (GCS) model (Thernisien et al., 2006, 2009), applied to the multi-spacecraft coronagraph observations from STB (left), SOHO (center), and STA (right) between 3:15 and 3:21 UT. The top panel shows difference images and the bottom panel shows the same with the reconstructed CME structures overplotted, where the green flux rope represents CME1 and the orange one CME2. Based on the GCS reconstruction we derive that both CMEs propagate in the same direction (CME1: E140S25 and CME2: E150S15 - uncertainties are within 10 degrees for longitude and latitude) which makes an interaction between their apexes highly likely. To derive the time and height of the apex interaction, we present in Fig. 7 the reconstructed CME fronts as function of their angular width (nb: the tilt, which is off by 80 degrees between the two flux ropes, is not taken into account) and marked the interaction with a red flash. From the model results we find that the apexes interact shortly after 3:15 UT at a height of 4 R. Although Fig. 7 suggests an interaction of the flanks before 3:05 UT at 3 R we note that the tilt between the two flux ropes of about 80 degrees makes the flank interaction rather unlikely. The radio observations, discussed in Sect. 3.1, show a drifting type-II-like structure with clear high- and low-frequency cutoffs containing type-III-like beams between 3:15 and 4:17 UT. The heights of these cutoffs are 2.1 and 4.7 R (using a 1x Saito model, Saito et al., 1970) making it likely that these radio signatures are caused by the interaction of the CME apexes. Furthermore, the type-III-like beams inside this drifting structure could be the signature of accelerated electrons between the two magnetic structures approaching each other. The two type-II radio bursts during the beginning of the event ( 3:15 UT) indicate the presence of two distinct shocks. These shocks are likely associated to each of the two CMEs. We note that, while the flux ropes of the two CMEs are not able to penetrate through each other, the CME-driven shock of CME2 might pass through the slower CME1 (Vandas et al., 1997).
3.2 In-situ observations
Solar energetic particle observations
From top to bottom the energy increases showing near-relativistic electrons (55-105 keV) in the top left panel, and relativistic electrons (0.7-2,8 MeV and 2,8-4 MeV) below.
We note that in the following we always use the terms ‘near-relativistic’ and ‘relativistic’ electrons for the 55-105 keV and 0.7-2,8 MeV energy channels, respectively.
The top right panel shows 13-16 MeV protons, the middle panel 25-60 MeV and the bottom right panel 60 MeV and 50-60 MeV for the two STEREO spacecraft and SOHO/ERNE, respectively.
We note that the ACE/EPAM and SOHO/EPHIN electron observations have been multiplied by intercalibration factors of 1/1.3 and 1/13. and EPHIN proton observations with a factor of 1.1 following Lario et al. (2013).
Although the two STEREO spacecraft are separated by 59 degrees, the time series of the SEP intensities of electrons and protons are very similar at both spacecraft. Even at the position of Earth, longitudinally separated by 150 degrees to each of the STEREO spacecraft, clear energetic particle increases are observed up to relativistic energies ( MeV for electrons). However, as shown by the bottom right panel, no protons 60 MeV are observed close to Earth while protons in the highest available energy channel of 60-100 MeV are observed at both STEREO spacecraft. The time profiles at all energies show long rise times of several hours up to almost a day.
The dashed black lines mark the onset of the associated type-III radio burst at 3:02 UT. The onset times of the near-relativistic electrons (top left panel) are 3:45 UT (STA), 3:53 UT (STB), and 4:10 UT (ACE). Assuming that the onset of the type-III burst marks the injection of the SEPs at the Sun, these electrons arrive at the spacecraft with a delay of 23 (STA), 31 (STB), and 48 minutes (ACE), respectively, compared to the scatter-free transport along a nominal Parker spiral.
It is important to note, the onsets of the relativistic electrons are later than those of the near-relativistic electrons (see Table 1). The highest-energy particles shown in the bottom panels arrive much later. The near-relativistic electrons and low-energy protons show a more or less rapid rise followed by a more gradual increase. The higher the energy, the less prominent the first steep increase and the more prominent the second component which is accompanied by a new steepening of the intensity time series (see e.g., the middle panels). The gray shaded range marks the approximate time of this steepening, indicating the onset of this second component, between 7:15 UT and 8:30 UT, that is, 4 hours later than the first component. We note that the onset of this second component is uncertain (and might be even earlier) because of the first component masking the onset. Interestingly, the first component nearly vanishes at 60-100 MeV protons and 3-4 MeV electrons (bottom panels) and only an increase corresponding to the second component is observed at these high energies (bottom panels). Even the energetic particle observations at the spacecraft close to Earth tend to show this break, suggesting the global character of this phenomenon which is observed all around the Sun.
Figure 9 displays the sectored intensity measurements of near-relativistic electrons as observed by STA (left), STB (center), and Wind (right). The top panel of each plot shows the pitch-angle-dependent intensity distribution where the color-coding corresponds to the electron intensity. White areas denote pitch angle ranges which were not covered by the telescopes taking into account the opening angles of the apertures. The second panel shows the pitch angles of the centers of the available viewing directions, the third panel displays the corresponding intensity measured in those viewing directions, and the bottom panel shows the first order anisotropy index as computed from the above observations (see e.g., Dresing et al., 2014). Figure 9 clearly shows that all three spacecraft observed significant anisotropies. At the two STEREO spacecraft the anisotropy is larger during the first few hours of the event and then reduces to a lower level. At Wind, however, the pitch angle distribution is rather isotropic during the first phase of the event but becomes more anisotropic around 7 UT. The reason for this could either be different propagation conditions during the two phases or a change of the source size, meaning that it gets closer to the magnetic field line connecting to Wind. We note, however, that significant anisotropy can be even observed without a direct magnetic connection to the source region close to the Sun (e.g., Strauss et al., 2017).
Figure 9 also shows that the total anisotropic period in the near-relativistic electron event extends over many hours at all three viewpoints. STA observes at least 14 hours of anisotropic flux. Unfortunately, after this time ion contamination begins to alter the electron measurement, so that it is no longer possible to determine a reliable anisotropy. For STB the ion contamination sets in much later so that an anisotropy lasting for a period of over a day can be confirmed. Even at Wind, which is situated on the far side of the associated activity region at the Sun (cf. Fig. 2), significant anisotropy is observed over nearly 12 hours.