Microwave Study of a Solar Circular Ribbon Flare
Jeongwoo Lee, Stephen M. White, Xingyao Chen, Yao Chen, Hao Ning, Bo Li, Satoshi Masuda
aa r X i v : . [ a s t r o - ph . S R ] S e p Draft version September 28, 2020
Typeset using L A TEX default style in AASTeX63
Microwave Study of a Solar Circular Ribbon Flare
Jeongwoo Lee, Stephen M. White, Xingyao Chen, Yao Chen, Hao Ning, Bo Li, and Satoshi Masuda Institute for Space Weather Sciences, New Jersey Institute of Technology, University Heights, Newark, NJ, USA Space Vehicles Directorate, Air Force Research Laboratory, Albuquerque, NM, USA CAS Key Laboratory of Solar Activity, National Astronomical Observatories of Chinese Academy of Sciences, Beijing 100101, China Institute of Space Sciences, Shandong University, Weihai, Shandong 264209, China Institute for Space-Earth Environmental Research, Nagoya University, Aichi 464-8601, Japan
ABSTRACTA circular ribbon flare SOL2014-12-17T04:51 is studied using the 17/34 GHz maps from theNobeyama Radioheliograph (NoRH) along with (E)UV and magnetic data from the Solar Dynam-ics Observatory (SDO). We report the following three findings as important features of the microwaveCRF. (1) The first preflare activation comes in the form of a gradual increase of the 17 GHz fluxwithout a counterpart at 34 GHz, which indicates thermal preheating. The first sign of nonthermalactivity occurs in the form of stepwise flux increases at both 17 and 34 GHz about 4 min before theimpulsive phase. (2) Until the impulsive phase, the microwave emission over the entire active region isin a single polarization state matching the magnetic polarity of the surrounding fields. During and afterthe impulsive phase, the sign of the 17 GHz polarization state reverses in the core region, which impliesa magnetic breakout–type eruption in a fan-spine magnetic structure. (3) The 17 GHz flux around thetime of the eruption shows quasi-periodic variations with periods of 1–2 min. The pre-eruption oscil-lation is more obvious in total intensity at one end of the flare loop, and the post-eruption oscillation,more obvious in the polarized intensity at a region near the inner spine. We interpret this transitionas transfer of oscillatory power from kink mode oscillation to torsional Alfv´en waves propagating alongthe spine field after the eruption. We argue that these three processes are inter-related and indicate abreakout process in a fan-spine structure. INTRODUCTIONCircular ribbon flares (CRFs) occur in a special magnetic configuration where a central parasitic magnetic field issurrounded by closed ribbons with the opposite magnetic polarity, implying an overlying dome-shaped fan separatrix(Masson et al. 2009, Sun et al. 2013). The implied fan-spine configuration has motivated solar MHD theorists tochallenge the observed CRF phenomenologies (Lau & Finn 1990, Schrijver & Title 2002, T¨or¨ok et al. 2009, Pontin etal. 2013, Rickard & Titov 1996, Galsgaard & Nordlund 1997, Galsgaard et al. 2003, Pontin & Galsgaard 2007, Pontinet al. 2007, Pariat et al. 2009, 2010; Karpen et al. 2017; Wyper et al. 2016, 2017, 2018). On the other hand, thefirst observational study of CRFs was made using the TRACE 1600 ˚A UV continuum images of a confined C8.6 flare(Masson et al. 2009). Later H α blue-wing images obtained from the digitized films of Big Bear Solar Observatory(BBSO) were used to study five CRFs exhibiting jets (Wang & Liu 2012). Hard X-ray (HXR) observations withthe Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI) were used to study the high-energy electroncontent of a CRF (Reid et al. 2012). EUV observations with the Atmospheric Imaging Assembly (AIA) instrumenton board the Solar Dynamics Observatory (SDO), in combination with field-line extrapolation, suggested additionalideas such as hyperbolic flux tube reconnection (Masson et al. 2017), spine-fan reconnection (Liu et al. 2019) as wellas the late phase extreme-ultraviolet (EUV) phases (Woods et al. 2011) in a non-eruptive CRF (Masson et al. 2017)and hot spine loops and the nature of a late phase in terms of a cooling process (Sun et al. 2013). Eruptions of fluxropes embedded inside the CRFs have also been studied (Liu et al. 2013, 2019).It is often said that CRFs form an important class of solar flares because they imply truly three–dimensional (3D)magnetic reconnection. This 3D nature is clear in theory but may be harder to identify in observations. Since allflares are actually 3D, it is not sufficient simply to image a large–scale 3D structure around the reconnection point.The so-called standard solar flare model is understood within a two-dimensional (2D) framework because the ribbonmotion away from the magnetic polarity inversion line (PIL) can adequately be described by a 2D picture (Kopp &Pneuman 1976, Priest & Demoulin 1995, Demoulin et al. 1996). A good example is the famous Bastille day flarethat exhibited a visually impressive structure, but the resulting arcade of newly-formed loops can still be explainedby 2D reconnection. On the other hand, H α brightness running along the circular ribbon in a CRF indicates that thereconnection involves a structure beyond axial symmetry that is not reducible to 2D physics. Such a structure on thescale of quasiseparatrix layer (QSL) is yet unresolved, and more observational tools are needed in order to address thesmall-scale physics.This Letter presents the study of a CRF focusing on microwave emission. The value of microwave observationsas diagnostics of CRFs is an open question. The most commonly cited microwave diagnostic is that, in the case ofgyroresonance emission, the observing frequency divided by the effective harmonic number gives the field strength ofthe outmost, optically-thick region (Gary & Hurford 2004, Lee et al. 1993a). On the other hand, magnetic reconnectionstudies require information on magnetic topology rather than field strength. A shortcoming of microwave images inthis respect is that they do not show morphological details as readily as EUV images, which can play an essential role intracing field lines carrying significant density. We expect that a couple of other properties of microwaves will be usefulin this problem. One is the sensitivity of microwave radiation to energetic electrons: gyrosynchrotron radiation candetect small numbers of nonthermal electrons thanks to the presence of magnetic fields (Rybicki & Lightman 1979).Another diagnostic is polarization, which is closely related to the coronal magnetic polarity (Zheleznyakov,1970; Leeat al. 1993b; Zheleznyakov et al. 1996; Lee et al. 1998). These two properties may be utilized as a unique diagnostictool for exploring magnetic reconnection in the fan–spine structure. MORPHOLOGIES AT MICROWAVE AND (E)UV CHANNELSThe target we select for this study is the SOL2014-12-17T04:51 flare that occurred in NOAA active region (AR)12242 at heliographic coordinates S20E09. In this event, a striking circular ribbon structure is clearly visible in bothEUV and microwave images. This event has already attracted several studies, including a magnetic field analysis (Liuet al. 2019) and studies of quasi-periodic pulsations (QPP) in the range 1.2–2.0 GHz from the Mingantu SpectralRadioheliograph (Chen et al. 2019) and of the thermal structure (Lee et al. 2020). Here we focus on the microwavedata for the event, obtained at 1.0–9.4 GHz from the Nobeyama Radiopolarimeter (NoRP) and at 17/34 GHz fromthe Nobeyama Radioheliograph (NoRH). The set of imaging observations at both microwave and (E)UV wavelengthsoffers a rare opportunity for studying microwave properties of a CRF.Figure 1 shows the NoRH 17 GHz maps as contours at six different times. The background images are all different(E)UV channels at the corresponding times to give an idea of how the CRF appears at different wavelengths. In theaccompanying animation, only 94 ˚A and 131 ˚A are used as background showing EUV evolution as well as 17 GHzevolution. In the preflare phase (Fig. 1a–c) the circular structure of the active region is evident. (a) The emission ismostly confined within the circular region. Comparison of the 17 GHz and 94 ˚A images confirms that that microwaveemission also outlines the circular-shaped area. The structure inside the circle may be called an “anemone” structure.In the (E)UV channels, the 94 ˚A image shows a hemispheric structure suggestive of the dome-shaped quasi-separatrixlayer (QSL) postulated for CRFs, as does the outer spine structure at the western edge of the frame. (b) Close to theflare time, the local region in the north brightens, while the circular ribbon is more obvious in the south of the AR. Inthe background EUV images, the outer spine halo structure is best visible at 94 and 131 ˚A and less apparent in otherchannels, which means that it is hot and tenuous (Lee et al. 2020). (c) Near the onset of the impulsive phase, thebrightness is more concentrated in an elongated shape connected to the center of the AR where the strongest magneticfields are located. (d) During the flare, the 17 GHz emission is highly concentrated in that region, and the extendedsource appears to be a flare loop. (e) Limited dynamic range in the 17 GHz images makes the southern part of thecircular ribbon less prominent in the images when the flare is bright, but it comes back as the flare diminishes. The304 ˚A image shows the circular ribbons most clearly. (f) A long decay phase follows during which the EUV sourceexpands and also other areas on the fan surface are visible again at 17 GHz recovering the anemone structure. The1600 ˚A image shows the inner and outer flare ribbons where the deposition of the flare energy into the chromosphereis concentrated. They appear to be conjugate footpoints in view of the loop-like structure in the 17 GHz map (Fig.1d).Figure 2 shows contours of the 17 GHz polarized intensity V = R − L at the same times as in Figure 1, except for thelast panel. V >
V < I = R + L , is plotted in yellow contours overthe line-of-sight Helioseismic and Magnetic Imager (HMI) magnetograms. The total intensity was initially concentrated Figure 1.
EUV and microwave images of the CRF SOL2014-12-17T04:51 in NOAA AR 12242 at six different times: (a)–(c)preflare phase, (d) impulsive phase, and (e)–(f) postflare phase. Contours in the top panels are at [1.9, 7.0, 26,99]% of themaximum in each frame, and those in the bottom panels are at [0.5, 1.9, 7.0, 26,99]% of each maximum. Background grayscaleimages are AIA images in six different channels at the corresponding times. The animated NoRH 17 GHz and SDO/AIA imagesinclude the entire flare event, running from 04:10 to 05:10 UT, with the 17 GHz contour levels at [2, 10, 50, 95]% of the maximumin each frame. over the central sunspot with positive magnetic polarity, and with time expands eastward and also northward to forma loop-like structure in the impulsive phase (Fig. 2a–c). The absence of red contours in Figures 2a–c shows thatall sources are LHCP in the preflare phase (top panels), while the region over the positive-polarity sunspot becomesRHCP during the impulsive phase and remains so during decay (Figs. 2d–f). Since the northern sources lie abovethe negative-polarity region, their natural polarization (corresponding to the extraordinary mode) is expected to beLHCP. However, the central sources over the positive magnetic polarity sunspot should be RHCP (Ratcliffe 1959;Zhelznyakov 1970; Melrose 1975, 1985; Dulk 1985). We thus regard the initial LHCP over this region to be reversedfrom its nominal polarization, RHCP, in the preflare phase. This is a new phenomenon, perhaps unique to CRFs, andis likely to be associated with a drastic change in the fan-spine structure, which we discuss further below. CORRELATIONS WITH EUV AND SOFT X-RAYSIn Figure 3, we plot the time profiles of microwave fluxes from NoRP and NoRH along with the AIA EUV andGOES soft X-ray lightcurves. Figure 3 a shows the three local regions selected for investigation: the flare loop ( L ),the southern section of the circular ribbon ( R ), and the outer spine ( S ). The grayscale image is a 17 GHz map inlogarithmic scale down to the 1% level of the maximum intensity, and the contours are the 304 ˚A intensity also downto 1% of its maximum intensity. Note that both of them show that the edge of the enhanced microwave emissioncoincides with the circular ribbons as represented by the 304 ˚A intensity. Although microwave maps do not show the Figure 2.
Contours of the total and signed polarized intensities at 17 GHz plotted over the HMI line-of-sight magnetograms.The region shown is the main flare site, on the northern edge of the circular ribbon structure evident in Fig. 1. The yellowcontours are 17 GHz total intensity plotted at [10, 50, 100]% of its maximum at each time. The blue (red) contours representthe polarized intensity in LHCP (RHCP) in absolute levels, [10, 50, 100]% of ± circular ribbon itself, it can be inferred from the boundary of enhanced emission from the hotter plasma inside thefan. Thus any locally enhanced features at 17 GHz are superimposed on a faint background circular disk produced byhot and dense plasmas inside the fan surface.Figure 3 b shows the fluxes from L , which dominate over those from the other regions. The flux time profiles at 17GHz and five EUV lines at 94 ˚A, 131 ˚A, 193 ˚A, 211 ˚A, and 335 ˚A are all normalized to unity, since we are mainlyinterested in relative timing. The prominent feature in L is a strong impulsive 17 GHz peak followed by gradually-increasing EUV fluxes and a secondary peak in the 17 GHz flux at the time of the EUV maxima. The impulsive 17GHz peak is therefore attributed to nonthermal gyrosynchrotron emission by accelerated electrons. The coincidenceof the secondary 17 GHz peak at 04:57 UT ( t ) with the EUV maxima supports the idea that the gradual 17 GHzflux around t is thermal. Figures 3 c, d shows fluxes from R and S , respectively. The 17 GHz fluxes of these regionsare weaker so that the signal to noise ratio (SNR) is lower than in L . Some of the EUV fluxes from R show a stepwiseflux enhancement at t (04:19 UT), signaling the circular ribbon activation. Those fluxes remain enhanced for about9 min and start to rise again at 04:28 UT ( t ). The flux enhancement during t ≤ t ≤ t is mostly in the relativelylow-temperature passbands, 211 ˚A, 335 ˚A, and 304 ˚A as well as 193 ˚A, implying thermal emission at standard coronaltemperatures. In S (Figure 3 d ), all EUV fluxes rise at nearly constant rates with the 131 ˚A flux leading and the 211 ˚Aflux rising last. The different AIA channels typically have responses peaking at several temperatures, so interpretingthe order of the different AIA channels is not straightforward: e.g., 131 ˚A has both Fe VIII and Fe XXI, so the earlyrise in 131 ˚A could be either hot or cold material: for a flare, it makes more sense to assume a dominant hot component.It might be that the rising magnetic field lines carry hotter plasma and the subsequently cooler plasma follows from Figure 3.
Local EUV and 17 GHz fluxes as a function of time: (a) A 17 GHz image (greyscale) showing three regions ofinterest: the flare loop region ( L ), southern section of the circular ribbon ( R ), and the outer spine region ( S ). The red contoursare the AIA 304 ˚A intensity at the 10% level of its maximum. (b)–(d) show the normalized EUV fluxes and the NoRH 17GHz fluxes versus time, computed for the three regions, L , R , and S , respectively. t and t mark the times of stepwise fluxvariations in L, and t is the time of flare maximum at 17 GHz. (e) GOES soft X-ray lightcurves are compared with the NoRP(1.0-9.4 GHz) microwave flux time profiles. (f) The NoRH 17 and 34 GHz fluxes agree with each other on and after t (04:57UT). Time profiles of the GOES temperature and EM are also shown. Note that the 17/34 GHz fluxes, the EUV fluxes fromL, and GOES EM all simultaneously reach their local maxima at t . behind. The 17 GHz flux is consistent with this trend of EUV fluxes. Thus the results in Figure 3 b – d indicate thatthe local microwave fluxes match the behavior of their EUV counterparts when thermal emission dominates. Theimpulsive nonthermal 17 GHz emission is a tracer of chromospheric heating by precipitating nonthermal electrons,and the consequent rise in EUV emissions, increasing at the onset of nonthermal radiation and continuing to riseafterwards, is consistent with the well known Neupert effect (Neupert, 1968).Figure 3 e, f show the evolution of multiple microwave frequencies and soft X-rays. In Figure 3 e , the lower-frequency(non-imaging) NoRP fluxes show impulsive peaks concentrated around t as in the NoRH 17 GHz flux from L , butthe lower frequencies do not show the second peak at t . A plausible interpretation is that at these frequencies theradio emission is optically thick and therefore the flux represents the temporal evolution of effective temperature,whereas the optically thin emission at ≥
17 GHz traces the total emission measure of thermal electrons present in thesource. In line with this interpretation, the GOES soft X-ray fluxes start to rise at about t and show the fastestvariation, indicating the primary energy release, at t (Neupert 1968). It is also notable that the GOES SXR peaksoccur around t where the second maximum of microwave flux is observed. Microwave emission with this type of timeprofile consisting of an impulsive burst followed by a gradual burst at 17 and 34 GHz has been suggested to be asignature for compound flares (Ning et al. 2018). Figure 3 f shows the 17 and 34 GHz fluxes from L computed fromthe NoRH maps and the temperature and emission measure (EM) calculated from the GOES soft X-rays. The GOESEM peaks around t , while temperature reaches its maximum much earlier and decreases monotonically through t .The 34 GHz flux is lower than the 17 GHz flux all the way until around t , after which they are very similar. A flatspectrum is expected when the radiation is dominated by optically-thin thermal free-free emission (e.g., Dulk 1985),and the time profile is mostly dominated by the EM variation. The above finding of the EUV line fluxes simultaneouslyreaching their maxima at t (Fig. 3b) is also in line with the conclusion that the microwave emission during the secondmaximum at t is mainly a thermal free-free emission.Despite the lower SNR of the 17 GHz intensity in R and S , this comparison sheds some light on the relationshipbetween the local microwave emissions and the corresponding EUV emissions, which represent different thermal com-ponents. It appears that the NoRH 17 GHz fluxes may show spatio-temporal variations similar to those of the AIAEUV fluxes when the 17 GHz fluxes are dominated by thermal emission (e.g., at t and t ). When nonthermal emis-sion is dominant ( t and t ), the 17 GHz fluxes show impulsive behaviors but the EUV fluxes increase only gradually,reaching their maxima much later. Figure 4.
Activation of and pulsations in the flare loop. (a) Local regions, A–D, set for calculation of local intensities aremarked over the inverted NoRH 17 GHz map at t . (b) The 131 ˚A time-distance stackplot shows two eruption features asdenoted with the guidelines and speeds. The slit denoted in (a) is used to construct this stackplot. (c) The 17/34 GHz timeprofiles from D are plotted with three major transition times, t – t , marked by the vertical dotted lines. (d) and (f) showtime profiles of the total and the polarized 17 GHz intensities calculated from A–C, respectively, with arrows pointing to thetemporally local peaks. 4. FINE STRUCTURES IN THE FLARE LOOP ACTIVITIESWe investigate fine structures in the localized time variations of the microwave bursts by focusing on the strongestemission region L . As shown in Figure 4 a , we set four subregions marked on the inverted 17 GHz intensity map at t .A and C are presumably the loop footpoints conjugate to each other, and B is likely to be the looptop location. Dincludes all the three regions and thus the flare loop. To calculate local flux from each region, we add up all brightnesstemperatures over the region, and divide it by the number of pixels within the region. The quantities shown in Figure4 therefore correspond to spatially–averaged local brightness temperatures ( T b ). We also mark a slit (dashed purpleline) for constructing the time-distance stackplot of the 131 ˚A intensity displayed in Figure 4 b . The slit distance startsfrom the tip denoted as s = 0 and increases southward to extend over the total distance of 120 ′′ . In this time-distancestackplot, two eruption features are noticeable, although faint. Their speeds are estimated as ∼
300 km s − or higheras denoted by the guidelines. The first feature is likely to have started at t , suggesting that this eruption may berelated to energy release responsible for the initial ribbon activation. Start time of the second eruption feature couldbe either t or t , which we can hardly discern because of the bright features on the stackplot. In any case, we notethat there are indeed eruption-like EUV features corresponding to the CRF activation times, t , , or the most intenseenergy release time, t (cf. Liu et al. 2019).Figure 4 c shows the 17 GHz and 34 GHz T b calculated over D. The 17 GHz T b starts to rise at t , which we countedas the first activation time based on the EUV lightcurves (Figure 3). Note however that at this time there is nocorresponding increase in the 34 GHz T b . The second activation occurs only at t in the form of an impulsive increaseof both 17 GHz and 34 GHz T b . The two activations at t and t may indicate different levels of energy release, withthe latter being more intense. During the flare, both 17 GHz and 34 GHz T b impulsively rise at t . Shown in the resttwo panels are T b for the total (Fig. 4 d ) and polarized intensities (Fig. 4 e ) measured from the subregions, A–C. Themultiple peaks on these time profiles, as marked by the arrows, yield an impression of oscillations superimposed onthe flare lightcurve, very similar to the phenomenon called quasi-periodic pulsations (QPPs) as thoroughly studied byChen et al. (2019) for this event. The periodicity is not particularly clear and we would not claim the multiple peaksas QPPs, but instead call them quasi-oscillations. The quasi-oscillation in total intensity, I (Fig. 4 d ) is more obviousin source C, and less obvious in A. Namely, the farther from the inner spine, the more clearly the quasi-oscillation isvisible. The polarized intensity, V (Fig. 4 e ) show a similar behavior with those of I but with a few differences. Thequasi-oscillation of V appear not only in the preflare phase but continue and are stronger during the impulsive phase.Spatially, the quasi-oscillation of V are more obvious in A, whereas the quasi-oscillation of I are stronger in C.We were able to count up to five peaks in the lightcurves and measure the time separations between the adjacentpeaks. The mean and standard deviation come out as 1.3 ± ± t with a tendency that the former is more obvious in the start and thelatter lasts longer. As a comparison, Chen et al. (2019) reported 2 min QPPs in the frequency range 1.2–2.0 GHzaround the flaring region during the impulsive phase, 3 min EUV QPPs along the circular ribbon during the preflarephase, and 2 min UV QPPs near the center of the active region from the preflare phase to the impulsive phase (04:00to 04:45 UT). In terms of the spatial location, the 17 GHz oscillation power residing in the flare loop is an almostidentical result with those of the QPP sources at 2 GHz and UV channels (Chen et al. 2019), except that we used ahigher resolution to resolve the loop structure. The EUV QPP are found in the circular ribbons of the AR, differentfrom other QPP sources. However, we must note that the oscillations in the 17 GHz and UV radiations are found inthe flare loop region, since those emissions are mostly concentrated there. By contrast, the EUV emission in the flareloop region is so strongly saturated that the EUV QPP could not be detected there anyway. Therefore we presumethat the oscillatory phenomenon is everywhere, but the different locations of QPP at EUV channels and 17 GHz maysimply be a matter of which region is more favorable for detecting subtle variations of the radiation. LATE-PHASE MICROWAVE ACTIVITYWe finally explore the late-phase. Figure 5(a) shows time profiles of T b at the 17 GHz and 34 GHz averaged overregion D. Four key transition times are marked again: t for the thermal activation, t for the nonthermal activation,and t for the maximum number of nonthermal electrons. The late phase activity of this event is then characterizedby the gradual rise and fall of T b around t , the time of the second maximum T b . Note that the four-fold T b at 34GHz (gray colored curve) tends to agree to the T b at 17 GHz after ∼ △ T b ) in the three local Figure 5.
The late phase. (a) The time profiles of T b at 17/34 GHz from D are plotted along with the four major transitiontimes, t – t , marked by the vertical dotted lines. (b) NoRH 17 GHz map at the start time of the second rise is shown as invertedgrayscale image and the 34 GHz map as red contours in the levels of [50, 75]% of its maximum T b . (c) Same as (b) for the secondmaximum time. (d) Time profiles of the relative T b increase in the three local regions show that the largest △ T b occurred in B. regions in reference to the minimum brightness time between t and t . Since the dominant radiation mechanism attime period is optically-thin free-free emission, △ T b is a measure. The result that △ T b in the looptop (B) is largerthan in the footpoints (A or C) indicates high density accumulation in the looptop at t . This result is consistent withthe finding that the local EUV fluxes in the flare region reach their maxima at t (Fig. 2b).Post-flare microwave emission from the top of a flaring loop has been detected in many events, and interpreted to bedue to several reasons: trapping of nonthermal electrons in flare loops (Reznikova et al. 2009), enhancement of plasmaflows along supra-arcade structures (Kim et al. 2014), and strong heating near the X-point (Chen et al. 2016, 2017).However, none of these may apply to the present event, because this second flux enhancement is distinctively wellseparated from the impulsive peak ( t − t ≈
25 min). Such a property can be more appropriately considered withinthe context of the EUV late phase activity (Woods et al. 2011; Hock et al. 2012). Phenomenology, the secondarymicrowave flux enhancements coincident with the GOES soft X-ray peaks was simply regarded as late phase thermalactivity in compound flares (Lee et al. 2017, Ning et al. 2018). DISCUSSIONWe have studied the circular ribbon flare, SOL2014-12-17T04:51, mainly using the 17/34 GHz NoRH maps to findnew properties inherent to microwave radiation. They are: (1) two activation times detected in the form of fluxincreases at 17 GHz and 34 GHz, (2) 17 GHz polarization sign reversal at the flare maximum time, (3) 17 GHz QPPof I in the preflare phase and QPP of V in the flare concentrated in different locations, and (4) the second maximumof 17/34 GHz fluxes is due to a density increase in the flare loop, which has implications for the nature of the lateEUV phase study (cf. Lee et al. 2020). We mainly discuss the first three results in relation to the eruption.6.1. Preflare Activations: thermal and nonthermal
Impulsive microwave bursts during the flare can be regarded as nonthermal gyrosynchrotron radiation. But thepreflare and the postflare activities may be due to other mechanisms, which include free-free emission and gyroresonantradiation (Dulk 1985). The earliest ribbon activation at t in this event comes in the form of a gradual rise of the 17GHz flux without being accompanied by the 34 GHz flux. Such a distinct response at two separate frequencies cannotbe explained by either the thermal free-free or the nonthermal gyrosynchrotron mechanisms, because the spectra ofthese radiations are broadband (Zheleznyakov 1970). The very discrete frequency dependence is a characteristic ofthe thermal gyroresonance mechanism (Zheleznyakov 1962, Zheleznyakov & Zlotnik 1964, 1988). It occurs becausegyroresonance opacity is limited up to a few low harmonics ( n = f /f B = 1 , , , f B = 2 . B g MHz is thegyrofrequency of electrons and B g is field strength in gauss) above which the opacity drops significantly. Therefore,for a given frequency (in the present case either 17 or 34 GHz) there is a minimum magnetic field required to makethe local opacity significant, given by B g ≥ f [GHz] / . n (see, for further explanations, Gary & Hurford 2004). Theeffective harmonic number, n , is determined by temperature and density. For a typical coronal temperature ( ∼ cm − ), n = 3 is the highest harmonic that has significant opacity. On very vigorous active regionswith higher temperatures, n = 4 may also have significant opacity (White et al. 1992; Lee et al. 1993ab; White &Kundu 1997). For 17 GHz, this means that the local coronal magnetic field strength should be above 1350 gauss,and for 34 GHz, 2700 G is required. Per field strengths from the HMI magnetogram, the former field strength ispossibly available in the inner spine, but the latter is not. Note, however, that thermal gyroresonance opacity even at34 GHz was reported for a record-breaking strong coronal magnetic field (Anfinogentov et al. 2019). This propertyof gyroresonance opacity explains how the activation at t can be seen at 17 GHz, but not at 34 GHz. On the otherhand, the second activation at t occurs simultaneously at 17 GHz and 34 GHz. This can be explained by nonthermalgyrosynchrotron emission, which is emitted over a wide range of higher harmonics (Zheleznyakov 1970). The thermaland nonthermal nature of the two activations is also consistent with the temporal behaviors in that the 17 GHz fluxincreases gradually at t ≤ t ≤ t and impulsively at t .6.2. Rapid Change of Microwave Polarization
The 17 GHz polarization reversal during this CRF can be a yet unknown feature inherent to the fan–spine structure,where magnetic polarity around the null point varies so rapidly as to affect the propagation of microwave polarization.A way to possibly explain this polarization change is to view it as a mode-coupling phenomenon, the process by whichthe rays reverse their original sense of polarization while passing through a quasi-transverse field region along theline of sight from the radiation source to the observer, depending on the degree of mode coupling there (Cohen 1960,Zheleznyakov 1970, Melrose 1975, White et al. 1992). This is an attractive scenario for a fan–spine structure, becausethe fan surface may well act as a quasi-transverse layer for the rays emitted underneath. To think about an idealfan-spine structure with a flux rope inside, in this configuration, the magnetic fields above the fan surface are all in thenegative magnetic polarity, and the rays emitted from either magnetic polarity underneath will be observed as LHCPeverywhere. Therefore, the LHCP observed everywhere before the flare can simply be due to the fan-spine structure,without any strong mode-coupling phenomenon. On the other hand, if a magnetic flux rope rises to reconnect with theoverlying fan field, the fan surface may partially open up to let the flux rope erupt out. Such a change of magnetic fieldstructure can explain the instant reversal of the 17 GHz polarization at t more naturally. The reconnection betweenthe magnetic fields inside and outside of the fan will occur across a current sheet, the so-called breakout current sheet(BCS), and the newly open field lines amount to the lower part of the rising and expanding BCS (see, e.g., Lynch etal. 2016, Karpen et al. 2017). A sustained BCS over the active region might affect the microwave polarization, asmode coupling across a current sheet is still a debatable issue (Zheleznyakov et al. 1996; Lee et al. 1998; Lee 2007).We here offer only the simplest interpretation, according to which the change from LHCP to RHCP of the 17 GHzemission over the inner ribbon is not just a signature for any magnetic field perturbation, but may indicate a specificform of a breakout eruption out of the closed fan structure. The implied magnetic field reconfiguration is in line withthe recently reported decay of the coronal magnetic field at the flare site by Fleishman et al. (2020).6.3. Trigger of the Eruption t . However, it is t that the 17/34 GHzoscillations started, at which the first eruption signature in the 131 ˚A channel also started (Fig. 4). Other importanttransitions at t are also reported by independent studies: the start time of the 2 GHz QPP (Chen et al. 2019) andthat of the eruption signature in 94 ˚A (Liu et al. 2019). Their quasi-periods lie in the range of 1.3–4.0 min and are inproportion to the length of loops. It is thus likely that an external driver was applied to this fan-spine structure at t and all closed field lines within the dome underwent the kink oscillations (Aschwanden et al. 2002, Zhang et al. 2020),which caused the null point to deform itself into current sheet, and in about 10 min, the eruption broke out (see Lee etal. 2020). These oscillations started before the eruption and continued after it, which suggests that the erupted fieldlines serve as a conduit for the waves propagating along the spine. Among many numerical simulations for fan-spinereconnection (Karpen et al. 2012, 2017; Pariat et al. 2009, 2010, 2015, 2016; Wyper et al. 2016, 2017, 2018), the latestworks (Wyper et al. 2016, Karpen et al. 2017) predict that reconnection at the null launches torsional Alfv´en wavestravelling along the outer spine. The waves are driven by the magnetic twist accumulated elsewhere and released atthe reconnection point with the magnetic torque as a restoring force, consistent with the present observation that thedominant oscillatory power moves from a footpoint (C in Fig. 4) to the inner spine (A) at the eruption. Figure 6.
Schematic illustration of the observed polarization intensity map (top) and magnetic field configuration (bottom).Thick blue (red) lines stand for a flux rope in the negative (positive) magnetic polarity, sky-blue lines for the fan fields, orangedot for a null point, dashed circles for the breakout current sheet, and the curvy lines for the loop oscillations.7.
CONCLUDING REMARKSWe present three specific phenomena of this microwave CRF: (1) the nonthermal preflare activation, (2) the suddenchange of local polarization during the flare, and (3) the oscillation before and during the eruption, as the characteristicfeatures of magnetic reconnection in a fan-spine morphology. Among these the most obvious evidence for the eruption1is the sudden and permanent change of the 17 GHz polarization in the AR center. The fan-like structure is impliedby the 17 GHz preflare emission appearing as a single polarization state over the region with mixed magnetic polarity.The polarization change restricted to the core region then implies that the central part breaks out, letting the innerspine field erupt and revealing a structural change around the inner spine associated with magnetic eruption out ofthe fan–spine system. This conclusion is solely based on observation and does not refer to a particular model.The other two pieces are connected to this eruption under the afore-mentioned models designed for a fan-spinereconnection (Karpen et al. 2012, 2017; Pariat et al. 2009, 2010, 2015, 2016; Wyper et al. 2016, 2017, 2018).Especially, we interpreted the post-eruption oscillations in favor of torsional Alfv´en waves based on specific models(Wyper et al. 2016, Karpen et al. 2017), although briefly discussed other modes as well. We want to stress here thatthe most crucial part in this argument is not the exact mode, but the coupling of the preflare and postflare oscillations.The latter judged by their comparable periods implies the transfer of the oscillatory power from the closed loop to theopen fields. The nonthermal activation time, t , at which the oscillation also starts is another important signature forthe breakout current sheet formation, since most of these models predict transformation of a null point to a currentsheet before the eruption (Karpen et al. 2012, 2017; Wyper et al. 2017, 2018). In this context, the observed timegap t − t ≈ t (Fig, 4b) to cause the null point to deform itself into a current sheet, which is not reallyviolent but gentle to trigger thermal heating mostly in the inner spine (Fig. 4c). The rays emitted from there musthave been RHCP, but changes to LHCP while passing through the overlying fan field. (b) The next reconnection occursat t between the magnetic fields inside and outside of the fan across a breakout current sheet, and the microwavesource extends to the flux rope. Since this energy release is more intense, the nonthermal phase sets in (Fig. 4c) andthe oscillations start everywhere (Fig. 4d). Again, the rays emitted from the two footpoints, initially polarized indifferent senses, become all LHCP while passing through the fan field. (c) The breakout eruption follows at t to causethe flare and let the flux rope erupt. The rays from the positive field will be observed as RHCP as the overlying fanfield is now open (Fig. 2). Continuation of the oscillations through this transition (Fig. 4e) implies that they are themagnetic-untwisting waves (torsional Alfven waves) propagating along the spine as guided by the erupted field lines. REFERENCES
Anfinogentov, S. A., Stupishin, A. G., & Mysh’yakov, I. I. et al.2019, ApJ, 880L, 29Aschwanden, M.J., DePontieu, B., Schrijver, C.J., et al. 2002,SoPh, 206, 99Chen, X., Yan, Y., Tan, B. et al. 2019, ApJ, 878, 78Chen, Y., Du, G., Zhao, D., et al. 2016, ApJ, 820L, 37Chen, Y., Wu, Z., Liu, W., et al. 2017, ApJ, 843, 8Cohen, M. H. 1960, ApJ, 131, 664Demoulin, P., Henoux, J. C., Priest, E. R., et al. 1996, A&A, 308,643 Dulk, G. A. 1985, ARA&A, 23, 169Fleishman, G. D., Gary, D. E. & Chen, B. et al. 2020, Science,367, 278Galsgaard K. & Nordlund, A. 1997, J. Geophys. Res., 102, 231Galsgaard, K., Priest, E. R., & Titov, V. S. 2003 Journal ofGeophysical Research (Space Physics), 108, 1042Gary, D. E. & Hurford, G. J. 2004, in Solar and Space WeatherRadiophysics, Astrophysics and Space Science Library, Vol.314, ed. D. E. Gary & C. U. Keller (Dordrecht: Springer), 712