Generalisation of the Magnetic Field Configuration of typical and atypical Confined Flares
Navin Chandra Joshi, Xiaoshuai Zhu, Brigitte Schmieder, Guillaume Aulanier, Miho Janvier, Bhuwan Joshi, Tetsuya Magara, Ramesh Chandra, Satoshi Inoue
aa r X i v : . [ a s t r o - ph . S R ] N ov Generalisation of the Magnetic Field Configuration of typical andatypical Confined Flares
Navin Chandra Joshi , , Xiaoshuai Zhu , Brigitte Schmieder , , Guillaume Aulanier , MihoJanvier , Bhuwan Joshi , Tetsuya Magara , Ramesh Chandra , Satoshi Inoue ABSTRACT
Atypical flares cannot be naturally explained with standard models. To pre-dict such flares, we need to define their physical characteristics, in particular,their magnetic environment, and identify pairs of reconnected loops. Here, wepresent in detail a case–study of a confined flare preceded by flux cancellationthat leads to the formation of a filament. The slow rise of the non–eruptive fil-ament favours the growth and reconnection of overlying loops. The flare is onlyof C5.0 class but it is a long duration event. The reason is that it is comprised ofthree successive stages of reconnection. A non–linear force–free field extrapola-tion and a magnetic topology analysis allow us to identify the loops involved inthe reconnection process and build a reliable scenario for this atypical confinedflare. The main result is that a curved magnetic polarity inversion line in ac-tive regions is a key ingredient for producing such atypical flares. A comparisonwith previous extrapolations for typical and atypical confined flares leads us topropose a cartoon for generalizing the concept.
Subject headings:
Sun: Flare - Sun: Magnetic Reconnection - Sun: MagneticField Udaipur Solar Observatory, Physical Research Laboratory, Udaipur 313 001, India School of Space Research, Kyung Hee University, Yongin, Gyeonggi-Do, 446-701, Republic of Korea Max-Planck Institute for Solar System Research, Goettingen 37077, Germany LESIA, Observatoire de Paris, PSL Research University, CNRS Sorbonne Universit´e, Univ. Paris 06,Univ. Paris Diderot, Sorbonne Paris Cit´e, 5 place Jules Janssen, F-92195 Meudon, France Institut d’Astrophysique Spatiale, CNRS, Univ. Paris-Sud, Universit´e Paris-Saclay, Bt. 121, 91405Orsay CEDEX, France Department of Physics, Kumaun University, DSB Campus, Nainital 263001, India Institute for Space-Earth Environmental Research, Nagoya University, Furo-cho, Chikusa-ku, Nagoya,464-8601, Japan
1. Introduction
Solar flares are the manifestation of the violent energy release taking place in the solarcorona via magnetic reconnection. The energy powering these flares is stored in the highlysheared, current–carrying magnetic field lines in the corona. Then, instabilities and photo-spheric motions can lead to the sudden release of this energy via the reconnection of coronalmagnetic field lines (Shibata & Magara 2011). The initial configuration of the coronal fieldsignificantly influences how the flare proceeds. Various categories of flares exist: eruptiveflares, flares with partial or failed eruptions, and confined (or compact) flares. Flares withfull or partial eruptions produce coronal mass ejections (CMEs) while flares with failed erup-tion and compact flares are not associated with CMEs (Schmieder et al. 2013, and referencestherein).Over the years, many observational studies have investigated the physical conditionsthat determine the evolution of eruptive and confined flares. These studies can be classifiedinto two different categories, based on statistical or case–study approach. Both approacheshave their own merit. With statistics, reliable parameters have been determined to explainthe occurrence of energetic flares. By exploring the connection between the photosphericmagnetic field and solar flares, it was found that there exists a strong correlation betweennon–potentiality and flare strength. Different parameters have been used in such studiese.g., the length of the polarity inversion line (PIL), the strong shear of the transverse mag-netic field, the degree of non–potentiality of the active region, as well as the decay index(Hagyard et al. 1990; Leka et al. 1993; Falconer 2001; Falconer et al. 2003; Abramenko 2005;Schrijver 2007; Joshi et al. 2014; Zuccarello et al. 2015; Joshi et al. 2014). For example, therecent study of Vasantharaju et al. (2018) shows statistically a good correlation betweennon–potentiality and flare strength. In this study, a total of 77 cases were studied, with theequal number of events in both confined and eruptive categories. However, the study didnot show a clear parameter to predict which flare would be eruptive or compact. While astatistical analysis provides probabilities that may be indicative of flare characteristics, sucha study does not provide clarity about the underlying physical processes, e.g., the triggeringmechanism and the magnetic configuration of reconnection.The case–studies allow us to investigate the physical mechanisms responsible for re-connection and eruption. The case–study approach is complementary to statistical studies,which aim at identifying macroscopic and global tendencies. The statistical approach hashelped in building several models of solar flares. For example, the two–dimensional (2D)model of flare known as the “CSHKP” model (Carmichael 1964; Sturrock 1966; Hirayama1974; Kopp & Pneuman 1976) or the three–dimensional (3D) standard flare model (see,Aulanier et al. 2012; Janvier et al. 2015), explain that flares are the consequences of recon- 3 –nection between magnetic field lines, in particular, the legs of surrounding arcades, below theeruptive filament or/and flux rope (Aulanier et al. 2012). But it is the case–study approachthat helped to build other scenarios such as the 3D “tether cutting model” (Moore et al.2001), or testing the so–called breakout model of Antiochos et al. (1999) (Aulanier et al.2000; Ugarte-Urra et al. 2007). Departure from such so–called “standard” models providesthe possibility to better understand other scenarios.The case study approach has revealed, in particular, the importance of magnetic fieldenvironments. Magnetic field configuration of active regions is often the key to explaineruptions and flares (e.g., Mandrini et al. 2014; Masson et al. 2009; Joshi et al. 2015). Thesestudies point towards the importance of a null point above an emerging flux region in thepre–existing fan–spine magnetic field configuration. However, null points are not alwaysefficient for driving eruptive flares as shown by the work of Zuccarello et al. (2017) whichpresents such a case with a null point. They investigated 10 flares from a single AR andfound that during the first day flares were eruptive, while the events occurring on the nextday were confined. The only possibility that they found to explain this different behaviourwas the respective orientation of the emerging field lines with that of the overlying field.More generally the analysis of the topology of an active region leads to informationabout the regions where reconnection is possible. These regions are 3D fine layers wherereconnection can occur because the connectivity of magnetic field lines can change drasti-cally. These regions are called quasi–separatrix layers (QSLs) (D´emoulin et al. 1996, 1997).During flares, strong electric currents develop in QSLs. Flare ribbons correspond to the QSLfootprints in the photosphere. With the QSL analysis, one may able to understand whichmagnetic field lines may reconnect.The identification of QSLs is possible via magnetic field extrapolation using observedmagnetograms of the photosphere. Confined flares are usually studied by this method.The QSL footprints in the photosphere correspond to multiple ribbons in compact flares(Mandrini et al. 1996; Chandra et al. 2009; Dalmasse et al. 2015). In fact, there exists a se-ries of atypical flares that do not match with either eruptive, null point or usual anti–parallelQSL models (Schmieder et al. 1997; Dalmasse et al. 2015). Only a topological analysis al-lows us to understand the 3D magnetic configurations and propose convincing scenarios ofreconnection leading to such atypical flare.In this article, we study an atypical confined flare of GOES class C5.0 that leads toa long duration event (LDE) in soft X–rays (SXR) and provides a topological model tounderstand such exceptional cases. By carefully studying multi–wavelength data set, we aimto decipher the physical mechanism for this event and draw comparisons with previous cases(Schmieder et al. 1997; Dalmasse et al. 2015). By doing so, our objective is to generalize 4 –Fig. 1.— (a)
SDO /HMI photospheric magnetogram on 2014 May 15 at 02:36:00 UT. (b)
SDO /AIA 211 ˚A image on 2014 May 15 at 02:36:01 UT. (c)
SDO /HMI continuum image at02:36:37 UT on 2014 May 15. The yellow box shows the field of view of Figure 13 includingARs 12058 and 12060 and the black box shows the field of view of Figure 3.the conditions that produce such atypical flares. The structure of the paper is as follows.Section 2 describes the instruments from which the data set come. Section 3 describes the 5 –Fig. 2.— GOES and RHESSI X–ray profiles between 02:10:01 UT and 04:26:33 UT on 2014May 15. The vertical lines from left to right represent the start ( ≈ ≈
2. Observational data set
We used a multi–wavelength data set to analyze this event. The analysis is primarilybased on the observations taken by the
Atmospheric Imaging Assembly (AIA; Lemen et al.2012), on board the
Solar Dynamics Observatory ( SDO ; Pesnell et al. 2012). AIA observesthe Sun in seven extreme ultraviolet (EUV) wavelength channels, two ultraviolet (UV) andone white light channel. The AIA takes images in EUV with a cadence of 12 s, UV with acadence of 24 s and white light channel with a cadence of 1 hr having a pixel size of 0 . ′′ . 6 –The images from the hottest AIA channels (131 ˚A , T ∼ K; 94 ˚A , T ∼ K) are used toanalyze the flare and the evolution of flare loops. We further explore data from AIA 304 ˚A(T ∼ ∼ K) data to analyze the flare ribbon dynamics. It isnoteworthy that the AIA 1600 ˚A bandpass includes a part of the continuum formed in thetemperature minimum region at the temperature of T ∼ ∼ K) (Brekke & Kjeldseth-Moe 1994).At flare ribbons, emission in this passband is significantly enhanced largely due to increasedC IV line emissions. For the magnetic field analysis, we used data from the
Helioseismic andMagnetic Imager ( HMI ; Schou et al. 2012) onboard the
SDO . The observational cadence ofthis instrument is 45 s (for the line–of–sight photospheric magnetogram) and 720 s (for thevector magnetic field data) and the pixel size of the images is 0 . ′′ . The flare - associatedX–ray sources have been identified with the observations made by the Reuven Ramaty HighEnergy Solar Spectroscopic Imager ( RHESSI ; Lin et al. 2002). For the filament detection,we used the observations provided by the spectroheliograph of the Observatoire de Paris inMeudon which are archived in the database of the solar survey . Full disk spectroheliogramsin H α and Ca H lines are obtained for several wavelengths in each line in a daily cadence.
3. Multi–wavelength observations and analysis3.1. General configuration of the active region
The flaring activity concerns two NOAA active regions (ARs) namely 12058 and 12060.AR 12058 consists of elongated plages or intense network areas. The AR 12060 appearedin the middle of the AR 12058. The magnetic classes of the ARs 12058 and 12060 were β and βγ , respectively on the day of the event. The ARs lie near the solar center with averagepositions S10W26 (AR 12058) and S14W19 (AR 12060). Both ARs have a leading regionwith a strong positive polarity and the following region with dispersed negative polarities(Figure 1).A long duration C5.0 class flare occurred on 2014 May 15 which persisted for ≈ ≈ ≈ ≈ http : //bass .obspm.f r/home.php ? lang = f r SDO /AIA 304 ˚A image on 2014 May 15 at 02:00:07 UT, showingthe northern (F1) and southern (F2) filaments. The white circle represents the area wherethe north–west footpoint of filament F2 and the western footpoint of filament F1 is anchored.The green and yellow circles show the eastern footpoint of F1 and south–east footpoint ofF2, respectively. Bottom panel (b):
SDO /HMI line–of–sight photospheric magnetogram on2014 May 15 at 02:00:37 UT. The overplotted red lines represent the center axes of thefilaments. These red lines are traced from AIA 304 ˚A image shown in panel (a). The blackbox indicates the field of view of Figure 5(c).and F2 are observed in the AIA 304 ˚A image (Figure 3(a)). F1 is well visible as a broaddark elongated area in 211 ˚A (Figure 1(b)) and in 304 ˚A (Figure 3(a)) images, while F2 isvisible as a narrow dark lane only in 304 ˚A image (Figure 3(a)). We traced the center lineof both filaments observed in 304 ˚A and overplotted on the HMI photospheric magnetogram(red lines in Figure 3(b)). The filaments correspond to the two main polarity inversion lines(PIL) that make an angle of ≈
45 degrees from each other. We will show that the curvatureof the composite PIL is a key to understand the flare build–up.The large–scale evolution of the filaments during three days before the flare is presentedin panels (a)–(c) of Figures 4 which are the zoomed images taken from the daily H α spectro- 8 –Fig. 4.— H α spectroheliograms of the Meudon survey on 2014 May 13, 14, and 15, showingthe formation of the filament F2.heliograms (Figure 4). F1 was already present on May 13 while F2 was gradually formingbetween May 14 and May 15. It can be clearly seen (Figure 3(b)) that the western ends ofboth the filaments, F1 and F2, are anchored at the same positive polarity region (indicatedby the white circle), implying the junction of the two PILs. The other end of F1 lies in anegative region of weak and dispersed magnetic field (large green circle in Figure 3(b)) whilethe other end of F2 is at the negative region in the north of the major positive polarity(yellow circle in Figure 3(b)). The photospheric magnetic field changes can be seen in Figure 5 that present the zoomedimages of the region where the two filaments have their common end in the positive polarity(white circle of Figure 3(b)). On the left of the positive polarity, we see the evolution ofnegative polarity structures that seem to approach towards the positive polarity region overthe period of three days before the flare, which cancels the existing positive polarity. Wefurther note the flux emergence within the positive polarity region. Both changes are shown 9 –Fig. 5.—
SDO /HMI line–of–sight photospheric magnetograms on 2014 May 13, 14, and15. The area shown by the boxes in each panel is the flux cancellation region during theformation of the filament F2. The region shown by the red circle in panel (c) is the regionof flux emergence a few hours before the flare onset. The field of view is 150 ′′ by 150 ′′ .in Figure 5(c) by a black box and a red circle, respectively. They can be followed in theSDO/HMI movie attached with Figure 5. Small and continuous cancellations of magneticpolarities and emerging flux can push the existing positive polarities towards the PIL. Suchconverging motions could be the cause of the gradual growth of the filament F2 as observedbetween May 13 and May 15 (Figure 4(a)–(c)). This growth of F2 suggests that the magneticloops that are surrounding F2 grow and naturally rise as proposed in flux cancellation modelsof filament formation (van Ballegooijen & Martens 1989; Aulanier et al. 2010). During the long–duration of the flare, three different phases can be identified accordingto the evolution of the sets of flare ribbons and loops system that provide evidence forthree phases of reconnection. We call these three phases as stage 1, stage 2, and stage 3, 10 –Fig. 6.— The sequence of selected
SDO /AIA 304 ˚A images showing the overall dynamicsof the formation of different sets of flare ribbons and associated brightenings. White circlein panels (d) and (e) represent the area of the remote brightening (RB). The white box inpanel (c) represents the field of view of Figure 7(a).respectively.
The dynamic evolution of the flare ribbons can be seen in AIA 304 ˚A images (Figure 6)and associated movie. Two pairs of flare ribbons on each side of the filament F2 (R1e/R1wand R2e/R2w) are clearly visible at ≈ ≈ ≈ SDO /AIA 304 ˚A images, showing the moving brightnessalong the ribbon R2e. The dashed line show the cut along which the distance–time map hasbeen made. (b) The distance–time map of the moving brightness. The speed of the movingbrightness is ≈
200 km s − . The field of view of the panel (a) is shown in Figure 6(c) by thewhite rectangle.We observe bright kernels moving along the ribbon R2e (Figure 6(c)). Such slippingkernels have been studied in detail as the subsequent motion of flare loops following 3D recon-nection in Dud´ık et al. (2014) (c.f. Figures 6(b)–(c) and 8(b)–(c)). The speed measurementof the kernels along the flare ribbon in 304 ˚A is shown in Figure 7. Figure 7(b) shows thetime–distance plot of the moving brightness along the trajectory shown by the white dashedline in Figure 7(a). The average speed of kernels along the ribbon R2e is ≈
200 km s − .According to Dud´ık et al. (2014) the propagation of brightening along a flare ribbon can be 12 –Fig. 8.— The sequence of selected SDO /AIA 131 ˚A images, showing the overall dynamicsof the flare loops in the coronal region. In panel (a), we see the flare loops forming by thereconnection during the stage 1 of the event. In panel (b), along with the flare loops of stage1, we also see the formation of new set of flare loops forming during the stage 2 of the event,which became brighter with time (as seen in panel (c)). The remote brightening (RB) areais marked by the white circle in panel (d). The post–flare loops of stage 2 and the newlyformed flare loops of stage 3 of the event can be seen in panel (f).attributed to the rapid change of connectivity that flare loops undergo during the recon-nection process. This mechanism is also referred to as slipping or slip–running loops dueto the intrinsic nature of reconnection in 3D (see Aulanier et al. 2006). These complicatedmultiple ribbons do not allow us to understand the magnetic reconnection during the flarewith clarity. For that, we had to investigate the evolution of the bright loops seen in 131 ˚Aand 94 ˚A images of AIA. 13 –
Figure 8 represents the sequence of selected images in the AIA 131 ˚A channel, showingthe development of flare loop systems. In stage 1, we find bright loops joining footpointregions E (east) and S (south) in the south–west of the region at ≈ ≈ SDO /HMI photospheric magnetogram to investigatetheir locations compared with the overall magnetic field (Figure 9(b)). It can be seen thatthe eastern (R1e) and western (R1w) ribbons lie on the negative and positive polarities,respectively. The set of loops observed in the hottest channels of AIA (i.e., 94 and 131 ˚A)are the post–reconnected loop system that joins the flare ribbons (Figure 9(c)–(d)). At ≈ ≈ ≈ The evolution of the third stage of magnetic reconnection of the flare and formation ofsubsequent loop system can be readily observed from the AIA 171 ˚A and 131 ˚A images (and 14 –Fig. 9.— (a)
SDO /AIA 304 ˚A image showing the flare ribbons formed during the stage 1 ofthe flare. (b)
SDO /HMI line–of–sight photospheric magnetogram at 02:29:07 UT on 2014May 15. The overplotted red lines show the flare ribbons locations. These red lines are tracedfrom the AIA 304 ˚A image shown in panel (a).
SDO /AIA 131 ˚A and 94 ˚A images showingthe flare loops connecting the flare ribbons are shown in the panels (c) and (d), respectively.The red contours in panel (c) are the RHESSI X–ray contours of 6–12 keV energy band.The contours levels are the 15%, 25%, 50%, 75%, and 95% of the peak intensity.corresponding AIA 171 ˚A movie). The loop system formed following this stage is shownby red arrows in different panels of Figure 11. We find that the newly developed loopssystem connects the E and S regions, and undergoes a rapid evolution in terms of altitudeas well as lateral expansion. In particular, the lateral expansion of the loop system towardthe east is quite evident (shown by yellow arrows in panels f, g, h). This is similar to theformation of nearly identical loops during the stage 1 which connect flare ribbons R1e andR1w (Figure 9 and 171 ˚A movie). Due to observing limitations, it is hard to ascertain 15 –Fig. 10.— (a)
SDO /AIA 304 ˚A image showing both the first (R1e and R1w) and second (R2eand R2w) set of flare ribbons formed during the stages 1 and 2 of the event, respectively. (b)
SDO /HMI line–of–sight photospheric magnetogram at 02:35:07 UT on 2014 May 15. Theover plotted red and green lines show the first and second set of ribbons, respectively. Theseribbons are traced from the AIA 304 ˚A image that is shown in panel (a). (c)
SDO /AIA 131˚A image showing the flare loops connecting the flare ribbons. In panel (c) the loops joiningthe regions E to N are the reconnected loops formed during stage 2 of the flare. (d) The
SDO /AIA 1600 ˚A image at 02:35:04 UT, showing the flare ribbon brightenings. The yellow,red and blue contours represent the RHESSI X–ray contours of 3–6, 6–12, and 12–25 keVenergy bands. Contours levels are 15%, 25%, 50%, 75%, and 95% of the peak intensity.whether parts of the loops system observed during stage 3 developed during stage 1 or not.This uncertainty in distinguishing the formation stages of the two apparently different loopsystems at nearly same locations is due to the fact that the AIA 131 ˚A channel is sensitiveto two widely spaced temperature regions, 10 MK and < RHESSI observations are explored to investigate the spatial changes in the magneticreconnection site as well as the temporal characteristics of the energy release during variousphases of this atypical confined flare. The RHESSI covered the rise and maximum phasesof the flare (i.e., stages 1 and 2) while it missed part of the decay phase (i.e., stage 3).It is worthwhile to note that RHESSI observed the rise phase of the event with the A0attenuator state, i.e., the observations are recorded with the highest sensitivity. For theRHESSI image reconstruction, we have used the computationally expensive PIXON algo-rithm (Hurford et al. 2002) and selected detectors 3F–9F with an integration time of 20s. We find that during stage 1, there is an X–ray source at 6–12 keV energies that connectthe region E with regions S and W (Figure 9(c)). Notably, this source presents two clearcentroids that lie over the closest parts of the conjugate flare ribbons besides emission fromhot loops formed in–between these ribbons (see also Figure 12(a)). The bright, distinctcentroids provide clear evidence for the particle acceleration at the foot points of the looparcade, in response to the coronal magnetic reconnection associated with stage 1 (see, e.g.,Joshi et al. 2011; Guo et al. 2012; Joshi et al. 2017). During the second phase ( ≈ > < SDO /AIA 131 ˚A (left column) and 171 ˚A (right column)images, showing the overall flare loops dynamics during stage 3 of the flare event. The rightcolumn is available as an electronic animation.A0 attenuator state of the RHESSI. 18 –Fig. 12.— ((a)–(d)) RHESSI X–ray sources overplotted on the
SDO /AIA 131 ˚A images.The yellow, red and blue contours represent the RHESSI X–ray contours of 3–6, 6–12, and12–25 keV energy bands. Contours levels are 15%, 25%, 50%, 75%, and 95% of the peakintensity.
4. Magnetic field modeling4.1. NLFFF extrapolation
An NLFFF field extrapolation was performed using an MHD relaxation method de-scribed in Zhu et al. (2013, 2016), to compute the magnetohydrostatic state of the solaratmosphere. We adopt the computational domain of 709 × ×
258 Mm with a resolutionof about 720 km pixel − of binned data of SDO /HMI vector magnetograms (Hoeksema et al.2014). The vector field is obtained through the Very Fast Inversion of Stokes Vector(VFISV; Borrero et al. 2011) which is a Milne–Eddington based algorithm. A minimumenergy method (Metcalf 1994; Leka et al. 2009) is used to resolve the 180 ◦ ambiguity in thetransverse field. 19 –Magnetic field modelling is required to analyze the magnetic topology of the AR as-sociated with the flare event. For eruptive flares, it is necessary to follow the evolution ofthe magnetic field configuration because a CME leads to complete restructuration of coronalmagnetic fields (Savcheva et al. 2012, 2016; Janvier et al. 2016). On the other hand, themagnetic field configuration does not change drastically during confined flares. Therefore,single magnetogram can be used for fitting the observed flare loops involved before and af-ter the flare (Mandrini et al. 1996; Dalmasse et al. 2015; Zuccarello et al. 2017; Green et al.2017).In the present case, we confirmed that, before, during, and after the flare, the large–scale topology did not change, hence, we concentrate our study with the magnetogramobtained before the flare at ≈ SDO /AIA 304 ˚A image overplotted with the selected field lines from themagnetic field extrapolation introduced in Section 4.1. Blue lines are the small arcades ofthe two filaments. Yellow and white lines are larger arcades of the filament F1 and F2,respectively. (c)
SDO /HMI line–of–sight magnetogram with the overplotted modeled fieldlines. Yellow lines correspond to the bright loops observed during the flare reconnection atAIA 94 ˚A wavelength image in panel (b).
SDO /AIA 94 ˚A (b) and 211 ˚A (d) images at ≈ SDO /AIA 131 ˚A (left panel) and 304 ˚A (right panel) images with overplotted Qmap (10 < Q < ). White ovals show the areas where the QSLs are matching with theflare ribbons. QSLs are defined as robust thin volumes in the corona indicating where the magneticfield gradient is the strongest. They generalize the concept of separatrix and separators in3D (D´emoulin et al. 1996, 1997). The footprints of QSLs are the locations where the con-nectivity of field lines changes drastically. This means that field lines anchored at a polarityin a neighbouring area will see their opposite footpoints anchored at drastically different lo-cations. QSLs are preferential locations for strong electric currents to arise (Aulanier et al.2005; Janvier et al. 2013). The squashing degree Q is a parameter which indicates the gra-dient of connectivity change in the magnetic field volume under consideration (Titov et al.2002). As high squashing factor Q indicates the degree of magnetic field distortion, andtherefore an increased probability of finding high electric current densities, the regions ofhigh–Q are associated with locations where magnetic reconnection is the most likely to takeplace. As QSLs are properties of a large–scale magnetic field volume, extrapolations pro-vide a mean to calculate the locations of these QSLs, and hence of potential reconnectionsites. However in many case studies, a one–to–one perfect match between the QSL footprintsand ribbons is difficult to achieve (Dalmasse et al. 2015; Zuccarello et al. 2017; Green et al.2017). This is because QSLs are estimated from magnetic field extrapolations that have theirown limitations, while flare ribbons are related with particle beams and chromospheric dy-namics. Nevertheless, it is possible locally to match them in some cases e.g., for flares withcircular ribbons (Masson et al. 2009; Janvier et al. 2016). Dalmasse et al. (2015) demon-strated the relative robustness of the QSL location even if they are computed with an LFFFextrapolation. In the present work, we used a NLFFF extrapolation, so that the extrapo-lated magnetic field connectivity does depend, to some extent, on the existence of the twofilaments F1 and F2 in the active region. 22 –Fig. 15.—
SDO /HMI line–of–sight photospheric magnetogram overplotted with the modeledfield lines and Q map (10 < Q < ), shows the selected field lines and magnetic topologychange during the first ((a)–(b)) and second ((c)–(d)) stages of the flare event. Greenand blue lines are the pre–reconnected lines, while the red and yellow lines are the post–reconnected lines, respectively.Figure 14 displays the SDO /AIA 131 ˚A and 304 ˚A images at ≈ ≈ ≈ We have identified the flare ribbons in AIA 304 ˚A and the bright flare loops in the hotAIA 94 ˚A and 131 ˚A filters (see Section 3). In Section 4.1, we found many magnetic fieldlines which globally fit the observed active region loops. In order to understand which mag-netic field line reconnects with another one, we employ a method based on the relationshipbetween QSLs and flaring loops which has been applied in many studies (Mandrini et al.1996; D´emoulin et al. 1997; Schmieder et al. 1997).In Figure 15, we present the results of the analysis where the set of preflare field linesare in blue and green while the set of post–flare field lines are in red and yellow. To obtainthese sets of field lines, we start with the observed bright loops for each stage of the flare.We plot a series of typical field lines rooted at the edges of the QSLs where these brightflare loops are observed. This defines two footpoints per field line. For example, the brightobserved loops ES in stage 1 (Figure 8(a)) correspond to the red lines of the panel (b) inFigure 15. Two footpoints of these field lines are defined at E and S locations which arelocated on one side of the QSL. Then we start from point S but on the external side of the 24 –Fig. 16.— Schematic representation of the different stages of the flare event, showing themagnetic topology changes. Green and blue lines are the pre–reconnected lines, while thered and yellow lines are the post–reconnected lines, respectively. Pink stars show the regionsof magnetic reconnection. Changes in the magnetic field lines during stage 3 are identical tothat of stage 1.QSL and integrate one of the two field lines that were involved in forming these red loopsby reconnection. It goes towards the east, they are represented by the green field lines ofFigure 15(a). The long field line connecting the east side to W corresponds reasonably tothe observed long loops from location E. On the other side of the QSL, we find a short fieldline joining the QSL on the west and that arrive to the point W. It is represented by theblue lines in Figure 15(a). The blue and green field lines give us footpoints of two flare loopsinvolved in the reconnection. We, therefore, recover the four points. For stage 2, the sameprocedure is applied. We start from the red magnetic field line in Figure 15(d) and finallyobtain the two sets of pre- and post–flare loops. For stage 3, the reconnected loops areobserved between the locations E and S (see Figure 11(g) and 11(h)) which are expected tobe similar with the loops identified in stage 1 (see Figure 15(b)). This is because the QSLsshould have moved slightly, while this motion is not shown by our unique extrapolation.In order to understand how the flare was initiated and how it progressed, we propose a 25 –Fig. 17.— Cartoon summarizing the magnetic configuration of compact flares with typicalquasi–anti–parallel ((a)–(b)) and atypical quasi–parallel ((c)–(d)) reconnection. The flarepresently studied is of the later type.cartoon (Figure 16) summarizing the evolution of the magnetic field configuration obtainedfrom the NLFFF extrapolation. Note that in this figure, we have respected the color schemeused for the different set of flare loops in Figure 15 for an easy comparison. There existtwo nearby filaments F1 and F2 with common footpoints anchored at the same polarityregion, which are overlaid by a low–lying loop system (blue lines in Figure 13). The fluxcancellation underneath the filament F2 (c.f. Figure 5) presumably led to its formation,and to the gradual expansion of its overlying loops (blue loops; Figure 15(a) and 16(a)) asshown by MHD simulations (van Ballegooijen & Martens 1989; Aulanier et al. 2010). As theexpansion progressed, a set of blue loops began to reconnect with the green overlying lines(see Figures 15(a) and 16(a)): this is the first stage of the flare. The reconnection regionassociated with stage 1 is shown by the pink star in Figure 16(a). As a result, the blue loopsand the green overlying lines changed their connectivities and formed the flare loops (redcolor) and the yellow overlying lines (Figure 15(b) and 16(b)). The flare loops and ribbons(marked as R1e and R1w in Figure 9) are shown in Figures 15(b) and 16(b). The expansion 26 –of overlying arcades associated with the filament F2 continued in the later phase (see bluelines in Figure 16(c)).In the second stage, the northward expanding arcades of the filament F2 reconnectedwith the long loops over the filament F1 (c.f. Figures 15(c) and 16(c)). The field lineschanged their connectivity, and a new set of loops formed on the northern side of the flaringregion (red flare loops and the long yellow loops; Figures 15(d) and 16(d)). The newlyformed post-reconnected yellow field lines shown in Figure 16(d) may also include slippingloops. These loops can be identified in the observations (Figures 6(b)–(c), 7(b)–(c) and 10).Following the reconnection in stage 2, described above, the ribbons (namely R2e and R2w)were formed subsequently (c.f. Figures 15(d), 16(d), and 10).During the third phase, the rearrangement in magnetic connectivity occurred again aswe observed long loops joining region E to the western regions W and S (see Figure 11).Considering that all the loops are quasi-parallel, the energy release in the reconnection cannotbe strong and could naturally explain why the flare was weak.
5. Results and discussions
In this work, we present the case study of a small and non–eruptive C class flare thatshowed significant departure from a “classical” confined flare in terms of long duration ofenergy release and development of multiple flare ribbons. The event occurred in compositeARs 12058 and 12060 and involved loops systems across the edges of two adjacent filamentsF1 and F2 that remained undisturbed during the flare energy release. The comprehensiveanalysis of the flare in (E)UV wavelengths and their comparison with modelled coronalmagnetic field configuration by a NLFFF technique reveal successive stages of reconnectioninvolving magnetic field lines overlying the two neighbouring filaments. The spatial andtemporal evolution of the RHESSI X–ray emission during the rise and maximum phases ofthe flare provide clear evidence that the magnetic reconnection sequentially proceeded attwo adjacent locations within the core region of the flaring environment.Flux cancellation and photospheric motions have already been observed to be themain sources for compact flares (Hanaoka 1997; Schmieder et al. 1997; Nishio et al. 1997;Chandra et al. 2006; Dalmasse et al. 2015). Here, we observed the formation of filament F2from 2014 May 13 to May 15 during the continuous flux cancellation in the region under-neath the filament (see Figures 4 and 5 and associated animation). During the formation offlux rope F2, we assume that the loops in the active region should evolve as shown in themodel of Aulanier et al. (2010). The expanding loops overlying filament F2 reconnect with 27 –the surrounding field as well as with the overlying loops of filament F1 successively in threedifferent phases. Notably, the spatial evolution of the X–ray footpoint sources from E–W toE–N directions, with the relatively small spatial shift in the eastern footpoint gives credenceto the interpretation above, in which magnetic reconnection successively involves adjoiningloop systems.Confined/compact flares ensue when the reconnection occurs between two groups ofloops (e.g., Hanaoka 1997) that must involve QSLs (Mandrini et al. 1996; Schmieder et al.1997; D´emoulin et al. 1997) or at a coronal null point with a single spine that emerges awayfrom the fan surface anchored in a remote region (e.g., Masson et al. 2009; Hernandez-Perez et al.2017). Confined flares are also observed in the active region where the overlying magneticfield is too strong to allow the filament (flux rope) to erupt (e.g., Amari & Luciani 1999;T¨or¨ok & Kliem 2005; Guo et al. 2010; Kushwaha et al. 2014; Joshi et al. 2014; Sun et al.2015; Kushwaha et al. 2014, 2015).In the present work, we want to generalize the magnetic configuration conditions inwhich long duration confined flares can occur without eruptions. Figure 17 summarizesdifferent typical conditions for getting the interaction between adjacent loops. Recently,Dalmasse et al. (2015), studied a similar type of atypical compact flare and found out theinteraction between different loop systems from a deep analysis of QSLs. We also computedthe QSLs to find a match with the observed ribbons which are the footpoints of the ar-cade loops. The scenario is different from the standard models of solar flares, where thereconnection occurs underneath the erupting filaments among the legs of erupting arcades.Differently, here we observe the formation of flare arcades over the filaments systems (seeFigure 6), without any disturbance and eruption of either filament. In the flare scenario, wealso believe that the moving bright kernels seen during the flare are the signatures of slip-ping, and maybe slip–running reconnection, as this process is intrinsic to 3D reconnectionat the core of the flaring mechanism (see Aulanier et al. 2006).We summarize our scenario for such compact flare configurations in Figure 17 aftercomparing our case with more usual set-ups used for compact flares from extrapolation(Mandrini et al. 1996; Schmieder et al. 1997; D´emoulin et al. 1997) and MHD simulations(Moreno-Insertis et al. 2008; Pariat et al. 2009; T¨or¨ok et al. 2009). Commonly, we get thesame “usual” 3D configuration in various confined flare events with a strongly curved PILand a straight QSL footprint surrounded by a horse–shoe (or arc–shaped) QSL footprint.Depending on the location of the rising loops (that can arise due to flux-emergence, orunderlying flux rope build–up like in our case), we can get:- quasi–anti–parallel reconnection leading to one set of flare loops (as commonly observed)- quasi–parallel reconnection leading to two sets of flare loops (like in our case) 28 –The latter may lead to a false impression of an eruptive two–ribbon flare. The present study,thus adds up a new well–identified case of so–called atypical flares.After having in hand different case–studies of confined flares (Liu et al. 2014; Dalmasse et al.2015) and the present study that together form an ensemble of well–identified and varioussituations for atypical flares, we can conclude on the key mechanism occurring in such flares.These events look like two ribbons flares but, in fact, the forcing lies in the interaction ofquasi–parallel loops. We show the importance of the curvature of the PIL, that creates apurely 3D topological effect, which in turn allows reconnection between quasi–parallel fieldlines, and which exists neither in 2D models nor in models of failed eruptions.We are grateful to the referee for his/her valuable comments and suggestions that sig-nificantly improve the scientific content and presentation of the paper. We thank
SDO /AIA,
SDO /HMI,
GOES and
RHESSI teams for providing their data for the present study. Thiswork is supported by the BK21 plus program through the National Research Foundation(NRF) funded by the Ministry of Education of Korea. BS thanks Prof. Magara to in-vite her to Kyung Hee University on the BK21 program where this project started to bediscussed. RC and BJ acknowledge the support from SERB-DST, New Delhi project No.SERB/F/7455/2017-17.
REFERENCES
Abramenko, V. I. 2005, ApJ, 629, 1141Amari, T., & Luciani, J. F. 1999, ApJ, 515, L81Antiochos, S. K., DeVore, C. R., & Klimchuk, J. A. 1999, ApJ, 510, 485Aulanier, G., DeLuca, E. E., Antiochos, S. K., McMullen, R. A., & Golub, L. 2000, ApJ,540, 1126Aulanier, G., Janvier, M., & Schmieder, B. 2012, A&A, 543, A110Aulanier, G., Pariat, E., & D´emoulin, P. 2005, A&A, 444, 961Aulanier, G., Pariat, E., D´emoulin, P., & DeVore, C. R. 2006, Sol. Phys., 238, 347Aulanier, G., T¨or¨ok, T., D´emoulin, P., & DeLuca, E. E. 2010, ApJ, 708, 314Boerner, P., et al. 2012, Sol. Phys., 275, 41 29 –Borrero, J. M., Tomczyk, S., Kubo, M., Socas-Navarro, H., Schou, J., Couvidat, S., &Bogart, R. 2011, Sol. Phys., 273, 267Brekke, P., & Kjeldseth-Moe, O. 1994, Sol. Phys., 150, 19Carmichael, H. 1964, NASA Special Publication, 50, 451Chandra, R., Jain, R., Uddin, W., Yoshimura, K., Kosugi, T., Sakao, T., Joshi, A., &Deshpande, M. R. 2006, Sol. Phys., 239, 239Chandra, R., Schmieder, B., Aulanier, G., & Malherbe, J. M. 2009, Sol. Phys., 258, 53Dalmasse, K., Chandra, R., Schmieder, B., & Aulanier, G. 2015, A&A, 574, A37D´emoulin, P., Bagala, L. G., Mandrini, C. H., H´enoux, J. C., & Rovira, M. G. 1997, A&A,325, 305D´emoulin, P., Priest, E. R., & Lonie, D. P. 1996, J. Geophys. Res., 101, 7631Dud´ık, J., Janvier, M., Aulanier, G., Del Zanna, G., Karlick´y, M., Mason, H. E., &Schmieder, B. 2014, ApJ, 784, 144Falconer, D. A. 2001, J. Geophys. Res., 106, 25185Falconer, D. A., Moore, R. L., & Gary, G. A. 2003, Journal of Geophysical Research (SpacePhysics), 108, 1380Green, L. M., Valori, G., Zuccarello, F. P., Zharkov, S., Matthews, S. A., & Guglielmino,S. L. 2017, ApJ, 849, 40Guo, Y., Ding, M. D., Schmieder, B., D´emoulin, P., & Li, H. 2012, ApJ, 746, 17Guo, Y., Ding, M. D., Schmieder, B., Li, H., T¨or¨ok, T., & Wiegelmann, T. 2010, ApJ, 725,L38Hagyard, M. J., Venkatakrishnan, P., & Smith, J. B., Jr. 1990, ApJS, 73, 159Hanaoka, Y. 1997, Sol. Phys., 173, 319Hernandez-Perez, A., Thalmann, J. K., Veronig, A. M., Su, Y., G¨om¨ory, P., & Dickson,E. C. 2017, ApJ, 847, 124Hirayama, T. 1974, Sol. Phys., 34, 323Hoeksema, J. T., et al. 2014, Sol. Phys., 289, 3483 30 –Hurford, G. J., et al. 2002, Sol. Phys., 210, 61Janvier, M., Aulanier, G., & D´emoulin, P. 2015, Sol. Phys., 290, 3425Janvier, M., Aulanier, G., Pariat, E., & D´emoulin, P. 2013, A&A, 555, A77Janvier, M., et al. 2016, A&A, 591, A141Joshi, B., Thalmann, J. K., Mitra, P. K., Chandra, R., & Veronig, A. M. 2017, ApJ, 851, 29Joshi, B., Veronig, A. M., Lee, J., Bong, S.-C., Tiwari, S. K., & Cho, K.-S. 2011, ApJ, 743,195Joshi, N. C., Liu, C., Sun, X., Wang, H., Magara, T., & Moon, Y.-J. 2015, ApJ, 812, 50Joshi, N. C., Magara, T., & Inoue, S. 2014, ApJ, 795, 4Joshi, N. C., Srivastava, A. K., Filippov, B., Kayshap, P., Uddin, W., Chandra, R., PrasadChoudhary, D., & Dwivedi, B. N. 2014, ApJ, 787, 11Kopp, R. A., & Pneuman, G. W. 1976, Sol. Phys., 50, 85Kushwaha, U., Joshi, B., Cho, K.-S., Veronig, A., Tiwari, S. K., & Mathew, S. K. 2014,ApJ, 791, 23Kushwaha, U., Joshi, B., Veronig, A. M., & Moon, Y.-J. 2015, ApJ, 807, 101Leka, K. D., Barnes, G., Crouch, A. D., Metcalf, T. R., Gary, G. A., Jing, J., & Liu, Y.2009, Sol. Phys., 260, 83Leka, K. D., Canfield, R. C., McClymont, A. N., de La Beaujardiere, J.-F., Fan, Y., & Tang,F. 1993, ApJ, 411, 370Lemen, J. R., et al. 2012, Sol. Phys., 275, 17Lin, R. P., et al. 2002, Sol. Phys., 210, 3Liu, R., Titov, V. S., Gou, T., Wang, Y., Liu, K., & Wang, H. 2014, ApJ, 790, 8Mandrini, C. H., D´emoulin, P., van Driel-Gesztelyi, L., Schmieder, B., Cauzzi, G., & Hof-mann, A. 1996, Sol. Phys., 168, 115Mandrini, C. H., Schmieder, B., D´emoulin, P., Guo, Y., & Cristiani, G. D. 2014, Sol. Phys.,289, 2041 31 –Masson, S., Pariat, E., Aulanier, G., & Schrijver, C. J. 2009, ApJ, 700, 559Metcalf, T. R. 1994, Sol. Phys., 155, 235Moore, R. L., Sterling, A. C., Hudson, H. S., & Lemen, J. R. 2001, ApJ, 552, 833Moreno-Insertis, F., Galsgaard, K., & Ugarte-Urra, I. 2008, ApJ, 673, L211Nishio, M., Yaji, K., Kosugi, T., Nakajima, H., & Sakurai, T. 1997, ApJ, 489, 976Pariat, E., Antiochos, S. K., & DeVore, C. R. 2009, ApJ, 691, 61Pesnell, W. D., Thompson, B. J., & Chamberlin, P. C. 2012, Sol. Phys., 275, 3Savcheva, A., Pariat, E., McKillop, S., McCauley, P., Hanson, E., Su, Y., & DeLuca, E. E.2016, ApJ, 817, 43Savcheva, A., Pariat, E., van Ballegooijen, A., Aulanier, G., & DeLuca, E. 2012, ApJ, 750,15Schmieder, B., Aulanier, G., D´emoulin, P., van Driel-Gesztelyi, L., Roudier, T., Nitta, N.,& Cauzzi, G. 1997, A&A, 325, 1213Schmieder, B., D´emoulin, P., & Aulanier, G. 2013, Advances in Space Research, 51, 1967Schou, J., et al. 2012, Sol. Phys., 275, 229Schrijver, C. J. 2007, ApJ, 655, L117Shibata, K., & Magara, T. 2011, Living Reviews in Solar Physics, 8, 6Sturrock, P. A. 1966, Nature, 211, 695Sun, X., et al. 2015, ApJ, 804, L28Titov, V. S., Hornig, G., & D´emoulin, P. 2002, Journal of Geophysical Research (SpacePhysics), 107, 1164T¨or¨ok, T., Aulanier, G., Schmieder, B., Reeves, K. K., & Golub, L. 2009, ApJ, 704, 485T¨or¨ok, T., & Kliem, B. 2005, ApJ, 630, L97Ugarte-Urra, I., Warren, H. P., & Winebarger, A. R. 2007, ApJ, 662, 1293van Ballegooijen, A. A., & Martens, P. C. H. 1989, ApJ, 343, 971 32 –Vasantharaju, N., Vemareddy, P., Ravindra, B., & Doddamani, V. H. 2018, ApJ, 860, 58Zhu, X., Wang, H., Du, Z., & He, H. 2016, ApJ, 826, 51Zhu, X. S., Wang, H. N., Du, Z. L., & Fan, Y. L. 2013, ApJ, 768, 119Zuccarello, F. P., Aulanier, G., & Gilchrist, S. A. 2015, ApJ, 814, 126Zuccarello, F. P., Chandra, R., Schmieder, B., Aulanier, G., & Joshi, R. 2017, A&A, 601,A26