Pre-eruptive Magnetic Reconnection within a Multi-flux-rope System in the Solar Corona

We analyzed the EUV images obtained by the Atmospheric Imaging Assembly (AIA; Lemen et al. 2012) on board the Solar Dynamics Observatory (SDO). AIA provides uninterrupted observations of the full-disk Sun with a pixel size of 06 and a cadence of 12 s. The AIA's six EUV passbands in 94, 131, 171, 193, 211, and 335 Å have distinctive temperature responses and cover a wide temperature range from 0.5–30 MK, which enables us to reconstruct the temperature distribution of plasma emitting along the line of sight in the optically thin corona, known as the differential emission measure (DEM). Here we employed an algorithm providing positive definite DEM solutions by solving a linear system based on the concept of sparsity (Cheung et al. 2015).

The DEM-weighted mean temperature (T_DEM) is defined conventionally as follows,

$$\begin{eqnarray}&&{T}_{\mathrm{DEM}}=\displaystyle \frac{\sum \,\mathrm{DEM}(T)\times T}{\sum \,\mathrm{DEM}(T)},\end{eqnarray} \tag{ 1 }$$

which gives the total emission measure $\mathrm{EM}=\sum \mathrm{DEM}(T){dT}$, with a binning $d\mathrm{log}T=0.1$ in this study. The thermal energy content in a region of interest is

$$\begin{eqnarray}&&{E}_{\mathrm{th}}=3{k}_{B}T\sqrt{\mathrm{EM}\,{fV}},\end{eqnarray} \tag{ 2 }$$

where k_B is the Boltzmann constant, f is the filling factor assumed to be unity in this study, and V is the volume of plasma.

For the morphological investigation, we mainly used three passbands, i.e., 131 Å (Fe xxi with peak response temperature log T = 7.05; Fe viii, log T = 5.6), 171 Å (Fe ix, log T =5.85), and 304 Å (He ii, log T = 4.7). To highlight the braided structure in the EUV images, we applied the unsharp masking technique on 131 Å images: a pseudo background image (the mask) is generated by smoothing the original image with a box-car of 10 × 10 pixels (6'' × 6''); the enhanced image is obtained by subtracting the background from the original image.

We also analyzed Ca ii images obtained by the Solar Optical Telescope (SOT) on board Hinode (Kosugi et al. 2007) and Si iv 1400 Å images from Interface Region Imaging Spectrograph (IRIS; De Pontieu et al. 2014) to investigate the lower atmosphere response to the energy release during the flare.

The HXR emission from the flaring region is recorded by Rueven Ramaty High-energy Solar Spectroscopic Imager (RHESSI; Lin et al. 2002). We synthesized HXR images with detectors 1, 3, 5, 6, 7, and 9, employing the PIXON algorithm (Hurford et al. 2002). Because RHESSI crosses the South Atlantic Anomaly, HXR data is only available from 16:29 UT on 2015 June 22. We prepared HXR spectra during 16:29 UT–16:52 UT with a time bin of 32 s, and then performed forward fitting with a theoretical photon spectrum combining isothermal and thick-target bremsstrahlung models available in the SPectral EXecutive (SPEX) package within the SolarSoftWare (SSW) distribution (Freeland & Handy 2012). The fitting procedure aims to minimize the reduced χ² value to unity by iterations. The best-fit theoretical spectrum thus provides thermal and nonthermal characteristics of the source plasma. We further derived the thermal and nonthermal energy content during the flare. The thermal energy released is calculated with Equation (2). The volume V of the flaring plasma is approximated to be A^3/2, where A denotes the area enclosing pixels with EM > 3× 10²⁶ cm⁻⁵, a number chosen by trial and error to best represent the emitting region in the EM maps of 5–20 MK, as derived from AIA data. The thermal energy is overestimated because we assume the unknown filling factor (f) to be unity. The energy available in nonthermal electrons is derived by employing the function calc_nontherm_electron_energy_flux.pro in SPEX, using parameters obtained from the thick-target fitting, namely, electron flux, negative spectral index, and low- and high-energy cutoff.

We studied the magnetic-field configuration by examining magnetograms from the Helioseismic Magnetic Imager (HMI; Scherrer et al. 2012) on board SDO. To extrapolate the coronal magnetic field, we employed Space-Weather HMI Active Region Patches (data product of hmi.sharp_cea series) vector magnetograms at 12 minute cadence. The vector magnetograms are preprocessed to best suit the force-free condition before being fed into the "weighted optimization" NLFFF code as the photospheric boundary (Wiegelmann et al. 2006). Here we built the NLFFF over a uniform grid of 840 × 452 × 452 pixels (pixel size 0.36 Mm) and investigated magnetic connectivities by tracing field lines pointwise on the bottom of a tenfold-refined grid with a fourth-order Runge–Kutta method, using footpoint positions of field lines to calculate the squashing factor Q (Titov et al. 2002). Simultaneously, we mapped twist number T_w by integrating the local density of T_w, ${\rm{\nabla }}\times {\boldsymbol{B}}\cdot {\boldsymbol{B}}/4\pi {B}^{2}$, along each field line (Liu et al. 2016b).

The C1.1-class flare of interest occurs at 16:45 UT (peak time) on 2015 June 22 in the NOAA active region 12371, located close to the disk-center (N13W14). This is a compact flare without causing any CME, also known as a simple-loop flare, to be differentiated from the classical two-ribbon flares containing numerous flaring loops. Following the C-class flare, two more episodes of precursor emission at 17:24 and 17:42 UT (Wang et al. 2017a) precede an M6.5-class flare at 18:23 UT (Jing et al. 2016, 2017), a major eruption associated with a full-halo CME observed by the Large Angle and Spectrometric Coronagraph Experiment on board the Solar and Heliospheric Observatory. All the above mentioned activities happen in the close vicinity of the polarity inversion line (PIL) that separates two major sunspots of opposite polarity in the center of the active region (Figure 1). Around this major PIL, we found no significant flux emergence or cancellation, and no significant photospheric shearing or converging motions within 16 hr before the C-class flare, hence we focus on the corona in our investigations as elaborated below.

Zoom In Zoom Out Reset image size

Magnetic-field restructuring during the precursor phase is understood to play a key role in triggering the impending flare (e.g., Wang et al. 2017a). We employed the NLFFF extrapolation method to model the coronal magnetic field, and selected a rectangular region covering the major PIL where the precursor emission is concentrated (Figure 1(a)) to derive the maps of squashing factor Q and twist number T_w (Figures 1(b) and (c)). A composite of Q and T_w maps in the X–Z plane at 16:34:25 UT and 17:22:25 UT, respectively, is plotted in Figures 1(d) and (e) (see also Figure 2). Based on these maps and field-line tracing, we identified a system comprised of at least five flux-rope branches separated in altitude, labeled FRB1 (green), FRB2 (orange), FRB3 (yellow), FRB4 (cyan), and FRB5 (violet), following the sequence of low-to-high altitude. Each branch consists of twisted field lines displaying similar winding and footpoint regions. Suspended in the corona, FRB2 is a coherent rope displaying an oval with enhanced T_w fully enclosed by a QSL, similar to the rope in Liu et al. (2016b), while the lower branch FRB1 is less coherent and apparently attached to the surface. Braiding with each other (Figure 3), the high-altitude set of the three branches (FRB3–5) are roughly bounded by a QSL (see also Figure 2), which is not as well defined as the one enclosing FRB2. At 16:34:25 UT the combined map of T_w and Q (Figure 1(d); see also Figure 2) shows an opposite (positive) twist region beneath FRB3 and FRB2, where the two high-Q layers intersect, a favorable site for 3D magnetic reconnection (Démoulin 2006). We traced a few representative field lines (black) threading these positive-twist regions, which have twist numbers of 0.41 ± 0.14. Compared with the post-flare map at 17:22:25 UT (Figures 1(e) and 2), the positive-twist region beneath FRB3 largely disappears, supposedly via the cancellation with the dominant negative twist.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It is well known that reconnection-related changes in the coronal field can be noticed from comparing the NLFFF before and after (e.g., Liu et al. 2016b), as NLFFF extrapolation reconstructs magnetic topology in active regions with high fidelity (e.g., Liu et al. 2014, 2016a), while magnetic reconnection changes topology. As follows, we analyze multiwavelength coronal observations to seek reconnection signatures, of which the most sought-after are plasma heating, reconnection outflows, and nonthermal particle acceleration.

Small, intermittent enhancements can be seen in the soft X-ray (SXR) 1–8 Å light curve as early as ∼1 hr before the C-class flare at 16:34 UT (Figure 4(a)), but we focused on activities from ∼16:18 UT when the emission level becomes persistently elevated. Snapshots of the flare during the precursor and main phase are shown in 131 Å (Figure 4; see also the accompanying animation). Superimposing the line-of-sight component of the magnetic field (contours) onto the 131 Å image (Figure 4(b1)), one can see that the EUV emission during the precursor phase is dominated by multiple threads apparently entangled and aligned along the PIL of interest (Figure 4(b1)). The main phase of the flare is pronounced in the form of multiple overlying loops (Figure 4(b2)), which evolve into a thick loop with enhanced emission during the gradual phase (Figure 4(b3)). Furthermore, we obtained the EM of the flaring plasma at 5–10 (Figures 4(c1)–(c3)) and 10–20 MK (Figures 4(d1)–(d3)), respectively. It is clear that the threads along the PIL are heated up to ∼20 MK during the precursor phase.

Overplotting the field lines of FRB1 (green), FRB2 (orange), and FRB3 (yellow; representing the high-altitude flux-rope branches for simplicity) on the 131 Å image (Figure 5(a)), one can see the clear spatial association between the flux-rope branches and the entangled threads. The two ends of this system (labeled NFP and SFP) are associated with extended surface brightenings in AIA 304 Å, IRIS 1400 Å, and SOT Ca ii. At ∼16:30 UT (Figures 5(e)–(g)), three brightened emission kernels (labeled K1, K2, and K3) are seen at both 131 and 304 Å. K1 is also seen in the Ca ii image, suggesting that it is a footpoint emission in the low atmosphere. On the contrary, K2 is missing in the Ca ii image, suggesting it occurs relatively high in the corona. These emission kernels are associated with the enhanced emission cospatial to NFP, as distinctly seen in the 1400 Å image (white arrow), as well as to SFP (yellow arrow), which is in agreement with the scenario of reconnection within different flux-rope branches: energy is released at the reconnection site as indicated by nonthermal HXR emission (see Figure 5(e) and below) and further deposited at the flux-rope footpoints. Furthermore, in the wake of the reconnection episode, the bright threads become more braided than before, with some threads apparently crossing each other (Figures 5(i)–(o)), indicating an ongoing magnetic reconfiguration. The braiding may further contribute to energizing coronal plasma by the dissipation of currents induced by the entangled field lines (Parker 1983).

Zoom In Zoom Out Reset image size

We further analyze the kinematics and thermal characteristics of several brightening features produced as a consequence of reconnection. Figure 6 presents snapshots of 171 Å images representing the brightening activities followed by the reconnection episode, as shown in Figure 5. The images reveal that some blobs originate from the close vicinity of the reconnection site as indicated by nonthermal HXR emission (see Figure 6(b) and below) and propagate away along the spinal direction of the flux-rope system. Among many episodes of flows representing distinctive blobs, we focus on one prominent episode with the flow onsetting at the peak of a small, yet impulsive HXR bump at 16:30 UT. The flow speed of ∼72 km s⁻¹ is estimated from the time–distance diagrams (Figures 6(j) and (k)). These are made by taking slices off the running-difference images along the curved flow path (dashed curve in Figure 6(a)) and then stacking them up chronologically. A further increase in speed to ∼176 km s⁻¹ is noted prior to the onset of the flare main phase at 16:35 UT. The time–distance diagrams also reveal a counterflow at 38 km s⁻¹, whose lower speed is likely due to projection effects. We interpret this set of bidirectional flow as a signature of reconnection outflow.

Zoom In Zoom Out Reset image size

To investigate the thermal characteristics of the reconnection outflows, we tracked one blob distinctly observed during 16:31–16:34 UT (marked by red boxes in Figure 6) and solved DEM solutions over 0.5–30 MK within a 4'' wide box enclosing the blob. The DEM distribution reveals a hot component peaking at ∼10 MK due to reconnection-induced heating, and a "cold" component at ∼1.5 MK that is mainly attributed to the "quiet" corona in the foreground and background of the blob along the line of sight. From the evolution of DEM-weighted mean temperature T_DEM (Figure 6) one can see that the blob temperature attains a maximum value ∼10 MK in the beginning and decays to ∼8 MK, as far as it can be identified, due presumably to cooling. Its emission measure (EM) peaks approximately one minute later than the temperature maximum and varies between [0.5–2.5]  ×10³⁰ cm⁻⁵. The thermal energy content of each individual blob is estimated by assuming its volume as a cube of width 4'', which effectively encloses the blob. The results vary in the range of 4–9 × 10²⁷ erg (Figure 6(l)), amounts to a subflare or the largest nanoflare (Parker 1988).

The spatial and spectral evolution of HXR emission is studied in conjunction with the reconnection episode. Superimposing HXR sources over AIA 131 Å images (Figures 7(a)–(d)), one can see that during the precursor phase the HXR emission is cospatial to the EUV enhancement in the center of the braided threads aligned along the PIL. The HXR source in 12–25 keV at 16:30 UT (Figure 7(b)) corresponds to the power-law component of the photon spectrum (Figure 7(f)). The spectral fitting reveals the presence of nonthermal electron flux with a spectral index (δ) of 6.3 and of hot plasma at 20 MK, consistent with the DEM analysis. This precursor HXR emission also coincides in time with the onset of outflowing plasma blobs (Figure 6). The bidirectional outflows and the presence of nonthermal electrons along with the high-temperature plasma argue strongly for the occurrence of magnetic reconnection within the flux-rope system, as the outflows are directed along its spinal direction. In contrast, during the main phase of the flare, the HXR spectra can be better fitted by an exponential function depicting thermal bremsstrahlung at a lower temperature than during the precursor phase (Figure 7(h)), and the corresponding HXR source takes the shape of a thick loop similar to its EUV counterpart arching over the braided threads (Figure 7(d)). We found that the nonthermal electron energy content is generally sufficient to energize the thermal emission during the flare (Figure 8), despite that the electron spectra is significantly harder during the precursor phase than other phases. The nonthermal energy released by reconnection within the flux-rope system is expected to be deposited at its footpoints, where the dense chromospheric plasma is heated and expands along the flux-rope field lines undergoing reconnection into the corona. We conjecture that these twisted field lines later relax into the less twisted, i.e., sheared field lines (blue; Figure 7(c)) to produce the post-flare loops emitting thermal X-rays and EUV (Figure 7(d)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

After the C-class flare, this multi-flux-rope system continues to evolve (Figure 2) and erupts about 1 hr later as a full-halo CME propagating at about 1200 km s⁻¹ in the outer corona (Figure 9). Such CMEs are known to be responsible for the majority of the most intense geomagnetic storms (Webb & Howard 2012), but the current one only causes a moderate geomagnetic storm (${\mathrm{Dst}}_{\min }\approx -80$ nT), when it arrives at the Earth three days later as a shock-driven ejecta (Figure 10). Its characteristics are typical of ICMEs (Zurbuchen & Richardson 2006): the speed declines smoothly like a single stream expanding as a whole, the plasma β (ratio of thermal and magnetic pressure) and proton temperature T_p are depressed, so is ${T}_{p}/{T}_{\exp }$, indicating that the ejecta expands faster than the ambient solar wind, as T_exp is given by the well-established correlation between the solar-wind speed and temperature; on the other hand, the average Fe charge state $\langle Q{\rangle }_{\mathrm{Fe}}$ is enhanced, ${{\rm{O}}}^{7+}$/O⁶⁺ also shows a bump inside the ejecta. Frozen in as the CME expands into the outer corona, "hot" ionic charge states are reliable indicators of ICME plasma. However, the ejecta's magnetic field is very irregular, making it impossible to identify the individual components of the source structure. However, it is unlikely that the ejecta could result from successive CMEs merging together, because it lasts only ∼22 hr, the typical size of a single CME expected at 1 au, which is distinct from those long-duration (typically over two days) events in which the CME–CME interaction is supposedly at play (Burlaga et al. 2002; Wang et al. 2003). In the LASCO CME catalog,⁵ we found no candidate that had a fair likelihood to interact with the CME of interest. It is also highly unlikely that the spacecraft only made a glancing encounter with this fast, earth-directed CME. Thus, the identity of the individual flux-rope branches at the Sun must have gradually lost as they continue to interact with each other and with the solar wind during the propagation from Sun to Earth. The resultant field irregularity explains why this CME causes no severe geoeffects.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size