"Dandelion" Filament Eruption and Coronal Waves Associated with a Solar Flare on 2011 February 16

Figure 4 presents the temporal evolution of the flare at various wavelengths. Columns from left to right are images in soft X-rays Thin-Be filter, in EUV at 171 and 304 Å, and in the Hα line center; and were taken by XRT, AIA, and FMT, respectively. Before the flare starts, the so-called "sigmoid" or S-shaped magnetic field structure (e.g., Rust & Kumar 1996; Canfield et al. 1999) that lies in the east–west direction is observed in soft X-rays (see the top left panel in Figure 4).

Zoom In Zoom Out Reset image size

During the time of the flare, NOAA 11158 was composed of several sunspots, as discussed in detail by Toriumi et al. (2013) and Sun et al. (2012). In particular, two neighboring fast flux emergences (P₀–N₀ and P₁–N₁ pairs) appeared in line along the east–west direction and formed a quadrupolar distribution (see the top panel of the Hα −0.8 Å column in Figure 5). Between the following negative spot N₀ of the western bipole and the preceding positive spot P₁ of the eastern bipole, a prominent filamentary structure appeared along the magnetic neutral line on February 13, which developed for about three days. The X-class flare on February 15 occurred along the elongated neutral line between N₀ and P₁. This basic magnetic configuration remained even after the X-class flare and displayed the sigmoid structure seen in the XRT images.

Zoom In Zoom Out Reset image size

The M-class flare in question, however, occurred at the eastern edge of the main neutral line between N₀ and P₁. When the impulsive phase of the flare just begins (∼14:22 UT), a small bright X-ray feature appears at the eastern end of the sigmoid structure. The brightening has counterparts in the Hα and 304 Å images, which are mainly from footpoints of the flaring loops. As of 14:31 UT, the filament eruption can be distinguished in the EUV and Hα images (see the bottom panels of Figure 4).

The starting point of the filament eruption is also located at the eastern edge of the neutral line between N₀ and P₁. This filament eruption is clearly discernible both in the Hα and EUV images, as shown in Figure 5. It was activated near 14:25 UT and displays an enlarging dark feature in the Hα blue-wing images, like a dandelion, whereas no manifestation is exhibited in the Hα red-wing until 14:32 UT.

Here, we describe how we used Hα images to derive the three-dimensional velocity field of the erupting filament. For the line-of-sight (LOS) velocity, we modified the method of Morimoto & Kurokawa (2003a). They computed the LOS velocities of erupting filaments observed with FMT using the method based on Beckers' cloud model (Beckers 1964; Mein & Mein 1988). Morimoto & Kurokawa (2003a) adopted the single cloud model. Because FMT has only three (center and ±0.8 Å) wavelength data points for the four unknowns (source function, optical depth, Doppler velocity, and Doppler width) of a cloud, they further assumed a fixed Doppler width in their calculation. Our method, on the other hand, deals even with the Doppler width as a free parameter. This can be implemented as follows: in the quiescent phase of the filament, we derive the parameters assuming the value of the Doppler width, as was done in Morimoto & Kurokawa (2003a). To analyze a time series, we search for the best-fit solution around the quiescent parameter values found in the initial time step, and select the "nearest" local minimum in the parameter field. This selection of nearest values is based on the assumption that the physical parameters in a single cloud do not change so quickly. Next, we search the parameters in the next time step around the parameters derived in the previous time step, and so on. We did not set a "pre-defined range" in our method; we used the hybrid method for the fitting (Powell 1970). The error of the derived LOS velocity is roughly comparable with the errors obtained with the method of Morimoto & Kurokawa (2003a) and is about ±10 km s⁻¹, while the derived LOS velocity tends to be overestimated for the fixed Doppler width, since this is thought to be larger for an erupting filament. The error could be, however, a little worse due to misalignment of the wavelengths of the FMT filters from the nominal values. Comparing the wing data, the wavelength of the red wing (+0.8 Å) seems to be set closer to the line center than that of the blue wing (−0.8 Å). The error of the LOS velocity, therefore, is roughly ±15 km s⁻¹.

The left column in Figure 6 shows the derived temporal evolution maps of the LOS velocity. The colors show blueshift (i.e., moving toward us) and redshift (i.e., moving away from us) of the filament. Near 14:30 UT, the filament begins to erupt northward and simultaneously shows blueshift features, especially at the leading part, with an LOS velocity of about −30 km s⁻¹. At the root of the filament, in contrast, it shows a redshift with a velocity of about 20 km s⁻¹. The filament eruption is clearly seen until 14:35 UT. At this time, the LOS velocity at the leading edge is nearly −25 km s⁻¹. Next, the front edge gradually becomes invisible at 14:45 UT, whereas near 14:55 UT, the filament shows a dominant redshift pattern. The observed blueshift velocity of about −30 km s⁻¹ is slightly larger than the result by Veronig et al. (2011), while the temporal evolution is roughly consistent.

Zoom In Zoom Out Reset image size

We confirm the temporal LOS motion with the side views of the filament eruption observed from STEREO-A, since it was located at 868 ahead of the Earth. In the EUV 195 Å images of STEREO-A/SECCHI/EUVI, we see movement of the cold plasma as a dark feature (second column in Figure 5). The dark feature is ejected from the flare site, and the traveling velocity of the front (i.e., the fastest) part is about 280 km s⁻¹. The horizontal component of the velocity (i.e., the velocity in the direction from the Sun to the Earth) is about 230 km s⁻¹. FMT cannot detect clouds with velocity greater than 50 km s⁻¹, because the wavelengths of the wing data are set at ±0.8 Å. Therefore, FMT missed the front part of the ejection.

By comparing the EUVI 195 Å and FMT Hα −0.8 Å images (Figure 5), the Hα filament seems to correspond to the main (i.e., the darkest) part of the filament in the 195 Å images. The main part of the filament has a velocity of about 110 km s⁻¹, and the horizontal component has a velocity of about 90 km s⁻¹. The velocity is much larger than the LOS velocity derived by FMT (30 km s⁻¹). LOS velocities derived from the cloud model reflect representative values of moving clouds, and the fine LOS velocity field in the clouds could be lost. Besides, the velocity in the Earth–Sun direction is derived by following dark features seen in EUVI images, and is not derived by tracing the moving ejecta. Alternatively, we could detect a slower component of the erupting filament, because the velocity of about 90 km s⁻¹ far exceeds the detection limit of FMT.

For the tangential velocity of the erupting filament (i.e., the velocity of the filament in the plane of the sky), we traced the apparent motion of filament blobs using the local correlation tracking (LCT) method. The middle column in Figure 6 shows the temporal evolution of the erupting filament with arrows that show the orientation and magnitude of the tangential velocity. The filament blobs are moving upward (i.e., northward) with a tangential velocity of about 10 km s⁻¹. The dark feature seen in the EUVI 195 Å images has a northward velocity of about 65 km s⁻¹, which is faster than the tangential velocity derived from FMT. This result is explained by the excessively large mesh size used for the LCT method, which means that features moving at such high speed could not be detected even if they were present. The seeing condition prevents us from reducing the mesh size for the FMT images. Therefore, the derived tangential velocity is a somewhat averaged value of the main part of the ejected filament.

In the right column of Figure 6, we show the temporal evolution of the inclination of the ejecta. The inclination angle is set to zero for the upward direction normal to the solar surface. In the early phase of the ejection (around 14:30 UT), the direction is nearly horizontal with respect to the solar surface, with a slight upward inclination. During the main phase of the ejection, which occurs around 14:35 UT, the direction is still nearly horizontal but with a slight downward inclination. Next, the inclination distribution becomes complex, showing upward and downward motions, probably because of the coexistence of upward and downward moving features in the later phase. Although the error in the velocity measurements is probably large, we believe that the ejecta move nearly horizontally with respect to the solar surface. Further, in Figure 5 (EUVI panel at 14:30 UT), we depict the plane of the solar surface (yellow arrows over the heliographic coordinates) and a vertical component perpendicular to the surface (arrow labeled "vertical"). The dashed green arrow highlights the direction of the filament eruption as seen by STEREO-A, and φ is the inclination angle with respect to the vertical on the solar surface. From EUV observations (193 and 195 Å) we infer the geometrical components of the erupting material, which leads us to estimate φ ≈ 58°. So the filament in EUVI was ejected having an elevation angle of about 32°.

In the Hα data, we also observe winking (activation and/or oscillation) of filaments associated with the flare. As we see in Figure 2, some filaments are labeled f₁ to f₅. These filaments are located far from the flare site. We identify a clear winking feature for filament f₂ and a much fainter signature for filament f₅ (see also Figure 3).

The time series of the winking filament f₂ is displayed in Figure 7(a) at Hα −0.8 Å (left), Hα center (middle), and Hα +0.8 Å (right) wavelengths. The field of view of the f₂ region is the same as in Figure 2. At 14:40 UT, the filament starts winking. First, a dark feature becomes prominent in the red-wing images, as shown by the arrow labeled 2r in Figure 7(a). The feature is dominant until 15:05 UT, and then it fades slowly away. However, at about 14:50 UT, a dark structure starts to appear in the blue-wing images, as pointed out by another arrow (2b). This structure is noticeable until 15:10 UT, when it disappears. In the Hα center, in contrast, no significant changes are observed.

Zoom In Zoom Out Reset image size

Figure 7(c) plots the temporal variations of the intensities calculated for the f₂ region at three wavelengths. In the red-wing plot (dashed–dotted line), we clearly identify a decrease of intensity associated with the appearance of the dark feature in the red-wing images. The decrease starts at about 14:36 UT and reaches maximal depletion around 14:45 UT, after which it recovers. The signal in the blue-wing plot (dotted line), however, remains roughly constant until 14:55 UT and then starts to decrease gradually. In the Hα center plot (solid line), the signal slightly increases.

A similar analysis for filament f₅ is presented in Figures 7(b) and (d). Unlike f₂, the dark structure appears mainly in the blue wing (left panels). The dark feature is evident from 14:40 UT until 15:05 UT. Conversely, at 15:00 UT, we recognize a very weak feature in the red wing (see arrow labeled 5r in Figure 7(b)). The temporal variation of the intensities is plotted in Figure 7(d). The winking feature dominates during the darkening in the blue-wing plot (dotted line), whereas the red-wing (dashed–dotted line) and Hα center (solid line) plots show gradual variation and no clear signals caused by filament oscillation. For the other filaments f₁, f₃, and f₄, we verified the intensity profiles and found no winking patterns.

We also examine the temporal evolution of the coronal waves associated with the 2011 February 16 flare and the relation with the oscillating filaments. The time sequences of the EUV waves produced by the flare are presented in the top panels of Figure 8 (panels (a)–(d)). These are AIA 193 Å running difference (20 minutes difference) images. The white arrows highlight the moving wavefronts, whereas the locations of the oscillating filaments f₂ and f₅ are marked by the boxes. The distances of the filaments (f₂ and f₅) measured from the flare site along the spherical solar surface are ≈5.1 × 10⁵ and ≈6.4 × 10⁵ km, respectively. We derive the speeds of the EUV wave by following the leading edge of the wavefronts consecutively along the spherical lines to the filaments f₂ (path 2) and f₅ (path 1) directions. The mean velocities computed from a linear fit are 430 ± 29 km s⁻¹ and 672 ± 24 km s⁻¹, respectively. The EUV waves have been studied by several authors (Harra et al. 2011; Veronig et al. 2011; Long et al. 2013; Nitta et al. 2013), and the propagating speed ranges from 500 to 700 km s⁻¹, which changes depending on the propagating direction. Therefore, the velocities derived in this study are comparable with those results.

Zoom In Zoom Out Reset image size

At 14:28 UT (Figure 8(a)), a well-defined wavefront propagates mainly toward the north, and a bright erupting material is observed that corresponds to the dandelion filament. At about 14:31 UT (Figure 8(b)), the central part of the wavefront seems to stop and to exhibit a stronger amplitude: this occurs just when the wave is approaching the edges of a weak active region that is located north of NOAA 11158. The observed enhanced pattern and the changes in the wavefront progression are tentatively attributed to the plasma compression generated by the interaction between the coronal wave and the confined magnetic system of the active region (Uchida 1974). We plot the potential magnetic field configuration derived from SOHO/MDI in Figure 8(e). The red triangle indicates the location of the northern weak active region. We confirm that the propagating wavefront tends to stop there. Note that, at 14:35 UT (Figure 8(c)), the leading part of the wavefront decelerates, perhaps retained by the magnetic field configuration, whereas the west part is slightly refracted and continues traveling northwest (Figure 8(d)).

Figure 8(f) shows the progression of the most prominent fronts of the EUV wave from 14:27 to 14:40 UT, the positions of the weak active region, and the oscillating filaments f₂ and f₅. At 14:37 UT, the wavefront labeled 5 approaches f₅ and, at 14:40 UT, wavefront 6 reaches f₂. These times are consistent with those identified both in the images and in the intensity plots of f₂ and f₅ (Figure 7). Figure 9 presents the time–distance diagram of these temporal behaviors. We measured the temporal evolution of the wavefronts along the paths, as shown by the projected dashed lines in Figure 8(f). The squares $(\square )$ and circles (◦) plotted together with a linear fit (dotted lines) in Figure 9 show the times and the positions of the wavefronts as they move along paths 1 and 2, respectively. Diamonds $(\diamond )$ indicate the oscillating filaments f₅ and f₂, which are on a straight line, as shown by the dotted lines. Therefore, the observed winking filaments are assumed to have been triggered by the passage of the EUV coronal wave.

Zoom In Zoom Out Reset image size