Astron. Astrophys. 328, 371-380 (1997)

5. Ribbon motions

5.1. Vertical motion (line-of-sight velocity)

We acquired spectra with the slit of the USG positioned successively in S1 and S2 as indicated in Fig. 2-b. The slits intersect regions of the ribbons R1 and R2 formed during the impulsive phase, and are tangent to sections of the ribbons newly formed during the decay phase. With the slit of the USG in the position S1, we acquired spectra from 15:44:48 until 15:46:52 UT, with a time resolution ranging between 5 and 30 s. We show in Fig. 6-a an enlargement of the ribbon R1 intersected by the slit in this interval: at 15:44:49 the slit is tangent to the inner edge of the structure B1 and crosses the outer edge of the structure C1. We recall that both B1 and C1 developed during the impulsive phase, and had a new brightening around this time. Examples of the corresponding spectra are shown in Fig. 6-b and 6-c respectively. On the inner edge of B1, both the Ca II K line and the H   line (not shown in the figure) are in emission, symmetric, with a small red-shift of the line center (Fig. 6-b), while on the outer edge of C1 both lines show a strong red asymmetry in the wings (Fig. 6-c). This asymmetry is stronger for the outermost points of the structure.

Fig. 6. a H 1.5 Å image of the flaring ribbons showing how slit S1 intersects the bright kernels. The image has been slightly rotated with respect to Figs. 2 and 3. The slit crosses C1 in its outer edge, and B1 in its inner edge; b, c CaII K profiles in a quiet region (thin line) and in kernels B1 and C1 (thick line). The dashed line is the rest wavelength of CaII K

At the time of the next spectrum (15:45:22 UT, see Fig. 7-a), the kernel A1 is well developed, almost at its maximum emission, and the USG slit is tangent to its outer edge. No spectral asymmetry was detected in these positions before A1 brightened, but at this later time both the Ca II K line and the H   line are in emission and show a red shift of the maximum emission (Fig. 7-b). Along the inner edge of B1 still no asymmetry is detectable, whereas in the points on the outer edge of C1 the red asymmetry is still present (Fig. 7-c,d).

Fig. 7. As Fig. 6, around the time of simultaneous maximum between A1 and A2. Now the slit touches also A1 in its outer edge. a H 1.5 Å image; b, c, d CaII K profiles in a quiet region (thin line), and in kernels A1, B1, C1 (thick line). In panel b the A1 pre-flaring profile (dotted) is also shown

Interpreting the red shift as Doppler velocity (for the method used see Cauzzi et al., 1996), we obtain downward velocities on the outer edges of A1 and C1 that are between 10 and 30 ( 2) km s^-1, and almost no velocity in the points of the inner edge of B1. This is clearly summarized in Fig. 8-a and 8-b, where we show the velocity measured along the slit, at the times reported above, together with the intensity in the center of Ca II K line as an indication of the position occupied by the ribbon. Fig. 8-b clearly exemplifies the case of A1, where a substantial downflow is visible even before the ribbon itself becomes visible in the center of CaII K.

Fig. 8. CaII K central intensity along the slit (solid line), and chromospheric velocity in the corresponding points (crosses). The letters indicate the position of the kernels. a 15:44:49, the time of Fig. 6; b 15:45:22, the time of Fig. 7; c 15:43:12, at the end of the impulsive phase

We acquired spectra with the slit in position S2 (see Fig. 2-b) from 15:46:10 until 15:48 UT. In these spectra the various bright features are all connected, and is not easy to distinguish their inner and outer edges. However, we note that the slit is tangent to A2 along its outer edge, before its second peak emission. The two spectra obtained between 15:46:10 and 15:46:20 UT along the outer edge of A2, show the Ca II K and the H   lines in emission with an unshifted component and a weaker and wider red shifted component. The derived velocity of this shifted, weak component is 50 km s^-1. The next spectrum,approximately on the same position, was obtained more than one minute later (15:47:27): here both the Ca II K and the H   line show a red asymmetry, which, if interpreted as Doppler shift, gives a velocity of about 15 km s^-1. The presence of this "edge" effect seems therefore to exist for all the considered kernels, although the temporal sequence is not so clear for A2 as it is for A1.

To check for the presence of stronger velocities in the outermost points of the ribbons during other phases of the flare, we considered also the spectra acquired during the impulsive phase. The slit intersected ribbon R1 in a portion heated by conduction from the hot corona (see Fig. 4 of Paper I, and the discussion in it). Six spectra were acquired in a 30 s interval (15:43:00 to 15:43:30). For all of them the stronger downflows (20-40 km s^-1) are found in the outer points of the ribbon, while the inner edge shows very small or zero velocity. Fig. 8-c shows this trend at 15:43:12 UT.

5.2. Separation motion

Also during the decaying phase, the two (new) ribbons move apart from each other, away from the magnetic neutral line (see Paper I for motion during the impulsive phase). We followed the displacements of the different bright kernels in time, and traced them in Fig. 3-a with solid short lines. The arrows indicate the direction of the motion.

At the time of their maximum, both A1 and A2 move at a high separation horizontal speed ( 40 km s^-1  at 15:45:30 UT). Kernels B1 and C1, that slowed down immediately after the impulsive phase, now show an average velocity of about 30 km s^-1. A2 is the only kernel that continues to move away from the neutral line also in the final phase of the flare (after 15:49:00). Some of these horizontal velocities are shown in Fig. 9.

Fig. 9. Line-of-sight velocities (open symbols) in the outer edge of flaring kernels A1 and C1 and in the inner edge of B1; positive values represent downflows. Velocities away from the magnetic neutral line (filled triangles) of the same flaring kernels are also shown

The kernels displacement velocities reported above are consistent with values found in the literature for two-ribbon flares. These vary from the particular case of no velocity at all (Kurokawa et al., 1992; de La Beaujardière et al., 1995), to values of about 40 km/s (Kitahara and Kurokawa, 1990). It must be said that often the reported values refer to an average over the whole ribbon, while individual kernels presumably can display faster horizontal motions, as it appears to be the case described in this paper.

5.3. What are these flows?

In Fig. 9 we summarize the ribbons dynamics, relying on kernels A1, B1 and C1 for which we have clear evidence. These kernels are all moving away from the magnetic neutral line during our observations; their separation horizontal velocities are shown in Fig. 9. On the outer edges of the flare ribbons (kernels A1 and C1, Fig. 9) we observe downflows of the order of tens of km s^-1, over a narrow region of 2- size. These downflows seem to decrease in amplitude when the flaring kernel is slowing down (and eventually stopping) on the solar surface, as exemplified by the case of kernel A1 (Fig. 9). On the inner edge of a moving ribbon, on the contrary, we observe only very weak downflows, regardless of the horizontal velocities of the ribbon itself (kernel B1, Fig. 9). This behaviour is strongly supported by the observations in R1 during the impulsive phase (Fig. 8-c), although the short interval for which we have spectral data does not allow us to study its temporal evolution.

These signatures are consistent with the picture of two-ribbon flares, where successive magnetic reconnection episodes provide sufficient energy to heat and compress the chromosperic plasma in the outer edge of the ribbons. For the case described in this paper, the mechanism of energy transport from the coronal to the chromospheric layers is most probably conduction. In fact the presence of a hot plasma (T 20 MK) in the upper part of a loop, that thermally emits in the 14-23 keV energy bands (Sect. 4.), supports the idea that the kernel A2 and B1 are heated by conduction, and suggests that conduction is the main energy transport mechanism also for other kernels like A1 and C1. We know that also the points of R1 for which we measured strong downflows (Fig. 8-c) were heated by conduction.

Reconnection models that explicity deal with the heat conduction effect (see e.g. Forbes et al., 1989; Forbes & Acton, 1996; Yokoyama & Shibata, 1997) show that the conduction front directly maps the boundary magnetic field lines. This leads to observations of hot cusp-like structures in soft X-ray (Magara et al., 1996), and of systematically higher temperatures in the outer loops of a flaring system (Tsuneta, 1996). We believe that the chromospheric downflows that we observed in the narrow regions at the outer edge of the flare ribbons map this conduction front as well. Gan et al. (1991) showed that indeed the effect of a conduction front on the chromosphere is the ablation and a simultaneous compression of the plasma, with downflows up to 100 km s^-1, a value consistent with our observations.

Forbes (1996, private communication), on the base of theoretical considerations, indicated a thickness of few arcsec for the locus of the downflows mapping to the conduction front. This value is consistent with our observations, that show a width of the chromospheric condensation at the edge of the ribbons of 2- .

© European Southern Observatory (ESO) 1997

Online publication: March 24, 1998

helpdesk.link@springer.de