Astron. Astrophys. 325, 305-317 (1997)

5. Conclusion

When trying to understand very complex physical phenomena like solar flares, a strong interaction between observations and theory is certainly required. In particular, observations allow us to choose between all the possible models, which always oversimplify reality, and guide us in selecting the most relevant approximations. On the other hand, models help us in extracting the most relevant features from the mass of observations. Some years ago, we began to investigate the flare problem in this spirit. Magnetic reconnection seemed to be the most likely candidate for energy release during flares but the debate was, and still is, largely opened since 3D MHD modelsare needed. Due to the intrinsic difficulties to build up a time-dependent 3D model of the field with the observed data as boundary conditions, we restricted ourselves to the study of 3D magnetic field equilibria and we focussed our work on the spatial distribution of flare by-products.

The 2D and 2 D studies of magnetic reconnection show the importance of separatrices, energy is released there giving as results plasma jets with high velocities and accelerated particles. At the time we began flare studies, following the work of Baum & Brathenal (1980) and Gorbachev & Somov (1988), the location of separatrices in 3D magnetic-field configurations was only known when the field was modeled using a set of sources. We further improved the method (called SM, for Source Method) by developing a numerical algorithm to find the separatrices, by fitting by least-squares the parameters of the sources to the observed data, by comparing charge and dipole representations and by taking into account the observed magnetic shear. This allowed us to show that, in very different configurations, both H and UV flare kernels are linked to the topology of the active-region magnetic field (see references in the Introduction).

Since the SM is based on the magnetic linkage between sub-photospheric sources, some readers may have some doubt on the results because flares are purely coronal events. We have then attempted to overcome this limitation (see Paper I) in several ways. In particular, we have found that flares are not necessarily associated to the presence of magnetic null points, nor to field lines tangentially touching the photosphere (or chromosphere). That is to say, that flares are not always related to the coronal separatrices in a classical sense. In Paper I we extended the notion of separatrices to the notion of quasi-separatrix layers (QSLs), which are regions where the field-line linkage is drastically changed. In theoretical configurations, we have shown that QSLs extend only along parts of the separatrices computed with the SM and we have described them in typical quadrupolar and bipolar regions.

In the present paper, we computed QSLs in flaring active regions, extrapolating the original observed photospheric field by a linear force-free field. We have found that the feature common to the various flaring regions studied is the presence of QSLs. The H or UV kernels are found lying close to them in zones where the magnetic field is in general greater than 100 G. This confirms and precises previous results obtained with the SM, in the sense that flare kernels are not observed all along separatrices computed with the SM but only on the portion obtained when computing QSLs. Our finding precise the locations where ideal MHD can break down in the theory of Hesse & Schindler (1988). We have further shown that two-ribbon flares have basically the same field-line connectivity as flares with three or four ribbons. The studied flares are found to be fed by only one electric current loop but they imply interactions between several magnetic structures, and none of the studied flares correspond to a single flaring flux-tube. All the studied flares have a parasitic bipole located in between a main bipole. The main difference between these regions is the relative orientation of the two bipoles. These results confirm that flares are coronal events where the release of free magnetic-energy is due to the presence of regions where the magnetic field-line linkage changes drastically. They agree with the observational results of Yohkoh satellite (e.g. Tsuneta 1993, Hanaoka 1994, Masuda et al., 1994, Shimizu et al., 1994) obtained on a completely independent base and strongly support the hypothesis that 3-D magnetic reconnection is at work in solar flares, even if more theoretical investigations on the physics of 3-D magnetic reconnection are still needed.

© European Southern Observatory (ESO) 1997

Online publication: May 5, 1998

helpdesk.link@springer.de