A still open debate about the way magnetic reconnection should work in 3D has arisen in the last years: do some 2D and 2 D properties extend to 3D configurations or are the 2D and 2 D cases so singular that a new theory, including other properties not revealed in previous analyses, needs to be constructed? The 2D and 2 D approaches to reconnection are singular, in the sense that 2D and 2 D reconnection can be defined in several ways that cannot be directly generalized to 3D. Basically 2D and 2 D reconnection theory is intimately linked to separatrices and to their intersection (the separator). Hesse & Schindler (1988) have shown that separatrices are structurally unstable features because they disappear when slightly perturbing a 2 D configuration to a 3D one. In their view, 3D reconnection can occur at any location provided that there is a break down of idealness. Another point of view has been given by Priest & Forbes (1989) who proposed that the imposed boundary flows select the particular line where reconnection takes place. On the other hand Bagalá et al., 1995; Démoulin et al., 1993, 1994; Mandrini et al., 1991, 1993, 1995; van Driel-Gesztelyi et al., 1994 show that the location of the energy release site is associated with the topology of the coronal magnetic field. In these works the photospheric field is extrapolated to the corona using a series of subphotospheric magnetic sources and the method, called Source Method (SM), used to determine the location of separatrices, is based on the connectivity between the sources.
However, the SM intrinsically needs the description of the magnetic field by subphotospheric sources. Priest & Démoulin (1995) have explored a way of generalizing the concept of separatrices to magnetic configurations without field-line linkage discontinuities. They propose that magnetic reconnection can also occur in 3D in the absence of null points at "quasi-separatrix layers" (QSLs), which are flat volumes where there is a rapid change in field-line linkage. They give an example of a sheared X-field where nearly any smooth and weak flow imposed on the boundary produces strong flows at the QSLs. Their results have been extended to typical theoretical flaring configurations built by four magnetic sources (Démoulin et al., 1996b, hereafter Paper I) and an algorithm, called QSLM (for quasi-separatrix layers method), has been developed in order to determine the locations of QSLs. The QSLM finds elongated regions that are in general located along small portions of the separatrices defined by the SM, and, in the limit of very concentrated photospheric fields, both methods give the same result (exept for the regions where field lines are tangent to the photosphere). In bipolar magnetic configurations, the trace of QSL at chromospheric level are formed by two elongated regions located at both sides of the longitudinal inversion line, while in quadrupolar configurations four appear. The thickness of QSLs has been shown to be determined by the character (bipolar or quadrupolar) of the magnetic region, the intensity of the coronal currents and by the size of the photospheric field concentrations.
The next step, presented in this paper, is to apply the QSLM to different observed flaring configurations and to compare its results to observed features like flare ribbons or flare kernels. The QSLM and the extrapolation technique are briefly described in Sect. 2. We study a variety of configurations, ranging from quadrupolar to bipolar and from nearly potential to sheared ones, in order to show that QSLs are indeed a feature common to all flaring regions. This is illustrated here by applying the QSLM to five regions selected from the set previously studied with the SM (Sect. 3). We then confront present flare models to our results (Sect. 4). We conclude that energy is released in flares by magnetic reconnection as described in the recent 3D theoretical developments (Sect. 5).
© European Southern Observatory (ESO) 1997
Online publication: May 5, 1998