The 1991 March 13 flare around 08 UTC occurred in the NOAA active region 6545 at S10-E44. Fig. 1 displays the temporal evolution of the H intensity for two kernels N1 and N2 (see Sect. 3.1 below), of the HXR count rate and of the microwave flux densities. The HXR and microwave time profiles exhibit two impulsive bursts. The first and weaker burst B1 started at 08:01:25 UTC, while the second and more intense one B2, started at 08:03:30 UTC. The H emission shows two rises corresponding to those of the HXR count rate and of the microwave emission, but decays on much longer time scales.
3.1. H observations
Fig. 2 shows snapshot maps of the H emission taken during B1 and B2. The first frame displays the Kitt Peak magnetogram rotated to the time of the H observations, and overlaid on the preflare H image. During B1 and B2, the bulk of the H emission arises from two ribbons, located on each side of a magnetic neutral line (dotted line in the frame at 08:01:48.6 UTC). Bright kernels are visible within the two ribbons. They are labelled N1 and N2 for the northern ribbon and S1, S2 for the southern ribbon which respectively overlay positive and negative magnetic polarities. As the flare evolves the two ribbons expand further away from the magnetic neutral line. H full Sun patrol observations obtained every minute by the 3 H heliograph at Paris-Meudon observatory (courtesy of Z. Mouradian) indicate that until 08:05 UTC the H flaring emission is entirely contained in the FOV of the fast H camera. After 08:05 UTC, i.e. at the end of B2, the northern ribbon extends south-westward from N2, outside the FOV of the fast camera. The intensity of each of the four kernels has been computed as a function of time by averaging over a square of 4.8 arcsec2 (6 pixels). A change of the averaging area by a factor of 2 only influences the amplitude but not the shape of the temporal evolution. The time evolution of the H intensity, shown for N1 and N2 in Fig. 1, is globally similar for the four kernels. In particular, the emission of each of the four kernels shows rapid rises corresponding to those of B1 and B2.
3.2. HXR/GR and microwave observations
Spectral analysis of the PHEBUS data has been performed for count spectra accumulated over the four intervals of time (T1 to T4) marked by vertical dashed line on the 73 keV time profile shown in Fig. 1. During T1 and T2, which correspond to B1 and the initial slow rise of B2, no significant emission is detected for photon energies (h) above 0.2 MeV. The photon spectrum was fitted to a single power law given by , where h is in MeV, A (photons cm-2 MeV-1 s-1) is the photon flux at 0.1 MeV and the power law index. During T3 and T4, GR emission is detected up to 8 to 10 MeV, and marginally significant 2.23 MeV and prompt gamma-ray line emissions are present. This time period (T3 and T4) starts with the rapid rise of the 73 keV emission which follows the initial slow rise of B2 and covers the two main peaks of B2 till the slow decay of the HXR/GR and microwave emissions. For these two peaks, the photon spectrum of the continuum is well represented by a broken power law. The corresponding parameters are then A and as before, the break energy and the power law index for h . The values of A, , and are given in Table 1 for the four intervals of time under consideration. Table 1 also shows the power P100 (resp. P20) deposited in the chromosphere by 100 keV (resp. 20 keV) HXR producing electrons. P100 has been estimated by assuming that the bulk of the HXR emission is the result of thick target bremsstrahlung of 100 keV electrons and by using the low energy part of the photon spectrum and thick target calculations by Hudson, Canfield & Kane (1978). P20 has been obtained by extrapolating the 73 keV HXR photon spectrum down to 20 keV.
Fig. 3 displays the microwave spectrum during the five intervals of time labelled T1, T2, T3, T4 and T5 (Fig. 1). A detailed discussion of the time evolution of the microwave spectrum, which is beyond the scope of this paper, has been given in Melnikov & Magun (1998) for the second burst B2. Fig. 3 shows that the radio emission at 35 and 50 GHz remains optically thin during the whole flare. Taking the shape of the optically thin part of the microwave spectrum as (where is the radio frequency) we determined from the ratio of the flux densities measured at 35 and 50 GHz. The spectral index of the microwave producing electrons is approximatly given by (Dulk & Marsh 1982). For thick target interactions the spectal index of the HXR/GR producing electrons is given by where is the slope of the HXR/GR photon spectrum (e.g. Brown 1971). As a first approximation we take below and above (see Trottet et al. 1998).The values of , , and obtained during T1, T2, T3 and T4 are given in Table 2 which also provides the turnover frequency of the corresponding microwave spectrum. The comparison of with indicates that:
These characteristics are similar to those found in other large HXR/GR flares (e.g. Trottet et al. 1998and references therein). Similar results have also been deduced from the comparison of the HXR and radio millimeter emissions during smaller flares for which the sensitivity of HXR/GR detectors is not sufficient to measure the high energy part (h) of the HXR/GR emission (e.g. Kundu et al. 1994; Silva et al. 1997; Raulin et al. 1999).
Fig. 3 shows that the shape of the microwave spectrum of B1 and B2 is indicative for an inhomogeneous radio source region that, for example, consists of magnetic field loops of different sizes. Indeed: (i) except during T1 the spectral rise is flater than (resp. ) as would be expected for an emission from relativistic (resp. midly-relativistic) electrons in a uniform magnetic field (e.g.Dulk & Marsh 1982); (ii) during T1, as well as during T3 and T5, the spectrum is rather flat around the turnover frequency which, again, is a signature of an inhomogeneous source (e.g. Klein & Trottet 1984). The turnover frequency at 20 GHz during B1 (T1) is shifted towards lower frequencies ( 12 GHz) during B2 (T2, T3, T4). Such a shift of requires that self absorption, Razin suppression or free-free absorption in the ambient plasma, become less effective at lower frequencies during B2. Thus, the shift of to lower frequencies reflects most likely a decrease of the magnetic field, of the ambient density and of the density of emitting electrons along the line of sight or a combination of the three. In any case this indicates that most of the microwave emission detected during B2 arises from more extended sources than during B1. It should be noted that during the late decay of B2 (T5) drifts towards even lower frequencies ( 8 GHz) and that the spectrum of microwave emitting electrons hardens (). This has been discussed by Melnikov & Magun (1998) in the framework of coronal trapping of the emitting electrons. During T5, the HXR/GR rates are too low to enable us to determine . However, is found to decrease with time which is also an indication for trapping of the hard X-ray emitting electrons with lower energies than those responsible for the optically thin microwave emission.
In summary, the above comparison between the HXR/GR and microwave emissions provides evidence indicating that:
3.3. Comparison between H and HXR/GR observations
As stated in Sect. 3.1 the four H kernels exhibit a similar time evolution. During both B1 and B2 the H time profile of each kernel exhibits a fast component that is superposed on a slowly varying component. The fast component shows two rises corresponding to those of the 73 keV HXR count rate. This led us to assume that at each time t in a given interval of time from t1 to t2, and for each kernel i, the time history of the relative H intensity is related to the HXR time evolution by:
where is the relative H intensity of kernel i, its relative level before the burst (B1 or B2), and are constants, the instantaneous 73 keV HXR count rate and its maximum value between t1 and t2. The indices i=1, 2, 3, 4 mark the kernels N2, N1, S1 and S2 respectively.
The above relation assumes that:
Eq. 1 has been applied separately for B1 and B2. The free parameters and have been determined for each kernel by using a minimization algorithm.
The interval of time considered for B1 is T1. Fig. 4 (left) shows that, during B1, the observed H time profile (thin line) of each of the four kernels is nicely reproduced by that modelled (thick line) with the simple two parameter model defined by Eq. 1 ( 0.4 - 0.7). The dashed lines show the slowly varying component modelled for each kernel.
During B2 we have considered two successive injections of HXR emitting electrons, taken as (t) over T2 and over T3 to T5 respectively. The first injection corresponds to the initial slow rise of B2 whereas the second one, which starts with the rapid HXR and GR rise, covers the most energetic part of B2. However, only a single slowly varying term has been computed, as due to the low HXR count rate during T2, relative to that during T3 and T4, the splitting of the slowly varying component into two parts would not significantly affect the results. For each kernel the free parameters are then: , for the fast components during T2 and T3, T4 and T5 respectively and for the slow component during the whole B2 burst. The results obtained for B2 are shown in Fig. 4 (right). Here again, there is a good agreement ( 0.5-0.9) between observed and computed time profiles. It should be noticed that ( 2.5) increases significantly when only one single electron injection over the time period, consisting of T2, T3, T4 and T5, is taken into acount.
© European Southern Observatory (ESO) 2000
Online publication: April 17, 2000