1 Introduction

The discovery of impulsive loop-top hard X-ray (HXR) emission sources (Masuda 1994; Masuda et al. 1994, 1995) is one of the most important results of the Yohkoh mission. Although the sources show different observational properties (Masuda 1994), the loop-top source detected during the impulsive phase of the M2.0 flare of 13 January 1992, often called "the Masuda flare'', has been considered to be a typical example. This source was located well above the apex of the soft X-ray (SXR) loop, where the bright loop-top kernel was developed, inside a high-temperature region seen in a temperature map derived from the Soft X-ray Telescope (SXT) images. The reported source had an impulsive time profile similar to that of the footpoint HXR sources. Its HXR energy spectrum, derived from the Hard X-ray Telescope (HXT) images which were synthesized within approximately 40-s time period starting from 17:26:52 UT, showed high-energy characteristics: 203 MK temperature (if a thermal model is assumed) or single power-law distribution with the index of 2.6 (if a non-thermal model is assumed). The simultaneously observed two HXR emission sources at footpoints were about five times stronger but had similar spectra.

Metcalf et al. (1996) pointed out that the quality of HXT images can be spoiled by spurious sources, especially in the case of weak sources having a low signal-to-noise ratio. Therefore, they have applied for the HXT image reconstruction the pixon algorithm (Pi $\tilde{\rm n}$a & Puetter 1993) which gives fewer spurious sources than the classical Maximum Entropy Method. Using the pixon algorithm Alexander & Metcalf (1997) have repeated the HXT image reconstruction for the Masuda flare. These authors have confirmed the existence of the high-energy loop-top source, but they note that analysis of this source requires long integration times ($\sim$ a few tens of second) due to poor statistics. As a consequence, it is impossible to investigate the impulsive behavior of this source.

For several years investigations of many researchers aimed to propose a consistent physical model which would be able to explain all observational aspects of the Masuda flare (e.g. Shibata et al. 1995; Tsuneta 1996; Aschwanden et al. 1996a; Tsuneta et al. 1997; Hori et al. 1997; Yokoyama & Shibata 1998). Nevertheless, there is no commonly accepted model of this flare. The main problem is matching the features of the impulsive loop-top HXR emission source with other observations.

 

 

Table 1: List of investigated behind-the-limb flares with occulted footpoints

		GOES data					Number of	Height [km]
No.	Date				Location	NOAA	degrees be-	of the	Refe-
		Time of	Class	Duration		AR	hind the so-	SXR loop-	rences
		max		[min]			lar limb	top kernel
1	21-Oct.-91	12:56	C7.8	20	S13Elimb	6891	$3^{\circ}$	4 103	a, b
2	10-Dec.-91	04:04	C9.3	20	S14Elimb	6968	$3^{\circ}$	4 103
3	13-Jan.-92	19:13	M1.3	100	S10Elimb	7012	$5^{\circ}$	1.0 104
4	6-Feb.-92	21:02	M4.1	140	N05Wlimb	7030	$4^{\circ}$	1.0 104
5	19-Feb.-92	14:58	C9.2	30	N06Elimb	7070	$5^{\circ}$	7 103
6	18-Jul.-92	13:44	M2.0	25	S11Wlimb	7222	$>0^{\circ}$	> 4 103
7	24-Nov.-92	10:05	C6.9	60	S07Wlimb	7342	$5^{\circ}$	7 103	c
8	24-Nov.-92	14:29	C5.9	80	S07Wlimb	7342	$7^{\circ}$	1.0 104	c
9	24-Nov.-92	16:04	C5.4	75	S07Wlimb	7342	$8^{\circ}$	1.4 104	c
10	24-Nov.-92	20:38	C6.4	140	S07Wlimb	7342	$10^{\circ}$	1.9 104	c
11	1-Feb.-93	02:04	M2.2	110	S10Elimb	7416	$6^{\circ}$	7 103	a
12	11-Nov.-93	11:26	C9.7	40	N10Elimb	7618	$5^{\circ}$	8 103	d
13	29-Jan.-94	11:29	M2.4	65	N07Wlimb	7654	$7^{\circ}$	1.8 104	a
14	29-Jan.-94	13:08	C6.5	55	N07Wlimb	7654	$8^{\circ}$	1.6 104

$^{\rm a}$ Mariska et al. (1996).

$^{\rm b}$ Mariska & Doschek (1997).

$^{\rm c}$ Khan et al. (1995).

$^{\rm d}$ Ohyama & Shibata (1997).

This may be due to instrumental reasons, i.e., the HXT image reconstruction methods may distort the actual parameters of the impulsive loop-top HXR emission source. Alexander & Metcalf (1997) identified the effect of weak source suppression during the HXT image reconstruction and found that even the best method of reconstruction cannot eliminate it completely. They warned that due to this effect the ability to study the weak sources of HXR emission in the presence of strong sources is problematic. This applies to the impulsive phase of the majority of flares, e.g. the Masuda flare, where the footpoints sources are distinctly stronger than the loop-top source.

As long as the available methods of the HXT image reconstruction cannot eliminate completely the effect of weak source suppression, the only safe way to study the impulsive loop-top hard X-ray emission sources in detail is to exclude somehow the stronger sources. In the case of the behind-the-limb flares the solar limb works as a screen which occults the lower part of the flaring structure and leaves emission of the higher part only.

Several papers analyzing individual behind-the-limb flares have been published (Khan et al. 1995; Mariska et al. 1996; Sterling et al. 1996; Mariska & Doschek 1997; Ohyama & Shibata 1997; Mariska & McTiernan 1999). The authors based their analyses mainly on the Yohkoh/Bragg Crystal Spectrometer (BCS) data. The main goal of this paper is to use such a specific geometrical configuration for a closer look at the loop-top HXR emission sources using the HXT data.