Astron. Astrophys. 334, 299-313 (1998)

3. Observations

The WATCH solar burst catalogue (Astronomy and Astrophysics Supplement Series) was created by systematically going through the WATCH-0 database with the following selection criteria:

1) The modulation pattern must correspond to the solar one (i.e. it must be centered at 0, see Sect.2.2.2).

2) There must be a signature in the GOES data (a classified GOES flare or a significant flux enhancement). The corresponding WATCH peak is not necessarily at the same time as that of the GOES event. Note that in general this second criterium is redundant with the first one.

3) The peak count rate of the WATCH event must be above the three sigma background level.

4) The event must not be significantly contaminated by particles.

A total of 1551 events were found, where approximately 45 percent of the events are associated with GOES X-ray flare classification (607 C-flares, 84 M-flares and 2 X-flares). 27 of these events were observed to have signals above background in both the low- and high-energy channels.

Furthermore, each WATCH solar burst was tentatively associated with an active region with the following selection criteria:
- the WATCH event is associated with a reported GOES flare associated with an active region in Solar Geophysical Data (comprehensive reports).
- the WATCH event is not associated with a reported GOES flare, but is simultaneously detected with an optical flare. The X-ray burst may begin a few minutes before the optical flare ending either during or after it, or it may begin during the optical flare and end before it. Only one optical flare must occur at the same time as the WATCH event for the association to be considered as valid.

A more detailed description of all the parameters accompanies the catalogue.

The top pannel of Fig. 4 shows the flaring rate as deduced from the WATCH-0 observations, where the on-time (total duration of WATCH observations for the month) is computed in days. The bottom pannel is the flaring rate as deduced from GOES. The flaring rate recorded by WATCH in the second observing period (month 16 to 23) decreases as expected from the shift of the energy channels described in Sect. 2.3.

Fig. 4. Top pannel: The monthly flaring rate recorded by WATCH-0. Bottom pannel: The corresponding monthly flaring rate recorded by GOES.

3.1. Solar X-ray bursts observed by WATCH

The bottom of Fig. 5 shows the time profile recorded in the low-energy channel of WATCH on the 19 June 1990 between 15:00 and 16:20 UT (the dotted line represents the background). The two upper pannels of the figure show complementary GOES observations in two energy channels (the dotted line also represents the background - see below in Sect. 3.2). No signal above background is observed in the high-energy channel of WATCH. The first burst (a) observed at peaktime 15:23:24 UT is associated with an optical flare and some significant GOES emission, which is however not reported as a GOES flare in 'Solar Geophysical Data'. The two bursts that follow (peaktimes: 15:57:16 UT (b) and 16:14:26 UT (c)) are both associated with a C1.6 GOES flare and a common optical flare. The first two peaks (1 and 2) in the burst (b) are seen as bumps in the GOES time profile, whereas the third peak (3) is close to the peak of the event observed at lower energies by GOES. Note that for this burst as for others the peak count rate as well as the peak time indicated in the WATCH catalogue corresponds to the largest peak detected in the low-energy channel of WATCH (here peak 2).

Fig. 5. Time profiles of the three WATCH bursts (bottom) observed between 15:00 and 16:20 UT on 1990 June 19 with the corresponding GOES time profiles (see text for details).

For some events, complementary observations at energies above 25 keV have been found originating e.g. from the Burst and Transient Source Experiment (BATSE) aboard the Gamma-Ray Observatory (GRO) (Fishman et al. 1989). The WATCH event presented in Fig. 6 corresponds to an impulsive non-thermal weak and short duration burst also observed with BATSE in the rising phase of the GOES flare. However no signal in the high-energy channel of WATCH was observed, probably because of the smaller detector area of WATCH compared to BATSE.

Fig. 6. Time profiles of a burst (1991 August 17) observed simultaneously by WATCH and BATSE in the rising phase of a GOES flare.

3.2. Nature of the Deka-keV emission observed by WATCH

As a first investigation of the nature of the emission observed by WATCH above 10 keV, it is checked whether the emission results only from the contribution of the thermal plasma measured by GOES. Fig. 7 is a schematic representation of how this was done.

Fig. 7. A schematic representation of the simulation of the combined WATCH and GOES observations.

Assuming that the emission measured in the two GOES soft X-ray ranges (1.5-12.4 keV and 3.1-24.8 keV) is produced by a homogeneous isothermal plasma, those observations are used to determine the parameters (temperature and emission measure) of this soft X-ray emitting plasma. Formulae relevant for the GOES-1 satellite data have been derived by Thomas et al. (1985). However a significant difference in detector efficiencies between the experiments aboard the different GOES satellites has been found by Sylwester et al. (1995), who have derived conversion factors between the different experiments. Following the suggestion by Garcia (1994), the GOES wavelength-averaged efficiencies given by Sylwester et al. (1995) are used to convert the GOES-7 fluxes into GOES-1 fluxes before they are used in the formulae of Thomas et al. (1985).

Two methods are used to determine the pre-event background of the GOES flux observations. 1) A rough estimation of the pre-event flux level which represents an upper limit of the background level. 2) The technique developed by Bornmann (1990) which determines the background by stating that the temperature, emission measure and flux should all increase at the start of the soft X-ray brightenings and that the flaring X-ray sources is hotter than the average of all background sources. It is found that the background levels and thus the temperatures and emission measures determined with the above two methods are consistent.

The temperature and emission measure inferred from GOES give an estimate of the thermal spectrum impinging on the WATCH detector as a function of time. Using the detector response that was described in the previous section, it is then possible to simulate the time evolution of the WATCH count rate produced by the thermal plasma. For all cases that were analyzed, the simulated count rate was always smaller than the observed one indicating that, as expected, the emission cannot be represented by an isothermal plasma alone. The simulated time profile is then subtracted from the measured WATCH count rate and what is left over is called the 'residual' (Fig. 7). Fig. 8 shows the measured count rate in the low-energy channel (full-line) compared with the simulated one (dotted line) for four events. In Fig. 8a-c the simulated count rates represent respectively 50 , 20 , 0 of the detected count rate. The ratio between the detected and simulated count rates thus vary from one event to the other. It may also vary a lot in the course of a burst (as for the 21-03-1992 burst in Fig. 8d) and is found to be generally higher at the rise phase of a burst. This confirms either the usual multithermal nature of the flaring plasma or shows the frequent production of suprathermal electrons. It also confirms the variation from one burst to the other or even in the course of a burst of the relative contribution of hot plasmas or of suprathermal electrons.

Fig. 8. Measured (full line) and simulated (dotted line) WATCH count rates for four different events (see text for details).

In some cases (see Fig. 8a) up to 50 of the low-energy count-rate of WATCH can be produced by the contribution of the thermal plasma measured by GOES and no signature is detected in the WATCH high energy channel. In those cases, two models have been used to analyse the combined GOES and WATCH data (see Crosby et al. 1996 and Crosby 1996 for details and examples to reproduce the combined observations of WATCH and GOES). The first of this model is based on the assumption that the flaring plasma can be represented by two thermal components and the two sets of temperature and emission measure are determined using the following guidelines:

- the high temperature component must not give a significant signal in the high-energy WATCH channel.
- the low temperature component accounts for the low-energy GOES channel.
- both components account for the 3.1-24.8 keV GOES channel and the low-energy channel of WATCH.
This analysis has been performed for the two events discussed in Crosby et al. (1996) (see Table 3a) and it was found that the hot plasma could be represented by a temperature in the range of 1.1-1.4 10⁷ K with an emission measure of 3-12 10⁴⁷ cm^-3, while the "cold" plasma had a temperature ranging between 5 and 7 10⁶ K with an emission measure roughly ten times larger than the hot plasma. The temperatures of the hot plasma are found to be in the same range as those deduced by combined YOHKOH HXT and SXT observations (Hudson et al. 1994) for impulsive footpoint brightenings.

Table 3. Spectral analysis of the two WATCH events described in Crosby et al. (1996)

In some cases (e.g. the first peak of the 1992-03-21 event shown on Fig. 8d), the combined observations of GOES and WATCH cannot be reproduced by this above model (see also Crosby et al., 1996). The 'residual' in the low-energy WATCH channel must then be interpreted with the alternative model in terms of a non-thermal photon spectrum. In the following, this second model is used to analyse in a systematic way the WATCH observations. The non-thermal spectrum is determined using the following constraints:

- the residual in the low-energy WATCH channel must be reproduced as well as the observations in the high-energy channel.
- if no significant emission is detected in the WATCH high-energy channel, only a limit of the photon spectral index is provided (see Table 3b) unless simultaneous observations of the event by other experiments are used to bring further constraints.

Table 3b summarize the results of the above analysis for the events studied in Crosby et al (1996) where no emission in the WATCH high-energy channel was observed and where simultaneous detection of the first peak of the 1992-03-21 event by BATSE was used as a further constraint. The results of the spectral analysis of all the WATCH events with significant flux in the high energy channels are similarly summarized in Table 4.

Table 4. Spectral analysis of purely non-thermal events (observed in both energy channels of WATCH.

Several peaks can be analyzed in these events, but we restrict ourselves to those for which no instrumental effects such as overflow, dead-time, or pulse pile-up are present. The slope of the non-thermal photon spectrum is found to be between -5.5 and -3.5 in agreement with values previously found in the literature in this energy range (Kane & Anderson 1970, Pan et al. 1984) as well as above 20 keV (Crosby et al. 1993; Bromund et al. 1995). The peak hard X-ray flux lies between 1 to 10 photons/(cm² s keV) at 20 keV for these events and is about one order of magnitude lower for the events studied in Crosby et al (1996) (Table 3b). These values lie towards the lower end of what was observed for HXRBS (Crosby et al. 1993). The peak energy flux in electrons above 25 keV is also computed assuming a thick-target model for the X-ray production. The values above 25 keV are found to lie around 10²⁶ - 10²⁷ ergs/s. Compared to the values found for the HXRBS observations (Crosby et al. 1993), the hardest bursts detected by WATCH correspond to an energy content at the lower limit or below the range detected by HXRBS. This suggests that the smallest and/or softest of the events detected by WATCH may belong to the same scale of events as the microflares detected by Lin et al. (1981).

Comparable time profiles sometimes observed around 10 keV with WATCH and at higher energy also suggest that the emission may be attributed to non-thermal origin even down to 10 keV. The Neupert effect generalized as the relationship between the time derivative of the soft X-ray time profile with the hard X-ray light curve (Dennis & Zarro 1993) is also observed in some cases between WATCH time profiles and the GOES ones around a few keV (Vollmer 1995). Fig. 9 illustrates such an example for an event where significant emission is observed in both channels of WATCH and a purely non-thermal photon spectrum has been derived (see Table 4). The correlation coefficient between the derivative of the two GOES channels with the low-energy channel of WATCH is found to be around 0.9, while it is around 0.76 with the high-energy channel of WATCH. This shows an apparent causal relationship between the soft X-ray emission observed by GOES and the deka-keV emission observed by WATCH. Such a causal relationship has been extensively studied at higher X-ray energies (e.g. Dennis & Zarro, 1993). A commonly proposed interpretation of this effect suggests that the non-thermal energetic electrons radiating hard X-ray emission are also at the origin of the heating and subsequent thermal bremsstrahlung observed in soft X-rays (e.g. Dennis & Zarro, 1993). It must however be pointed out that the relationship between the time profiles is not sufficient to support the causal relationship which must be further demonstrated by a comparison of the energy contained in the non-thermal electrons and radiated by the hot plasma. The present observations of a Neupert effect between the deka-keV and the lower energy time profiles may simply suggest that in some events the observations in the deka-keV range are a good indicator of the primary energy release in the flare and not of the thermal response of the medium. Thus the deka-keV X-ray energy range can be potentially used as a diagnostic of the flare energy release process.

Fig. 9. The derivative of the two GOES time profiles compared with the WATCH count rates for the 1992 January 29 event. The top two figures (a and b) represent the low-energy time profile of WATCH (full line) and the bottom two figures (c and d) the high-energy time profiles of WATCH (full line). The dotted lines are the derivatives of respectively the low-energy channel of GOES (right-side) and of the high-energy one (left-side).

3.3. Conclusions

The present section shows some examples of solar bursts observed by WATCH around 10 keV. Depending on the event or on parts of the event, the light curves resemble either those of the soft X-ray emission observed by GOES or those of the non-thermal hard X-ray emissions observed e.g. by BATSE. Several deka-keV WATCH events may be observed during a single GOES soft X-ray flare. The injection of energy above 10 keV occurs thus throughout a soft X-ray flare and not only at the rise phase of the flare. This was sometimes observed at higher X-ray energies by HXRBS/SMM, but less systematically. A lot of small soft X-ray enhancements not classified as GOES flares are associated with small bursts observed by WATCH around 10 keV. The production of suprathermal populations or of hot components in the corona is thus not limited to the "usual" GOES soft X-ray flares classified in 'Solar Geophysical Data'. The spectral analysis performed on some events clearly shows that the emission around 10 keV does not only result from the plasma detected around a few keV by GOES. A hotter component or a non-thermal population is required in the solar corona to produce this emission. As already suggested from the previous analysis of a few flares (Hernandez et al. 1986, Kane et al. 1992), the non-thermal electron population thus extends down to 10 keV in many events.

© European Southern Observatory (ESO) 1998

Online publication: May 12, 1998

helpdesk.link@springer.de