Astron. Astrophys. 336, 57-62 (1998)

1. Introduction

Despite the advances carried out so far, the origin of the gamma-ray bursts (hereafter GRBs) remains unknown. The identification of absorption lines in the optical spectrum of GRB 970508 strongly supports models arising from sources at cosmological distances (Metzger et al. 1997), but there is still a lack of knowledge on the mechanisms originating these enigmatic phenomena. One of the most important clues that could clarify the nature of the GRBs would be the detection of a repeater behaviour.

Initial studies showed an apparent evidence of repetition for the BATSE 1B catalogue (Quashnock and Lamb 1993), suggesting that it would be possible to have an excess of pairs of GRBs clustered in both time and space (Wang and Lingenfelter 1995). This fact was not confirmed by the work carried out using the BATSE 2B catalogue (Brainerd et al. 1995), although other studies provided marginal evidence for both temporal and angular clustering (Petrosian and Efron 1995). Analyses based on autocorrelations with data from the BATSE 3B catalogue did not find any evidence of repetition (Bennett and Rhie 1996) and have imposed several constraints to the number of repeaters (Tegmark et al. 1996). Finally, recent studies confirm the lack of repetition in the 4B catalogue and lead to an upper limit to the repetition rate of 0.04 burst source^-1 yr^-1 (Hakkila et al. 1997).

The BATSE 4B catalogue was obtained by the BATSE experiment on board the CGRO satellite and contains 1637 GRBs detected from April 1991 to August 1996 (Paciesas et al. 1998). The BATSE experiment consists of eight identical detector modules, placed at the corners of the CGRO spacecraft and covering energy channels from keV to 2 MeV. It provides error boxes with a minimum radius of (1 confidence level, Fishman et al. 1994). BATSE is detecting bursts at a rate of 0.8 bursts per day. The bursts are daily added to the so-called Current GRB Catalogue, which contains the BATSE 4B catalogue plus all bursts detected after August 1996. When this study was started, the catalogue contained 1905 sources; this sample constitutes the basis of the present work.

The WATCH X-ray all-sky monitor is based on the rotation modulation principle (Lund 1986). The instrument has a circular field of view of 4 steradians and an effective area of 30 cm² (averaged over the field of view). Position sensitivity is achieved using the rotation collimator principle, with the collimator grids rotating with a frequency =1 Hz. The phoswich detectors consist of interleaved scintillator-strips of NaI and CsI crystals. The geometric area of the scintillator is 95 cm². Four units were mounted on board the Soviet GRANAT satellite in a tetrahedral configuration covering the whole sky, and one unit on board the European Space Agency EURECA spacecraft. The total energy range is 8-80 keV, therefore overlapping with the lower BATSE energy band. WATCH/GRANAT detected bursts in 1990-94 and WATCH/EURECA in 1992-93, thus both experiments also overlapped in time with BATSE. One of the main advantages of WATCH was the capability of locating bursts with relatively small error boxes ( error radii with 1^o) (Brandt et al. 1990). WATCH/GRANAT detected 47 GRBs in this period and WATCH/EURECA 12 (Castro-Tirado et al. 1994, Brandt et al. 1994, Sazonov et al. 1998). Two GRBs (GRB 920814 and GRB 921022) were detected by both the WATCH/GRANAT and WATCH/EURECA experiments. Therefore, the sample of WATCH GRBs used in this study comprises 57 GRBs: 45 WATCH/GRANAT bursts, 10 WATCH/EURECA bursts and the above-mentioned two GRBs. BATSE also detected 27 of them. Fig. 1 shows the sample of 57 WATCH GRBs used in this study.

Fig. 1. Error boxes for the 57 GRBs detected by WATCH, represented in galactic coordinates. The sample contains 45 GRBs detected by WATCH/GRANAT, 10 by WATCH/EURECA and two localized by both experiments at the same time. The typical radii of the error boxes are , with a confidence level.

The distribution of time amplitudes for GRBs shows two classes of bursts: a) durations shorter than 2 s and b) longer than 2 s (Kouveliotou et al. 1993). It was noticed that the energy spectra of the short bursts were generally harder than those of the long ones (Kouveliotou et al. 1993, Lestrade et al. 1993).

The fraction of short events in the WATCH sample is smaller than that in the 4B catalogue. This fact can be justified by at least three selection effects:

i) The availability of WATCH for localizing sources is governed by the rotation speed of the collimator grids (1 Hz). So, a source needs to be bright enough for at least one rotation of the modulation collimator in order to be localized, implying a burst duration longer than 1 s. In contrast, the BATSE experiment is able to detect bursts with durations as short as 64 ms.

ii) The low energy band of the WATCH experiment (8-20 KeV) is sensitive to the soft GRBs, below the BATSE lower limit (25 KeV), which generally belong to the class of bursts with durations longer than 2 s.

iii) On the other hand, since WATCH is about an order of magnitude less sensitive than the large-area detectors of BATSE, the WATCH catalogue contains bursts which are brighter than those in the BATSE sample.

The above three reasons explain why the GRBs in the WATCH sample are longer, softer and brighter than the average BATSE 4B bursts.

This study is the first known attempt to search for repeaters combining data of -ray experiments flying on board different satellites. The method proposed makes use of the so-called "simultaneous bursts" and is suitable to correlate GRB data provided by experiments that overlap partially or totally in time. In the future, this work could also be used to detect systematic pointing errors between different -ray experiments, allowing to improve their capability for locating GRBs.

© European Southern Observatory (ESO) 1998

Online publication: July 7, 1998
helpdesk.link@springer.de