Introduction

The detection of gamma ray bursts (GRBs) and other transients with BATSE is controlled by a real-time burst detection (or trigger) algorithm that runs onboard the spacecraft. When the onboard computer signals a burst trigger, data are recorded at high temporal and spectral resolution for a limited time interval. For all times when the detectors are on and telemetry is available, the count rates in the detectors are recorded at lower resolution in the continuous (or ``background'') data types (designated CONT, DISCSP, and DISCLA). The availability of nearly continuous archival data from the BATSE detectors permits a search for astronomically interesting transients that did not activate the onboard burst trigger. We refer to such transients as ``non-triggered events''.

A GRB or other transient may fail to cause an onboard burst trigger for any of several reasons. It may be too faint in the monitored energy range to meet the onboard burst trigger's minimum count rate threshold; it may occur while the onboard trigger is disabled for technical reasons; or it may bias the onboard background estimate by rising slowly compared to the 17.4 s onboard background averaging. Some of these non-triggered events can be detected by searching the archival continuous data with lower detection thresholds and different background averaging schemes than the ones used onboard the satellite. The resulting catalog of non-triggered GRBs can increase the total number of GRBs detected with BATSE and extend knowledge of the peak flux distribution to lower peak fluxes. It can also provide some understanding of the extent to which the details of the onboard trigger mechanism have affected the sample of GRBs listed in the BATSE catalogs.

We have searched 6 years of archival DISCLA data from BATSE for GRBs and other astronomically interesting transients that did not activate the onboard burst trigger. This web site contains an on-line presentation of some of the events we detected.

The Off-line Search

The strategy for searching the archival continuous data is much like the one used onboard: examine the count rates in the detectors and look for statistically significant increases. The procedures we used are described the Ph.D. thesis by Jeff Kommers (PDF [2328 Kb], PostScript [7758 Kb]). Briefly, we use lower thresholds (in terms of statistical signifcance) so that we can detect GRBs and other transients that have peak fluxes (on the 1.024 s time scale) that are a factor of about 2 fainter than can be detected onboard. We also search on longer time scales, 4.096 seconds and 8.192 seconds, for even greater sensitivity to long GRBs.

The scope of the off-line search is the 6 years of DISCLA data recorded between 9 December 1992 and 17 December 1997. This covers 2200 days, or 1.90E8 seconds, of the mission (TJDs 8600.0 to 10800.0). The total time for which data are available to search (using our methods) is less owing to data gaps: 1.33E8 seconds. The sky exposure of the search is anisotropic owing to Earth blockage and spacecraft passages through the SAA. It is given in tabular form on the sky exposure page.

Results

The off-line search detected 2265 events that appear to be GRBs. Of these, 1392 are onboard triggers that are already listed as GRBs in the BATSE trigger catalog. The remaining 873 events that did not activate the onboard burst trigger are the non-triggered GRB candidates. During the time period covered by the off-line search the onboard burst trigger detected 1815 GRBs. The off-line search did not detect 423 of these owing to gaps in the DISCLA data and the fact that the 1.024 second time resolution reduces the statistical significance of bursts with durations shorter than about 1 s. Thus the off-line search increases the number of GRBs detected with BATSE by 48% during the time period of the search.

One of the goals of the off-line search was to study the intensity distribution of GRBs using peak fluxes below BATSE's onboard detection threshold. On the 1.024 second time scale, the off-line search is sensitive to bursts that are about a factor of 2 fainter than can be detected with the onboard trigger in its usual configuration. A detailed look at the intensity distribution of the GRBs detected with our search is given in the manuscript of our paper, ``The Intensity Distribution of Faint Gamma-Ray Bursts Detected with BATSE,'' in press at the Astrophysical Journal (PostScript [343 Kb], PDF [250 Kb]).

Briefly, we find a surprising deficit of faint bursts that are detected only on the longer time scales of our search (4.096 seconds and 8.192 seconds). The mean value of the V/Vmax statistic (in Euclidean space) for the GRBs detected with the off-line search is 0.177 +/- 0.006. This is the lowest value ever found for a global sample of GRBs. It implies that increasing the volume within which we can detect long (duration greater than 8 seconds) bursts does not reveal many new burst sources. Of the 2265 GRBs detected with the off-line search, only 105 are detected exclusively on the 4.096 second or 8.192 second timescales. Put another way, searching on longer time scales only revealed 105 new bursts that were not already detected by the (less sensitive) 1.024 s search. A detailed comparison of the (1.024 second) peak flux distribution with cosmological models shows that it is consistent with scenarios in which the co-moving rate of GRBs traces the star formation history of the Universe. For these models, a fairly broad luminosity function is required. These results suggest that instruments more sensitive than BATSE are unlikely to uncover large numbers of ``classical'' GRBs below the BATSE threshold.

The off-line search also detected a modest number (24) of bursts from SGR 1806-20 and a modest number (50) of burst-like transients of unknown origin that could be spectrally soft GRBs, bursts and flares from X-ray binaries, or other phenomena. The 50 bursts of ``unknown'' origin have an isotropic direction distribution and their instrumental hardness ratios are consistent with all of them being the low-energy ``tail'' of the GRB spectral distribution. They do not occur uniformly in time, however, so this class of events may be a combination of very soft GRBs, bursts and flares from X-ray binaries, and Earth magneotspheric particle precipiations.