5. Observational results for WATCH GRB 920925c
As it was mentioned in the previous section, the search for emission from GRB 920925c had a surprising result. An excess was found searching for emission in coincidence with the WATCH trigger on all short time scales. No excess has been found either in the search for delayed emission or in the 1 hour time window. The results for the four time scales, in the two search modes (coincident and delayed emission), are shown in Fig. 4. Again the cumulative number of trials is plotted as a function of -log and the dashed line shows the results for the MC sample. In this case significant deviations from the expectations are observed, especially in the 4 minute time scale (coincident emission), which shows a deviation with a chance probability .
The search interval with the smallest chance probability (4 minutes time scale, coincident emission) is centered at , (J2000) and UT = 22:45:21. Therefore it precedes the WATCH trigger by less than one minute and is about 9° away from the WATCH location. For this search interval we observe 11 events in 4 minutes while 0.93 events are expected. The chance probability for the background to yield such an excess, computed according to the formula shown above, is 7.7 and the Li and Ma significance (Li & Ma 1983) is 5.4 (in both cases we take into account only 10 events inside the search interval, see above). This is the most significant excess seen in the whole data sample (compare Fig. 4 with Fig. 2). Correcting this probability with a trial factor (using the MC) due to the search for the four GRBs in Table 2 on several time scales and in a large solid angle region, yields a final probability of 3.3 (2.7), i.e. the CL for this excess to be related to WATCH burst is 99.7 (neglecting the possibility of a second independent burst). The data registered with the HEGRA scintillator array does not show any excess, which may be due to its higher energy threshold and worse angular resolution. The time distribution of the events registered by AIROBICC in the position of the excess for the night of the GRB is plotted in Fig. 5. The distribution of the 11 events yielding the smallest chance probability is shown in detail in the inner part of the figure. Note that 7 out of the 11 burst events come within 22 seconds, 3 of which arrive within 0.25 seconds.
Assuming that this excess is due to high energy emission we can calculate the integral flux in the same way as the upper limits, but replacing with . In this case the tentative mean integral flux above 16 TeV during the 4 minutes window is (9 4) cm-2 s-1.
The burst as seen by the space detectors had a duration of 5 minutes exhibiting two main peaks. It was observed by WATCH and ULYSSES, thus reducing the possible locations to an IPN (InterPlanetary Network) annulus (Hurley 1996). This annulus is obtained using the relative time of detection of both spacecrafts. The IPN ring passes through the WATCH 3 error circle and is 3° away from the excess observed with the AIROBICC array, see Fig. 6.
Spectral data from WATCH in the interval 6-100 keV can be fitted by a power law spectrum. The fit yields a spectral index of 2.50.2 which, if naively extrapolated, does not predict any TeV flux detectable by AIROBICC. However, the TeV spectrum may differ significantly from what one expects from the 100 keV extrapolation. Furthermore, the hypothetical emission at TeV energies precedes the WATCH observations and may therefore be due to another production mechanism.
© European Southern Observatory (ESO) 1998
Online publication: August 6, 1998