2 The INTEGRAL Burst Alert System IBAS  

  

Figure 1: The IBAS Software Architecture at ISDC

INTEGRAL will not have a GRB detection and triggering system on board. However, it will downlink its acquired data continuously to earth allowing for constant near-real time monitoring. At the ISDC all data will be automatically analyzed to detect any transient events. In addition, a fast analysis (Table 2) will be performed by the INTEGRAL Burst Alert System IBAS [(Mereghetti et al. 1999)]. This "on ground'' approach to detection not only allows for the application of larger computational power than available on board a spacecraft, but also permits the implementation of several detection algorithms running in parallel (Fig.1).

  

Figure 2: Light curve and deconvolved images of a simulated GRB in the ISGRI fully coded field of view (ISGRI is a detector layer of IBIS). The upper image row displays 15-300 keV images of 1s integration each, while the images in the lower row have been obtained integrating over increasingly longer time intervals (from 1 to 10s). This shows how rapidly it is possible to obtain a good signal to noise level for a typical GRB

After receiving the INTEGRAL telemetry at ISDC, the IBAS relevant data is extracted and fed into the attitude determination and into the several detection processes D_n running in parallel. As soon as a GRB candidate event is detected, it must pass a verification process and a final screening, which is additionally in charge of spawning a more detailed offline analysis of the burst; the exact verification algorithm is still to be defined. The GRB position and trigger time then reach the alert generation process, and the information is broadcast electronically.

Ongoing IBAS simulations currently concentrate on GRB detection with IBIS, since its photon by photon mode data is expected to yield the best position accuracy. During the simulations photon arrival times are generated based on light curves actually observed by BATSE. The simulations shown in Fig.2 e.g. reveal the GRB to be localizable less than 2s after the trigger time $(S/N \sim 20)$.This has been achieved using a simplified prototype code: The countrates of the incoming events are binned, averaged in time ($R_\mathrm{bin}$, $t_\mathrm{bin}$ and $R_\mathrm{avg}$, $t_\mathrm{avg}$ respectively) and compared, taking the deviation $\sigma$ into account. The trigger time $t_\mathrm{trig}$ is recorded as soon as a threshold is passed ($R_\mathrm{bin}\gt R_\mathrm{avg} + n\sigma$). Images are deconvolved (preburst and integrated ones over $t_\mathrm{image}$time bins) and the position is found. The simulation parameters for Fig.2 were n = 7; $t_\mathrm{bin} = 50$ms; $t_\mathrm{trig} = 11.7$s; $t_\mathrm{image} = 1$s. It was based on BATSE trigger #2321 featuring a peak flux of 0.85ph/(cm²s) in the 50-300 keV channel. A broken power law spectrum is assumed with photon index $\alpha_{1}=0.7$ and $\alpha_2=2.0$ and with break energy $E_\mathrm{break}=100$ keV.

 

Table 2:   The breakdown of the anticipated full delay with IBAS. The largest uncertainty in this chain lies in the time it will take the groundstations (GS) to transfer the spacecraft's (S/C) telemetry frames to ESA's Mission Operation Centre, and from MOC to the INTEGRAL Science Data Centre ISDC

INTEGRAL is expected to detect about 20 GRBs per year within the IBIS and SPI fields of view [(Pedersen et al. 1997)]. Localization accuracy is a function of the event's S/N ratio, the spacecraft attitude and stability, the instrument to star-tracker alignment and the instrument angular resolution. The attitude accuracy will be $\leq$30$^{\prime\prime}$ during stable pointings, i.e. for most of the time (during slews it will be $\sim 10\hbox{$^\prime$}$). For a $10\sigma$ source detected with IBIS the source location accuracy is $\sim 1\hbox{$^\prime$}$.The use of OMC data for improved attitude information is under study, as OMC will provide offset values $\Delta Y$ and $\Delta Z$ to the central source position with an anticipated accuracy of $<17\hbox{$^{\prime\prime}$}$.

Although SPI's localization accuracy will be significantly worse than that of IBIS, SPI data will be used to assess the validity of the event. Additionally, SPI may provide localization for those bursts at large off axis angles. For those GRBs SPI's sensitivity is better than that of IBIS due to the larger fully coded field of view.

In the relevant time frame of 2001 to 2003 INTEGRAL seems to be the satellite best suited as the Interplanetary Network's near earth node. As an optimized input to the IPN, SPI's anticoincidence shield (ACS) will take data in time bins of 50ms, time tagged to an accuracy of 1ms. Thus the data of $\sim 300$ ($5\sigma$) bursts per year, located mainly perpendicular to the instruments' fields of view, can usefully contribute to the IPN ([Hurley 1999]; [Teegarden & Sturner 1999]).