THE SECOND FERMI GBM GAMMA-RAY BURST CATALOG: THE FIRST FOUR YEARS

The ability of the GBM to observe GRBs in the energy range of the maximum energy release in the instrument reference system and to provide energy coverage up to energies of the main instrument is achieved by employing two different kinds of scintillation detectors. In the energy range from 8 keV to 1 MeV, sodium iodide detectors (NaI) read out by a photomultiplier tube are adopted. The capability to coarsely determine locations of triggered GRBs over the full unocculted sky is obtained by using 12 disk shaped NaI crystals, 12.7 cm diameter by 1.27 cm thick, each of which have a quasi-cosine response, and by arranging the NaI detectors around the spacecraft in such a way that each detector is observing the sky at a different inclination. The location of a GRB is calculated by comparing the measured individual detector counting rates with a lookup table (LUT), containing a list of relative detector rates for a grid of simulated sky locations. The on-board and on-ground LUTs have resolutions of 5° and 1°, respectively. With this method the limiting accuracy is approximately 8° for on-board locations and approximately 4° for on-ground locations. A detailed investigation of the GBM location accuracy can be found in Connaughton et al. (2014). For the detection of the prompt gamma-ray emission in the MeV-range, between ∼200 keV and 40 MeV, detectors employing the high Z high density scintillation material bismuth germanate (BGO) are used. Two detectors using large cylindrical BGO crystals, 12.7 cm diameter by 12.7 cm thick, each viewed by two photomultipliers, are mounted on opposite sides of the spacecraft, allowing observations of the full unocculted sky and providing spectral information up to the MeV regime for all GBM detected bright and hard GRBs. The GBM instrument is described in more detail in Meegan et al. (2009).

The GBM trigger algorithms implemented in the flight software (FSW) monitor the background count rates of all NaI detectors for the occurrence of a significant count rate increase in different energy ranges and timescales with adjustable sensitivities. A trigger is only generated in case of simultaneous exceedance of the trigger threshold of at least two detectors, thus reducing the probability for false triggers. The concept of the trigger algorithm was adopted from the predecessor instrument, the Burst and Transient Source Experiment (BATSE) on the Compton Gamma-ray Observatory, but with advancement in the number of parallel running algorithms. Compared to the BATSE FSW which allowed only for three algorithms, running at different timescales in one commandable energy band (Paciesas et al. 1999), the GBM FSW supports up to 119 trigger algorithms, 28 of which are currently in use. The parameters of all algorithms, i.e., integration time, energy channel range and time offset (see below), are adjustable by command. With the large number of algorithms and flexibility it is possible to investigate if the population of BATSE observed GRBs was eventually biased by the limited number of trigger algorithms. In addition, the capability was added to run a copy of a search algorithm which is offset in time compared to the original algorithm. From this an improvement in the trigger sensitivity (Band 2002; Band et al. 2004) is expected. The standard setting of the offset is half the timescale of the original algorithm. A summary of the actual settings (by 2012 July) and the changes in the first four years of the mission is shown in Table 2.

Since GBM triggers on events with a broad range of origins in addition to GRBs, the FSW performs an automatic event classification by using a Bayesian approach that considers the event localization, spectral hardness, and the spacecraft geomagnetic latitude (Meegan et al. 2009). This information is very important and useful for the automated follow up observations. Furthermore the capabilities of the instrument to detect events other than GRB events were improved by tuning dedicated trigger algorithms.

In case of a trigger the most important parameters for rapid ground based observations, i.e., onboard localization, event classification, burst intensity and background rates, are downlinked as TRIGDAT data by opening a real-time communication channel through the Tracking and Data Relay Satellite System. In addition, these data are used in near real-time by the Burst Alert Processor (BAP), redundant copies of which are running at the Fermi Mission Operations Center at GSFC and the GBM Instrument Operations Center (GIOC) at the National Space Science and Technology Center (NSSTC) in Huntsville, Alabama. Relative to the GBM FSW, the BAP provides improved locations, since it uses a finer angular grid (1°resolution) and accounts for differences in the burst spectra and more accurately for atmospheric and spacecraft scattering. Users worldwide are quickly informed within seconds about the flight and automatic ground locations and other important parameters by the automatic dissemination of notices (see Table 3 for the different kind of GBM notices) via the GRB Coordinates Network (GCN¹⁴). The GBM burst advocates (BAs), working in alternating 12 hr shifts at the GIOC and at the operations center MGIOC at the Max Planck Institute for Extraterrestrial Physics in Garching, Germany, use the TRIGDAT data to promptly confirm the event classification and generate refined localizations by applying improved background models. Unless a more precise localization of the same GRB has been reported by another instrument, a GCN notice with the final position and classification is disseminated by the BA. In addition the BAs compute preliminary durations, peak fluxes, fluences and spectral parameters, and report the results in a GCN circular in case of a bright event or a GRB that was already detected by another instrument. Trigger times and locations of GBM triggered GRBs are also passed directly to the LAT in order to launch dedicated onboard burst search algorithms for the detection of accompanying high-energy emission. In case of a sufficiently intense GRB, which exceeds a specific threshold for peak flux or fluence, a request for an autonomous repoint (ARR) of the spacecraft is transmitted to the LAT and forwarded to the spacecraft. This observation mode maintains the burst location in the LAT field of view for an extended duration (currently 2.5 hr, subject to Earth limb constraints), to search for delayed high-energy emission. Table 1 lists the number of ARRs which occurred in the first four years.

Table 3. GBM GCN Notice Types

GCN/FERMI	Sequence of Notices	Content/Purpose	Issues
_GBM_ALERT	1st, occurs directly after GBM trigger	Date, time, trigger criteria, trigger detection significance, algorithm used to make the detection	1
_GBM_FLT_POSITION	2nd	R.A., decl. GRB location, calculated by on-board flight software	1–5
_GBM_GND_POSITION	3rd	R.A., decl. GRB location, calculated by automated ground software	⩾0
_GBM_FINAL_POSITION	4th	H-i-t-l location. If the trigger is a GRB and it is not detected by an instrument with better location accuracy, a GBM final notice is sent within 2 hr	⩾0
_SC_SLEW	Only in case of an ARR	Indicates whether or not the spacecraft determined if it will slew to this burst. ∼1–3 month$^{{^{-1}}}$	1

Note. For more details see: http://gcn.gsfc.nasa.gov/fermi.html#tc2.

Download table as:  ASCIITypeset image

The continuous background count rates, recorded by each detector are downlinked as two complementary data types, the 256 ms high temporal resolution CTIME data with eight energy channels and the 4 s low temporal resolution CSPEC data with full spectral resolution of 128 energy channels which is used for spectroscopy. The LUTs used to define the boundaries of the CTIME and CSPEC spectral energy channels are pseudo-logarithmic so that the widths are commensurate with the detector resolution as a function of energy. In case of an on-board trigger the temporal resolutions of CTIME and CSPEC data are increased to 64 ms and 1 s, respectively, a mode lasting nominally for 600 s after the trigger time.

Moreover, high temporal and spectral resolution data are downlinked for each triggered event. These time-tagged event (TTE) data consist of individually recorded pulse height events with 2 μs temporal resolution and 128 channel spectral resolution from each of the 14 GBM detectors, recorded for 300 s after and about 30 s before trigger time. The benefit of this data type is the flexibility to adjust the temporal resolution to an optimal value with sufficient statistics for the analysis in question.

The GBM instrument, which is primarily designed to detect cosmic GRBs, additionally detects bursts originating from other cosmic sources, such as SFs and SGRs, as well as extremely short but spectrally hard TGFs observed from the Earth's atmosphere, which have been associated with lightning events in thunderstorms. Table 1 summarizes the numbers of triggers assigned to these additional event classes, showing that their total number is of the same order as the total number of triggered GRBs. Approximately 10% of the triggers are due mostly to cosmic rays or trapped particles; the latter typically occur in the entry region of the South Atlantic Anomaly (SAA) or at high geomagnetic latitude. In rare cases outbursts from known Galactic sources have caused triggers. Finally, ∼6% of the GBM triggers are generated accidentally by statistical fluctuations or are too weak to be confidently classified. The monthly trigger statistics over the first four years of the mission is graphically represented in Figure 1. The rate of GRBs is slightly lower in the second two years because at the beginning in 2011 July triggers were disabled during times when the spacecraft was at high geomagnetic latitude. It is evident from Figure 1 that the major bursting activity from SGR sources took place in the beginning of the mission, mainly in 2008 and 2009. In addition to emission from previously known SGR sources (von Kienlin et al. 2012; van der Horst et al. 2012; Lin et al. 2011), GBM also detected a new SGR source (van der Horst et al. 2010). It is also obvious from the figure that the rate of monthly detected triggers on TGF events has increased by a factor of ∼8 to about two per week, after the upload of the new FSW version on 2009 November 10 (Fishman et al. 2011). This version includes additional trigger algorithms that monitor the detector count rates of the BGO detectors in the 2–40 Mev energy range (see Table 2). This is advantageous because the TGF bursts show very hard spectra up 40 MeV, which also increases the deadtime in the NaI detectors (Briggs et al. 2013).

Zoom In Zoom Out Reset image size

Table 4 summarizes which trigger algorithm has triggered first on bursts or flares from the different object classes. Once a trigger has occurred the FSW continues to check the other trigger algorithms and ultimately sends back the information in TRIGDAT data as list of trigger times for all algorithms that triggered. This detailed information was already used in Paciesas et al. (2012) to investigate the apparent improvement in trigger sensitivity relative to BATSE. A breakdown of GBM GRBs which triggered on BATSE- and non-BATSE-like trigger algorithms, individually listed for the first and second catalog periods is shown in Table 5. It was found that mainly GBM's additional longer trigger timescales triggers (>1.024 s) in the 50–300 keV energy range were able to detect GRB events which wouldn't have triggered the BATSE experiment. These observations are confirmed by analyzing the current full four year data set. Furthermore we ascribe the improved trigger sensitivity to the in general lower trigger threshold of 4.5–5.0σ (see Table 2) compared to the BATSE settings (see Table 1 in Paciesas et al. 1999). The longest timescale trigger algorithms in the 50–300 keV energy range, running at ∼16 s (20, 21) and ∼8 s (18, 19) were disabled in the beginning of the mission (see Table 2), since no event triggered algorithms 20 and 21 and only three GRBs algorithms 18 and 19. The algorithms running on energy channel 2 (25–50 keV) with timescales higher than 128 ms were disabled, since they were mostly triggered by non GRB (and non SGR) events. The short timescale algorithms in the 25–50 keV energy range (22–26) were kept, mainly for the detection of SGR bursts, which are short and have soft energy spectra. The new algorithms above 100 keV did not increase the GRB detection rate. They were disabled with the exception of the shortest timescale algorithms running at 16 ms, particularly suitable for the detection of TGFs. Table 2 clearly shows the capabilities of the newly introduced "BGO"-trigger algorithm 116–119 for TGF detection.

At various times during Years 3 and 4 the instrument configuration was temporarily changed in two ways that affect the GRB data: (1) some or all of the trigger algorithms were disabled, and (2) the low-level energy thresholds (LLT) were raised on the sun-facing detectors (NaI 0–5).¹⁵ It is evident in Figure 1 that the number of triggers due to SFLs increased significantly around the beginning of 2011, an effect which is consistent with entering an active phase in the 11-year solar cycle. SFs typically have very soft spectra, producing high rates of low energy events in the solar-facing GBM detectors. Due to concerns that this might result in an unacceptable amount of TTE data, the LLTs were raised during two intervals when solar activity was high, a solar Target of Opportunity pointing on 2011 September 8–10 and an interval of high solar activity beginning on 2012 July 11 and continuing beyond the end of the period covered by this catalog.

The potential for high rates of soft solar X-rays was also a concern for a series of nadir pointings intended to detect TGFs in the LAT. During those intervals all triggering was disabled but TTE generation was turned on continuously so that a sensitive search for GBM TGFs coincident with the LAT could be performed. Again, in order to mitigate against unacceptably high rates of TTE, the LLTs in the sun-facing NaI detectors were raised above the nominal.

GBM TTE data suffer from timing glitches arising from rare conditions in the FPGA logic that produces GBM science data onboard. Every effort is made on the ground to correct these glitches but some are not cleanly reparable using pipeline software logic, and the TTE data files occasionally show the effects of these glitches, which can be seen in the TTE lightcurves.¹⁶ During pointed Target of Opportunity observations of the Crab Nebula between 2012 July 5 and 9, the on-board electronics were subjected to unusually low temperatures that caused a higher than normal rate of the TTE timing glitches.

The success of GBM in detecting TGFs led to great efforts to further increase the number of detected events (Briggs et al. 2013). A fundamental hardware limitation of the onboard triggering process is the minimum integration time of 16 ms. This integration time is much longer than the duration of a typical TGF of about 0.1 ms, which adds unnecessary background data and reduces the trigger sensitivity. These limitations are circumvented by downlinking the GBM photon data as continuous TTE data and by conducting a ground-based search for TGFs. In order to limit the data volume the GBM photon TTE data were only gathered over select parts of orbit (moving boxes) where the highest seasonal thunderstorm activity is expected. This mode was first implemented on 2010 July 15 and is typically enabled for ∼20% of the observing time. The resulting increase of the TGF detection rate is a factor of 10 compared to the rate of TGF triggers.

The GRB locations listed in Table 6 are adopted from the BA analysis results, uploaded to the GBM trigger catalog at the NSSTC (with a copy at the FSSC). Non-GBM locations are listed for bursts that were detected by an instrument providing a better location accuracy, such as Swift Burst Alert Telescope (BAT) (Barthelmy et al. 2005) or X-ray Telescope (XRT) (Burrows et al. 2005), or were localized more precisely by the Inter-Planetary Network (IPN; Hurley et al. 2013).

For each GRB the individual detector response matrices (DRMs) needed for analysis of the science data were generated for the best location using version GBMRSP v1.9 or v2.0 of the response generator and version 2 of the GBM DRM database. The detector response is dependent on incident photon energy, the measured detector output energy, and the detector-source angle. Two sets of DRMs are generated, one for eight-channel (CTIME) data and one for 128-channel (CSPEC and TTE) data. The Earth-source-spacecraft geometry is also considered in order to account for contributions from Earth's atmosphere scattering. In case of relatively long duration GRBs RSP2 response files with multiple DRMs are used, which provide a new DRM every 2° of satellite slew.

The determination of a reliable location is quite important since all analysis results depend on the response files generated for the particular GRB location. Systematic errors of the localizations are evaluated by comparing GBM locations with "true" locations from higher spatial resolution instruments or the IPN (Connaughton et al. 2014).

The analysis performed to derive the duration, peak flux and fluence of each burst is based on an automatic batch fit routine implemented within the RMFIT software.¹⁷

The data typically used for this analysis are either the CTIME or CTTE¹⁸ data from those NaI detectors that view the burst with an angle of incidence less than 60°, without significant blockage by the spacecraft or LAT components. Detectors may be omitted if they have rapidly varying backgrounds (e.g., due to solar activity). If available, CTTE data are used if the burst is too short to resolve with CTIME or if the peak is before the trigger time. The energy range is set to ∼10 keV to ∼1 MeV by selecting all but the first and last energy channels. The temporal resolution is typically set to 256 ms, but may be as short as 64 ms if necessary to resolve very short bursts. Source and background time intervals are then selected. The source interval covers the burst emission time plus approximately equal intervals of background before and after the burst (generally at least ∼20 s on either side). Background intervals are selected before and after the burst, about twice as wide as the burst emission and having a good overlap with the source interval. In the case of a burst with quiescent times between pulses, additional background intervals may be selected within the source interval. RMFIT computes a background model by fitting a polynomial of up to fourth order to the selected background intervals, separately for each detector and energy channel. Depending on the background variability, the lowest order polynomial that gives a good fit is selected.

RMFIT then deconvolves the counts spectrum of each time bin in the source interval, yielding a photon flux history over the selected energy range. For this analysis, the "Comptonized" (COMP) photon model was used:

$$\begin{eqnarray*} &&f_{{\rm COMP}}(E) = A \ \left (\frac{E}{E_{{\rm piv}}}\right) ^{\alpha } \exp \left [ -\frac{(\alpha +2) \ E}{E_{{\rm peak}}} \right ], \end{eqnarray*}$$

characterized by the parameters: amplitude A, the low energy spectral index α and peak energy E_peak (the parameter E_piv is fixed to 100 keV).

Figure 2 shows the light curve as measured by a single NaI detector of a relatively long burst consisting of two major emission periods, with the selected source and background intervals highlighted. Figure 3 shows a plot of the increase of the integrated flux in the 50–300 keV energy range derived from the model fitting for all time bins within the source interval. The three plateaus are the time intervals where no burst emission is observed. This function is used to determine the T50 (T90) burst duration from the interval between the times where the burst has reached 25% (5%) and 75% (95%) of its fluence, as illustrated by the horizontal and vertical dashed lines.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Peak fluxes and fluences are obtained in the same analysis, using the same choices of detector subset, source and background intervals, and background model fits. The peak flux is computed for three different time intervals: 64 ms, 256 ms and 1.024 s in the energy range 10–1000 keV and, for comparison purposes with the results presented in the BATSE catalog (Paciesas et al. 1999), in the 50–300 keV energy range. The burst fluence is also determined in the same two energy ranges. The RMFIT analysis results presented above are stored in a BCAT fits file: glg_bcat_all_bnyymmddttt_vxx.fit with specified wildcards for the year (yy), month (mm), day (dd), fraction of a day (ttt) and version number (xx).