GBM Data

Where is the GBM data?

• GBM data can be obtained using HEASARC BROWSE. (See Browse Interface for GBM Tables help page.)

Multiplicity of the GBM files

• Each burst results in a large number of files. The GBM data and the file name conventions are described in Cicerone. (See Data - GBM Data Products.)

Channels to be fit

Most binned spectra have high and low energy channels that are beyond the instrument response function's believable energy range, are affected by non-linearities, or are overflow channels (all counts above a certain energy end up in that channel). These channels should not be included in fitting spectra; in XSPEC terminology these channels are 'ignored'. Assuming that the GBM channels run from 0 to 127 the following channels should always be ignored:

• NaI spectra—all channels below 8 keV, and then channel 127. Note that 8 keV corresponds to different channels in different NaI detectors; therefore, the low energy channels that should be ignored will vary from detector to detector.

• BGO spectra—channels 0 and 127

These are aggressive ranges (i.e., ignoring as few channels as possible); more realistically, it will be necesary to ignore a few additional channels on either end. If you find that the residual (fit minus actual count numbers) for a few channels at either end of the fitted spectrum are particularly large, those channels are probably affected by systematics and should be ignored.

Updated Files

The GBM team will update files as better data become available. Thus response files (files with names such as 'glg_cspec_n0_bn080808123_v00.rsp') may be updated as better burst positions become available (e.g., because Swift or the IPN provide a more accurate position than the GBM position). Background spectra (in files with names such as 'glg_cspec_n0_bn080808123_v00.bak') also require non-automated processing and will not appear immediately in the archive. A typical delay of 1-2 days is anticipated.

Detectors to be Fit

The GBM burst data include files for all detectors, whether or not they were oriented to accumulate many counts from a burst. The NaI detectors that triggered should have accumulated the most burst counts, although, if FERMI slews to point towards a long burst, other detectors will record significant burst flux.

The triggered NaI detectors will be listed in the burst catalog. But they also can be determined from the burst data files. For example, look for the 'DET_MASK' keyword in the primary header of the file that begins 'glg_tridat' (see the example below). This keyword provides a string that indicates which detectors triggered: the first character is '1' if the first NaI detector—NaI #0—triggered, and '0' if it did not, etc. Therefore, the string '110000001100' means that NaI detectors 0, 1, 8 and 9 triggered.

Time

Time in the SAE and in the FERMI data files is Mission Elapsed Time (MET), i.e., the number of seconds since midnight January 1, 2001 (UTC). (See Cicerone: Data - Time in FERMI Data Analysis.) Consequently, times will be numbers with many digits such as 239113751.4. The xTime Date/Time Conversion Utility converts between FERMI's MET and other time systems.

For analyzing GBM data, the GBM's trigger time is needed. This is also provided in the BROWSE catalogs and in the primary header of the GBM burst data files as the keyword 'TRIGTIME'.

Tips:

• The fv tool allows you to read the headers of the different FITS extensions.

• Plotting lightcurves as counts vs FERMI's natural time system (i.e., MET) usually results in an x-axis that is not very informative because of the large difference in scale between burst durations (10's of seconds) and MET. It is therefore useful to plot lightcurves as counts vs. time relative to the trigger time.