Fermi Gamma-ray Space Telescope

GBM Terrestrial Gamma-ray Flashes (TGF) Catalog

These tables contain the data of the second Fermi GBM TGF catalog. The catalog contains 4144 TGFs detected between 2008 July 11 and 2016 July 31. It is composed of TGFs bright enough to trigger on board ass well as TGFs recovered in an offline search for weaker events. The on-board triggering algorithms evolved throughout the period covered by the catalog, and the offline search was enabled by the availability of high temporal resolution data in 2011. This means the catalog is inhomogenous in its limiting sensitivity, as described below. The catalog is the result of the efforts of the GBM TGF Team with assistance from the Fermi LAT TGF Team.

Several methods detected these TGFs, with increasing sensitivity with time. The initial method was in-flight detection using only the GBM NaI detectors, proceeding to in-flight detection also using the GBM BGO detectors, then to an offline search using time-tagged event (TTE) data. No TGF detection method has ever been removed. The time-tagged event datatype returns information on each photon received by the GBM detectors. With TTE data we can search for TGFs at higher temporal resolution than possible in orbit and achieve a much higher detection rate. Originally the TTE data was collected for portions of the orbit of Fermi, changing on 2012 Nov 26 to production for the full orbit, termed continuous TTE (CTTE). The catalog is more uniform after this date. The initial method was finding ten TGFs per year, now with the offline search of the CTTE we are finding more than 800 TGFs per year.

We also provide software (with documentation) designed for analyzing GBM TTE data, with an emphasis on TGF research.

For further information about this catalog, see The First Fermi-GBM Terrestrial Gamma-ray Flash Catalog", O. J. Roberts, et al., Journal of Geophysical Research -- Space Physics, 2018. Please use that as the citation for this catalog. For questions, please contact Michael.Briggs@uah.edu

The Catalog

The catalog consists of six tables and datasets. The same TGF can appear in several tables. Different ID styles are used to distinguish between the tables: “oTGF” is used for TGFs in the Offline Search Table, “tTGF” for TGFs in the Trigger Table and “TEB” for TGFs in the Terrestrial Electron Beams Table, even when the entries are the same TGF.

  • Offline Search Table. The Offline Search Table contains information for 4135 TGFs detected by the ground-based offline search of the TTE data. Most TGFs of the Trigger Table are also included in this table, however, eight of the triggered TGFs are not included because they are not found by the offline search. The parameters in the Offline Search Table are described below.

  • Trigger Table. The Trigger Table contains information for 686 brighter TGFs that were detected in orbit by the GBM flight software. The content of the Trigger Table is described below.

  • Terrestrial Electron Beams Table. The TEB table lists the 30 TGFs that might have been detected as electron/positron beam events. One entry in the table is the reliability of the classification as a TEB. Also included are maps of the lightning activity underneath Fermi and at the magnetic footprint. The TEB table is documented below.

  • Comments and Special Cases Table. There are a few TGFs with missing values (which are denoted with “NULL” in the tables) due to unusual analysis issues or missing data. These cases are described in this table. If we are able to obtain a missing value, or if corrections are needed, updated tables will be posted.

  • WWLLN Associations Table. The WWLLN Associations Table contains results for 1544 TGFs for which temporally-coincident radio signals of the World Wide Lightning Network (WWLLN) were found. These associations provide accurate localizations of the TGFs. More information on this table is provided below.

  • WWLLN Lightning Maps and Data files. This dataset consists of maps of the lightning detections made by WWLLN for ±10 minutes relative to the TGF time. Also included are text data files with the positions and times of the lightning sferics on the maps. The content of these maps and data files are described below.

GBM Mission Elapsed Time (MET) is the number of seconds since the epoch of January 1, 2001, at 00h:00m:00s in the Coordinated Universal Time (UTC) system.

This TGF catalog uses the WGS84 system for geographic coordinates. Some other GBM data products use geographic coordinates for a spherical Earth, which may seem discrepant from these values.

Some events identified as TGFs in the GBM data are, actually, produced by incident cosmic rays. Several tests using GBM and Fermi LAT data remove cosmic rays from this sample. Nevertheless, there may still be some contamination of the sample, which we estimate to be approximately 1%.

Table and Dataset Contents

All tables are in comma-separated value (CSV) format.

Offline Search Table

The Offline Search Table has parameters for 4135 TGFs. The items in this table are:

  1. OS_ID: the offline search identification string, which is a date code “oTGFyymmddfff”. “fff” is the fraction of the day. If two or more TGFs otherwise would have the same ID, a letter suffix is added.
  2. MET: the GBM Mission Elapsed Time of the start the “discovery bin”.
  3. File: The name of the TTE FITS file with data for this TGF.
  4. BGO_0_N: The counts recorded by BGO detector 0 within the “discovery bin”.
  5. BGO_1_N: The counts recorded by BGO 1.
  6. NAI_N: The counts of the NaI detectors, summed.
  7. Date: The UTC date of the TGF.
  8. UTC: The UTC time of the start of the discovery bin.
  9. Width_ms: The width of the discovery bin in ms.
  10. P2: The corrected, joint Poisson probability of the event. See Briggs et al. (2013). However, for better consistency the correction is now always done for the number of search bins in one hour.
  11. Lon: The WGS84 East longitude of Fermi at the time of the TGF, in degrees,
  12. Lat: The WGS84 latitude of Fermi,
  13. Alt: The WGS84 altitude of Fermi, in km.
  14. LST: The local solar time, as fraction of a day.
  15. TRIG_ID: If this TGF was detected in-orbit as a trigger, the Trigger ID “tTGFyymmddff”. This item serves as a key into the Trigger Table.

The "discovery bin" is the interval for which the search program found the highest statistical signifcance. On timescales below 1 ms the background rate is typically low so that statistical significance is typically maximized by including most of the counts of a TGF. This leads to a time interval which typically approaches the duration of the TGF. The program only searches for TGFs on sixteen timescales from 25 microseconds to 16 ms. The program searches for rate increases in a sequence of bins. With these constraints, the "discovery bin width" is an approximation to the duration of the TGF and the counts in the "discovery" bin is an approximation to the observed fluence.

Some entries are unavailable due to missing data or other issues – these cases are denoted with “NULL”. Consult the Comments and Special Cases Table for explanations of such entries. Eight triggered TGFs are not included in this table because they were not detected by the offline search. Most of the non-detections by the offline search are because the offline search calculates a background using five second long intervals before and after the TGF, while in-orbit triggering uses a 17 second long pre-trigger background interval. The offline search method is unable to calculate a background, and therefore unable to detect a TGF, if the data ends within a few seconds after the TGF. One of the triggered TGFs may not have been found because the detector pattern was very asymmetric, so that it failed to meet the requirement of at least four counts in each of the BGO detectors.

Trigger Table

The Triggered TGFs table contains data for the 686 TGFs that were detected in-orbit by the GBM flight software. All but eight of these TGFs are also present in the Offline Search Table, which provides additional data on their properties. The items in this table are:

  1. TRIG_ID: Trigger ID, which is a date code “tTGFyymmddfff”. “fff” is the fraction of the day.
  2. MET: the GBM Mission Elapsed Time of the end of the 16 ms interval in which the GBM flight software detected the TGF.
  3. Date: The UTC date of the TGF.
  4. UTC: The UTC time of the end time of the 16 ms interval.
  5. Lon: The WGS84 East longitude of Fermi at the time of the TGF, in degrees,
  6. Lat: The WGS84 latitude of Fermi,
  7. Alt: The WGS84 altitude of Fermi, in km.
  8. Algor: The number of the flight software algorithm by which the TGF was detected. See Table 2 of von Kienlin et al. (2014) for the properties of each algorithm.
  9. TSig: The significance of the detection, as signal-to-noise ratio.
  10. LST_Frac: The local solar time, as fraction of day.
  11. LST: The local solar time, as hours and minutes.
  12. OS_ID: The ground search ID of the form “oTGFyymmddfff”, which serves as a key into that table. A few triggers had more than one detection by the ground search – these IDs are separated by colons. A few triggered TGFs are not detected with the criteria used by the ground search (typically because one BGO detector has fewer than four counts), these have “NULL” as the ID.

Triggered TGFs are associated with TGFs found by the ground search (OS_ID) if the discovery bin of the ground search TGF starts within 19 ms of the trigger time (i.e., end time of 16 ms interval) of the triggered TGF.

Terrestrial Electron Beams Table

The TEB table lists 30 TGFs, of which 16 are reliably classified as TEBs and the other 8 have some characteristics of TEBs. These events are also in the Offline Search Table if they were found by the the offline search. Those events that triggered GBM are also included in the Trigger Table.

Triggers with durations ≥1 ms or that weren't found by the offline search were evaluated to determine whether they were TEBs. We included triggered TGFs not found by the offline search in the candidates because a common reason for the TEB search not finding a bright TGF is a large differences in the signal strengths of the various detectors, which is also a signature of TEBs in GBM.

The candidates were evaluated using the following criteria:

  • presence/absence of lightning, as shown by WWLLN sferics during within ±10 minutes, at the Fermi nadir and at the footprints of the magnetic field line through Fermi,
  • presence/absence of a positron annihilation line,
  • large difference in the signal levels in the detectors, particularly between BGO 0 and BGO 1.

In a few cases one item is decisive (e.g., presence of annihilation line), but in most cases the preponderance of evidence was used. Typically the most useful criterion were the WWLLN sferic maps: many cases could be classified as TEBs when there was no lightning at nadir and high activity at a footprint. Conversely other cases could be identified as TGFs if there was lightning activity at the nadir but none at the footprints. However, if lightning map criterion is indeterminate if there is lightning at both the nadir and a footprint. For 20 of the events we consider the evidence to be sufficiently strong to consider the TEB classification as reliable. For eight of the events the evidence suggests a TEB classification but is inconclusive. Events that we determined to be gamma-ray TGFs are not listed in this table.

We also provide 60 PDF maps, showing the lightning activity in the vicinity of the Fermi nadir and near the magnetic footprint, as observed with WWLLN. These maps are described in the WWLLN Lightning Maps section below.

  1. TEB_ID: identifier. “TEB” is used in the identifier regardless of the reliability of the classification.
  2. MET: GBM Mission Elapsed Time,
  3. Reliability: 1 for reliable classifications as TEBs, 0 for possible TEBs.
  4. Date: The UTC date,
  5. UTC: The UTC time,
  6. Fermi_Lon: WGS84 East longitude of Fermi, in degrees,
  7. Fermi_Lat: WGS84 latitude of Fermi,
  8. Fermi_Alt: The WGS84 altitude of Fermi, in km.
  9. Footprint_Lon: WGS84 East longitude of the likely magnetic footprint,
  10. Footprint_Lat: WGS84 latitude of the likely magnetic footprint,
  11. OS_ID: If this event was found by the offline search, the identification string “oTGFyymmddfff”. This ID serves as a key into the Offline Search Table.
  12. TRG_ID: If this TGF was detected in-orbit as a trigger, the Trigger ID “oTGFyymmddfff”. This item serves as a key into the Trigger Table.

Comments and Special Cases Table

For some events, there are missing values in the tables which are denoted with "NULL". These arise due to unusual analysis or missing data and are documented in this table. If we are able to obtain a missing value, or if corrections are needed, updated tables will be posted.

  1. ID: Event identification string, which can be used to index the other data tables. The format of the string indicates what table it refers to, e.g. “oTGFyymmddfff“ for offline table, “tTGFyymmddfff” for triggered table.
  2. Comment: Text describing the special case for this event.

WWLLN Associations Table

The WWLLN Associations Table has data on the 1342 TGFs for which a close association between a GBM TGF and WWLLN radio signal is found. A window of ±3.5 ms is used to define an association. Only associations with a chance probability of ≤1% based on the WWLLN data within ± 1000 s are included. Close separations of less than 200 μs are considered to be simultaneous – most of the simultaneous radio signals originate from TGFs rather than from associated lightning (Connaughton et al., 2013). Some TGFs have several associations. The table contains 1544 associations, including 1328 simultaneous ones.

  1. ID: an identification string, which is either the the Offline Search ID, which is a date code “oTGFyymmddfff” or Trigger ID, which is a date code “tTGFyymmdd.fff”.
  2. MET: GBM Mission Elapsed Time,
  3. Date: The UTC date,
  4. UTC: The UTC time,
  5. Fermi_Lon: WGS84 East longitude of Fermi, in degrees,
  6. Fermi_Lat: WGS84 latitude of Fermi,
  7. Count: count of WWLLN associations for this TGF,
  8. WWLLN_Lon: WGS84 East longitude of WWLLN sferic,
  9. WWLLN_Lat: WGS84 latitude of WWLLN sferic,
  10. Separation_ms: time separation after correction for light travel time between WWLLN and Fermi, in ms.
  11. Offset_km: the offset of the WWLLN position from the Fermi nadir in km,
  12. Fermi_Azimuth: Azimuth angle in spacecraft coordinate system,
  13. Fermi_Elevation: Elevation angle coordinate in spacecraft coordinate system,
  14. Right_Ascension: equatorial coordinate, J2000 system,
  15. Declination: equatorial coordinate, J2000 system.
  16. Optional comment:

When a TGF has several WWLLN associations, it is listed with more than row in the table. The item “count” counts the associations per TGF.

Typically WWLLN localization uncertainties are ≈10 km.

This table only has results for the region under Fermi and so won't include WWLLN associations for TEBs originating from distance magnetic footprints – for TEBs see the Terrestrial Electron Beams Table section.

The response of the GBM detectors is a function of the position of a source in spacecraft coordinates, azimuth and elevation. While it would seem to make little sense to locate TGFs in an astronomical coordinate system (i.e., J2000 equatorial coordinates), those coordinates are a required input of the GBM tool that calculates Detector Response Matrices.

WWLLN Lightning Maps and Data

This data set includes 4144 PDF maps of the WWLLN sferics underneath Fermi at the times of TGFs to show the lightning and storm activity. Sferics within ±10 minutes of the TGF time are included. The circles on the maps are of 300 and 800 km radii centered on the Fermi nadirs. Most triggered TGFs are within the inner circle and almost all TGFs within the outer. Sferics within the time window are plotted in green. If a sferic is associated with a TGF, and has a low chance probability (<= 1%) it is plotted in magenta. If the association has a low chance probability and is simultaneous with the TGF it is plotted in red.

This data set only analyzes the region under Fermi and so won't find lightning related to TEBs – for maps for TEBs see Terrestrial Electron Beams Table section. Also included are text files containing the geographic and time coordinates of the sferics on the maps.

The exact content of these files will vary depending on whether or not a given TGF has an association or not, but all will obey the following format. First will be a header:

  • network_name
  • TGF_name
  • Fermi/footprint coordinates, whether or not association is good (as T/F)
  • Coordinates for center of map
  • Number of associations, number of sferics

The main body will follow and will contain two lists, the length of which is given in the line above.

Associations list: Contains data for associated sferics.

  • count, longitude, latitude, light travel time corrected time separation, whether or not simultaneous

Sferics list: Contains data for all sferics within time window.

  • count, longitude, latitude, time separation, dist from simultaneous, time from simultaneous

The last two columns in this list will only be present if there is one simultaneous association.


M. S. Briggs, S. Xiong, V. Connaughton, et al., (2013), Terrestrial Gamma-ray Flashes in the Fermi Era: Improved Observations and Analysis Methods, J. Geophys. Res. Space Physics, 118, 3804, http://onlinelibrary.wiley.com/doi/10.1002/jgra.50205/abstract

V. Connaughton, M. S. Briggs, S. Xiong, et al., (2013), Radio signals from electron beams in Terrestrial Gamma-ray Flashes, J. Geophys. Res. Space Physics, 118, 2313, http://onlinelibrary.wiley.com/doi/10.1029/2012JA018288/abstract

V. Connaughton, M. S. Briggs, R. H. Holzworth, et al., (2010), Associations between Fermi GBM Terrestrial Gamma-ray Flashes and sferics from the WWLLN, J. Geophys. Res. Space Physics, 115, A12307, http://onlinelibrary.wiley.com/doi/10.1029/2010JA015681/abstract

A. von Kienlin, C. A. Meegan, W. S. Paciesas, et al., (2014), The Second Fermi GBM Gamma-Ray Burst Catalog: The First Four Years, Astrophys. J. Suppl., 211, 13, http://stacks.iop.org/0067-0049/211/i=1/a=13


Development of this catalog was supported by Fermi Guest Investigation grants NNXllAE96G and NNX13AO89G. In addition to Fermi data, we used lightning data from the World Wide Lightning Location Network, a collaboration among over 50 universities and institutions. We thank Lisa Gibby (Jacobs Engineering) and Bill Cleveland (USRA) for their assistance in conducting the offline search for GBM TGFs.