Fermi GBM Online Targeted Search

About the Fermi GBM Targeted Search

The Fermi Gamma-ray Burst Monitor (GBM) Targeted Search is a ground-based analysis designed to find weak, short gamma-ray transients in GBM continuous data that may be too faint, too short, or viewed from an unfavorable spacecraft geometry to trigger GBM onboard. The method is especially useful for searching near the times of external triggers, such as gravitational-wave candidates or short gamma-ray bursts detected by other missions.

Fundamental concept

Instead of treating each GBM detector independently, the targeted search combines information from all 14 GBM detectors using the instrument response. A real celestial source should produce a predictable pattern of counts across the NaI and BGO detectors that depends on the source location, photon spectrum, and spacecraft orientation. The search compares that predicted detector pattern to the observed counts and asks whether a signal-plus-background model is preferred over background alone.

1
Start with a target time

For externally triggered searches, the analysis examines a short time window around a candidate event, such as a gravitational-wave trigger or a Swift/BAT trigger.

2
Model the background

The local detector background is estimated for each detector and energy channel so that any short excess can be measured relative to the expected count rate.

3
Predict detector counts

For each trial sky location and spectral template, the expected photon spectrum is folded through the direction- and energy-dependent response of each GBM detector.

4
Rank candidates

The predicted counts are compared to the observed counts. A likelihood ratio ranks whether a signal-plus-background model is preferred over background alone.

How the analysis proceeds

Input dataGBM CTTE or CTIME data from the NaI and BGO detectors.
Time searchTest nested durations from fractions of a second to several seconds, with multiple phase offsets.
Spectral searchTest soft, normal, and hard GRB-like spectral templates.
Sky searchEvaluate a grid of possible source directions, or use an external localization map when available.
OutputReport the most significant candidate time, duration, spectrum, localization, likelihood ratio, and false alarm probability.

The key advantage is that a real celestial signal should illuminate the GBM detectors in a pattern that depends on source direction, photon spectrum, and spacecraft geometry. Random background fluctuations generally do not match that coherent detector pattern as well.

Event displays

The figure below shows examples of short GRBs that did not trigger GBM onboard, but were recovered in the GBM continuous data by the targeted search. The top panels show Swift/BAT light curves, the middle panels show GBM light curves, and the lower panels show the likelihood ratio as a function of time and tested duration. Darker regions indicate time-duration combinations where the signal is more significant above the local background.

BAT and GBM light curves for sub-threshold bursts recovered by the targeted search.
Adapted from Kocevski et al. 2018, Figure 2. The targeted search can recover short-timescale signals in GBM continuous data that were below the onboard triggering threshold.

Interpreting significance

The likelihood ratio measures how strongly the data favor a burst-like signal over background alone for a particular time, duration, spectrum, and sky position. Because the search tests many combinations of time bins, spectral templates, and sky positions, the raw likelihood ratio is not the whole story.

To account for these trials, the search builds a false alarm rate distribution by running the same analysis on background intervals. This estimates how often GBM data produce background transients with comparable significance. That false alarm rate is then used to calculate the chance probability of finding a candidate in the searched time window.

Why this matters

The targeted search increases GBM sensitivity to weak, sub-threshold short GRBs by using the full detector array coherently. In the Kocevski et al. analysis, the method recovered most Swift/BAT short GRBs that were in GBM's view, including events that did not trigger GBM onboard. The paper also showed that this added sensitivity is important for weak nearby events like GRB 170817A and for searches for gamma-ray counterparts to gravitational-wave candidates.

Related papers

Figure images are adapted from Kocevski et al. 2018, published under the Creative Commons Attribution 3.0 license. Please retain attribution when reusing this page or its figures.