Skip banner and navigation tools.

 |  site map

About Swift-XRT tiled observations and analysis

For some GRBs with large errors circles (such those detected by Fermi-LAT), Swift will perform a series of tiled pointings to search for a counterpart. Automated analysis of the XRT data of these fields will be posted online; this page explains how new X-ray sources are announced, how it is determined whether an afterglow has been detected and what the web pages show.

List of GRBs.



Swift observes the error circle of the GRB in a series of typically 4 overlapping pointings (Fig. 1). Ordinarily each field is observed on each spacecraft orbit, so the exposure time is accumulated evenly across the error circle. During a ground-station pass (see the latest schedule) the data are telemetered to the ground, and are typically available on the quick look site a couple of hours later.

Exposure map of 4-tile field

Fig. 1. An exposure map of a typical 4-tile observation set.

Once data are received at the quick-look site, they are automatically analysed by software at the UK Swift Science Data Centre. This first checks to see if data have been received for all of the tiles, and if not waits for up to 5 minutes for those data to arrive. Then all of the tiles are combined into a single image and the source detection system developed for the 1SXPS catalogue (see Evans et al. 2014) is applied. This identifies sources, localises them, and assigns each one a detection flag which can be Good, Reasonable or Poor. These indicate the probability that the source is spurious (see the 1SXPS documentation). A light curve and spectrum is constructed for each source, using the tools described by Evans et al.(2007, 2009). The source positions determined above use the astrometric information from Swift's on-board star trackers, however it is sometimes possible to improve on this. We attempt this in two ways. The first uses field stars in the Swift UV/optical telescope to derive the astrometry,so called ‘position enhancement’ — see Goad et al. (2007); Evans et al. (2009). The second matches the detected XRT sources with the 2MASS catalogue to derive the astrometry (see Evans et al. 2014). These methods do not always work for every source, and the position with the smallest error is that which is reported.

The position of each source is compared with the X-ray master catalogue, the 1SXPS catalogue and SIMBAD. For any source not present in this catalogues, a GCN Counterpart notice is produced. This notice contains a field, ID_CONF, which indicates the probability that the source is the afterglow. Based on the 1SXPS catalogue we expect on average ~7 serendipitous sources in a 4-tile observation with 2 ks of data per tile. Therefore initially each source is assigned an ID_CONF of 0.15 (i.e. approximately 15% probability of the source being the afterglow). Two tests (described below) are performed on each source to attempt to distinguish the afterglow from the serendipitous source, and depending on the outcome of these, the ID_CONF may be increased. In this case, the Swift team are alerted that a source is found which is likely to be the afterglow: they will verify this manually and issue a GCN Circular if appropriate.

[Back to top]

Afterglow checks

Two tests are automatically carried out on each uncatalogued X-ray source, to determine whether it is a strong candidate for the afterglow. If either of these tests comes back positive, the ID_CONF in the GCN Counterpart notice is increased to 0.8; if both tests are positive it is set to 0.95. These are intended to indicate that this source is probably the afterglow, but should not be taken as quantitative probabilities.

The checks are thus:

Is the source fading?
If the last bin in the source light curve is at least 3-σ below the first bin, the source is considered to be fading, and thus is probably the GRB afterglow.
Is the source "bright"?
If the source is more than 1-σ above the RASS 3-σ upper limit and yet not detected by Rosat, it is assumed to be a new source and therefore potentially the afterglow. The Rosat upper limit is derived using the RASS images and background maps; the XRT count-rate is converted to a Rosat PSPC rate using PIMMS and the power-law fit to the XRT spectrum, and then compared to this upper limit.

[Back to top]

Web pages

For each GRB observed using the tiled approach, a web page is created which collates in one place the analysis of each field. This web page is linked to in the table on the XRT Products index and is given in the GCN Circular produced by the Swift team in response to the initial observation.

The web page begins with a summary of the GRB -- which mission detected it, its location, the trigger time and the distance from the Sun/Moon at the time of the detection (and a link to calculate these distances every day for a year). This is followed by a statement indicating how many fields have been observed so far, and for how many of these the data have been downlinked and analysed. A list then follows giving the times of the first and last observation of each field as well as the useable exposure time per field. This can be lower than the exposure time of the actual observations, as some data may be rejected by screening, particularly that which removes times of contamination by light reflected off the bright Earth. In some cases it may be possible to bypass the bright-Earth removal phase (i.e. if the level of contamination can be confirmed by eye to be low): this has to be specified manually, and a warning will appear on the web page in such cases.

After the list of observations, details of the sources detected (if any) are given. These appear in one or two tables. The first contains any sources which passed the afterglow checks; if there are no such sources this table does not appear. The second table lists all detected sources.

The tables have the same format: the source number (which serves as a link to a page detailing just that source), the position and error, the detection flag, the distance from the position reported by the mission which detected the GRB, and then any notes (e.g. details of catalogue matches). Below the source number is a link (“details”) which, when clicked, opens a new pane in the table. This contains more details: the mean and peak XRT count-rate of the source (based on the light curve); the Rosat upper limit, and the XRT mean and peak count-rate converted into Rosat PSPC count rates (using the XRT spectrum); the mean and peak XRT flux (derived from the XRT spectrum and light curve); and links to the various products available for that source. The table rows are colour coded. Most rows will have a white background (apart from the detection flag which is coloured according to the flag); sources which match catalogued objects are on red rows, and sources which pass one or both afterglow checks have cyan backgrounds.

After these tables, two images are given. The first shows the XRT data, with the locations of the detected sources marked, along with the fields of the individual tiles. The second image shows the exposure map (i.e. pixel values indicate the exposure at that location). These images are typically too large to fit on a web page, so a zoomed-out view is shown: however this can makes it hard to see individual pixels due to the resampling involved. There is therefore an option above the image to magnify the image and make is scrollable. If you select this the image is zoomed to 100% but cropped to fit on the screen: you can click and drag the image around however to explore it.

[Back to top]

Number of serendipitous sources

When we detect an X-ray source we need to quantify the probability of this being a serendipitous source unrelated to the trigger. This is done indendently for each detected source, and the value is reported both in the ‘details’ panel and on the specific page for each source. The number reports how many sources are expected in the entire tiled dataset collected so far, with a count-rate at least that of the detected object.

We calculate this number as follows:

  1. Determine the minimum exposure necessary to detect the XRT source.
  2. Calculate the sky area observed with at least this exposure.
  3. Determine the sky density of sources at least as bright as the XRT source, using the log N-log S distribution of Mateos et al. (2008).
  4. Multiple this by the sky area from #2 above, to give the expected number of sources.

The log N-logS was created based on XMM data, and gives the distribution of extragalactic source fluxes. We convert the XRT count-rates to fluxes assuming a typical AGN spectrum (NH=3×1020 cm-2, Γ=1.7).

Note that this statistic cannot be read independently of the other information. In particular, the quality flag must be taken into account. For example, a ‘poor’ detection, has a ~35% probability of it being a spurious detection (see the 1SXPS quality flagging documentation for details); if it is spurious, the fact that it is unlikely to be a serendipitous source is irrelevant.

[Back to top]