Skip banner and navigation tools.

 |  site map

On-demand positions — Online Documentation

Contents

Overview

There are three types of position that can be produced through our web tools. All of them are based on a PSF-fit that uses the exposure map to correct for effects such as bad columns on the CCD and the edge of the field of view; the difference between them lies in how the astrometric data for the field is derived:

Position
type		Astrometric
source		Systematic
Error (90%)
StandardOnboard star-trackers3.5″
EnhancedUVOT+USNO-B11.4″
AstrometricXRT+2MASS0″

In each case, the systematic error is included in the final position error reported on the web pages. Each of these positions is described briefly in the following sections. The Standard Position algorithm is used as part of both the Enhanced and Astrometric positions, therefore we recommend you read that section first, even if you are not using a standard position.

Back to contents | Back to product generator.

Standard positions

The standard position is the most basic position, and can be created for any source which is detected in the XRT data. The astrometric reference for the positions comes from the star-trackers attached to the Swift XRT, which have an uncertainty of 3.5″ (radius, 90% confidence), so standard positions cannot have errors smaller than this.

Creating standard positions

In its most basic form, if the “Detect only” method was selected, the standard position is derived simply by minimising the C-statistic calculated by comparing the model Swift-XRT PSF multiplied by the exposure map, with the image. The statistical uncertainty from the fit is added in quadrature with the 3.5″ systematic associated with the star trackers to give the position error.

If source detection is also requested, the standard position algorithm can follow one of two approaches. If the ‘Single pass’ option is selected, source detection is performed using a simple sliding-cell detection approach. A cell of 21×21 pixels is stepped (in steps of 7 pixels) across the detector, the background is estimated from a colocated cell of 51×51 pixels. If a significant excess above the background is detected, the cell is divided into successively smaller sub-cells to identify the position and cell size which maximises the signal-to-noise ratio of the source. Once the source has been located, the PSF fit is performed, using the cell-detect position as the starting point. The Chandra Detect Reference Manual provides an excellent description of how sliding-cell detection works.

If “Iterative” source detection is requested, the process is more complex. An initial sliding-cell detection as above is employed to identify the brightest sources; these are then screened out and a background map is created. A second cell-detect pass is then called, but this time the background is taken from the map rather than the larger cell. This process of detecting sources, screening them and making a background map, and repeating the detection, is repeated until no new sources are found; then a PSF fit is performed (this is also a multi-stage approach, as the PSF fit uses the background map). A detailed overview of this algorithm can be found in the 1SXPS catalogue documentation and Evans et al., (2014), section 3.

The algorithms described above are used by the Enhanced and Astrometric positions. For the latter, you can specify whether the single-pass or iterative method is used. The enhanced position always uses the iterative process, however due to the way enhanced positions are created, this is never applied to the full dataset.

Back to contents | Back to product generator.

Standard position web pages

Once a standard position has been produced, the top-level web page for your product request will show the position, and supply a link to the more detailed results page. That page will give the best-fitting position (in decimal and sexagesimal format), and will identify which of the PSF models was used to derive the position. There are 8 PSF models available: the standard model from the CALDB, and 7 further ones which represent the PSF for sources which are increasingly piled up. The latter are named pu_psf$RATE.fits where $RATE is the fiducial count-rate at which the model was calibrated. If the PSF used is clearly inappropriate (e.g. the rate for a 5 ct/s source is used, but the source was very faint) then it is likely that the PSF fit got stuck in a local minimum, and the reported position should be viewed with caution.

Also supplied is an XRT image of the dataset(s) used to determine the position, with a circle marking the input position within which the source was sought. A second circle shows the standard position. This image should be used both as a sanity check of the result (does the position look right compared to the image? Has the code centroided on the correct source?) and as a potential debugging guide: for example, if the input position circle does not encompass the actual source position, you should repeat the position build, first adjusting your input position or radius accordingly.

If the object and dataset(s) you selected correspond exactly to an entry in the 1SXPS catalogue the position will not be redetermined, instead you will be directed to a page summarising the catalogue result, and providing links to your object in that catalogue.

Back to contents | Back to product generator.

Enhanced positions

‘Enhanced’ positions are produced by using the known mapping between the XRT and UVOT detectors, and the astrometric solution derived by aligning the objects detected by the UVOT with the USNO-B1 catalogue. This procedure does not require the XRT source to be detected by the UVOT. In addition to the statistical errors on the XRT position, XRT to UVOT mapping and UVOT astrometric solution, there is a futher 1.4″ systematic associated with enhanced positions, which is added in quadrature to the statistical errors before the results are reported.

Creating enhanced positions

Because the UVOT filters are lenticular, and Swift never remains perfectly stationary, the aspect solution derived by aligning UVOT-detected objects with sources in the USNO-B1 catalogue is different for each UVOT image. So, to create an enhanced position, the XRT data are first split up to create one image for each UVOT exposure. For each of these XRT-UVOT ‘overlaps’, the iterative source detection procedure described above is used to detect and localise the XRT source. If the source is not bright enough to be detected, the overlap is rejected. If the source is detected, then the UVOT image is aligned with the USNO-B1 catalogue to derive the aspect solution for that overlap. This solution is used in conjunction with the mapping between the XRT and UVOT detectors to give the corrected position for that overlap. Once this has been done for all overlaps, a weighted mean of all the overlap positions is calculated. Any outliers are rejected, and the mean is recalculated, giving the final enhanced position.

Enhanced positions cannot always be produced. If the source is too faint to detect in a single overlap then a position cannot be determined. Also, sometimes the UVOT astrometric solution cannot be derived (particularly in fields which are either very sparse, or very crowded), or the astrometric solution is degenerate and every overlap position ends up being rejected as an outlier.

The accuracy and systematic error of the enhanced positions has been determined using either only the optical UVOT filters (v, b, white), in which case the systematic error is 1.4″; or using only the UV filters (u, uvm2, uvw1, uvw2) which has a 2.1″ systematic error. Initially, the software will attempt to build an enhanced position using the former filter set, and if this fails, it will repeat the attempt using the latter set. Full details of the enhanced position creation is given in Goad et al., (2007), and revisions to that method are given in Evans et al., (2009).

Back to contents | Back to product generator.

Enhanced position web pages

Once an enhanced position has been produced, the top-level web page for your product request will show the position, and supply a link to the more detailed results page. If the position cannot be produced, this web page also exists, and can be used to determine why the position was not produced.

At the top of the page the position (if available) is given, in sexagesimal and decimal formats, along with the number of overlaps from which this was derived. This is followed by a breakdown of the individual overlaps, showing why those which were not included in the final position were rejected. Finally, there is an image of the UVOT data with the enhanced XRT position shown: if there is a UVOT counterpart it may be visible in this image. Below this a second image shows the same UVOT data, this time with the XRT position show, and the positions derived from the individual overlaps also marked. Green circles mark those positions which were combined to give the enhanced position, the yellow circles are overlaps where positions were determined, but rejected.

Back to contents | Back to product generator.

Astrometric positions

Astrometric positions are similar to the standard positions, except that we attempt to identify X-ray sources with objects in the 2MASS catalogue, to improve the astrometric solution from the star trackers. There is no systematic error associated with the astrometric positions, but if the number of detected X-ray sources is low, the statistical error in the astrometric solution can be very large.

Creating astrometric positions

To create an astrometric position first the source detection and centroiding described under the Standard Positions is executed, with the position derived for every source in the field of view. If more than two such positions are found, a list of 2MASS sources is downloaded from the Vizier facility. 2MASS sources are considered potential counterparts to an X-ray source if they lie within 20″ of the source; we then perform a Maximum Likelihood fit to identify the best aspect solution. Full details are given in Evans et al., (2014), section 3.7.

Back to contents | Back to product generator.

Astrometric position web pages

Once an astrometric position has been produced, the top-level web page for your product request will show the position, and supply a link to the more detailed results page. This page shows the position (in decimal and sexagesimal format) and uncertainty; it then shows the size of the astrometric correction calculated, in the RA and declination axis and the rotational correction. Beneath this is an image of the XRT field of view, with the astrometry derived from the star trackers. Green circles mark the positions of the X-ray sources, and cyan boxes show the 2MASS objects. A link above the image allows you to toggle between this, and an equivalent image after the astrometric correction has been applied: i.e. the XRT image (and X-ray source locations) are plotted using the derived astrometry, rather than that from the star trackers.

If the object and dataset(s) you selected correspond exactly to an entry in the 1SXPS catalogue the position will not be redetermined, instead you will be directed to a page summarising the catalogue result, and providing links to your object in that catalogue.

Back to contents | Back to product generator.