WP-A5.2: X-ray Pilot Design

Task: A5.Y1.Q2-11 from Section 5.2.0 of the WP-A5 project description.

1. Overall aim

Cross-correlate XMM-Newton X-ray source catalogue with optical object list in order to:

• Identify optical counterparts of X-ray sources
• Establish X-ray – optical parameter sets for identified objects
• Provide means to explore parameter space and select sub-samples for a variety of astrophysical projects

This is a worthwhile AstroGrid pilot project because:

• It will produce something useful for the community
• It is representative of a wider range of federations
• The relevant expertise with the input data exists within the AstroGrid consortium

2. AstroGrid issues

From the AstroGrid viewpoint the goal must be to investigate the generic issues that arise and not just find a solution to this relatively straightforward problem. Presumably these generic issues will be determined in the first phase of the programme, or will emerge from other AstroGrid work, but we can anticipate some of these:

• Storage requirements for catalogue information in a database to optimise cross-correlation
• Handling one-to-many cross-correlation matches and figure-of-merit algorithms to order matches
• Visualisation tools required to make cross-correlation results useful
• Reporting options required to make cross-correlation results useful
• Strategies for cross-correlation between remote databases and/or co-located databases
• Strategies for handling cross-correlated databases effectively as a new database, e.g. extracting subsamples based on selection criteria applied to parameters from both databases
• Real-time and/or dynamically updated database correlations

It should be noted that the XMM and WFC object lists discussed below are not currently available within any database system. Instead the component information is stored in individual files. This makes a “clean slate” approach possible! The SuperCOSMOS object lists are stored in a database at Edinburgh which provides the infrastructure for a variety of on-line services.

3. General description of XMM-Newton catalogue

The XMM-Newton spacecraft is equipped with 3 co-aligned X-ray telescopes with 3 ‘EPIC’ X-ray cameras in their focal planes. There are two different camera designs: two EPIC MOS cameras and one EPIC pn camera distinguished by somewhat different physical characteristics and performance but these differences produce only detailed differences in the X-ray catalogues discussed below. The EPIC cameras have a field of view of ~30 arcmin. diameter with pixels sizes of 1.1 arcsec and 4 arcsec for the MOS and pn cameras respectively. The angular resolution of the XMM-Newton telescopes is ~10 arcsec. Both types of EPIC camera provide intrinsic energy resolution E/E ~ 5-10 at the lowest energies increasing to E/E >50 at the highest energies. The band-pass of the cameras is ~0.1-12 keV. Standard pipeline processing of XMM-Newton scientific data is the responsibility of the XMM-Newton SSC with Leicester acting as the central processing site. Processing includes the production of a set of images and derived source lists for each of the EPIC cameras.

The XMM-Newton catalogue will be based on a compilation of the source lists for individual XMM-Newton observations where each source list contains X-ray sources detected serendipitously in the XMM-Newton EPIC cameras from a single observation, i.e. field objects distinct from original target. The detection procedure is a multi-part process based on sky images and associated exposure maps. It involves several distinct stages:

• Local ‘box’ detect: sources detected as maxima with respect to local background
• Estimation of ‘global’ background from ‘source-free’ image created by masking out brighter sources found in local box detect stage. Global background is determined by 2-D spline function fit to image.
• Refined ‘box’ detect: sources detected as maxima with respect to global background.
• Maximum likelihood fitting and parameterisation of detected sources.

For each detected source the following principal parameters are determined (essentially from the last ‘ML’ stage in the detect process described above):

• Celestial coordinates; coords in various detector frames
• X-ray count, count rate and flux in multiple energy bands (currently 5 primary bands and 2 additional bands from combination of primary bands)
• Likelihood parameter related to probability of source existence
• Spatial extent parameter (not yet implemented in pipeline, but relevant code exists and has been tested)
• Time variability parameter (not yet implemented)

These parameters are accompanied by additional parameters such as:

• Effective exposure time for each source (vignetting dependent)
• Local background value estimated for each source
• Cross-reference to ‘box detect’ source lists
• Error estimates for all relevant parameters

Typical values for XMM-Newton source lists and the catalogue:

• Sky density of XMM sources ~ 100 – 1000 deg-2
• Position error (statistical) ~ 1-3 arcsec
• Position error (systematic for whole field) < 4 arcsec (can be corrected to <1 arcsec accuracy by use of correlations of X-ray sources with all-sky optical catalogues, e.g. USNO A2.0)
• Minimum detectable X-ray flux ~ 10-15 erg cm-2 s-1 (0.1-10 keV; depends on exposure time of observation, background conditions etc.)
• Number of XMM-Newton observations per year ~700
• Total number of sources detected in existing XMM-Newton datasets ~100,000 –200,000 (depending on TBD thresholds)

Special issues

• Choice of source list (how to combine lists from 3 different cameras)
• CCD gaps in EPIC cameras – low or zero sensitivity regions
• False detections at CCD edges, close to bright objects etc.
• Optimum astrometric correction of source lists

Catalogue status. The task of merging the XMM-Newton source lists into a usable catalogue began at the end of 2001. The main issues relating to the catalogue creation relate to the quality and reliability of the catalogue entries. For public release a substantial programme of quality control and testing is required, together with visual screening of the data and other activities. In the interim reasonable quality compilations of source lists already exist and there is substantial experience of the main quality and reliability issues, making it possible to easily construct trial samples for development purposes. It is planned that the first public version of the catalogue will be released towards the end of 2002.

4. General description of potential optical catalogues

4.1 INT WFC catalogues

The INT Wide Field Camera (WFC) is a wide field of view optical CCD camera used at the prime focus on the INT 2.5m telescope. The camera consists of 4 CCDs, each 4K x 2K, formatted so as to cover most of a ~30 arcmin field of view. The pixel size is 0.33 arcsec and the camera is equipped with a filter wheel and autoguider system. The INT WFC has been used quite extensively to image XMM-Newton fields as part of the SSC XID programme. Currently >100 XMM-Newton fields have been imaged in one or more colours.

Cambridge (CASU) have developed a standard processing pipeline for INT WFC data. This pipeline includes bias correction & flat-fielding and astrometric registration of the images. Object catalogues are created, and merged catalogues produced by matching objects in separate observations of the same field in different colours.

The CASU object catalogues are organised as one catalogue per camera CCD, i.e. there are 4 catalogues per WFC image. They contain the following primary parameters for each detected object:

• Celestial coordinates; CCD coords *Intensity parameters: a variety of different measures of the count within the object profile
• Extent flag: unresolved, extended or defect
• Extent parameters: size, ellipticity, orientation
• Signal-to-noise statistic

The merged catalogues are created by combining catalogues of the same field in different colours. The merging is based on positional coincidence between objects alone. All objects are included, even those with no match in other colours. The content of the merged catalogues is very similar to the individual filter object catalogues, although somewhat idiosyncratically, the celestial coordinated are omitted and have to be recreated from FITS metadata.

Photometric calibration is not explicitly included in the catalogues, but average calibration can be applied via parameters mostly stored as FITS metadata in the catalogues (i.e. exposure time, photometric zeropoints, aperture corrections, airmass correction).

Typical values for INT WFC images and object lists:

• Sky density ~ 104- 105 deg-2 (105 – 106 for low latitudes)
• Position error ~ 0.25 arcsec (median rms)
• Magnitude limits ~ 23-25 mag. (600/1200 sec. exposures; depending on band, seeing, sky brightness)
• Relative photometric accuracy <0.1 mag.
• Absolute photometric accuracy – depends on calibration strategy

Special issues

• Correct handling of inter-chip gaps (i.e. regions for which no detections possible – distinguish from no detection made!)
• False and missing detections close to bright and/or bright extended objects

4.2 SuperCOSMOS catalogues

The SuperCOSMOS scans of the southern photographic sky surveys have been used to produce a digitised sky atlas and associated object catalogues. The scans are not yet complete in all colours but they already half the sky in 2 colours. The SuperCOSMOS scans and derived object catalogues have the reputation of being amongst the best, if not the best, products derived from the photographic sky surveys.

The SuperCOSMOS catalogue data is broadly rather similar to the INT WFC data in terms of content. Object parameterisation is probably superior. SuperCOSMOS catalogues contain calibrated multi-band magnitudes as well as proper motions derived from the multi-epoch plates.

Typical values for SuperCOSMOS catalogues:

• Sky density ~ 103 deg-2
• Current sky coverage: southern sky in 2 colours, smaller regions in more colours
• Position error < 0.3 arcsec ??
• Magnitude limits ~ 20-22 mag. (depending on band, plate emulsion)
• Relative photometric accuracy <0.05 mag.
• Absolute photometric accuracy ~0.3 mag.

5. Catalogue cross-correlation issues

The XMM – optical cross-correlation is a typical example of a common situation where one is trying to match objects in two catalogues with rather different sky densities and positional accuracies. This can (and does in this specific case) lead to the possibility of one-to-many matches, e.g. several potential matches for each X-ray object considered. (The XMM-Newton positional accuracy is on fact good enough that this only happens for ~10% of the time at high galactic latitudes for images down to r~23-24 mag.)

Such cases generate the need to:

• Determine a figure-of-merit that can be used to identify the most likely counterpart match
• Handle multiple matches correctly in subsequent usage of the cross-correlated data

There is a well-established formalism for determining the most likely counterpart based on a likelihood estimator..This likelihood depends on the positional offsets relative to the positional error (and could also incorporate the object density as a function of magnitude CHECK). This algorithm is already well documented and various implementations exist and have been extensively exercised with real data.

A related but different issue arises where a single X-ray ‘object’ may be associated with multiple optical objects. The most obvious example is provided by clusters of galaxies where the cluster appears as a single extended X-ray source (at sufficient signal-to-noise) but the counterparts are the set of galaxies comprising the cluster. HII regions and star-formation regions may have similar issues. One can also envisage other similar cases, e.g. where an unresolved X-ray source (e.g. in an external galaxy) is associated with a complex of optical objects. The extent to which solutions can be or should be found for cases of this sort is unclear.

6. Context of AstroGrid activities

It is important that the relationship between AstroGrid activities and those of other parties are clearly defined and that responsibilities are unambiguous, as there are potentially substantial grey areas.

As a working model we should assume that:

• Catalogues are made available to AstroGrid in final form or close-to-final form with adequate documentation.
• Advice on the content of the catalogues, special issues relating to the data and in particular quality, reliability and calibration issues should be provided to the AstroGrid by the relevant external experts.

7. Proprietary rights

XMM-Newton observational data, including source lists, are proprietary for one year from the date that data are made available to the observer. Exceptions are calibration, performance verification and target of opportunity observations which have no proprietary period. Apart from these exceptions the first public release of XMM-Newton data will take place in April 2002 and a very large volume of data will be public by late 2002 (processing problems led to delays in the release of XMM data in 2001).

INT WFC data are also, I believe, proprietary for one year.

SuperCOSMOS data are fully in the public domain.

AstroGrid activities will need to respect proprietary rights, e.g. in terms of any public release of data, but there should be no significant difficulties in using proprietary data for testing, prototyping and development purposes.

-- BobMann - 15 Sep 2002 [In original form, by MikeWatson, on 2002-02-14]