XMM-Newton Optical Monitor Serendipitous UV Source Survey

Data Processing: Astrometry

XMM-OM source lists are compiled in raw detector coordinates in order to simplify the process of quality flagging. In this section, we describe the process of transforming raw coordinates within the detected source table to sky coordinates, J2000 Right Ascension (RA) and Declination (Dec).

Image distortion

As with most optical designs, the elements within the OM optical path introduce image distortion. Relative pixel locations of sources will depend on the pointing of the telescope. As a space-based detector, there is no gravity-related flexure across the telescope, so the distortion map used to correct source positions is constant over time. The correction is an offset in the two CCD dimensions, it varies with detector position, and is illustrated in Figure 1. The SAS data reduction pipeline characterizes the distortion as a 2-dimensional cubic polynomial. After correction, the plate scale is linear, which in the unbinned image case is 0.476513 arcsec pixel-1. Correction for optical distortion performed upon raw images. Fig 1: The right hand image illustrates the optical distortion correction that is performed on raw data. Colour-coding represents the magnitude of the positional offset, Δα, which is a function of detector position.

Coordinate transformation

The attitude history of the spacecraft during each observation, sampled typically every 10s, is telemetered to the ground with each pointing. It is determined by the star trackers, locked on to a sample of pre-determined guide stars. The stability of the star tracking is generally good to 1 arcsec over the duration of a pointing. The OM provides an additional on-board source tracking feature which measures the positions of several guide stars in the OM field-of-view, typically with 20s sampling. An onboard algorithm corrects all event locations relative to a reference image before summing events into a single image and telemetering to the ground. The only attitude data required for OM coordinate transformations are therefore those timestamps that correspond to the reference times (one per image) against which all recorded events are adjusted. There is no onboard roll adjustment, but flight experiments indicate these are unecessary. The shift-and-add process just described will not work if the OM cannot detect suitably bright reference stars. In such cases, no shift and add is performed and the attitude history for an image is averaged over the exposure duration.

There is a calibrated offset between the OM boresight and the star trackers which is taken into account before transforming distortion-corrected image pixel coordinates to the sky. The tranformation also includes a rotation about the spacecraft boresight to produce tangent plane coordinates.

Astrometric correction

While source positions within the transformed tables are accurate relative to each other, the resulting sky coordinates are only good to a few arcsec because of the intrinsic absolute accuracy of the star trackers. One single, additional transformation applied to all sources within an exposure corrects for this.

The fine-correction is performed relative to sources in the USNO-B1.0 catalogue. A grid is computed of detected OM source positions, shifted by a discrete range of offsets in RA and Dec. The best pair of offsets are chosen from the grid using a maximum-likelihood parameter. If sufficient stars are matched, then the offsets are further improved by a least-square fit against the USNO-B1.0 sources. In these cases a rotation correction can be performed, if necessary. For the majority of observations where the number of detected sources, Nsrc > 30, the requirement for 10 identified USNO sources was set. For the cases where Nsrc ≤ 30, the requirement was for max(4, Nsrc/4) correlated USNO sources. Figure 2 presents the full sample of offset corrections which meet the least-square matching criteria and consequently were included within the catalogue. Only those corrections performed using least-square fitting are included in the catalogue. Offset errors are calculated using Monte Carlo simulations and propagated into the RA and Dec uncertainties derived from Gaussian moments during the source detection process. Positional offset correction performed on XMM-OM images after
cross-correlating detected sources with the USNO-B1.0 catalogue. Fig 2: Using source correlations between the XMM-OM and USNO-B1.0 catalogues, systematic corrections are applied to the recorded XMM pointings and source positions. The above plot summarizes the full sample of offsets applied in both the RA and Dec directions (red dots). The black cross is the mean correction and the blue circles represent 1-, 2- and 3-σ radii from the mean.

No sources are removed from the catalogue based upon the magnitude of the astrometric corrections. The J2000 RA and Dec of each least-square-corrrected object are added to the source table, with 1σ errors and corresponding Galactic longitude and latitude.

SAS task: omatt
CCF product: OM_ASTROMET_000n.CCF (distortion)