VIMOS vs. CFHT12K astrometry

Executive summary: The positions of the targets from the VVDS catalog are overlaid on VIMOS image from the first commissioning run. Depending on how the VIMOS image is registered, the results are of varying quality:
  • If the World Coordinate System of the VIMOS image is computed using the USNO astrometric standard catalog, the results are unsatisfactory, even when allowances are made for distortion present in the VIMOS optics. The RMS position error is on the order 0.7 arcseconds. This is not a systematic shift but varies across the field.
  • If the WCS of the VIMOS image is computed using the VVDS catalog as the astrometric standard the results are much better, with the RMS position error being around 0.25 arcseconds.
  • Motivation: The VVDS catalogs are generated from CFH12K data. The individual images are stacked to form large mosaiced images. Galaxy catalogs are generated from these mosaics. The catalogs have a nominal astrometric accuracy of 0.3 arcseconds. However it is possible larger astrometric errors exist in some regions.  Particularly at the joins between images of the original images, shears may exist. Hence, if the spectroscopic masks are designed under the assumption that the catalog astrometry is perfect and only VIMOS distortion need to be accounted for, there is a possibility that targets may not end up in their slits.

    The screen grab at left shows the a section some of the  pre-imaging data taken during the first commissioning run. A linear WCS as computed for the field by comparing with the USNO catalog. The WCS in the image header was not used. Using this WCS, the targets in the F14 catalog were plotted on the image. Their positions are shown as red 3'' circles. There are a number of objects not centred in their circles. Note that a simple shift will not correct this. Some objects are well centred, some are not.
    There are of course distortions in the VLT focal plane which  cannot be corrected with a linear WCS. Therefore a 2nd order plate solution was computed using the USNO stars in the field. This should correct for the distortion.

    The situation is marginally better, but still sub-optimal. There are still several sources not in circles.

    If, instead of fitting the WCS of the VIMOS image to the USNO catalog, one fits to  the F14 catalog instead of the USNO catalog, the situation is much better. The sources are almost all centred.
    These results can be quantified. For every target in all the VIMOS images of the F14 field, the position determined from the VVDS catalog was compared to the position on the VIMOS frames: The RMS offsets for the three methods is shown below.
    of fit
    offset ('')
    Linear USNO  0.95
    2nd  USNO 0.70
    2nd VVDS 0.24



     There are two possible conclusions:
    • The astrometry in the VVDS catalogs is not perfect. This would mean that some sort of re-centering would have to be done using the pre-imaging data.
    • The WCS of the VIMOS images computed in the present analysis do not accurately describe the instrumental distortion. 

    Send comments/suggestions/problems to