Re: Astrometry errors

[<aah@nofs.navy.mil>]

Doug wrote about astrometry errors.  This has been covered
in emails and also the Technical Notes regarding the Mark IV
reductions.
  The primary astrometric error sources are optical distortions
(usually cubic terms) and optical aberrations (coma).  I think
both Michael and Andrew have shown that they can get ~0.5arcsec
accuracy over the entire frame.  Since this is about 1/10 pixel
or so, the consensus is that the astrometry is 'good enough'.
The image profile is about 2 pixels fwhm, dominated by the
optics and not seeing and so is relatively stable.  This is
good enough that profile sampling is adequate.  There is
some tracking error on some images which will increase the astrometric
error.  No one has done a real careful job on how the focus
changes with temperature or how the astrometry changes with
focus.  There are lots of basic tests that still need to be
performed; I think Michael and I will be starting those Real
Soon Now.
  I think 0.5arcsec is quite adequate.  The match will be with
respect to real astrometric catalogs like UCAC so you mainly
want to ensure you are matching the right star.
Arne