[fitswcs] FK5-ICRS

[Don Wells]

Andreas Wicenec writes:
 > The differences between FK5 J2000 mean equinox and equator and IRCS
 > should be less than 80 mas by definition of the ICRS (Arias et al., A&A,
 > 303, 604, 1995). This number is just the uncertainity of the FK5 system.
 > BUT: The *realisation* (reference frame) of the FK5 system at J2000 by
 > the GSC is actually poor, not only because of the quality of the GSC
 > reduction but also because of the mean epoch of the POSS and SERC plates
 > used to construct the GSC.

Thank you for the 80_mas number; I didn't know that.

I had lunch today with the two key people (Muxlow & Richards) in the
MERLIN/HFF comparison; it was Muxlow who gave the lunch talk here in
Charlottesville on Tuesday, and he will be speaking at STScI in
Baltimore tomorrow.  Eric Richards told me that they used a
recent-epoch Palomar 5-meter CCD image to make the transfer of
coordinates, using VLA coordinates of radio galaxies in the 10_arcmin
field as the reference coordinates. They then used faint galaxies to
transfer the coordinate system from the 5-meter image to the HFF
imagery. This means that the FK5-ICRS differences were eliminated, and
that GSC inadequacies didn't matter. It also means that proper motions
were not a problem, because the epoch difference was negligible and
because they used galaxies (zero proper motion).

I conclude that the coordinate differences which Muxlow and Richards
observed are probably due to inadequacies of modelling the geometries
of both the 5-meter and WFPC2 imagery, not due to reference frames.  I
am sure that Muxlow and Richards could improve their geometry
modelling techniques, but I still contend that they should not have to
know about these technical matters---they should be concentrating on
the astrophysical issues.  It is our duty to design and implement
universal solutions (astrometric software and interchange notations)
which will assure, with high probability, that Muxlow and Richards and
all other astronomers will not need to worry that astrometric problems
will corrupt their astrophysical conclusions.

 > Even with those new reductions going on with GSC and USNO-Ax the
 > realisation of the ICRS by those catalogs will be not very good, because
 > of the epoch problem mentioned above. Even the Hipparcos frame will
 > degrade very soon because of the errors in proper motion determination.

USNO combined the Hipparcos Tycho catalog with the Astrographic
Catalog to obtain good positions and proper motions for 10^6 stars. It
appears to me that USNO's ACT catalog it is currently the best single
overall star catalog for astrometric reductions at arbitrary epochs. 

Monet scanned the original POSS. If the POSS-II is also scanned, and
if the reduction of the combined datasets is put onto the system of
the ACT, it should produce an excellent reference catalog of 10^8
positions and proper motions.

I agree: there should be at least one more new astrometric satellite mission.

-Don