[fitswcs] TAN+poly & astrometric discussions in WCS paper-2

[Dr.Brian McLean]

Greetings,
I've just been reading the recent spate of email regarding the TAN+poly vs.
ARC+poly and would like to add a couple comments.

As you all know, the DSS-I used the TAN+poly approach and even though we
never fully complied with the FITS WCS (for reasons that are not worth going
into here), the popularity of the DSS was one of the drivers for adding
polynomial functions to the WCS.

Many of Peter's comments about the ARC projection for Schmidt plates are
correct and I stronly support the use of ARC+poly in the new standard....in
fact I was counting on it!!

In principle, the TAN+3rd order poly is comparable to the ARC+2nd order
polynomial BUT, we found that using the ARC, we were able to get more stable
correction functions (to correct systematics at plate edges -- which is why
we also support the pixel correction maps in the proposed standard).

Therefore, the new GSC-II is being reduced using an ARC+poly+pixel
correction map procedure and we intend to release new DSS-I and DSS-II
headers based on these new plate reductions in order to give the community
the best possible astrometry. We would like to be fully compliant with the
WCS standard this time.... :-)

	Cheers
		-Brian

Dr. Brian McLean, Branch Chief,
Catalogs and Surveys Branch, Space Telescope Science Institute,
3700 San Martin Dr., Baltimore, MD 21218, USA

URL: http://www-gsss.stsci.edu
Phone: +1-(410) 338-4900
Fax: +1-(410) 338-4767

Visiting Apache Point Observatory/Sloan Digitial Sky Survey, Sunspot, NM
Phone:  +1-(505) 437-6822