[fitswcs] WCS documents

[Stephane Guilloteau]

I have read Doug Tody proposal, and find quite a number of convincing
arguments
to decouple the WCS from the NAXIS problem.

However, as far as I understood, the ordering of the WCS axes must remain
the same as the ordering of the
Image axes (NAXISi). I think this is an unnecessary restriction which may
lead to cumbersome
code, and that we should look into the alternative, which is to specify a
mapping between the
WCS axes and the image axes, e.g. by a WCS keyword like

    WCSAXISi    = j

indicating that image axis J corresponds to WCS axis I

or vice-versa (which is even more convenient)
    WCSj          = i
indicating that the image axis J must take its WCS coordinate from WCS axis
I

The latter form allows to carry on very easily all the World Coordinates
whatever transposition is done on the data.
It also make things somewhat easier when a "dumb" image must be associated
to a WCS space with no  associated data as in Doug's example. Simple copy of
keywords would do the job. Keyword inheritance could be used to derive WCS
coordinates in IMAGE XTENSION, even if the image extensions represent
different things like a sky map first,
than a long slit spectrum.

        Stephane


Dr. Stephane GUILLOTEAU
ALMA European Project Scientist                Phone: (33) 476 82 49 43
(IRAM)
IRAM                                                         FAX:     (33)
476 51 59 38  (IRAM)
300 Rue de la Piscine
F-38406 Saint Martin d'Heres
        France                                                  E-Mail:
guillote at iram.fr    (IRAM)