spectral WCS in FITS

[Robert Hanisch]

----- Original Message ----- 
From: "Markus Dolensky" <Markus.Dolensky at eso.org>
To: <dal at ivoa.net>
Sent: Tuesday, November 25, 2003 7:11 AM
Subject: Re: spectral WCS in FITS


> It all boils down to the question:
>
> Is there a way to express a vector in a single cell of a FITS ASCII table,
right?
>
> If so, FITS BINTABLE structure proposed by Greisen et al. can be mapped to
> ASCII tables. And consequently XSLT can be used to transform between ASCII
> table and VOTable formats.
>
> I don't know the answer, but at least it should clarify what we are
looking for.
>
> Markus

The answer to "Is there a way to express a vector in a single cell of a FITS
ASCII table?" is no.  In ASCII tables, vectors map into columns.

An earlier draft of the spectral WCS used binary tables, but entire columns
rather than vector cells.  That format would indeed have been representable
in ASCII tables.

To respond to Anita, the spectral WCS proposal is quite intimately tied to
FITS.  The coordinate transformations, of course, can stand alone, but how
the data is structured does depend on the capabilities of FITS.  Which is,
again, the reason for my query to VO folks.  There are FITS representations
which map quite cleanly to VOTable, and there are FITS representations that
do not (and thus get wrapped by a VOTable).  It is fine to say let FITS be
FITS, we don't care, but for the case of simple 1-d spectra and SEDs where
the data structures Could be extremely similar, should we push to have them
so?

Bob