[fitsbits] FIT Keyword data dictionaries

[Mark Calabretta]

On Tue 2000/11/07 09:54:30 CDT, Doug Mink wrote
in a message to: fitsbits at cv.nrao.edu

Dear Doug,

>CTYPE1  = 'AZ---LIN'
>CTYPE2  = 'ALT--LIN'

Those dashes would lead me, and perhaps WCS interpreting software, to think
that this is intended as a non-linear FITS WCS constructed according to the
precepts of the three WCS papers.  But

   1) "LIN" is not a recognized cartographic or spectral transformation
      code.

   2) "AZ/ALT" is not a recognized spherical coordinate system or spectral
      code and doesn't fall into the "xLON/xLAT" or "yzLN/yzLT" mold.

So general WCS interpreting software wouldn't know what to make of it.
It ought to conclude that this is a non-linear WCS that it has not been
programmed for and refuse to process it.

However, if this header really is intended to represent a non-oblique
(linear) coordinate grid then just use the trusty Wells et al. (1981)
prescription:

   CTYPE1  = 'AZ'
   CTYPE2  = 'EL'

On the other hand, if you did want an oblique grid (see fig. 2 of WCS
paper 2) then you need to use the '-CAR' projection code and use, say
'VLON/VLAT' and set WCSNAME to "Azimuth / elevation".

(We defined coordinate codes only for commonly used systems since there is
no way (or need) to preempt the potentially enormous numbers of coordinate
systems which might be needed, e.g. longitude and latitude on the surface
of Europa.  The WCSNAME card can be used to label the spherical coordinates
in human-readable form.)

Mark Calabretta
ATNF