[fitsbits] WCS keywords in tables

[William Pence]

Steve,

In general terms, the Greenbank convention is commonly interpreted to 
mean that if you use a keyword to define some attribute of a single FITS 
image, then that keyword gets converted into a column name when you pack 
multiple images into a vector column of a binary table.  Suppose the 
keyword AIRTEMP = 17 is used to give the ambient temperature at the time 
the image is obtained. If you were to pack many such images into a 
vector column in a binary table, one image per row of the table, then 
you would record the temperature value for each image by creating 
another column in that table whose name is the same as the keyword, 
i.e., TTYPEn = 'AIRTEMP'.  Thus, under the Greenbank convention what was 
a single keyword for a simple FITS image becomes the name of a column in 
the binary table.

The current discussion involves a different situation, where a binary 
table contains multiple images in a vector column, and one uses the 
simple form of the WCS keywords instead of using the specific forms that 
were designed for this purpose (e.g., '1CTYP7', if the vector is stored 
in column 7 of the table, instead of CTYPE1).  By using the simple form 
of the WCS keywords one loses the information about which column of the 
table contains the image, but as long as there is only one vector 
column, or if there are several vector columns, but they all have the 
same WCS, then there is no ambiguity.  And to answer Arnold Rots' recent 
email, if both forms of the WCS keywords exist in the table header, then 
the column-specific form of the keywords would take precedence for that 
particular column.

I think the description of the Greenbank (or should it be Green Bank??) 
convention in the FITS User's Guide 
(http://archive.stsci.edu/fits/users_guide/node69.html#SECTION00723200000000000000 
) is rather confusing, so I agree with Lucio's recent post that this 
convention should be written up and submitted for inclusion in the FITS 
Registry.  If there are no other volunteers, I'll write up a draft.

Bill

Steve Allen wrote:
> On Tue 2010-04-06T12:56:25 -0400, William Pence hath writ:
>> So far there have been no objections to this simplifying WCS convention
>> for images stored in FITS binary tables.  If that's really the case,
>> then I will write up a short description (crediting Mark Calabretta as
>> the originator) and submit it for inclusion in the Registry of FITS
>> Conventions on the FITS support office web site.
> 
> I fear I have missed some subtle point about this.
> How is this not just another application of the Greenbank convention
> as applied to WCS keywords at least as early as 1995?
> http://www.cv.nrao.edu/fits/documents/drafts/wcs_keywords.ps
> Is there some substantive difference in this case?
> 
> --
> Steve Allen                 <sla at ucolick.org>                WGS-84 (GPS)
> UCO/Lick Observatory        Natural Sciences II, Room 165    Lat  +36.99855
> University of California    Voice: +1 831 459 3046           Lng -122.06015
> Santa Cruz, CA 95064        http://www.ucolick.org/~sla/     Hgt +250 m
-- 
____________________________________________________________________
Dr. William Pence                       William.Pence at nasa.gov
NASA/GSFC Code 662       HEASARC        +1-301-286-4599 (voice)
Greenbelt MD 20771                      +1-301-286-1684 (fax)