[daip] VLA A-array FITS and ps files - AL218 Snapshot Survey (Texas Survey Sources) (fwd)

[Eric Greisen]

Ray Lucas writes:
 > ---------- Forwarded message ----------
 > 
 > 1) the deprecated EPOCH keyword is used (EQUINOX is now recommended)

       Yes but EPOCH is well understood - I suppose we ought to
change for the future (frankly I thought we already had).

 > 
 > 2) the deprecated BLOCKED keyword is used. This was originally used to
 > describe tape files where more than one logical record was contained
 > within one tape record. As far as I know it has no meaning for disk FITS
 > files but its use is not recommended.

      I hated this keyword when it was proposed but then it was
required for any data set actually blocked.  AIPS can still write
tapes (e.g.DATs et al) and the word still has meaning.  Old packages
may still require it and I do not plan to drop it.  It never had any
real meaning anyway.

 > 
 > 3) Less important, the BSCALE and BZERO keywords are legally specified
 > but they serve no purpose. Including them will just mean that every data
 > point will be multiplied by 1.0 and 0.00 added. They were originally used
 > so floating point numbers could be specified using scaled integers.

    But one may still write scaled integers and there are good reasons
to do so (compression on carefully scaled/truncated images works very
well for network transmission - see FITAB in AIPS).  The keywords are
meaningless for floating-point images but we write them as (1,0) so
they are harmless there and they are essential with 16 or 32 bit
integer images.

 > 
 > In order to archive your files, I would suggest to change the EPOCH
 > keyword to EQUINOX and remove the BLOCKED keyword. The BSCALE and BZERO
 > are an unnecessary complication, but they won't do any harm. So, if you
 > don't want to remove them, that's fine as well. Do you think you could do
 > these changes?

This guy is being foolishly pedantic - if these are his worst problems
reading FITS he is leading a charmed life!
 > 
 > I was trying to understand the data structure. Your fits files contain
 > (1024x1024x1x1) arrays. I am used to having all the frequencies in one
 > large datacube. It seems that these files contain only one frequency each.
 > I guess, I should better have a look at your AAS file and the attached
 > paper. I'll do that this week.

Yes there is one polarization and one frequency in your image.
Optical guys have never understood - even after many many years of
explaining - how one should have 1-pixel axes so as to convey
coordinate information.

Eric Greisen