[fitsbits] Question about CROTA

[William Thompson]

Mark Calabretta wrote:

 > AIPS wrote FITS files with CROTAn attached to the latitude axis and
 > documented this usage in its memos and programming manuals.  It was an
 > informal usage, but the only one we know of and widely used in other
 > astronomical software packages (if you know of any other usage then
 > we would be extreeeemly interested to know).  This is fortunate because
 > if there had been two or more conflicting usages then interpretative
 > software would have been confronted with the problem of disambiguation.

We don't tend to use any of the software packages that you consider standard. 
We use the SolarSoft library, at

	http://www.lmsal.com/solarsoft/

which is really a vast collection of routines that have been aggregated together 
over the years, from multiple sources.  The treatment of CROTA values is not 
completely standardized.  Historically, sometimes it's been taken from the 
latitude axis, and other times it's been taken from the longitude axis.  That's 
why it's always been safest to put the angle in both axes.  I've been trying to 
make it more standardized, but it's an uphill battle.

The question is, what happens in AIPS and other software packages, if the CROTA 
value in the latitude axis is replicated in the longitude axis?  Is it simply 
ignored, or does it lead to unintended consequences?

> Thus the new WCS standard deprecates CROTAn because it has no formally
> defined interpretation.  Writing it with the new WCS keywords/values,
> whether on one axis or two, violates the spirit, if not always the
> letter, of the standard.

I completely agree, but getting people to abandon CROTA is also an uphill 
battle.  I do tell people they can use the old CROTA, or the new PC, or CD, but 
that they can't mix them.

Bill