[fitswcs] Stokes parameters in FITS files

[Eric Greisen]

Jason Quinn wrote:
> I read with interest the thread in the March 2008 archives about
> polarization in FITS files. I have recently been reading the WCS
> standards with an eye towards creating FITS images to store optical
> polarization data of supernovae. Not being intimately familiar with
> the standards, it took quite some time before I started to suspect
> that maybe FITS cannot gracefully bundle all four Stokes parameters
> together. This archived thread seemed to confirm that for me.
> 
> Have there been other relevant discussions about this issue since
> then? Or any other developments to which I could be pointed? Optical
> spectrophotometry is a growing field and I expect that a need for this
> will grow.

We use a STOKES axis all the time with 4 pixels and reference value
either 1 with increment 1 for I, Q, U, V or refernce value -1, increment 
-1 for RR, LL, RL, and LR.  Other combinations are obviously available - 
e.g. ref value -4, increment 1 for LR, RL, LL, RR.

I suspect these choices are shown in several of the basic FITS papers.

Eric Greisen