[fitsbits] [EXTERNAL] Re: Stokes parameters coordinate system

[Tim Pearson]

I am not aware of any standard convention for defining the polarization convention in use in a FITS file.

The position angle of the electric vector and hence Q and U may be referred to Galactic north, Equatorial north, or an instrument frame, and in practice the frame of the polarization (equatorial or Galactic for example) is not always the same as that used for the angular coordinates of the image.

The CMB community introduced further confusion by failing to follow the IAU convention for position angle and hence of Q and U. See:

https://www.iau.org/static/archives/announcements/pdf/ann16004a.pdf <https://www.iau.org/static/archives/announcements/pdf/ann16004a.pdf>

The deprecated "CMB" convention is used for most CMB-related HEALPIx images: see 

https://healpix.jpl.nasa.gov/html/intronode12.htm <https://healpix.jpl.nasa.gov/html/intronode12.htm>

A FITS header keyword POLCCONV is sometimes used to indicate the polarization coordinate convention; its value is either COSMO for files following the Healpix/CMB/Planck convention or IAU for those following the IAU convention. But this does not specify the zero point of position angle.

Tim Pearson

> On Mar 2, 2020, at 2:01 PM, Thompson, William T. (GSFC-671.0)[ADNET SYSTEMS INC] via fitsbits <fitsbits at listmgr.nrao.edu> wrote:
> 
> I don’t think people are understanding my question.  Table 7 of the WCS paper lists how to encode where the I, Q, U, and V values are stored, but they don’t say how they’re defined.
>  
> Let me put it this way.  Stokes Q is defined as the horizontal power minus the vertical power, but what does “horizontal” and “vertical” mean in this context?  Is it horizontal and vertical with respect to the telescope, or with respect to some kind of world coordinates such as RA and Dec?
>  
> More to the point, are there conventions for delineating which is used?
>  
> Bill Thompson
>  
> From: fitsbits <fitsbits-bounces at listmgr.nrao.edu> on behalf of Peter Teuben via fitsbits <fitsbits at listmgr.nrao.edu>
> Reply-To: Peter Teuben <teuben at astro.umd.edu>
> Date: Monday, March 2, 2020 at 4:52 PM
> To: "fitsbits at listmgr.nrao.edu" <fitsbits at listmgr.nrao.edu>
> Subject: [EXTERNAL] Re: [fitsbits] Stokes parameters coordinate system
>  
> Bill
> 
>       If an instrument doesn't measure one of the standard I,Q,U,V (stokes 1,2,3,4) or XX,YY,.... (stokes -5,-6,...) or RR,LL,... (stokes -1,-2,...) for which we have an integer convention (such that even CDELT3 etc. work), i'm inclined to say to assign it a non-standard integer (but negative), and indeed put in the comment fields that these are non-standard and not calibrated yet. For example -12,-13 might be nice if they somehow reflect Q and U ?
> 
> Hopefully the calibration routines will then output proper stokes Q and U. presumably the FITS header will have other parameters that the calibration routines know how to deal with....
> 
> just a wild guess
> 
> - peter
> 
>  
> 
> On 3/2/20 4:09 PM, Thompson, William T. (GSFC-671.0)[ADNET SYSTEMS INC] via fitsbits wrote:
>> Folks:
>>  
>> WCS Paper I describes a convention for Stokes parameters.  I’ve been asked how to best represent the coordinate system that these Stokes parameters are expressed in, specifically the Q and U parameters.  Is there a standard practice of how this should be done?  It appears that some space-based instruments define Q and U relative to instrument (pixel) coordinates.  However, the comment was made that this is not really practical for a ground-based alt-az telescope due to field rotation.  Another possible way to encode the Stokes parameters would be relative to the real-world coordinates of the data.
>>  
>> Suppose that one had a three dimensional cube with the following axis definitions:
>>  
>> CTYPE1 = ‘RA---TAN’
>> CTYPE2 = ‘DEC--TAN’
>> CTYPE3 = ‘Stokes’
>>  
>> And that the PC matrix had cross terms between axes 1 and 2, i.e. image rotation.  Is there a convention for how the Q and U values should be interpreted?  For example, if Q/I=1, would you interpret that as polarization aligned along the first pixel direction, or along the RA direction as defined by the PC rotation matrix?
>>  
>> Is there a better way to distinguish between these cases other than putting a comment into the header?
>>  
>> Thank you,
>>  
>> Bill Thompson
>>  
>> 
>> 
>> _______________________________________________
>> fitsbits mailing list
>> fitsbits at listmgr.nrao.edu <mailto:fitsbits at listmgr.nrao.edu>
>> https://listmgr.nrao.edu/mailman/listinfo/fitsbits <https://urldefense.proofpoint.com/v2/url?u=https-3A__listmgr.nrao.edu_mailman_listinfo_fitsbits&d=DwMDaQ&c=ApwzowJNAKKw3xye91w7BE1XMRKi2LN9kiMk5Csz9Zk&r=4qrOT2Lux_0sXV9XNPmKKlSxaCYiFna5mbzVK8hbN2E&m=pfFrBl-T-6p2Pn50qdNZ942KwzfwIhDrTxoOrOLCgHc&s=l98r3qSi7RcJefjajrwjYGwPaDknbbvRiCrMVE9npMc&e=>_______________________________________________
> fitsbits mailing list
> fitsbits at listmgr.nrao.edu
> https://listmgr.nrao.edu/mailman/listinfo/fitsbits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/fitsbits/attachments/20200302/5f538bce/attachment.html>