[fitswcs] Polarization codes

[William Thompson]

Steve Allen wrote:
> On Fri 2008-03-14T02:39:27 +0000, Paddy Leahy hath writ:
>> Example: From (IQUV) absorption spectra, calculate the optical depth
>> separately for polarized and unpolarized emission. The ratio (fractional
>> polarization of optical depth) tells you something interesting about the
>> absorbing material. The polarized absorption may also be anisotropic,
>> hence there may be an angle for the polarized optical depth. This can
>> happen in theory for absorption by Zeeman-split lines, although I have to
>> admit that I've not heard about a measurement in an astronomical context.
> 
> Isn't this pretty much the data used by a solar magnetograph?
> If so, we need input from practicing solar astronomers before we're done.

I can give you some information from my own experience.  There are three cases 
that I'm aware of where polarization is important:

1.  Line-of-sight or vector magnetographs
2.  Coronagraphs
3.  Solar radio astronomy

My own experience is with coronagraph data.  There are two quantities that are 
important in coronagraphic polarimetry, the total brightness (denoted B), and 
the linearly polarized brightness (pB).  Since the scattered light from the 
corona is always polarized tangential to the limb, the actual direction of 
polarization (mu) is not generally of interest, except as a check on the 
validity of the data.  I will point out, however, that the definition of the 
angle of polarization as we use it differs by a factor of two from that 
described in Paddy Leahy's original post.

The coronagraphs on STEREO take measurements at 3 polarization angles separated 
by 120 degrees.  These are stored separately as individual 2D FITS files, with 
the keyword POLAR in the header giving the angle of the linear polarizer for 
each exposure.  From image triplets the total brightness and polarized 
brightness images (or the ratio) are formed and stored in separate 2D FITS 
files.  I doubt it would really be of interest to define a "polarization axis" 
where both B and pB are combined into a single array.  In any case, one couldn't 
include the fractional polarization or angle in this scheme because of the 
different units.

I've talked to some of my colleagues about magnetometer data, which I'm not as 
familiar with.  Usually, magnetometer data is distributed already combined into 
physical quantities, such as magnetic field strength, or azimuth and elevation 
angles.  When the actual polarization state of the data are distributed, they're 
distributed as I,Q,U,V images, which are already well covered by the current 
standard.  (However, I'm not aware of anybody who actually *uses* that standard 
for magnetograph data--probably, they don't even know it exists.)  The actual 
filtergrams are not routinely distributed.

Finally, I have very little familiarity with solar radio data.  I'm given to 
understand that radio data are usually distributed as either separate left and 
right polarization data, or as Stokes I and V, both of which are covered by the 
current standard.  Some of my colleagues who are more familiar with radio 
observations are gone for the day, so I may have more to say on the subject 
later.  In any case, there's a lot of overlap between solar and celestial radio 
astronomy, e.g. both use the VSO, so I would expect that community practices 
would be very similar.

Personally, I think that one has to make a distinction between data axes and 
data types, and they should not be confused together.  The current Stokes 
standard in the WCS makes a certain amount of sense, because a "Stokes vector" 
is a well defined quantity.  These other parameters, such as B and pB, are 
quantities, and should be described as such, e.g. through a keyword such as 
BYTPE.  One should definitely *not* use degenerate axes for this kind of 
information.

Bill Thompson


-- 
William Thompson
NASA Goddard Space Flight Center
Code 671
Greenbelt, MD  20771
USA

301-286-2040
William.T.Thompson at nasa.gov

(Note changed email address)