[Difx-users] Handling lin-pol data with difx2fits
harro verkouter
hverkouter at gmail.com
Thu Apr 11 09:26:35 EDT 2019
I believe neither AIPS nor CASA have documented support for H/V
polarization labels. Checking AIPS memo 114 (revised 2016 version) and the
CASA Stokes.h source file (containing the enumerated polarization
combinations it recognizes) and it's only X/Y, R/L that's supported. CASA
has a little "win" here because it has built-in support for naming mixed
linear/circular polarization combinations. These are, by current definition
of the FITS-IDI standard, unrepresentable in FITS-IDI.
I agree that maybe it would be better if they had support for H/V, but that
might need a disproportionate investment of coding.
On Thu, 11 Apr 2019 at 14:45, Walter Brisken <wbrisken at nrao.edu> wrote:
>
> Three comments here.
>
> 1. For the polarizations to be properly propagated, a vex file correctly
> identifying the polarizations is needed from the start. Is it possible
> that the wrong polarization labels started with an incorrect vex file?
>
> 2. It is strictly true that H/V would be the more appropriate labels for
> alt-az antennas. However, for the few HaDec mounts. For other mount types
> (e.g., X over Y, Nasmyth, RadioAstron, ...) neither prescription is
> correct. So there is no pleasing everyone. I think the best thing to do
> is just consider the "X/Y" that comes from the correlator as "uncalibrated
> X/Y". Downstream calibration (when applying the parallactic angle
> correction) will then calibrate the polarization.
>
> 3. CASA also uses X/Y for all linear polarizations. The cost of changing
> to H/V would be large, with almost no tangible gain. I see no reason to
> pursue code changes to this effect.
>
> -Walter
>
> On Thu, 11 Apr 2019, Leonid Petrov wrote:
>
> > Dear colleagues,
> >
> >> I second Chris's note that X/Y labels seem to be getting propagated
> >> correctly via the FITS-IDI file to AIPS at the moment in DiFX-produced
> >> files.
> >
> > I have VGOS experiment b17339
> >
> > b17339/h_1001.input :
> >
> > ...
> > NUM REC POLS 31: 2
> > REC BAND 0 POL: X
> > REC BAND 0 INDEX: 0
> > REC BAND 1 POL: X
> > REC BAND 1 INDEX: 1
> > ...
> > REC BAND 9 POL: Y
> > REC BAND 9 INDEX: 1
> > REC BAND 10 POL: Y
> > REC BAND 10 INDEX: 2
> > ...
> >
> > and difx2fits puts 'R' in POLTYA and 'L' in POLTYB
> > I see these lines in applications/difx2fits/src/fitsAN.c (2.5.2):
> >
> > polTypeA = 'R';
> > polTypeB = 'L';
> >
> > i.e. polTypeA, polTypeB are just hardcoded! Ivan, can you look at
> > your version fitsAN.c? Maybe somebody has already applied a similar
> patch?
> >
> >> But I suppose that the data may be stored in multiple places in the
> >> FITS-IDI file, and it is entirely possible that one place is correct and
> >> another is not!
> >
> > How a program that parses FITS-IDI is supposed to learn the
> polarization??
> > Keep in mind, this is the first step. I have mixed polarization VLBI
> > experiments in my desk that difx just rejects. For this experiment
> incorrect
> > propagation of polarization labels will be disastrous.
> >
> > Ivan, thank you for your comment. I missed that some antennas really
> > record X and Y. In that case if AIPS expects X and Y polarization data
> and
> > if we will try to eat VGOS data that record H and V, it will produce
> totally
> > wrong result! If today AIPS does not support logic for handling H and V
> data,
> > we need ask Eric to add an error message like "Do not support H and V pol
> > data yet".
> >
> > Then the question, how the labels in *.input files are set an the first
> place?
> >
> > Leonid
> > 2019.04.11_07:38:29
> >
> > _______________________________________________
> > Difx-users mailing list
> > Difx-users at listmgr.nrao.edu
> > https://listmgr.nrao.edu/mailman/listinfo/difx-users
> >
>
> --
> -------------------------
> Walter Brisken
> NRAO
> Deputy Assistant Director for VLBA Development
> (575)-835-7133 (office)
> (505)-234-5912 (cell)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listmgr.nrao.edu/pipermail/difx-users/attachments/20190411/44955094/attachment.html>
More information about the Difx-users
mailing list