[daip] Error in parallactic angle

[Andy Biggs]

Hi Eric,

This is old VLA.

I know it's probably very unlikely to stem from a programming error, but 
I'm really lost to explain what I'm seeing. I suppose I'm wondering if 
an error might have been missed for many years because users generally 
avoid using PPA calibrators (or targets) that are close to the zenith. I 
have no choice as, for better or worse, 3C48 was the calibrator chosen 
by these observers and no attempt was made to avoid transit in what will 
have been dynamically scheduled observations. On the other hand, I have 
many epochs and so could remove those that are potentially affected and 
I think that's probably what I'll end up doing. Before I do that though, 
I'd just like to check every possibility.

In case it's relevant, these are A-configuration data and 3C48 is very 
resolved at X and U Band. To measure the PPA of this source I make maps 
and measure the PPA at the peak of the polarized emission. This is not 
ideal and hour-angle effects will play a role due to the varying beam 
and the averaging of the PPA which varies across the source, but I would 
think that the beam would be pretty circular around transit.

The source I'm comparing 3C48 to, 3C119, transits much further away from 
the zenith, is brighter and much less resolved so I'm pretty confident 
in the results derived from it.

Andy

On 09/01/17 19:56, Eric Greisen wrote:
> On 01/08/2017 11:32 AM, Andy Biggs wrote:
>> Hi there,
>>
>> I'm analysing VLA data from two monitoring campaigns that were observed
>> over the same 3-month period. One used 3C119 as the PPA calibrator and
>> the other 3C48. 3C119 is far superior as it's unresolved in polarized
>> flux and doesn't pass as close to the zenith. If I compare the R-L phase
>> corrections, there is a very interesting systematic offset between the
>> two within an hour or two of transit - see attached plot.
>>
>> I showed this to Rick and he suggested an error in the assumed
>> parallactic angle due to misalignment of the antenna pads and, sure
>> enough, the functional form of that agrees well with what I see (blue
>> line on plot). However, to match the data I have to assume an offset of
>> ~10' which is about an order of magnitude too big. Ken Sowinski says
>> that a few pads were erroneously built parallel to local gravity as
>> opposed to the centre of the array and that an error of ~10' would be
>> possible for a misaligned pad at the end of an arm, but repeating the
>> analysis deleting each antenna one by one makes no difference.
>>
>> Although very unlikely, I am wondering therefore if it might be possible
>> that AIPS is making an error with the source or array position when
>> correcting for the parallactic angle. This would be a very neat
>> explanation, particularly as it's not clear to me how random
>> pad-alignment errors would lead to a systematic offset - I would expect
>> them to all cancel out.
>>
>> Might such a thing be possible? Note that the 3C48 R-L corrections were
>> derived from Q and U fluxes measured from maps made using IMAGR.
>
> The current code recognizes the old VLA as being all at the same array
> longitude and latitude but does not recognize the EVLA in the same way.
> Thus each antenna will have its own parallactic angle changing
> systematically across the array.  Could that be the explanation?  It
> would be a bit uncertain how to change this since the array name is not
> in the COMMON that the PARANG routine uses and the JVLA is now 0,0,0
> like most other arrays.
>
> Eric
>
>

-- 

Andy Biggs
ARC Astronomer, ALMA Regional Centre
European Southern Observatory
Karl-Schwarzschild-Straße 2
D-85748 Garching
Germany

tel. +49 89 3200 6471
fax. +49 89 3200 6898