[daip] clean 3bit Ka data

[zadeh]

Hi Eric,
Thanks for the suggestions. I never thought that aips could do this but 
I will give it a try.
One other  possibility that you might want to consider is  to go through 
cubes of individual scans (1-5min) on a short time scale and make the 
flux of
the bight source (Sgr A*) constant using uvsub. This should work 
because  the rest of the flux is thermal and the flux does not change 
much with
frequency. This method works nicely as Doug came up with this 
suggestion  using  vla's archival  data. However,
the  new correlator is creating  head-ache for us  as there are too many 
channels and IFs to do uvsub. Once the flux of the non thermal point 
source is fluxed
in the uv data, then deconvolution and self-cal works much better. The 
real proble we have what to do with Q and U as  they can go positive and 
negative and
this technique does not work.

Anyways, I will certainly try your suggestion.
Cheers,
Farhad


On 1/23/14 11:44 AM, Eric Greisen wrote:
> zadeh wrote:
>> Hi Eric,
>> We have broad band  data sets at Ka band but the flux of the bright 
>> source varies as a function of frequeny as nu^0.3 across the 8GHz of
>> bandwidth. How do I clean this data correctly  or even  self-cal this 
>> data in aips? Any suggestions would be appreciated.
>>
>> Another question. What is the task name to do Faraday depth map and 
>> Faradat RM map? We have  images
>> at multiple frequencies across the band.
>
> Two excellent - nearly research level - questions.
>
> 1. If most of the flux has a single spectral index, then IMAGR will 
> allow you to correct it in the UV data work file using IMAGRPRM(2).
>
>   If there is real structure in spectral index that might matter, you 
> need to do a multiple step procedure.  First IMAGR the data in 
> spectral chunks of modest size.  Take the output images and make a 
> cube with MCUBE or FQUBE. then bring the planes of the cube to the 
> same resolution (usually just the fattest of your beams) with CONVL.  
> Then run SPECR on that cube.   Take the SPECR output as IN3NAME et al 
> (and perhaps IN4NAME if curvature is significant) in an IMAGR that 
> includes all channels/IFs.
> Use FQTOL to make this process run faster and also include the primary 
> beam (IMAGRPRM(1)) when you do this since it won't cost you anything.
>
> 2. RM analysis is definitely a research topic.  The current best 
> answer is to run FARS on your P and Q cubes (corrected for spectral 
> index) (task SPCOR).  Then run RMFIT.
>
> I am about to go back to work on RMFIT to add another parameter to be 
> fit having to do with the thickness of the Faraday screen. that will 
> take a while but you will have much to learn.  The Faraday rotation 
> measure synthesis, done with FFTs, has a very fat point-spread 
> function (beam) and so is no so useful.  But RMFIT is aided with the 
> FARS output and so you must wait for it (FARS is slowwwww).
>
> You may wish to talk to Larry Rudnick (larry at umn.edu) as he is expert 
> on the subject.
>
> Eric Greisen