[daip] some aips issues

[Eric Greisen]

Cao Hongmin wrote:
> Hi Eric,
> 
> If a source was found far from the its phase centre,  then 
> theoretically, whether we can refine the position measurement by adding 
> an extra step (which is correcting its position coordinates to the new 
> measurement values) before the fringe-fitting? 
> 
> because the fringe-fitting could not properly deal with the phases of 
> the structures far from the phase centre, especially their visibilities 
> on long baselines,  the averaging (in frequency and in time) could still 
> suffer from the correlation losses, which is the so called smearing, so 
> the data averaging after fringe-fitting could still shrink the FOV(?).
> 
> we found a source which is about one 1" away from its phase centre from 
> the pipe line output data where heavily averaging was made (one IF with 
> a width of 16 MHz was averaged into one channel and the averaging time 
> is 10s, that is, we used aparm 2,10,0,1,0,1 for task "split", the 
> estimated FOV is about 300 mas). I have not try your suggestions of 
> using imagr boxfile to solved the FOV issue that you noted before at the 
> moment.
> 
> after comparing the clean images of the position corrected one and the 
> uncorrected one, it turn out that the smearing effects does play some 
> roles but it seems not very serious. whether it is because the target 
> source is very weak?
> 
> i attach the clean images,
> attachment two: first-3_wtmod1.eps ( the position corrected one)
> attachment one: first-3_wtmod3.eps (produced by undone the correction)
> (first contour is at 3 sigma noise level)

The position corrected image is much better - the flux is 25% more and 
the low-level stripes are relatively lower.  I would say it is always 
better to correct for a known position offset before FRING and other 
tasks as well.
Even if the offset is not known perfectly, a correction which is close 
is a good idea.

Perhaps, you should run SPLIT after all the calibration is ready, then 
UVFIX, and then IMAGR (in a simple mode).  This large shift will change 
the u,v,w some and a better image still might result.

Cheers

Eric Greisen