[daip] Re: About AIPS task -- UVSUB

[Eric Greisen]

Jin Chengjin writes:

 > I have a question about the AIPS task UVSUB. I used it to shift the phase 
 > center of a VLBI multi-IFs data set by deviding data with a point source
 > model. But when the shiftting is very large, say, about hundreds of
 > milliarcsecs, then peak of the image made with the resulting data become
 > weaker. And if we just select one IF, the peak didn't become weaker. It
 > seems UVSUB doesn't dividing the data properly when the data has multi-IF. 
 > e.g., does UVSUB use different UV-coordinates for different IFs? 
 > 
 > The AIPS version I use here is AIPS_VERSION=/home/aips/aips/15APR99.  
 > 

      UVSUB does indeed correct for frequency.  Having said that, I
should point out that there have been numerous subtle corrections in
the routines made since your version of AIPS.  It is possible that one
of these corrections would affect your results.

      This is a large shift and it is done all in the image plane
tangent to the sphere at the original reference point.  This is less
than accurate if the shift is large and different frequencies would
emphasize this more.  I do not know if this affect alone is enough to
account for your problems.

      You do not say why you are doing this large shift - is it to
shift to another object or to correct for a bad calibrator position or
??  In the first case, the IFs would have parge phases wrt each other
after the shift, in the 2nd they would likely have large phases wrt
each other before the shift.  Following the shift, you do not say how
you are doing the imaging.  If it is with IMAGR with all IFs,
chhanels, and DO3D true frequency differences would be handled as well
as possible.  If it is by splitting and averaging IFs or other
software systems that would do that, then the phase differences would
cause a loss of amplitude.

That's about all I can say without more info.  Have you considered
UVFIX instead?  That is often used for vlbi data.

Eric Greisen