[daip] Assorted questions

[Eric Greisen]

Ian McHardy writes:

 > We're doing some wide field, deep, mapping at 20cm (4IF stuff). This is
 > causing a variety of rather awful problems, but also some more
 > basic probs, but for which I can't see the answer so I'll ask you here
 > if I may.

        I gather you have communicated with Frazer Owen on some of
this.  If AIPS tasks appear not to work, I do stand ready to try to
help.

 > 
 > We have some B-array (8hrs) maps and some A-array (24hrs) maps. Of course we 
 > can add the uvdata and remap but we are curious to see what happens when
 > we simply add in the map plane. That is a quick way of doing it (and as
 > we are still deeply involved in removing sidelobe noise from the A-array 
 > data, we'd  like to try the quick way).

        I am not sure exactly what you expect to achieve by this addition.
 > 
 > So how do we add the maps?
 > 
 > I assume that we need to first smooth the A-array to B-array resolution.
 > We found a task `smoth' - of 1984 VAX vintage. It doesn't work properly.
 > The noise rises linearly with the beamsize.

       If you want I can look at the failure of SMOTH if you provide
details.  The usual way to smooth an image is with task CONVL.  Note
that when you smooth an image, you down-weight all the long spacings
and so actually increase the noise.  You raise the signal of extended
sources but dilute the signal of point sources.

       Frankly, I would run OHGEO on the images to make a common grid,
run the RESCALE verb to put them on the same Jy/pixel scale and then
just add them without smoothing pehaps with different weights.  COMB
does care that the images are aligned since it just adds the arrays.

 > 
 > Assuming there is a smoothing task, what next? COMB is very picky if
 > maps aren't of exactly the same area so Hgeom (or ohgeo)?
 > I've been having difficulty getting that to work. Any hints?

       Again, what difficulties?

The 31DEC00 version of AIPS - already essentially frozen and soon to
be available for binary downloads and CDroms - has a much improved
IMAGR.  I would use that version to image the combined A and B data -
Frazer can advise on the weighting-control adverbs - and not waste
further time imaging the data sets separately.  The new tasks SETFC
will even prepare a BOXFILE of field locations - and recommend the
right cell and image sizes - to cover the primary beam (or any
circular area you define) and even to cover the Sun and NVSS sources
nearby if they are strong enough.  The more data you bring to bear on
the Clean modeling the better constrained it is.

Anyway, that's my two cents..

Eric Greisen