[daip] Regarding an issue in using 'spectral index' image with AIPS task IMAGR {External}

[egreisen]

On 2023-05-22 05:42, Subhashis Roy wrote:
> Hi Eric,
> 
> IMAGRPRM(17) in IMAGR is a very useful option for imaging wideband
> data, for which spectral index image has already been made by imaging
> from each sub-bands.


Thank you for such a detailed query.  I have been playing with various 
things
for the past 2 days and this is what I have  learned so far.

I think the spectral index code is doing what it should but it requires 
a
good spectral index image.  I made one with IMMOD and then things worked
better with than without the IN3NAME.  Not a lot however.  The imaging 
of
the southern point source works rather better than the northern when
doing a channel at a time or all together.  Some of this is due to 240
not being an exact pixel, but I have worked with a UVMOD data set that 
used
an exact pixel instead.  Some of your issue comes from the extra central
source which means that at the center the model of a single spectral 
index
is not true.  Curvature would be needed to describe that area.  I 
removed
that source from my models to avoid this question.

You know how to run IMAGR better than I do - I rarely use all the 
options
that you invoked.  Running IMAGR without taper and SDI and ROBUST=0, I 
found
when I imaged each channel to make a cube, that the apparent spectral 
index
of the central pixel of the southern source was +1 although when ISPEC 
took
the total flux over the source the net spectral index was as required 
-0.9.
AIPS cubes require that all planes have the same units (Jy per header 
beam)
and keep a CG table of the actual Clean beam parameters.  Right now 
SPIXR
gets very strange answers from such a cube and I will need to think 
about what
to do with it.  I will study the code some more to make sure the 
routines that
compute the multiple CC files do the right thing, but I suspect they do.
This wide a bandwidth does cause issues although the CASA people tell me 
that
they do even wider bandwidths.  They do store their cubes with different 
units
in each plane!

Eric Greisen