[daip] image versus u,v plane stacking

[Andrew Biggs]

Hi Eric. So, would the improvement stem just from the rotation of each
source to the phase centre? In this case, I presume that instead of
dbcon'ing the data together (which for 200 sources takes a long time) one
could repeat the image stacking, but using maps made from each uvfix'd
dataset - is that right? Or might the different u,v coverage of each
rotated source in the dbcon'd dataset produce some improvements as well?

Cheers,

Andy

On 15/07/2013 17:36, "Eric Greisen" <egreisen at nrao.edu> wrote:

>Andrew Biggs wrote:
>> Hi Eric. I have a quick question regarding stacking in radio images.
>>I've
>> done this with my GMRT data and got a detection so it all seems to work.
>> However, I saw a talk by some people in Sweden who claim that "u,v
>> stacking" is superior. By this I think they mean the equivalent of
>>running
>> UVFIX on each source to place them at the phase centre and then
>>combining
>> them into a single dataset (presumably with some hefty time averaging to
>> keep the data size reasonable). They claim that the SNR of their
>>detection
>> doubles, but I don't really understand this as I would have thought that
>> the two techniques would give the same answer. I was wondering if you
>>had
>> any comment on this - before I go to the bother of writing a script to
>>do
>> this in AIPS.
>
>There is already a script that should help - RUN STUFFR and HELP STUFFR
>
>This converts the data times into hour angles, does DBCON, and the
>averages the data in time in a baseline-length dependent mode.  It has
>been used to put together multiple days on the same field primarily.
>
>The image value at the origin is simply the sum of the visibilities
>(assuming no Clean).  Summing images or summing visibilities should be
>exactly equal - at the origin.  I guess I could see how uv stacking
>might help away from the origin.  There are issues - the UBAVG averaging
>depends on the u,v,w of a baseline/source staying similar for a while
>but if you have multiple sources combined that will not be the case.  I
>guess if the sources were separated into separate subarrays, then the
>baseline sort would separate them.  Perhaps you need to "stuff" each
>source separately and then combines them.
>
>Eric
>
>