Theory of restoring
Posted: Thu Feb 09, 2012 11:40 pm
So, I'm now arriving at the imaging phase of my HI imaging project. I have a large set of different configurations. Some are 12 hours long, some only a couple. To add to the fun, there are multiple pointing offsets such that mosaicing is required. My main question is: how can I produced the 'best' restored image. I'm following the steps from the PhD thesis by Jess O'Brien. I'll use the reduction for ESO115-G021 as an example
First a general invert of the image
invert vis=../reductions/*/*.uvl map=first.map beam=first.beam options=mosaic,double robust=0.5 line=velocity,91,350,3.3,3.3 stokes=ii
Followed by a first mosmem run. I only set 5 iterations because I'm only interested in the beamsize from restor.
mosmem map=first.map beam=first.beam out=first.model region=quarter niters=5
Restor command. Not setting a fwhm or PA.
restor map=first.map beam=first.beam model=first.model out=first.restor
Using Gaussian beam fwhm of 18.975 by 10.043 arcsec.
Position angle: -55.5 degrees.
If I follow the way Jess did her reductions I would need to pick the major axis of the beam, so 18.975 arcseconds, and then use that as an input for the next phase of the invert. However, I learned that the restor beam is only based on the first file that goes into invert. In this case a rather short observation in the 1.5B configuration.
My question thus is: What is the recommended strategy in finding a restor beam size? Is there a way to find some optimal value between all these beams? Like an average or so? I'm aiming for the smallest beam possible. So do I have to place my longest 6* observation as the first input for invert?
Cheers,
Stephan
First a general invert of the image
invert vis=../reductions/*/*.uvl map=first.map beam=first.beam options=mosaic,double robust=0.5 line=velocity,91,350,3.3,3.3 stokes=ii
Followed by a first mosmem run. I only set 5 iterations because I'm only interested in the beamsize from restor.
mosmem map=first.map beam=first.beam out=first.model region=quarter niters=5
Restor command. Not setting a fwhm or PA.
restor map=first.map beam=first.beam model=first.model out=first.restor
Using Gaussian beam fwhm of 18.975 by 10.043 arcsec.
Position angle: -55.5 degrees.
If I follow the way Jess did her reductions I would need to pick the major axis of the beam, so 18.975 arcseconds, and then use that as an input for the next phase of the invert. However, I learned that the restor beam is only based on the first file that goes into invert. In this case a rather short observation in the 1.5B configuration.
My question thus is: What is the recommended strategy in finding a restor beam size? Is there a way to find some optimal value between all these beams? Like an average or so? I'm aiming for the smallest beam possible. So do I have to place my longest 6* observation as the first input for invert?
Cheers,
Stephan