Imaging is a beast and we're using two different packages to do it. MSClean has failed us so far. Apparently our signal to noise isn't so good. Also, we want a beam that's double the size of our image.
Here are the steps:
- run IMAGR in AIPS.
- cellsize 1.5
- imsize 2720
- robust 5 (or 1.5)
- getn DBCON
- run SUBIM to cut out the piece of the cube and beam you want. AIPS will automatically round 2720 up to the nearest power of 2 (4096). Miriad can't handle files this big. You will have two steps: cut out a piece of the beam and a piece of the cube.
- DEFAULT SUBIM
- blc 1369 1369 0
- trc 2728 2728 0
- getn *ICL001; go subim
- DEFAULT SUBIM
- blc 689 689 <nchannels / 2>
- trc 3408 3408 <nchannels / 2>
- getn *IBM001; go subim
- Write these two files out to disk with FITTP (*.map.fits, *.beam.fits)
- Read the files into miriad using fits
- task fits
- in=*.map.fits
- op=xyin
- out=*.map
- go fits;
- task fits
- in=*.beam.fits
- op=xyin
- out=*.beam
- Measure the noise in a line free channel (kvis, make a box, press s)
- Clean your map lightly with clean
- task clean
- map = *.map
- beam = *.beam
- out = *.clean
- niters = ??
- cutoff = 3 sigma?
- Restore the clean components to the map. You will have to run this a total of three times; once for the clean components map (C), once for the residual map (R), and once for the C+R map.
- task restor
- map = *.map
- beam=*.beam
- model = *.clean
- mode=convol
- out = *.cc.restor
- go restor
- mode = resid
- out = *.resid.restor
- go restor
- unset mode
- out= *.restor
- go restor
- Smooth the *.cc.restor map with a gaussian twice the size of the beam
- task convol
- map=*.cc.restor
- out=*.cc.restor.convol
- fwhm=2*fwhm, 2*fwhm (from header.. just make this a circular beam, use the largest fwhm)
- options=final
- Hanning smooth in velocity space. We should try to find a faster method because this takes a long time.
- task hanning
- in=*.cc.restor.convol
- out=*.cc.restor.convol.hanning
- width=3
- Blank the smooth cc cube
- task maths
- exp=<*.cc.restor.convol.hanninng>
- mask = <*.cc.restor.convol.hanning>.gt.0.00001
- out = *.mask
- Use cgcurs to define clean regions. It will output a file named cgcurs.region
- task cgcurs
- in=*.mask
- type=pix
- device=/xwin
- range=0,0.001
- nxy=1
- options=region
- Rename the cgcurs.region file
- mv cgcurs.region *.region
- mask the mask cube using immask to get rid of any spurious clean components
- task immask
- in=*.mask
- region=@*.region
- logic=and
- flag=true
- Use this mask to re-clean your map. The inputs will be the same as before, except you will now have a mask and we will clean to 1 sigma (probably).
- task clean
- map = *.map
- clean = *.clean
- niters = 50000
- cutoff = 1*sigma
- out = *.mask.clean
- Restore your masked clean components, as before. We will run this three times (C, R, C+R maps).
- task restor
- map = *.map
- beam = *.beam
- model = *.mask.clean
- options = convol
- out = *.mask.cc.restor
- go restor
- options=resid
- out = *.mask. resid.restor
- go restor
- unset options
- out = *.mask.restor
- go restor
- copy the dirty map to new file
- apply the mask to the dirty map and the residual map
- task immask
- in=*.mask.map
- region=mask(*.mask)
- flag=true
- logic = and
- go immask
- in = *.mask.resid.restor
- go immask
Now we have a cleaned mask! To find fluxes, we need to use the Jorsater-van Moorsel method.See
DeterminingFluxes.
--
AdrienneStilp - 2009-03-03