Clean

[shf]

Hi
I am very new in this forum, therefore please direct me to the right address if I am posting my question to the wrong place.
I have an ATCA observation (both dirty map and dirty beam) and try to clean it. I use the following procedure to clean the map.
After importing the fits files. I performed a convulsion with fwhm=2xbeam size.
convol map=atca out=atca_conv fwhm=150 //the original beam size is 77 arcsec
I use the output of this task to create a mask of the region which should be cleaned. I used the following task to create the mask.
maths exp=atca_conv mask="atca_conv.gt.0.56" out=mask0 //create the first mask-0.56 is the map contour level I determined.
I made a moment0 of the dirty beam cube and normalized it. Then I ran the following task
clean map=atca beam=atca_beam0_normalized out=mod0 niters=1000000 cutoff=70E-03 region="mask(mask0)"
I would like to clean the map and create a model (in this case mod0). Restore it with the help of
restor map=atca beam=atca_beam0_normalized model=mod0 out=line.res0
And repeat the procedure.
The problem is, after I perform the
clean map=atca beam=atca_beam0_normalized out=mod0 niters=1000000 cutoff=70E-03 region="mask(mask0)"
a fatal error occurs "the region to deconvolve is too big".
However, if I do not define any region it starts to clean. So I think that the mask is the problem.
I resized the original atca cube so that to original cube is embedded in the center of a bigger cube (4x bigger and zeropadded)
The cubes that I am using are as following:
1.4GHz
ATCA-original:
'restoringbeam': {'major': {'unit': 'arcsec',
'value': 76.999998092760009},
'minor': {'unit': 'arcsec',
'value': 76.999998092760009},
'positionangle': {'unit': 'deg',
'value': 0.0}}},
'axisunits': array(['rad', 'rad', 'Hz']
'refpix': array([ 341., 320., 0.])
'shape': array([578, 610, 78]
'unit': 'Jy/beam'

ATCA-resized:
'restoringbeam': {'major': {'unit': 'arcsec',
'value': 76.999998092760009},
'minor': {'unit': 'arcsec',
'value': 76.999998092760009},
'positionangle': {'unit': 'deg',
'value': 0.0}}}
axisunits': array(['rad', 'rad', 'Hz']
'refpix': array([ 1364., 2440., 0.])
'shape': array([2312, 2440, 78], dtype=int32)
'unit': 'Jy/beam'

ATCA_beam0_normalized:
'axisunits': array(['rad', 'rad'],
'refpix': array([ 114., 114.]),
'refval': array([ 0.23998277, -1.27118147]),
'shape': array([229, 229, 1], dtype=int32),
'unit': 'Jy/beam.km/s'},
I appreciate any suggestion since it is very important for my work to get this done.

Thanks a lot
Shahram

[Mark.Wieringa]

Hi Shahram,

Clean likes to have a beam 4x bigger than the image area to clean. This is so it can correctly subtract a sidelobe from a source at the edge of the clean region.
You should either restrict yourself to a smaller area of the image, or make a larger beam, or failing that, you could try options=pad. This latter option will not clean things correctly but it may give an acceptable result if there is no strong emission near the edge of the clean area.

Cheers,

Mark

[shf]

Hi Mark,

thank you very much for your answer. However, I have another question regarding your answer. Unfortunately I do not have the visibility data of the observation that's why I have chosen the above approach to clean the data. Could you please explain to me if there is any reasonable way to enlarge the dirty beam (cube) without having the visibility data. Thanks again for your help.

Cheers
Shahram

[Mark.Wieringa]

Hi Shahram,

the 'pad' option in clean should allow you to clean nearly the whole image with a beam padded with zeroes to make it twice as big.
I haven't tried this myself, so let us know if you run into problems.

Cheers,

Mark