MFCLEAN problem

Is MIRIAD being a pain? Let us know your experience.

Moderator: Mark.Wieringa

Post Reply
AdamS
Posts: 1
Joined: Mon Mar 26, 2012 9:36 am

MFCLEAN problem

Post by AdamS »

Hi everyone,

I am attempting to produce an image from 16 cm CABB data (in 128 MHz wide chunks) obtained from the ATCA in late January with the array in the 6A configuration.
My problem arises (I think!) when I am using mfclean to deconvolve the beam pattern from my image. Mfclean appears to work well on all point sources in my field except for the two brightest sources. The effect is somewhat reduced by performing a deeper clean but the problem persists even when I clean down below the noise level (which introduces other issues). I have been defining small boxes around each source for cleaning. Defining these boxes works well if I perform the clean for each source individually but not when I try to clean all the sources at once. I have tried cleaning the two bright sources individually and then feeding the resulting cleanmap into mfclean as a model, however this seems to give me a lot of spurious extended emission and negative flux around my point sources.

To form my image I use the following tasks and parameters:
Invert:
vis=n641_25.1908
map=n641_25.1908.map
beam=n641_25.1908.beam
imsize=4096
cell=3
robust=3
stokes=i
options=systemp,sdb,mfs,double
slop=1
With all other parameters left unset.
==>Theoretical RMS noise is about 2.5E-4

Mfclean:
map= n641_25.1908.map
beam= n641_25.1908.beam
out=n641_25.1908.mfclean
gain=0.1
cutoff=0.0008
niters=5000
region=@./cleanbox (This is just a file with the boxes that I want cleaned)
Again, with all other parameters unset

This is then restored.
Image

Moreover, the background noise in my image is plagued by a series of fairly regular stripes at various angles. I have been told that these stripes are probably due to RFI but even a dangerously high amount of flagging (over 50% of the visibility data) does not help. I have used BLFLAG, PGFLAG and MIRFLAG (with both amp and rms based flagging).

Has anyone else come across any problems like these before, and if so how did you fix them?

Many thanks!
Adam
Mark.Wieringa
ATCA Expert
Posts: 297
Joined: Mon Feb 08, 2010 1:37 pm

Re: MFCLEAN problem

Post by Mark.Wieringa »

Hi Adam,

a few questions and suggestions:

- Is this a 'cuts' mode observation? I.e. many short integrations spread over 12h. This will result in images with limited dynamic range (100:1) and often shows radial stripes around strong sources.

- Could it be a calibration problem? Have you tried selfcal?

- Have you tried removing the two strongest sources from the data before imaging? You can use uvmodel or uvfit to subtract point sources from the uvdata. You may need to take spectral index into account.

Cheers,

Mark
saracmbr
Posts: 38
Joined: Fri Jan 20, 2012 3:40 pm

Re: MFCLEAN problem

Post by saracmbr »

Hi Adam,

I had the same problem. I had a 250 mJy source in the field (I had this problem with other less bright sources) which made horrible stripes in my image. There was not any problem with calibrator since the image of the calibrator was nice without any RFI or phase error. I tried to remove the source but because I had to give the clean component as a model it was contaminated from stripes so it did not work well. We had that observation in "cuts-mode" and it was in 6k configuration. I tried offset in "invert". I turned to be better but the actual issue still persists. Even I made a very big image imsize=13000 and cell=1 to get MFCLEAN to clean whole the ara of the interest but this issue still exist. But when I used option fwhm=30,40 it worked well but I lost the resolution.

Good luck,
Sara
Post Reply