Wide-band, wide-field imaging

Got an image problem? Let us help!

Moderator: Mark.Wieringa

Post Reply
thomas
Posts: 5
Joined: Thu Apr 07, 2011 10:36 am

Wide-band, wide-field imaging

Post by thomas »

Hi,

I have been thinking about how best to image some wide-band CABB data. I am using 800-MHz of data, from 1350 to 2150 MHz. I have many pointings which I need to combine to form a mosaic. To improve the accuracy of the primary beam correction, I have decided to split the band into 4 sub-bands so that a different primary beam model can be used for each sub-band.

One disadvantage with this is that it is harder to clean the individual sub-bands because of the poorer sensitivity and uv-coverage. I thought a better strategy would be to image the entire 800-MHz band and generate a model using mfclean, then somehow use that model to produce images of the separate sub-bands. I am not sure whether it is possible to this in Miriad.

One thing I tried is this: I imaged the whole 800-GHz band and generated a model using mfclean. I then imaged the sub-bands separately using the model I obtained for the whole band as an initial model of the deconvolved image. I wasn't able to do this as Miriad complained that the reference pixel values differ. I am guessing that this is because the reference frequency of the model differs from that of the dirty map and beam. I didn't think this would matter because I thought that Miriad would be able to generate the model for the required frequency range from the model of the 800-GHz band.

I was wondering what your thoughts are on this and if there is a better way of proceeding rather than just imaging and cleaning each sub-band separately.

Thanks
Tom
Mark.Wieringa
ATCA Expert
Posts: 297
Joined: Mon Feb 08, 2010 1:37 pm

Re: Wide-band, wide-field imaging

Post by Mark.Wieringa »

Hi Tom,

Yes this is turning into a common problem for those 'trying to do the right thing' with their CABB data.
In response to Julie's question in this forum I've got a first pass of linmos working that can use an I*alpha plane to calculate the image at a range of frequencies, correct each one with the appropriate primary beam and add them all together in mosaic fashion. This goes one step further than the bw parameter in the current linmos, which only averages the beam over frequency.
We'll need to do some simulating and checking against the divide and conquer approach to see if this improves things or not. I have now checked in my latest version of restor and linmos, they should appear tomorrow.

I had a look at mfclean - it insists that the coordinates on all three axes line up (or have integer pixel offsets). I could try relaxing this constraint for mfs models - it is not immediately obvious it will do the right thing, but it is worth a try. I'll give this a test first and let you know.

Cheers,

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

Re: Wide-band, wide-field imaging

Post by Mark.Wieringa »

Hi Tom,

I had a closer look at mfclean and found that the input model has to be produced by a previous run of mfclean with the same input image parameters. Mfclean doesn't evaluate the model and then subtract it, instead it assumes it is at the image frequency and convolves it with the beam to subtract it off. This can't really be changed.

To subtract a wide band model before making sub-band images you could use uvmodel to subtract the mfs model from the full uv dataset and then split and image (or use the line keyword to select sub-bands in invert). However this doesn't help if the aim is to make a better primary beam correction, because you'd run into the same problem when you try to restore the wide band mfs model (restor just convolves the model with a gaussian and adds it).

The route using restor with options=mfs and linmos with bw seems to be the only way you can use the full bandwidth mfs model.
CASA has a task called widebandpbcor which does a similar thing I tried to implement in linmos, it includes higher order terms but only does a single field.

Cheers,

Mark
alfredsoo
Posts: 1
Joined: Tue Apr 07, 2015 11:51 pm

Re: Wide-band, wide-field imaging

Post by alfredsoo »

This is because the wide-band system makes a much better synthesized ... and work is under way to integrate them fully with wide-field imaging techniques.
Post Reply