Page 1 of 1

options=mfs in restor

Posted: Mon Oct 28, 2013 7:12 pm
by minh25
Dear Mark, Jamie, et al.,

I'm doing some imaging of CABB 16cm data, splitting data into 256 MHz chunks to make things easier.

I'm wondering if anybody can tell me what options=mfs does in restor. I've never used that option before, but since I use mfclean and options=mfs in invert and selfcal I guess I should/must use it?

The puzzling thing is restor with options=mfs is giving me many negative pixels (some at highly negative numbers), while restor with options unset doesn't.

I've attached a kvis figure to show what I mean. The image to the left has options=mfs, and the one on the right does not. options=mfs results in pixels as negative as -300 mJy, while leaving options unset results in -4 mJy as the most highly negative pixel. The image rms seems almost identical if I look at some boxes away from bright sources. So other than the negative pixels I think the image quality is the same.

The helpfile says :" If the input model was produced by mfclean and contains a 2nd plane with alpha*I components, using "mfs" will cause restor to write a second plane in the output image containing the alpha*I model convolved with the Gaussian beam. Linmos will use this plane to do wideband primary beam correction if the bw parameter is used." So I think I want to use this, for correct linmos PB correction but shouldn't it not affect the first plane then?

I've included below my miriad imaging commands in case that illuminates anything.

thanks,
Minh

... previous invert/clean/selfcal cycles then
selfcal vis=pksX.1972 model=pksX.icmp4 clip=2.5e-3 interval=5 options=mfs,amp refant=4 nfbin=4
invert vis=pksX.1972 map=pksX.imap5 beam=pksX.ibeam5 imsize=7000 cell=0.5 robust=-0.5 stokes=i options=mfs,systemp,double,sdb
mfclean map=pksX.imap5 beam=pksX.ibeam5 out=pksX.icmp5a niters=2000 region=relcen,box\(-3200,-3200,3200,3200\)
restor model=pksX.icmp5a map=pksX.imap5 beam=pksX.ibeam5 out=pksX.icln5a options=mfs

Re: options=mfs in restor

Posted: Tue Oct 29, 2013 8:20 am
by Mark.Wieringa
Hi Minh,

The pixel histograms in kvis use data from both planes in the restor image, so the negative values are most likely coming from the off-centre strong source. It will have a negative spectral index due to the fact it sits on the slope of the primary beam. If its apparent spectral index is less than -1 it will cause pixels with negative values 'greater' than the peak of the source in the I image.

So I think all is well, I was surprised by this too when I first loaded a restor mfs image into kvis.
One a related note: I've recently updated linmos and mfspin so linmos can create a mosaiced I*alpha plane and mfspin can use this to produce the spectral index mosaic.

Cheers,

Mark