Which linmos PB correction is right for wide-band data?

Got an image problem? Let us help!

Moderator: Mark.Wieringa

Post Reply
abutler
Posts: 11
Joined: Fri Feb 06, 2015 1:28 pm

Which linmos PB correction is right for wide-band data?

Post by abutler »

Hi all,

I'm trying to understand the two different ways in which linmos performs the primary beam correction. Under the "bw" keyword in the linmos help file, it says:

"Bandwidth of the image in GHz, default 0. If specified the beam response will either be averaged across the frequency band before being applied to the image or, if the input images contain a spectral index plane (created with the mfs option of restor) the images will be evaluated and corrected across the band. Use this for wide band images to improve the accuracy of the correction."

Which correction should I use? I am trying to make a full-band mosaic using the full-band reduction of 471 pointings (which have spectral index planes) imaged at 1-3 GHz (2.0 GHz bandwidth, so delta_nu/nu~1).

Could you please briefly explain the difference between these two different ways that linmos does the primary beam correction and give a recommendation?

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

Re: Which linmos PB correction is right for wide-band data?

Post by Mark.Wieringa »

Hi Andrew,

Imaging the entire band from 1-3 GHz in a single mfs image is not recommended. The factor 3 change in primary beam size (and usable field size) is too large for Miriad's first order correction scheme and the results won't be very accurate. Generally you should try to keep the relative bandwidth to less than about 40%. CASA may let you go a bit wider than this with higher order terms, but not still not to 100%.

Linmos will pick between the two options (average beam response or evaluate images across bandwidth) based on the absence or presence of the spectral index plane as long as you set the bw parameter >0.
The first option takes the image as is and divides by a beam calculated as the average over 10 frequencies across the bandwidth. The second option uses the spectral index plane to compute a series of images across the bandwidth and divides each one by the corresponding beam at that frequency.

So my recommendation would be to divide your bandwidth up into 3 bands, image those with mfs and mosaic them separately (using restor with mfs option and linmos with bw>0) and then analyse the 3 mosaic images to determine the source flux-densities at 3 frequencies and the spectral parameters.

Cheers,

Mark
Post Reply