Any primary beam correction after MOSMEM/PMOSMEM?

[ryanleung]

Hi,

Anyone knows if primary beam corrections have been performed after restoring
from models using MOSMEM/PMOSMEM? My data has 5 full track observations, 3 of them have 8-pointings
while two of them have a single pointing. The resultant mosaic image has a brighter background at the
single-pointing part and falls off quickly beyond the primary beam of the single pointing.

Best Regards,
Ryan Leung

[Mark.Wieringa]

Hi Ryan,

Yes, mosmem (and pmosmem) normally correct for the primary beam. In fact invert should already have done this when making the dirty images. But it uses the same method as linmos with options=taper, which means the images have the correct flux between pointings, but are uncorrected at the edge of the mosaic. I have never checked what happens when one pointing has a lot more data - I would expect the noise to be lower in that area but not the background to be brighter. You could try making separate images and combining them with linmos with and without the taper option to see if you can work out what is happening. If you still think something is going wrong I'm happy to have a closer look at your data.

Cheers,

Mark

[ryanleung]

Hi Mark,

Thanks for your detailed explanation, It takes some time to generate images for linmos. By the way, I have attached my 3 cm image to illustrate the problem, there is a elliptical/circle shape around the 2 full-track data pointing while the upper part has no such edge. 6 cm data show similar edge. Statistics using Funtools are as follows :

reg net_counts error background berror area surf_bri surf_err
---- ------------ --------- ------------ --------- --------- --------- ---------
1 25.718 5.071 0.000 0.000 7357.12 0.003 0.001
2 17.151 4.141 0.000 0.000 7357.44 0.002 0.001

Best Regards,
Ryan Leung

[Mark.Wieringa]

Hi Ryan,

I did a few tests and I think the effect you see might be a mosmem 'feature' that produces a small offset at the level of the noise. I don't see the same effect with mossdi.
If I run mosmem on a mosaic with just noise I get an image like the one below.
I think it shows something very similar to the gain function: close to 1 between mosaic pointings and dropping to 0 at the edge of the mosaic (due to the taper on the primary beam correction).
Adding pointings with longer integration time (extreme left and right pointing in my simulation) seems to make little difference here, but you can see the reduced noise in the original image.
You can reduce the effect by specifying a low (but positive) value for the flux parameter in mosmem.

Cheers,

Mark