Hello everyone,
I'm reducing CABB data at 16cm and when I try to image my source or secondary, after using mfclean to clean my image, I still can see the beam pattern on the image and no matter I chnge the number of iterations or box size in region the beam shape is visible.
Is there any othet cleaning method to use with CABB?
Is there anything I can check to underestand the problem?
Mfclean problem in CABB
Moderator: Mark.Wieringa
Mfclean problem in CABB
- Attachments
-
- Beam shape
- Red Hat Enterprise Linux 5 (3)-2012-02-27-13-09-31.png (135.46 KiB) Viewed 30339 times
-
- Final image after using mfclean
- Red Hat Enterprise Linux 5 (3)-2012-02-27-13-08-32.png (152.01 KiB) Viewed 30339 times
-
ste616
- Site Admin
- Posts: 220
- Joined: Thu Feb 04, 2010 3:27 pm
- Location: Paul Wild Observatory Narrabri NSW
Re: Mfclean problem in CABB
Hi Sara,
Before I answer your question, I should note first that as Miriad stands currently, it is not possible to get good results when imaging the entire 16cm band. To my knowledge, there is no procedure to handle how the beam changes from 1.1 GHz to 3.1 GHz while cleaning a single MFS image - in any package, let alone Miriad.
To answer your question: the synthesised beam changes quite a lot at 16cm even over relatively modest bandwidths since the fraction bandwidth is higher at lower frequencies. In regular image-based clean, the synthesised beam is subtracted from the image at the location of the sources, assuming that the spectral index of the source is flat. If the spectral index of the source isn't flat, then the synthesised beam will not be an accurate representation of the telescope response. Using mfclean (after making the spectral dirty beam in invert) tries to use a measure of the source's spectral index to more accurately estimate the telescope response to the source. This still only really works though over small-ish fractional bandwidths because the changing size of the synthesised beam is not accounted for here.
I suspect that you are trying to image too wide a bandwidth, and so the synthesised dirty beam won't work over the entire range of emission you have. Please tell me if I'm wrong though! I suggest limiting (to begin with) your image's bandwidth to just 128 MHz and see if this helps at all. You can then try to expand your bandwidth if you want.
You may also want to try using uv-based cleaning, using CASApy's clean task. This will still not be able to cope with large fractional bandwidths though (at least I don't think it can handle it yet).
Before I answer your question, I should note first that as Miriad stands currently, it is not possible to get good results when imaging the entire 16cm band. To my knowledge, there is no procedure to handle how the beam changes from 1.1 GHz to 3.1 GHz while cleaning a single MFS image - in any package, let alone Miriad.
To answer your question: the synthesised beam changes quite a lot at 16cm even over relatively modest bandwidths since the fraction bandwidth is higher at lower frequencies. In regular image-based clean, the synthesised beam is subtracted from the image at the location of the sources, assuming that the spectral index of the source is flat. If the spectral index of the source isn't flat, then the synthesised beam will not be an accurate representation of the telescope response. Using mfclean (after making the spectral dirty beam in invert) tries to use a measure of the source's spectral index to more accurately estimate the telescope response to the source. This still only really works though over small-ish fractional bandwidths because the changing size of the synthesised beam is not accounted for here.
I suspect that you are trying to image too wide a bandwidth, and so the synthesised dirty beam won't work over the entire range of emission you have. Please tell me if I'm wrong though! I suggest limiting (to begin with) your image's bandwidth to just 128 MHz and see if this helps at all. You can then try to expand your bandwidth if you want.
You may also want to try using uv-based cleaning, using CASApy's clean task. This will still not be able to cope with large fractional bandwidths though (at least I don't think it can handle it yet).
cheers
Jamie Stevens
ATCA Senior System Scientist
Jamie Stevens
ATCA Senior System Scientist
Re: Mfclean problem in CABB
Hi Jamie,
Thanks for your answer. I've been trying to image the whole bandwith(2 GHz) using mfclean. I tried to redo imaging for 128 MHz and it seems it enhanced alot but it didnt clean one of the strong ponit sources in the field. I used CASA and the same problem appeared. I'm going to image in chunks of 500 MHz .
Thanks,
Sara
Thanks for your answer. I've been trying to image the whole bandwith(2 GHz) using mfclean. I tried to redo imaging for 128 MHz and it seems it enhanced alot but it didnt clean one of the strong ponit sources in the field. I used CASA and the same problem appeared. I'm going to image in chunks of 500 MHz .
Thanks,
Sara