Hi,
Me again! Lots of problems/questions today sorry...
I'm trying to run MOSSEN on my mosaics to get gain and sensitivity maps for them. It works fine on mosaics that I've created through the joint deconvolution method, but not on mosaics that I make via the individual method stitched together with LINMOS. That is, I've imaged, cleaned and restored each pointing, then LINMOSed them all together. The resulting mosaic looks fine. When I run MOSSEN on the mosaic and view the resulting gain/sensitivity maps in KVIS, all I see is the calculated gain/sens from the bottom right-most pointing in the mosaic, with the rest being blanked out (exactly zero). I was getting a warning from LINMOS during the mosaicing operation - "WARNING: Setting RMS to 1.0 for each image". Could this problem be related to that warning perhaps? Any help much appreciated.
Cheers,
Craig.
Running MOSSEN on LINMOSed mosaics
Moderator: Mark.Wieringa
-
- ATCA Expert
- Posts: 297
- Joined: Mon Feb 08, 2010 1:37 pm
Re: Running MOSSEN on LINMOSed mosaics
Hi Craig,
when you use invert to make a mosaic, it stores a mosaic table with the details of each pointing that went into the image (position, theoretical rms noise etc). Mossen uses this table to calculate the theoretical sensitivity image (or gain image).
When you use linmos this mosaic table doesn't get filled so mossen can't do its job, however linmos itself has an option to calculate the sensitivity pattern from all the input images. In your case this would not work either because you don't seem to have the rms values in your individual image headers.
To find out why you don't have the rms values you need to check the processing steps between invert and linmos. Do a prthd on every intermediate image and see where the rms value disappears.
Cheers,
Mark
when you use invert to make a mosaic, it stores a mosaic table with the details of each pointing that went into the image (position, theoretical rms noise etc). Mossen uses this table to calculate the theoretical sensitivity image (or gain image).
When you use linmos this mosaic table doesn't get filled so mossen can't do its job, however linmos itself has an option to calculate the sensitivity pattern from all the input images. In your case this would not work either because you don't seem to have the rms values in your individual image headers.
To find out why you don't have the rms values you need to check the processing steps between invert and linmos. Do a prthd on every intermediate image and see where the rms value disappears.
Cheers,
Mark
Re: Running MOSSEN on LINMOSed mosaics
Thanks Mark. I assume that the relevant parameter in the header is "Nominal Theoretical Rms:". If so, then using your suggestion, it seems that the RMS in dropping out of the header when I convolve the images to a common resolution using task CONVOL. I'm sticking my CLEANed and RESTORed image in, and using the parameters 'fwhm=85,35 pa=0 options=final' to generate a convolved image. Can you see how this might make be causing my problems? It's not immediately obvious looking at the task help file how to fix this. Is there a way to fix it that you can see?
Cheers,
Craig.
Cheers,
Craig.
-
- ATCA Expert
- Posts: 297
- Joined: Mon Feb 08, 2010 1:37 pm
Re: Running MOSSEN on LINMOSed mosaics
Hi Craig,
I guess convol does the right thing, since it can't really know what the theoretical noise is going to be after convolution.
You could either specify the noise for each image using the rms parameter in linmos, or do the convolution after the linmos step.
Ideally you'd get invert to generate the images at about the same resolution using its fwhm parameter and skip the convol step, but I don't know if that works in your case.
Cheers,
Mark
I guess convol does the right thing, since it can't really know what the theoretical noise is going to be after convolution.
You could either specify the noise for each image using the rms parameter in linmos, or do the convolution after the linmos step.
Ideally you'd get invert to generate the images at about the same resolution using its fwhm parameter and skip the convol step, but I don't know if that works in your case.
Cheers,
Mark