AARTFAAC QC tests

[hsuyeep]

This is a placeholder list of possible QC tests for AARTFAAC images. These might be different from those applicable to LOFAR images due to several reasons:

• Image noise dependent on LST of observation (location of galactic plane on the sky)
• Large variation of noise over the FoV.
• Lack of deconvolution in AARTFAAC images

From the AARTFAAC meeting of 29Jul15, the following tests were suggested:

• The mean fractional flux offsets and position offsets for the A-team sources wrt. catalog positions and fluxes. This is possible as the real-time calibration and imaging pipeline carries out an estimation of these parameters before the source finder (Creating a separate issue for its implementation).
• Calculate rms of a few fields around the A-team sources Vs. rms in empty sky, to detect residuals after subtraction of A-team sources.
• Calculate the rms of a piece of sky that is circumpolar and always available (e.g. NCP), and assign that rms value to the whole image. This rms might then be affected only by calibration errors and possibly not by A-team subtraction issues.

[ycendes]

Sooooo, fun stuff, I've been working on this issue today and have made some progress. I now have a script that can take an AARTFAAC imag from the directory @hsuyeep put on struis (that is a few minutes long) and spit out the ones with an RMS value that's greater than a given threshold. (If you look at the RMS of all the images first, most are normal and will have an RMS in the 1-2 range- whatever that means without a flux scale- so after toying around with it I arrived at the very scientific cutoff of saying all images > 2.) And lo and behold, every single one of these images that gets put on the list has a strong RFI source in the upper right hand quadrant of the image (not sure but I think this means it's an RFI source in the northeast?), which you can see in the attached image. ALL images with an RMS > 2 have this source!

I will add the script for this to the AARTFAAC tools in the next few days, there's still a (hopefully minor) bug that needs to be squashed before I do so to make sure we are getting all the images. But hey, it's a nice start!

[ycendes]

Script is now debugged, and is available here: https://github.com/transientskp/aartfaac-tools/blob/master/rms-check.py