Closed kfkaplan closed 5 years ago
@kfkaplan : Is this come from one of the commissioning data? Can you let me know the date and obsID? If not, will you be able to send me the files? I recently found that the PLP does not subtract that pattern as I intended (which is fixed in the development branch). Maybe this is related to that bug.
The above example is from the Sept. run, 20140923 for the combined frame 0209. You can see it for several extended ON-OFF targets on that night.
I am afraid that I don't have an access to September data.
Meanwhile, can you check if the below change make any difference?
https://github.com/leejjoon/plp/commit/895fc1e9bcc765b5991a8273ee1abe2dc13d6599
Yes that solves the problem! Must have been the bug you were referring to.
I have noticed an issue for the combined frames of extended ON-OFF targets that the columns of pixels near bright lines have higher than average counts across the whole detector. This issue does not arise for ABBA combined frames, only for extended ON-OFF combined frames.
To test if the issue was the detectors themselves, or the PLP, I manually created ON-OFF quads but the issue did not occur, so I suspect it has something to do with how the PLP zeros out detector noise for each column of pixels. Whatever algorithm is being used is confused when there is a single bright line across a column on the detector. I have tested using the median of pixel columns to zero out vertical detector noise and that seems to work fine, so that might be the solution.
See the screenshots below: