Check cases where there appears to be a vertical pattern of not-quite-connected masked pixels that might indicate a hot column that isn't quite triggering the automatic masking (credit @schlafly for noticing), e.g. from https://data.desi.lbl.gov/desi/spectro/redux/daily/nightqa/20240330/nightqa-20240330.html (DARK pulldown) -> z7 (lower left) and r3 (lower middle):
If these appear on multiple nights with different pixels they might disappear from median masks and might need a by-hand update to force a mask. There is some #survey-ops Slack discussion where @Waelthus inspected individual preproc files, but I'm transferring the issue here for tracking longer term.
Alternately, it could be useful to identify which fibers overlap these potentially problematic columns to see if they are causing any real problems in redshift distributions.
Check cases where there appears to be a vertical pattern of not-quite-connected masked pixels that might indicate a hot column that isn't quite triggering the automatic masking (credit @schlafly for noticing), e.g. from https://data.desi.lbl.gov/desi/spectro/redux/daily/nightqa/20240330/nightqa-20240330.html (DARK pulldown) -> z7 (lower left) and r3 (lower middle):
If these appear on multiple nights with different pixels they might disappear from median masks and might need a by-hand update to force a mask. There is some #survey-ops Slack discussion where @Waelthus inspected individual preproc files, but I'm transferring the issue here for tracking longer term.
Alternately, it could be useful to identify which fibers overlap these potentially problematic columns to see if they are causing any real problems in redshift distributions.