search

desihub / desispec

DESI spectral pipeline
BSD 3-Clause "New" or "Revised" License
33 stars 24 forks source link

Check z7=sm8z BADCOLUMNFIBERS; maybe too many masked #2288

Open sbailey opened 5 days ago

sbailey commented 5 days ago

Just before Jura, @julienguy and @Waelthus updated the z7=sm8z PIXMASK since it appeared that the previous mask was masking columns that were no longer bad in newer data. However, the BADCOLUMNFIBERS list is still quite long and is possibly extra-masking fibers that could now be fine:

BADCOLUMNFIBERS: 3615,3616,3696,3697,3698,3699,3700,3701,3702,3750,3751,3752,3753,3756,3757,3802,3803,3804,3805,3811,3812,3813,3817,3818,3819,3820,3821

Check that and update list if needed. Thanks to @schlafly for noticing that we are still masking lots of fibers despite updating the pixmask.

sbailey commented 4 days ago

I checked the latest mask (ccd/pixmask-sm8-z7-20240324.fits.gz) and some recent darks (preproc/20240325/00232302/preproc-z7-00232302.fits.gz, preproc/20240701/00240946/preproc-z7-00240946.fits.gz) and it appears that all of these BADCOLUMNFIBERS are now good.

At the time of ccd/pixmask-sm8-z7-20221129.fits.gz, comparing to preproc/20221129/00155639/preproc-z7-00155639.fits.gz, it appears that the mask and the BADCOLUMNFIBERS list are appropriate for that epoch.

However, there are 3 intermediate masks between those two where the situation is more subtle. e.g. inspecting ccd/pixmask-sm8-z7-20230620.fits.gz and preproc/20230620/00186046/preproc-z7-00186046.fits.gz, it looks like fiber 3615,3616 could be restored as good but fibers 3698,3699 should stay bad, and maybe 3696,3697,3700,3701,3702 should stay bad as well. 3750,3751 look ok, but 3752,5753,3756,3757 are polluted by extra-noisy columns that are no longer masked. 3802 might be ok, but 3803-3805 should remain masked. 3811-3813 and 3817-3821 appear to be ok.

For now, I propose to remove the BADCOLUMNFIBERS only for the latest config so that daily processing will no longer mask them, but I wait until after the collaboration meeting when @julienguy and I can review these cases in more detail with more eyes on it. @schlafly heads up.