search

legacysurvey / legacypipe

Image reduction pipeline for the DESI Legacy Imaging Surveys, using the Tractor framework
BSD 3-Clause "New" or "Revised" License
37 stars 22 forks source link

Remove bad Bok exposures from 2017-06-03 #381

Closed djschlegel closed 5 years ago

djschlegel commented 5 years ago

It looks like something is going on with the night of C20170603 for the first 48 exposures, between + including these: ksb_170604_053203_ooi_r_v1.fits.fz ksb_170604_070038_ooi_r_v1.fits.fz HDU #2 in all these image looks like the attached image. The following frames for the rest of the night look fine.

By luck, most of these are excluded from DR8, except for these 3: FILENAME RA DEC 90prime/BOK_CP/CP20170603/ksb_170604_053203_ooi_r_v1.fits.fz 249.4851 60.26233 90prime/BOK_CP/CP20170603/ksb_170604_053650_ooi_r_v1.fits.fz 251.4904 60.26239 90prime/BOK_CP/CP20170603/ksb_170604_054138_ooi_r_v1.fits.fz 251.69 59.26239 The other bad ones all have CCD_CUTS=16384 (==depth_cut).

What do we do? Let’s add these 48 exposures to the Bok bad exposure list. Then we could decide whether to re-reduce the few bricks these touch.

Not obvious to me what would have caused this. There’s a lot of moon that night (76%), though the whole night was pointed very far north (dec=+50 to +70) with the moon at dec= -3. Maybe a light accidentally in the dome?

Here is the list of 67 bricks (from DR6) that included the above 3 exposures. some of which are already processed in DR8: 2483p597 2483p600 2483p607 2484p602 2485p605 2487p597 2488p600 2488p607 2489p602 2490p605 2492p597 2493p600 2494p602 2494p607 2495p605 2497p597 2498p600 2499p602 2499p607 2500p605 2502p597 2503p600 2504p602 2504p607 2505p605 2507p597 2502p597 2503p600 2504p602 2504p607 2505p605 2507p597 2508p600 2509p602 2509p607 2510p605 2512p597 2513p600 2514p602 2514p607 2515p605 2517p597 2518p600 2519p602 2519p607 2520p605 2522p597 2523p600 2524p602 2524p607 2525p605 2527p597 2503p595 2506p590 2507p592 2507p597 2508p587 2508p595 2511p590 2512p587 2512p592 2512p597 2513p595 2516p590 2517p587 2517p592 2517p597 2518p595 2521p590 2522p587 2522p592 2522p597 2523p595 2526p590 2527p587 2527p592 2527p597 2528p595

Screen Shot 2019-05-06 at 9 02 12 PM
djschlegel commented 5 years ago

Frank points out that these effected exposures were due to a moth in the evening flats that were used for only those first 48 exposures (the later exposures used a different flat). The problematic CP frames all have: FLATFIL = 'BASS_20170603_8ebcf80-ksb170604023022rF[ccd1]' / Dome flat
The raw frames are fine, and could be reprocessed with the unaffected morning flats.

djschlegel commented 5 years ago

Frank re-processed 124 of the exposures with a good flat-field, replacing these in the NOAO archive and at NERSC in the project directory. These still have the v1 name. They will need all new calibs run for DR8. Then re-run the 2089 bricks touching the 124 updated exposures. Lists of exposures and bricks attached below. filelist.txt bricklist.txt

djschlegel commented 5 years ago

Eddie has updated the ccds file for the Bok on 6 June 2019: /global/homes/s/schlafly/legacysurvey/ccds/new-survey-ccds-90prime-updated.fits which can be copied into the two working directory locations: /global/cscratch1/sd/landriau/dr8/survey-ccds-90prime-dr8.fits.gz /global/project/projectdirs/cosmo/work/legacysurvey/dr8/survey-ccds-90prime-dr8.fits.gz We also need to generate the kd versions of the above files before re-processing these bricks.

djschlegel commented 5 years ago

This was completed for DR8.