search

desihub / desispec

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

Skyline residuals stronger and more frequent in ELG tile 80610 #1170

Open vruhlmannkleider opened 3 years ago

vruhlmannkleider commented 3 years ago

Compared to the first two deep ELG tiles, there is a trend for ELG spectra on tile 80610 to show more frequently strong skyline residuals, generally above 9300AA but in some cases already from 8300AA. All 50 files I’ve VIed have strong residuals in a large proportion of their spectra (roughly speaking half to 2/3). Hereafter are a few examples with different effects from the residuals (no effect on doublet up to fake doublet created by residuals). All plots of full spectra below are with a smoothing of 4 while all zooms are with no smearing. The difference with the other two tiles is not so much on the impact of residuals than on their strength and frequency, both significantly higher. Hope this helps Vanina

 File         Target id     Spectrum    comment 
 27     39627835555447513     32     doublet OK but strong residuals above 8300AA 
 58     39627865754440155     41     doublet OK but strong skyline residuals above 8600AA 
 46     39627853674841231     10     fake Hbeta and OIII lines from strong skyline residuals (present above 9300AA)
 10     39627817448639880      1     fake negative doublet with Dchi2=198 from skyline residuals (above 9300AA)
 27     39627835551255388      9     fake doublet with Dchi2=192 from skyline residuals (above 8300AA)
 13     39627823467467680     11     unsecure doublet in region of skyline residuals (above 8300AA)
 19     39627829503070197     16     unsecure doublet in region of skyline residuals (above 9300AA)
 54     39627859731418084     32     redshift biased by strong skyline residuals (above 8300AA) 
 58     39627865754441167     46     OIII-OII mismatch due to doublet distortion skyline residual (above 8600AA)
Capture d’écran 2021-02-26 à 11 32 31 Capture d’écran 2021-03-05 à 16 04 51 Capture d’écran 2021-03-03 à 12 32 30 Capture d’écran 2021-02-22 à 15 45 15 Capture d’écran 2021-02-26 à 11 12 31 Capture d’écran 2021-02-26 à 11 10 56 Capture d’écran 2021-02-23 à 11 12 16 Capture d’écran 2021-02-23 à 11 13 11 Capture d’écran 2021-02-24 à 16 48 57 Capture d’écran 2021-02-24 à 16 48 37 Capture d’écran 2021-03-04 à 16 05 23 Capture d’écran 2021-03-04 à 16 05 42 Capture d’écran 2021-03-04 à 16 05 34 Capture d’écran 2021-03-05 à 16 13 19 Capture d’écran 2021-03-05 à 11 21 58 Capture d’écran 2021-03-05 à 16 16 32
julienguy commented 3 years ago

Thank you Vanina, this is very useful. We can use this set of spectra to monitor future improvements on sky subtraction. Below are comparisons of the same spectra between the 'blanc' and 'cascades' release. It's better for most spectra but not for all. There is still some progress to make on this. green='blanc' :-) brown='cascades' 39627865754441167 39627865754440155 39627859731418084 39627853674841231 39627835555447513 39627835551255388 39627829503070197 39627823467467680 39627817448639880

How to produce these figures:

plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627835555447513
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627865754440155
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627853674841231
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627817448639880
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627835551255388
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627823467467680
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627829503070197
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627859731418084
plot_spectra -i $DESI_SPECTRO_REDUX/{blanc,cascades}/tiles/80610/deep/coadd-*.fits -t 39627865754441167
vruhlmannkleider commented 3 years ago

Thanks for the fast feedback, Julien. The improvement with Cascades is already great. If you need an enlarged control sample, please feel free to ask and I'll fetch some more (the tile is very rich of such spectra...).