julienguy
commented
2 years ago Fiber 1098 has an incorrect throughput correction term for the sky. The value is 0.67 for this fiber when one expects ~1.
import fitsio
tpcorr=fitsio.read("/global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/sky-b2-00070352.fits","THRPUTCORR")
print(tpcorr[[97,98,99]])
[0.96382457 0.6705732 1.0109111 ]Enforcing a value of 1 fixes the issue
cp /global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/sky-b2-00070352.fits sky-b2-00070352-bis.fits
import astropy.io.fits as pyfits
h=pyfits.open("sky-b2-00070352-bis.fits")
h["THRPUTCORR"].data[98]=1.
h.writeto("sky-b2-00070352-bis.fits",overwrite=True)
desi_process_exposure --infile /global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/frame-b2-00070352.fits --fiberflat /global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/fiberflatexp-b2-00070352.fits --sky sky-b2-00070352-bis.fits --calib /global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/fluxcalib-b2-00070352.fits --outfile cframe-b2-00070352-bis.fits --cosmics-nsig 6
plot_frame -i cframe-b2-00070352-bis.fits /global/cfs/cdirs/desi/spectro/redux/fuji/exposures/20210101/00070352/cframe-?2-*.fits --fiber 1098 --legend --rebin 8
This is a rare occurrence of an effect that will be fixed soon with the work of @schlafly on the sky subtraction improvements.
b-channel spectrum offset bright relative to the other channels.
akim@cori05:/global/project/projectdirs/desi/spectro/redux/fuji/tiles/pernight/80661/20210101> plot_spectra -i coadd*fits -t 39627757461701029