I'm looking at the Sloan Digital Sky Survey data. I'm primarily using the flux data, from the documentation I see the first extend (HDU 1) has flux
, loglam
, ivar
, and_mask
, or_mask
, etc.
There's been a lot of confusion here about what the or_mask
is doing exactly, and the documentation hasn't been extremely clear.
My understanding so far is that the or_mask
tells me when a particular pixel was flagged in one of many ways possible. I've understood from various random sources that this means the pixel is "bad" somehow. I haven't yet figured out how to know exactly "why" it is bad, my working assumption is that the possibilities are numerous, that there are many fields that have flags that might answer that question.
My primary question: Is every flux value with a non-zero or_mask
inherently untrustworthy?
About 80% of pixels have a non-zero or_mask
value, particularly on the bluer end of the spectrum.
Or is the or_mask
telling me more broadly that there's a flag somewhere, not necessarily even with the flux data, that I might want to consider?
HDU 1 (extname COADD): Coadded Spectrum from spPlate
Binary table with columns:
Required Columns
Name Type Comment
flux float32 coadded calibrated flux [10-17 ergs/s/cm2/Å]
loglam float32 log10(wavelength [Å])
ivar float32 inverse variance of flux
and_mask int32 AND mask
or_mask int32 OR mask
wdisp float32 wavelength dispersion in pixel=dloglam units
sky float32 subtracted sky flux [10-17 ergs/s/cm2/Å]
model float32 pipeline best model fit used for classification and redshift