LSSTDESC
Add missing keywords to raw FITS headers
For the ops-rehearsal-3 activities, a number of keywords describing the observations were noted as missing, specifically the null ones in this list (see the post in slack):
RA: 124.82881462031425
DEC: -14.893919091079661
RASTART: null
DECSTART: null
RAEND: null
DECEND: null
ROTPA: 116.0158594226007
ROTCOORD: sky
HASTART: -10.490392243587635
ELSTART: null
AZSTART: null
AMSTART: 1.0522574903781972
HAEND: -10.36504999786186
ELEND: null
AZEND: null
AMEND: 1.0522574903781972
TRACKSYS: null
RADESYS: null
However, the *END values aren't used by the metadata translator for the LSST pipelines code, so those can be omitted.
A couple other keyword entries that have been noted as missing:
ORIGIN
TELCODE
TELCODE describes the camera, i.e., MC for main camera and CC for ComCam. ORIGIN describes what produced the file, e.g., imSim v2.0. Also, the DAYOBS values probably need a 12 hour offset.
Also noted during ops-rehearsal-3 preparations: the DAYOBS header value should be defined with a 12 hour offset applied to MJD. Details TBC. See https://rubinobs.atlassian.net/browse/DM-43495
Why are they offsetting the DAYOBS by 12 hours? Seems not great to lie about what time it is. But maybe I don't understand something.
I think it's a question of whether the observing day starts at midnight or at Noon.
Isn't MJD an objective time though? I didn't think there was ambiguity in how it was defined.
https://scienceworld.wolfram.com/astronomy/ModifiedJulianDate.html#:~:text=A%20modified%20version%20of%20the,0%20of%20the%20Julian%20calendar.
The proposal is not to change MJD, but to compute DAYOBS from MJD with an offset. I don't think anyone cares about DAYOBS except the Rubin pipeines.
Some documentation on expected header keywords: https://docushare.lsst.org/docushare/dsweb/Get/LSE-400
I think the ones we absolutely need have been added here, but I'd like to do some additional checks against the new obs_lsst code with the LsstCamSim instrument model.