Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: meas_algorithms, obs_decam, obs_subaru
-
Labels:None
-
Story Points:4
-
Epic Link:
-
Sprint:AP S21-3 (February), AP S21-4 (March)
-
Team:Alert Production
-
Urgent?:No
Description
Eli Rykoff brought to my attention that the existing PhotoCalTask, Colorterm, and jointcal code regarding reference catalog filters are all inconsistent in their use of physical/band filter labels. The colorterm dicts and getCorrectedMagnitude code should be changed to explicitly use physicalLabel (colorterms are associated with piece of glass), while refcat loading should be clarified to be band (you want the refcat filter name that is like the band of your science data). Jointcal and PhotoCalTask will likely need to pass around filterLabels as necessary, which will require a bit of reworking of jointcal's data loading mechanism. Cleaning up the docs and variable/method names is part of this work.
As part of this, I will remove the now unnecessary filterMap (physical) and colortermDict (band) entries in obs_subaru and obs_cfht, to reduce confusion on this question.
Eli Rykoff: I think I've go this mostly working except for fgcmcal. _computeOffsetOneBand creates an exposure and does setFilter(afwImage.Filter(band)), which is still using the deprecated interface, and also needs to have a non-empty physicalLabel, so that the colorterms work. It looks like FgcmOutputProductsTask run() and generateTractOutputProducts() both need to be updated to pass physical and band labels to _computeReferenceOffsets which passes them to _computeOffsetOneBand. I suspect you'd be able to fix that much quicker than I would.
You'll want this ticket branch of pipe_tasks and obs_subaru to test it.