We decided to go ahead with merging the faro branch that adds the base class for metrics involving catalogs of matched sources, e.g., the pandas dataframe that is output by the isolatedStarAssociationTask.
Note that photometric and astrometric corrections are currently not applied to the parquet source catalogs. These corrections will be applied following the implementation of https://jira.lsstcorp.org/browse/DM-33959
For now, we decide not to add psfFlux and psfFluxErr columns to the matched source catalogs generated by isolatedStarAssociationTask in pipe_tasks and we will wait for https://jira.lsstcorp.org/browse/DM-33959 to be implemented to decide exactly how to include the columns with astrometric and photometric calibrations applied. The conversation on this topic from a draft PR in pipe_tasks is included below for reference.
In order to use the output of isolatedStarAssociationTask for faro photometric repeatability metrics the isolated_star_sources output should contain the columns psfFlux and psfFluxErr.
erykoff 3 days ago
Do we want the psfFlux or the psf instflux? Also you probably want the associated flag.
@psferguson psferguson 3 days ago •
I think the psfFlux if we are computing photometric repeatability from this (comparing measurements of a single source).
I guess we would want the psf_instflux If the calibration is going to be superseded and we should be applying updated zeropoints. (may need to think about this)
Agree on the flag.
@erykoff erykoff 3 days ago
If you are going to be applying external calibrations (and I believe you are!) then you need the instFluxes. Furthermore, at least at the current time, the "calibrated" fluxes aren't even calibrated according to the correct calibration (there's a ticket on this). In the farther future, you'll be able to get fluxes with the best calibrations out of the source tables, but we are still far from that point.
@bechtol bechtol 3 days ago
To confirm, is this the referenced ticket for consistent calibrations between the source table and calexp.photoCalib: https://jira.lsstcorp.org/browse/DM-34019 ?
Two more questions:
(1) Assuming the above ticket is resolved, should we be using local_PhotoCalib column for the source table, or should we be loading the photoCalib from the butler?
(2) Do the (ra, dec) columns in the source table represent the calibrated coordinates?
@erykoff erykoff 3 days ago
Yeah, that's the ticket!
(1) Until we are running source tables with the best calibrations (not coming imminently), faro is going to want to run the (fgcm) recalibration. And therefore should be taking in instfluxes rather than calibrated fluxes and the local photocalib, and then trying to remove the calibration before applying the new one.
2) The ra, dec are from the initial wcs in the single frame processing. So faro will want to apply jointcal recalibration to the x, y positions. (But because of the bug from that ticket, I believe that the local_wcs will not translate between x, y and ra, dec.)