search

LSSTDESC / dia_pipe

Difference Image Analysis pipeline using LSST DM Science Pipelines developed for DESC Data Challenges
BSD 3-Clause "New" or "Revised" License
3 stars 2 forks source link

Determine how to best get/use aperture corrections in DIASrc table photometry #21

Closed wmwv closed 3 years ago

wmwv commented 3 years ago

The measured values in DIA Source tables are off by an aperture correction. #20

But the forced photometry on DIA images based on DIAObject positions is correct.

This issue is to try to look into what's going on. At a minimum we should create a test case that we can share with the Rubin Observatory DM team as an issue to fix. More ambitiously we might read the code and try to figure out what's going on.

@rearmstr makes the reasonable argument that if there is an error here, then it's on the DM code side and not in dia_pipe.

The DIA catalogs are not explicitly recalibrated (because that's not really possible) so it sounds like there's something wrong at some stage in the DIA source measurement. Because the DIA source catalog photometry should be correctly calibrated to the original calexp and template. Given that the forced photometry on the same DIA image is correct, that implies that there is something not set in the original measurement task.

I will speculate that this is because the relevant code was focused on generating some instrumental magnitude that would calibrated later.

Note: For point sources Aperture correction and calibration are somewhat interchangeable here. If you have a known reference catalog, then you can calibrate your photometry to that catalog and never have to worry about an aperture correction. It might be nice to know the aperture correction if you want to get resolved-object photometry correct.

wmwv commented 3 years ago

This was just (independently) filed on the Project side as https://jira.lsstcorp.org/browse/DM-28498

Future discussion will happen there and we'll close this present Issue here once things are fixed.

wmwv commented 3 years ago

https://jira.lsstcorp.org/browse/DM-28498 has now been completed and closed.

There are some remaining issues with the diaSrc forced-photometry done as part of imageDifferencey.py that will be resolved in https://jira.lsstcorp.org/browse/DM-28641