Fix Version/s: None
Component/s: meas_algorithms, pipe_tasks
Sprint:Science Pipelines DM-S15-5, Science Pipelines DM-S15-6
We need to be able to coadd aperture corrections in much the same way we coadd PSFs.
See the HSC-side HSC-798 and HSC-897 implementation for a prototype:
with code here:
Note that we should also bring across HSC-975 (or equivalent functionality) in this story – that was originally slated for merge in
DM-2981, but didn't make it there.
I cherry-picked the commits for:
which add CoaddBoundedField and CoaddApCorrMap.
git cherry-pick 7cbd8df1c9 68cffd3ba9 a9ac67d006 a6ae568f537d fd7bae622f df26ee217c
I then squashed 6b5c958731 (a small fix to CoaddBoundedField.i) into the second of these to allow a the unit test to pass.
The oldest commit had some conflicts. Most were due to relying on things that had moved to meas_base, but one was due to lack of an object named BinnedWcs on the LSST side. This sounds like something LSST may want eventually, but which is not needed for this ticket and was added before the oldest commit above.
I cherry-picked commits for the following, which apply aperture correction on coadds:
- d5c2356, which apparently has no ticket branch in git but may be part of https://hsc-jira.astro.princeton.edu/jira/browse/HSC-976
git cherry-pick 7656d164 a0812f2f d5c2356d ba18e527
There were a few minor conflicts on inputRecorder.py: don't use afwImage.PARENT in calls to getBBox(), don't use getSerial() on the results of getId() and don't import math (since it's not used).
Bob Armstrong has agreed to review these changes, but he does not yet have a JIRA account, so I am leaving this in state "In Progress" for now.
Since Bob Armstrong still has not been given a JIRA account he kindly emailed his review to me. Here it is:
I went through the code and I think you ported over all the necessary changes that I could see. Since these are mostly ports straight from the HSC side I don't have any comments. Have you been able to run the code on any real HSC data? My one concern is that there is a piece that may be missing that is not fully captured by the unit tests. Unfortunately, running through real HSC coadds has been the only thing that gives me confidence that things are working as they should. Hopefully, we can work on changing that for the future. Otherwise, I think your changes are good to go.
I tested the changes using some HSC data and it all seems to run fine. Merged to master.
Does anyone have strong opinions on whether CoaddBoundedField should go in meas_algorithms or meas_base (or elsewhere)?
I plan to put makeCoaddBoundedField in the same package as CoaddBoundedField, since it uses that class and ApCorrMap and not much else.