Fix Version/s: None
Original ticket title: Investigate amp background level offsets in DECam postISRCCDs
This ticket updates the DECam ISR overscan correction config parameter fitType='MEDIAN_PER_ROW'. A large body of evidence (see comments below) point towards this being a superior approach at handling overscan corrections with DECam. This also brings Science Pipelines data processing in-line with the DECam Community Pipeline, which also fits per row.
Original ticket efforts by Kenneth Herner, original description below:
If you scroll through here, the problem is immediately apparent
Hat-tip to Lee Kelvin for noticing this problem and plotting the images. This feels very familiar to when ISR order of operations was inconsistent regarding when assembleCcd happens, but I don't remember the resolution.
Possibly relevant and/or wrong: https://community.lsst.org/t/a-change-to-the-order-of-operations-for-isr/3060
- relates to
DM-30703 Reprocess DECam HiTS data from scratch with background fixes
DM-33096 Enable empirical read noise calculation for DECam ISR
DM-35252 Add DECam config overrides into cpBias and cpFlat pipelines
DM-15152 crosstalk correction was moved above assembleCcd, which broke it
DM-30495 Troubleshoot DECam ISR failures
- split to
DM-33126 Investigate cpFlatNormalization level settings in DECam postISRCCD
- To Do
I created DM-33126 to continue work on the still-open question of level settings in cpFlatNormalization.
Jenkins picked up a couple of hard coded unit test values which also needed to be updated. I've updated these, pushed to the PR, and re-ran Jenkins here: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/35645/pipeline/47
Thank you again to Kenneth Herner for your review, and Robert Lupton for taking a closer look under-the-hood. I rebased this ticket onto main, and re-ran Jenkins to make sure nothing fails. Branch merged and deleted, thanks again all.
Thanks Meredith Rawls and Kenneth Herner, that's good to know!
Link to Jenkins: https://ci.lsst.codes/blue/organizations/jenkins/stack-os-matrix/detail/stack-os-matrix/35639/pipeline
Build products used for Jenkins are: lsst_distrib lsst_ci ap_verify_ci_hits2015. If there's another package that should be added to the build products to test this DECam config change, please do let me know!
Update: I have subsequently learned that adding ap_verify_ci_hits2015 to the list of build products should not make a difference - the ap_verify datasets are not scons-buildable. With that said, it shouldn't have any negative impact either, so the above Jenkins is still valid.