Fix Version/s: None
When importing meas_mosaic, the coordinate system for writing out the wcs/fcr files was not adapted to that expected from LSST (which always associates the detector origin with the electronics, whereas HSC's is such that a given detector's origin, pixel (0, 0) is associated with its LLC w.r.t. to the focal plane), but rather the tasks in meas_mosaic's updateExposure.py were adapted to account for the rotated CCDs. It was assumed that this was the only place those corrections were every used. This turns out not to be the case since the wcs used to create the coadd gets attached to it in coaddInputs. If meas_mosaic was run and doApplyUberCal=True (which are both the case for our HSC data processing), the wcs’s that are getting attached to the coaddInputs are not in the coordinate system appropriate for LSST for any nQuarter != 0 ccds.
Since this is already causing an issue in pipe_tasks's propogateVisitFlags.py, see issue highlighted in
DM-9383, (and could very well be causing issues elsewhere as yet undiscovered), we have decided these outputs need to be written out in the coordinate system expected by LSST.
DM-9870 Release hscPipe 5.0-beta1
DM-9907 Release hscPipe 5.0-beta2
DM-10129 Process HSC RC dataset using Stack version chosen for the full S17B reprocessing
- is triggering
DM-10236 Properly apply the meas_mosaic solution
- relates to
DM-9901 TanWcs::rotateImageBy90 and flipImage do not transform SIP terms
- To Do
DM-9383 Investigate propagation of visit flags for certain patches in HSC RC processing
Sounds good. Let me know when you have something ready for testing & we can iterate if necessary.
Lauren MacArthur, I think this is ready for you to try in meas_mosaic. You'll need u/jbosch/
DM-9862 of afw and meas_astrom; those should build against w.2017.10 and you should only need to rebuild those two, not any other afw dependencies, as this shouldn't be ABI-breaking.
The function you care about is lsst.meas.astrom.rotateWcsPixels(wcs, bbox, angle). It'll require a TanWcs, so you may need to use lsst.afw.image.TanWcs.cast(...) if you just have a regular Wcs.
Ok, I will do my initial test to check that it's working as expected and then will start the new runs of the RC dataset.
This should probably have been reviewed and merged a while ago, but dual ownership, distractions, and the fact that it still leaves us in a broken state (fixed on
DM-10236) got in the way.
Branches to review are tickets/
DM-9862 on each of afw, meas_astrom, and meas_mosaic. I'll make PRs and launch a Jenkins run shortly.
As we discussed yesterday, my plan is to put code to properly rotate a TanWcs with SIP terms in meas_astrom (since some of the machinery to do the work is already present there). I'll let you test that and plug it into meas_mosaic, and we can figure out how to refactor that code and deprecate the incomplete rotation code in afw on a future issue.