Alignment issue in Keck/DEIMOS stack

[muethela]

I attempted to re-reduce a Keck/DEIMOS dataset to produce a better WCS solution using the most recent version of POTPyRI on Burbidge. In the final stack, multiple copies of the same object appear in the vicinity of the actual objects. I inspected each individual frame (science and calibration) and have ensured that the data going in are ok. The field has a limited number of Gaia calibrators. The data and re-reduction attempt can be found at /data/Keck/FRB20230807_redux_v2/ on burbidge (note: I made a typo in the directory name, the FRB is actually named 20230827; the above path is correct though).

[charliekilpatrick]

Thank you for pointing this out, Alexa. This sounds very much like an alignment error. Have you tried looking at the reprojected images with the WCS locked to see if one or more of them is offset?

In any case, I can look closely at this soon.

[muethela]

Of the nine frames, three have good solutions across all four "panels" (detectors?): FRB20230827.R.ut230906.4.11.1378473764_reproj.fits, FRB20230827.R.ut230906.4.11.1378474869_reproj.fits,
FRB20230827.R.ut230906.4.11.1378475238_reproj.fits. In the other frames, the two right panels have good solutions and the left panels are off (see attached screenshot).

[charliekilpatrick]

Thanks @muethela - it looks like there is definitely a bad WCS here. Have you tried running with the --skip-gaia flag to determine whether that will fix the issue? The WCS module needs some work and DEIMOS imaging is hard to calibrate, but in the interest of a quick fix that may address this specific data set.

[muethela]

I've started a reduction with that flag (wasn't aware of that option) - will update on how it goes

[muethela]

That seemed to alleviate the object "copying" issue, but the wcs solution is still poor across the image, including in the region near the FRB (central and top ~third of "second" detector from right) where it matters most

[charliekilpatrick]

OK, thank you for performing this check @muethela. I will try to debug this issue via the next POTPyRI PR (#57) - can you let me know whether this is time critical for JWST?