Investigate low-ish LRG success rates on 20241220

[schlafly]

20241220 has low-ish LRG success rates on P1, P2, and P7. P7 also features significant sky redshift pile ups. The only thing I note about P7 there is that z7 seems to have particularly bad ~cross-fiber contamination correction; you see a lot more bright / faint sky fibers on the red edge that I attribute to imperfect modeling of the PSF wings in the red. But that's worse on bright tiles than dark, and I see the sky fiber issue on the dark tiles, so I am not sure if these are the same effect. I don't see a big impact on redshifts overall.

P1 was the one where we just changed the cryostat valve and we may need a new dark, and we're also overflagging cosmic rays in the dark images.

I can't think of anything remarkable about P2.

@ashleyjross

[ashleyjross]

In all cases, petal 1 seems to have low thrufrac associated with poor redshift success:

Some tiles that stick out:
Maybe long slew?
looking at petal 1 on tile 6026
number with no data 27
number with bad qa 314
The fraction of good LRGs is 0.9459459459459459 for 37 considered spectra
looking at petal 2 on tile 6026
number with no data 18
number with bad qa 67
The fraction of good LRGs is 0.8943089430894309 for 123 considered spectra

Also longish slew, variable thrufrac
looking at petal 1 on tile 7450
number with no data 25
number with bad qa 66
The fraction of good LRGs is 0.8925619834710744 for 121 considered spectra

Not a long slew, high extinction, many splits
looking at petal 1 on tile 7036
number with no data 17
number with bad qa 76
The fraction of good LRGs is 0.9390243902439024 for 82 considered spectra

Not a long slew, petal 1 has low thrufrac, 0.86
looking at petal 1 on tile 3895
number with no data 15
number with bad qa 72
The fraction of good LRGs is 0.9210526315789473 for 114 considered spectra