Satellite on EXPID=153715 (TILEID=22380)

[araichoor]

on night=20221116, I think a satellite crossed the exposure expid=153715 for tileid=22380. I don t know if some action should be taken here (maybe not). maybe some masking for the LSS catalog? (@ashleyjross).

see:

• the spacewatch movie (full version here: https://data.desi.lbl.gov/desi/users/raichoor/main-status/spacewatch/spacewatch-20221116.mp4):

https://user-images.githubusercontent.com/61986357/202979013-96284fab-d8d0-4987-a2b8-251ea438f991.mp4

the spacewatch image sampling is one every two minutes: the satellite is visible at the tile location on the one on 01:24:05 (UTC) the 153715 exposure started at 01:24:39 (UTC).

• the 3 bright sky fibers in petals 2, 3, and 5 of the sframesky (https://data.desi.lbl.gov/desi/spectro/redux/daily/nightqa/20221116/sframesky-20221116.pdf):

if I take the median flux of the sky fibers, flag fibers with FLUX > 10, I identify those three fibers: 1209, 1800, 2639 which are aligned, consistent with the satellite orientation (crude estimate : dec = 1.13 + 0.89 * (ra - 330)):

here is the cframe flux for those three fibers:

then, if I take for the 5000 fibers, the 24 ones with dec within +/- 0.01 deg of the line identified above, those consistently have very high fluxes:

[schlafly]

I know there has been a little discussion of using the new pointing camera information to identify and flag these (El Nino?). I haven't heard discussion of using the all sky camera to do the same. My impression is that the all sky camera doesn't have the precision necessary to identify the affected fibers a priori, though your investigation into the individual fluxes does look pretty convincing. 0.01 deg is pretty large relative to the PSF (36"); I'm surprised you see an effect that far away (for the most part).

FWIW, I think if we do something of this kind for LSS we should try to be a bit more systematic about it. Do you expect there are ~lots of examples of this?

[ashleyjross]

Something related to this that I think we have (I know it has at least been discussed) is total observed flux compared to expected flux from the photometry. We might be able to use that generally and remove outliers to screen bad observations?

[araichoor]

thanks both for the feedback. I ll put on my to-do-list to make some proper comparison betweeen the measured flux in spectro. and the FIBERTOTFLUX in ls-dr9. I ll tackle that later this week, or next week.

in the meantime, I can do this quick-n-dirty comparison, using FLUX_R_MED / FIBERTOTFLUX_R, ie the median spectro. flux in the r-camera divided by the ls-dr9 flux (so this doesn t account for the filter convolution; and maybe not for Gal. ext; but it should be right at the first order). flagging objects with (OBJTYPE="TGT") & (FIBERTOTFLUX_R > 0) & (FLUX_R_MED / FIBERTOTFLUX_R > 1) nicely highlights the satellite track:

about my dist < 0.01 deg criterion: it s a rough estimate, as my "line" estimate is not precise at all (drawn by eyes on the three sky fibers; however, e.g. the one in the middle has a lower flux than the other two, which could mean that the "line" should be closer to the two external fibers than to the one in the middle..)

[araichoor]

ps: from watching the spacewatch movies, those happen mostly at the beginning/end of the night (as we expect), so it should mostly affect bright tiles, not dark tiles.

ps2: I agree that the spacewatch astrometry is not precise enough to get the fiber identification; I just used it as an indication that something s there; @ashleyjross s suggestion to compare spectro/photo flux should nicely work.

[araichoor]

another case here (though less obvious): https://desisurvey.slack.com/archives/C01HNN87Y7J/p1674458401818369.

[ashleyjross]

Can we try to make a list of all (probably) affected fibers, at least for Y1?

[araichoor]

I didn t have time to investigate, so I just mention here for book-keeping: on 20230421:

• tileid=4377 (expid=177284): there likely is a satellite crossing this exposure
• tileid=22351 (expid=177289): there s a super-bright satellite passing just nearby; that s probably ok..

[araichoor]

another case in the dark tileid=2769 (expid=180742) on 20230514: