JoeB, M. Wade, A. Viets, L. Dartez

Maddie pointed out an issue with the computation of the kappa_TST starting since 10/13. After looking into it a bit further, Aaron found a noisy peak at 17.7 Hz just above the kappa_TST line (which is very close at 17.6 Hz). It turns out that the peak has been there for a while, but it got louder on 10/13 and has been interfering with the kappa_TST calculation since then. 

For a quick reference take a look at the top left plot on the calibration summary pages for October 12 and October 13. Taking a look at the DARM spectra for those days, JoeB, noticed a near 17.7Hz on 10/12 at about 23:30 UTC (darm_spect_oct12_2023.png). Interestingly, he noted that it looks like the 17.7Hz line, which was present before the glitch, got louder after the glitch (darm_spect_oct13_2023.png). 

I've attached an ndscope screenshot of the moment that the glitch happens (darm_17p7hz_glitch_kappaTST_uncertainty.png). Indeed, there is a glitch at around 23:30 on 10/12 and it is seen in by the KappaTST line and the line's uncertainty. Interestingly, after the glitch the TST line uncertainty stayed high by about 0.6% compared to its value before the glitch occurred. This 0.6% increase pushed the mean KappaTST line uncertainty above 1%, which is also the threshold applied by the GDS pipeline to determine when to begin gating that metric (see comment LHO:72944 for more info on the threshold itself).

It's not clear to us what caused the glitch or why the uncertainty stayed higher afterwards. I noticed that the glitch at 23:30 was preceded by a smaller glitch by a few hours. Oddly, the mean KappaTST uncertainty also increased (and stayed that way) then too. There are three distinct "steps" in the kappaTST uncertainty shown in the ndscope I attached. 

I'll note that I initially looked for changes to the 17.7Hz line before and after the TCS changes on 10/13 (LHO:73445) but did not find any evidence that the two are related.

==
Until we identify what is causing the 17.7Hz line and fix it, we'll need to do something to help the kappaTST estimation. I'd like to see if I can slightly increase the kappaTST line height in DARM to compensate for the presence of this noisy peak and improve the coherence of the L3 sus line TF to DARM.