Some of them may be caused by human activities (I haven't checked carefully yet.) But noise excess exist also early morning.
-----
Fig2 shows the noise spectrum of calibrated frequency noise in quiet situation (~19:00 5/30, solid line) and the situation that noise excess exists (~3:00 5/30, dashed line).
Fig3 shows the coherence with microphone signal at the same time of Fig2.
Because there is no difference of spectra of microphone signals between the quiet situation and the loud situation (See fig4),
some kind of the change in the coupling between the frequency noise and microphone signals probably occur.
According to Fig5, PMC and PLL signals also have coherence with microphone signals.
But they don't seem to be linked to the change in the excess on the IMC frequency noise.
Interesting. And strange... hmm... lets see