It can be seen directly on the spectrograms of the dark fringe (the cyan fog that is enlarging just below 100 Hz in fig. 1).
A comparison between a noisy and a normal period is shown in fig 2. The noise is really perfectly centered around 100 Hz.
Gabriele noticed that this noise seems to be correlated with low frequency noise. Indeed in fig. 3 you can see that it follows the level of the seismic activity among 0.2 Hz and 4 Hz, which correspond to the microsesim (0.2Hz-1Hz) plus the anthropic activity (1Hz-4Hz).
The coupling path is not at all clear. No correlation were found with IPS and UPS sniffer channels.
