Figure 1 shows the  projection of coherences calculated by Bruco on a subset of channels whose name cointains strings ENV, LC, Clock, BsX. The projection of BsX signals fits several peaks (as already exlained by BPC injections 39870). 

Fgure 2 zooms in the region between 300Hz and 450Hz where a few small new peaks are coherent with vibration noise, the winning accelerometer is that on DT tower wall.

I looked more carefully to the coherences for each one of these peaks, a few observations:

• the two peaks 322Hz (small) and 421 Hz (bigger) are similar in the sense that they have 0.2 coherence with accelerometers ENV_DT_ACC_Z, ENV_DT_CT_ACC_Z, ENV_SDB2_ACC_Y, but there is no corresponding peak on accelerometers (Figure 3 and Figure 4)
• other peaks at 401Hz, 402Hz, 412Hz (narrow) 448Hz, instead have some more pronounced coherence (0.6) with the same accelerometers but also (less) with other sensors in the hall (like microphones) which also have an associated peak (Figures 5, 6 and Figure 7)
• 452Hz peak is something in between, because it has has 0.4 coherence with ENV_DT_ACC_Z (not with DT_CT_ACC and not with SDB2_ACC)and maybe an associated peak in it, some coherence is seen also by the hall microphone,but no peak (Figure 8)