I analyzed the broadband acoustic injection data:

run3:
11:07:00 - (JST), OMC area, 5000 cnt, white

At first, I checked the amplitude spectral densities of the following channels:

• K1:OMC-TRANS_DC_SUM_OUT_DQ
• K1:OMC-LSC_ERR_IN1_DQ   
• K1:PEM-MIC_OMC_BOOTH_OMC_Z_OUT_DQ
• K1:PEM-SEIS_OMC_GND_X_OUT_DQ
• K1:PEM-SEIS_OMC_GND_Y_OUT_DQ
• K1:PEM-SEIS_OMC_GND_Z_OUT_DQ
• K1:PEM-ACC_OMC_VACTABLE_OMC_Y_OUT_DQ

The coherences between these channels and the microphone channel were also plotted.
In all channels, the coherence was increased by the acoustic injection. 

The coherence between the MIC and the OMC trans DCPD was also found. It means there is not only non-linear coupling but also liner coupling.
Such behavior was not found in the OMC commissioning with the single-arm-lock, performed this summer.

I also checked the coherence between the OMC trans DCPD and the squired-time-series of the OMC error (length) signal.
There was also significant coherence.
Below 10Hz, the coherence was decreased by the acoustic injection. It was strange because the speaker used today can't make these frequency sounds.

I evaluated the acoustic noise contribution in the Geophone signal and the OMC error signal, by the coupling function model (aberable for a liner coupling case).

These results (CF and projection) are plotted.
From them, the in-vac table vibration measured by the Geophone is almost originated by the sound.
The peaks around 100Hz in the OMC error signal are also due to the sound.

Finally, I evaluated the CF and projection from the Geophone to the OMC error signal. (It should be done by a ground shaker injection test, not by an acoustic injection, ideally)
This result also suggests the peaks around 100Hz in the OMC error signal are caused by the table vibration.
[See also klog30589, for a comparison]

I calculated the OMC error (linear and squared) projection into the OMC trans-PD, by the coherence (klog31295) of the reference time.
Here, coherence > 0.25 points are plotted.

Around 1-10Hz, 20Hz, 58Hz, 92Hz, 140-155Hz, and harmonics of the 60Hz, are dominant or close to the sensitivity.

I also calculated the OMC error (linear and squared) projection into the OMC trans-PD, by the ICA method
$H_{x\to y}(f) = \frac{<y(t),x(t)>}{<x(t),x(t)>}$
for the broadband acoustic injection data.

Here, coherence > 0.25 points are plotted.

Note that the projection by ICA may be overestimated because the OMC error signal was limited by the sensing noise and not the real OMC length above 100 Hz, in the reference data.

So I also made the projections by applying the 100Hz low-pass filter for the OMC error signal.