After the lockloss around 8:30 JST, precision of estimated time dependent coefficients of the calibration became quite worse.
Cause of lockloss seems to be an earthquake around t=-12 in Fig.1.
Figure 2 shows the lock status and estimated time dependent coefficients in recent these days.
Estimated \kappa_C showed roughly +/-15% before the lockloss by that EQ (t=-12h).
On the other hand, it shows now larger than +/-30% after t=-12h.
It appears that calibration lines at 27.650Hz and 28.670Hz are hidden by the noise excess on DARM around 20-30Hz.
Blue and red curves in Fig.3 represent a noise curve yesterday (it seems no excess yet) and today.
We can see large excess around 20-30Hz (t-cursols are the calibration lines).
On the other hand, 100-300Hz seems to become slightly quiet though I haven't checked it's really statistically significant or not.
Anyway, because of poor SNR of the calibration lines, estimation of time dependent coefficients doesn't work fine now.
Thanks to the repair work of SR2 control (klog#34117), noise excess around 20-30Hz was removed and calibration line can be seen with enough SNR.
Figure 1 shows the estimated time dependent coefficients and OpLev-Y of SR2. Precision of estimated time dependent coefficients was drastically improved at the same time as OpLev RMS decreased.
