We performed a χ^2 fitting the Shindo histogram during the O4a into an exponential + Gaussian function.

Sorry, the sentence was wrong:

We calculated the Shindo values for every 1 minute and categorized them as "Stand-by (not locked)", "Lock-keep", and "Lock-loss (within 2min before a lock-loss)".

-> every 2 minute

I updated the analysis for the O4a lock-loss and earthquakes.

Changed points are

• Shindo value is calculated for every minute. (old: every 2 minutes)
• Referring to the latest unlock segment from K1:GRD_UNLOCKED in Uchikata-san's summary. (old: Referring to the online lock-loss list)
• As a Shindo value for a lock-loss, the maximum value within 2 min before the lock-loss is chosen.

The number of lock-loss caused by earthquakes (Shindo > -6.0) events was 39.
I checked the waveforms of the seismometers and Shindo values (every 1 second) for all these events. (PDF1)
All of them look like typical earthquake signals.

By the way, 24 lock-loss events were labeled "EQ" in the online lock-loss list), but their Shindo values were less than -6.0.
I also checked the waveforms of the seismometers and Shindo values for all these events. (PDF2)
They don't look like earthquakes and should be categorized as "other seismic phenomena".
 

• non-sharp earthquakes? (far away? ) or slow-slips? 
  • 1369020402
  • 1369231065
  • 1369568030
  • 1370382859
  • 1370527221
  • 1370568696
  • 1370939656
  • 1370949941
  • 1371063948
  • 1371136503
  • 1371141866
  • 1371156938
  • 1371175429
  • 1371209557
  • 1371287335
• something seismic signal is found after a Lock-loss
  • 1369182750
  • 1369577616
  • 1370292318
  • 1370487998
  • 1371136503
• Local seismic signal only at IXV
  • 1369536350
• No significant seismic signal around LL time
  • 1369814805
  • 1370425722
  • 1370695955
  • 1371232565

I also plotted for non-categorized (unknown) lock-losses.

> No significant seismic signal around LL time
> 1369814805
> 1370425722
> 1370695955
> 1371232565
According to my memo, these events seemed to correspond to the following EQs.
- 1369814805 => https://earthquake.usgs.gov/earthquakes/eventpage/us7000k7dk/executive
- 1370425722 => https://earthquake.usgs.gov/earthquakes/eventpage/us7000k8xk/executive
- 1370695955 => https://earthquake.usgs.gov/earthquakes/eventpage/us7000k9va/executive
- 1371232565 => https://earthquake.usgs.gov/earthquakes/eventpage/us6000knfz/executive

These EQs can be seen on the BLRMS from 30mHz to 100mHz though there is no excess in a few Herz band (see also Fig.1 that show the 4th event).
Relative contribution of the o(10mHz) band to the 1Hz band in JMA Shindo calculation is likely to be smaller than one in BRLMS by a factor of (0.1*0.1)**2 or less. So BLRMS seems to be much more sensitive for EQs occurring in overseas.

Yes, actually in Hoshino-kun's study last summer, 0.01-0.1 Hz might bo a good window for oversea earthquakes.

JGW-G2315065
JGW-G2315131
JGW-G2315296