The earthquake we had last night was a very successful test of the SEI_ENV automated earthquake response. Attached trend gives the sequence of events. SEI_ENV state is the top left, SEI_CONF is top right, eq band ground is bottom left, ISC_LOCK state is bottom right.
Look like seismon/SEI_ENV got the notification just before the P/S wave arrival (where top left goes from 10 to 20), but that wasn't big enough to be a problem.
SEI_ENV switched SEI_CONF to the earthquake mode as the R waves started rolling in and the IFO actually stayed locked through the peak (at -33000 seconds), we didn't lose lock until ~400 seconds after, maybe from an aftershock? At least it wasn't from the peak or the SEI_CONF transition.
The time to transition back to nominal also seems pretty reasonable. This was a pretty good size earthquake, we probably haven't ridden out anything bigger to this point.
There are still edge cases to be thought about: the test for the SEI_CONF transitions are kind of tricked by very high winds and microseism. Down the road we may have to consider pausing SEI_ENV during big windstorms, to avoid spurious transitions. Still, great success.
SEI_ENV also worked great this evening! Switching us to EQ mode and back, keeping lock.
The 2020-02-28 08:34:15 UTC lockloss does seem strange though that we survived the massive 6000 peakmon peak and lost lock 8 minutes after plot here. Have you seen that before? There is also one of those quick (100-500Hz) DARM spikes 1s before the lockloss as you can see in the bottom of this zoomed plot.
