[time-nuts] GPS, USGS Early Earthquake Warning

[David]

On Fri, 27 Apr 2012 00:55:40 -0700, Hal Murray
<hmurray at megapathdsl.net> wrote:

>
>Fun talk at the USGS last (Thur) night:
>  ShakeAlert!
>  --building an earthquake early warning system for California
>  by Doug Given
>
>He's a good speaker.  If you get a chance to hear him, go for it.
>  http://online.wr.usgs.gov/calendar/
>The video should be up in their archives in a few days.
>  http://online.wr.usgs.gov/calendar/2012.html
>(scroll down to April)
>
>The basic idea is to detect an earthquake at location X, and then spread the 
>word using telecommunications.  Earthquakes propagate at 2 miles/sec so 
>phone/internet is much faster.  Ballpark is 30-60 seconds of warning.
>
>The reason this might be interesting for time-nuts is that he mentioned using 
>GPS to supplement seismometers.  The context is that they need the answer in 
>a few seconds.  I didn't catch any details.
>
>It's unlikely that they are doing any fancy post-processing.  I suppose it's 
>possible that they have streamlined the traditional post-processing setup so 
>that they can do it in a few seconds.  Maybe that is mostly getting an early 
>look at the data the traditional post-processing processing uses.

Through the 80s and 90s I monitored the W6FXN 2 meter amateur radio
repeater which was located near Cal Poly Pomona.  It was setup to
rebroadcast one of the USGS tone modulated seismic sensors when it
deviated from rest.  The warning time depended on the geometry but was
typically 5 to 10 seconds for Orange County since most of the
earthquakes like Whittier-Narrows, Loma Prieta, and Northridge
originated closer to the seismometer than where I lived.

It did not distinguish between local and far earthquakes but you could
hear the separation in the S and P waves and tell small from large
with a little experience.  I also heard the last couple of underground
atomic tests.

Latency in the warning was on the order of milliseconds.