Re: [Dews] USGS: ShakeAlert

[Matt Mathis <mattmathis@google.com>]

[-- Attachment #1: Type: text/plain, Size: 4632 bytes --]

I think the exercise at this point is just discovering all of the existing
related work: CAPs wasn't even on my radar.  I agree that there is no point
in duplicating any of that effort here.

But my sense from the very beginning is that the distributed sensors are
only the tip of the iceberg in terms of system cost and/or barriers to
deployment.  And although there is a certain elegance to fully distributed
solutions, I don't have much faith in multi-ISP multicast being
sufficiently robust at the ~100 km range needed for DEWS.  (Actually, any
one ISP for that matter).

OTOH, I know some people who run some data centers, ship a few browsers and
a few phones.  (BTW, As long as the epicenter is more than ~100 km from the
data center, it will survive long enough.)  This idea is a whole lot bigger
than my role, but perhaps I can draw the interest of the right people.

It is probably the case that the "right answer" is multiple DEWS systems
using different technologies, on the assumption that some will fail.

Thanks,
--MM--
The best way to predict the future is to create it.  - Alan Kay



On Sat, May 26, 2012 at 11:42 PM, Fred Baker <fred@cisco.com> wrote:

> I'm very very confused. The discussion on the bufferbloat list was about
> how to get a message from a system containing a cheap accelerometer - a CPE
> Router or a set-top box - to someone that wanted to receive them from
> zillions of sources. It seems to be morphing into a discussion of the
> Common Alerting Protocol (
> http://en.wikipedia.org/wiki/Common_Alerting_Protocol) or something like
> it to distribute alerts.
>
> Please pick one. If it's the CAP discussion, there is a lot of work going
> on and we don't need another place to comment on it; drop me off.
>
> On May 26, 2012, at 8:25 PM, Hal Murray wrote:
>
> >
> > In Apr, 2012, the monthly public lecture at the USGS Menlo Park campus
> > described their work in this area:
> >
> > ShakeAlert!
> > --building an earthquake early warning system for California
> > by Doug Given, USGS Earthquake Early Warning Coordinator
> >   * Millions of Japanese citizens received advance warning of the 2011
> > magnitude 9.0 Tohoku earthquake -- can such a system be built for use in
> > California?
> >   * University researchers and government agencies are working together
> to
> > create an Earthquake Early Warning system in California to reduce
> earthquake
> > losses
> >   * April is Earthquake Awareness Month in California -- how could you
> and
> > your family best prepare for severe ground shaking using 30 seconds of
> > advance warning?
> >
> > The video of the talk is online.  It's an hour and a half.
> > http://online.wr.usgs.gov/calendar/2012/apr12.html
> >
> > It's pretty good.  I think anybody interested in this topic should watch
> it.
> > If nothing else, it will give us a common reference.
> >
> >
> > Numbers and such from my memory:
> >
> > P waves travel at 3 miles/second.  S waves (the destructive ones) travel
> at
> > 2 miles/second.
> >
> > It takes about 10 seconds to verify that a quake exists and estimate how
> > big it is.
> >
> > They think they can get a 20-30 second warning.  That's for a big enough
> > quake, and far enough away but not too far.  If it isn't big, nobody
> cares.
> > If it's too far away nobody cares.  If it's too close, you don't get
> enough
> > warning time to be useful.  Quakes on the San Andreas Fault near Los
> Angeles
> > are likely to fit.  (The USGS people doing the work are located in
> Pasadena.)
> >
> > Half (or more) of the work is making contact with the people who want to
> > know that a quake is coming.  BART wants to stop their trains.  (BART
> trains
> > can carry 1000 people.A serious wreck would overload the emergency
> response
> > system even without any other earthquake damage.)
> >
> > He had lots of info from Japan.  They have a lot more/denser sensors
> than we
> > do.
> >
> > The USGS is looking for $50-100 million over 5 years to install more
> sensors
> > and $5-10 million/year for operations.
> >
> > --------
> >
> > The Moore foundation gave $6 million to CalTech, Berkeley, and Univ of
> > Washington for work in this area.
> > http://www.usgs.gov/newsroom/article_pf.asp?ID=3041
> > They are cooperating with the USGS.
> >
> >
> > --
> > These are my opinions.  I hate spam.
> >
> >
> >
> > _______________________________________________
> > Dews mailing list
> > Dews@lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/dews
>
> _______________________________________________
> Dews mailing list
> Dews@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/dews
>

[-- Attachment #2: Type: text/html, Size: 5998 bytes --]