NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/ca/earthqua/1497

Internet Message Format | 1992-12-22 | 5.5 KB

Path: sparky!uunet!stanford.edu!morrow.stanford.edu!pangea.Stanford.EDU!andy
From: [e-mail] (Andy Michael USGS Guest)
Newsgroups: ca.earthquakes
Subject: Re: Parkfield
Date: 22 Dec 1992 22:11:14 GMT
Organization: Stanford Univ. Earth Sciences
Lines: 86
Distribution: ca
Message-ID: <[e-mail]>
References: <[e-mail]>
NNTP-Posting-Host: pangea.stanford.edu

In article <[e-mail]> [e-mail] (Andrew L. Alden) writes:
>The Parkfield segment of the San Andreas fault, it was officially
>forecast in 1984, would almost certainly have a large, magnitude 6 or
>greater, earthquake on it before the end of 1992.  The stories around
>that time said that the fault had had regular M5-1/2 or M6 earthquakes,
>almost like clockwork, for the entire historical period of record--every
>22 years from 1857 to 1966, if you call the 1934 M6 event a "premature"
>quake that should have occurred in 1944.  Continuing the trend into the
>future appeared to mean another similar event in the 1988-1993 window.
To be fair the 1934 event was not just considered premature, it was
considered premature due to the occurrence of the 1934 foreshocks that
may have triggerred the event early.  This was a physical argument that
was published in a paper by Bakun and Lindh with accordant peer review.

>This simple model was, it seems, quietly discarded a while back.  Jim
No, it was disagreed with by Jim Savage.  Bakun, Lindh, and others did
not necessarily agree with him.  Discarded seems like a strange word.

>Savage (USGS) told the meeting that if 1934 is accepted at face value,
>the window would run until October 1991, not January 1993.  T.R.
>Toppozada (CDMG) described Parkfield earthquakes that weren't on the
>list in 1984; in fact, he said, the Parkfield segment shows a pattern of
>decreasing activity since the great 1857 quake, not a steady state of
>"characteristic events," a term I haven't seen as much as I used to.  At
It should also be mentioned that the damage and felt reports used by
Toppozada fail to make the 1922, 1934, and 1966 earthquakes the same size
even though he has the most data for them.  Given that we have strong
instrumental evidence that all three of these earthquakes were the same
size it is hard for me to give his other data, when there are many fewer
reports, any credence.

>a December 10 press conference, Bill Bakun and Al Lindh (USGS), authors
>of the clockwork hypothesis on which the Parkfield prediction was based,
>defended it stoutly but in a rather generic way: hypotheses are useless
>if they cannot fail, and the "regular recurrence hypothesis" will fail
>on December 31 if a M6 has not struck Parkfield; so science has marched
>a step further, and our new hypotheses are even better.
Well, in the early 80's this hypothesis did pass a lot of peer review,
so I think it is fair to say that we have learned something from this.

>So now we know that even on perhaps the world's straightest, cleanest
>stretch of strike-slip fault, you don't ever get a nice, regular set of
>stress releases like a violin string under a well-rosined bow (or a
>perfect planar fault in an infinite half-space).  Although the plate
>movement at depth in the lithosphere is in fact smooth and inexorable,
>the crust above it always responds in a complex, irregular way.  That's
>nice scientific progress...thing is, we didn't need the whole Parkfield
>experiment to learn that.
No one ever suggested that the Parkfield experiment was a test of the
characteristic earthquake hypothesis.  The instrumentation at Parkfield
is there to record details of the preparation period before an
earthquake, the earthquake itself, and the time period afterwards in
great detail.  The Parkfield site was picked for a variety of reasons,
one of which was that *every* statistical model places Parkfield at the
top of the list as the most likely place for a M6 or greater earthquake.

>The benefits from the project were great and worth the public money.
>Bakun pointed to our better knowledge of which instruments work best for
>monitoring faults and progress in informing the public about earthquake
>hazards.  Lindh noted that scientists, policymakers, emergency agencies,
>media, and the public are now talking together in a useful way.  Andy
>Michael and many others spoke up for it too, to good effect.
>
>I don't begrudge a dime paid for the experiment, even though the
>hypothesis ostensibly being tested was ridiculously simple.  My question
>is, was that hypothesis used because even Congress could understand it?
>Was the test question ignorantly simple or disingenuously simple?

Perhaps it was neither, but really a statement of a testable
hypothesis?  As for the second suggestion I consider it close to
libelous.  I know Bill Bakun and Al Lindh extremely well.  I consider
both of them ethically incapable of making a simple hypothesis look
good in order to get an experiment funded.  I don't know Tom McEvilly
as well but I can't believe he would participate in such subterfuge
either.

If you were listening to Jim Savage's talk carefully he placed the only
blame on the National Earthquake Prediction Evaluation Council (NEPEC)
for not considering alternative hypotheses carefully enough.  I
discussed this with Tom McEvilly who is now chairman of NEPEC (but who
was not on the panel when Parkfield was approved).  He pointed out that
the most recent probability documents to go through NEPEC are now sent
out to four separate uninvolved statisticians for review.  So this is
something that the seismological community is getting better at.

Andy