[CEUS-earthquake-hazards] Not that simple

[John Nichols]

Dear All:

 

Having worked on the statistics for deaths in earthquakes now for the better
part of a decade, it is nowhere near as simple as your thoughts would imply.

 

Firstly, the risk is not uniform for a given regional area, because the
losses are dependent on the population location, size and density, in simple
terms the higher population centres attract higher risk because they can
sustain small loss events and large loss events.

 

Secondly, the different types of buildings have different survivability
rates, so some buildings are a lot safer than others, we do not allow for
this yet in design codes.

 

Thirdly, the return period of the earthquake is an interesting number, but
the more interesting number is the return period on the combination of loss
of life and earthquake combination, so 2500 year return period events cause
a lot of the deaths. You need to get past just looking at the statistics of
earthquakes and look at the statistics of real loss.  The losses are
probably best described using a Generalized Poissonian Model, and the
published results do not indicate a great improvement in life safety in the
last century for a variety of reasons.  So for example, taking California
out of the potential large death pool really doesn't help that much cause it
is not really in the most significant part of the pool for a lot of reasons,
and really never has been.  

 

Fourthly, the most significant work on the best method to estimate a world
consistent loss model would be based on the work of Kafka and Levin in 2000
(SSA, 90(3)), I have estimated that this study would cost about $500,000
(now), and it would put a modelled risk on every urban area that was based
on a consistent model that allowed for all of the above factors.  I put a
proposal to the USGS for 250,000 once on this work, but it was declined.  I
am not going to waste time chasing this stuff, some day the insurance
companies will need the data, actually they do already but just haven't got
the model, and it is certainly a cheap method to quantify risk, till then I
prefer to potter on masonry building losses in third world countries as it
provides good data.

 

Just a few thoughts.

  

John Nichols Ph.D. MIE(Aust)

Assistant Professor

Room 429 A,  Building 0398

Texas A&M University

College Station, TX, 77843-3137

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geohazards.usgs.gov/pipermail/ceus-earthquake-hazards/attachments/20090507/8c2ef629/attachment-0001.html