[CEUS-earthquake-hazards] no "right" answer

[Jose M Pujol (jpujol)]

I appreciate the opportunity offered by this listserve to express my concerns
regarding the NMSZ probabilistic hazard analysis, summarized below.
Before proceeding, however, I want to make clear that I am not siding
with anybody involved in the ongoing debate, most of whom I don't know personally.


1) I have some problems  with the approach used to select the magnitude of 7.7
as the characteristic magnitude of the 1811-1812 New Madrid earthquakes. The
logic tree described in the 2002 USGS Open-File report 02-420 is based on the
following magnitudes and weights (in parentheses):  7.3 (0.15), 7.5 (0.2),
7.7 (0.5), 8.0 (0.15). This scheme produces a  mean hazard essentially
equivalent to that obtained by giving full weight to the magnitude 7.7.
This scheme was developed using input from a workshop and results from Bakun
and Hopper that determined a magnitude range of 7.0-7.7, and includes the
magnitude 8 from Johnston results (Geophys. J., 1996) and the lower
magnitudes determined by Hough et al. (JGR, 2000). I don't know why the
magnitude 7.7 was given the highest weight, but what I think is missing here
is a serious discussion of the arguments made by Hough et al. Their results
are essentially incompatible with those of Johnston, and it seems to me that
it is not scientifically sound to treat their respective magnitudes as if they
were samples from a random variable with weights that are assigned subjectively.

At the risk of oversimplifying things, the case made by Hough et al. is that
the strongest effects of the 1811-1812 earthquakes were felt in river valleys,
while hard-rock sites were much less affected. This clearly point to strong
site effects. Everybody knows that unconsolidated sediments can increase the
amplitudes of ground motion several fold with respect to a rock site. In addition,
the duration of shaking also increases, which is  significant for the following
two reasons. First, damage to buildings and other structures increases when
the duration of an earthquake increases. For example, a building may withstand
one of two cycles of strong shaking, but not several cycles of it. Second,
the possibility of liquefaction increases with duration because an increase
in the number of stress cycles lowers the intensity required for failure.
Damage has been used to assign intensities and liquefaction arguments have
been used to justify the larger magnitudes assigned to the 1811-1812 events.

It seems obvious to me that site effects must be taken into account, as proposed
by Hough et al. This is why I don't think that it is scientifically sound to treat
Hough et al. and Johnston results as more or less equally probable. In addition,
the Bakun and Hopper (BSSA, 2004) larger magnitude estimates were obtained using
their own intensity estimates; when the Hough et al. estimates are used the
magnitudes end up being somewhat smaller. In 2005 there was a meeting in Memphis
on the earthquake hazards in the central U.S. organized by the Applied Technology
Council and the USGS. One of the speakers was B. Schweig, who noted that the
1811-1812 earthquakes had "probably at least magnitude 7.5, and they shook like
magnitude 8.0". This statement clearly summarizes the uncertainty affecting the
magnitudes estimates.

Another point I want to make is that several other papers published after
the Hough et al. paper are consistent with their results (see Hough et al.,
SRL, 2005, v. 76, 373-386). As a disclaimer, this includes a paper written by
Mueller and me (BSSA, 2001, v. 91, 1563-1573), where we also argue for somewhat
lower magnitudes. Yet, the USGS sticks to its 7.7 magnitude.

In any case, what I think is needed is to investigate the magnitude problem
using a more scientific approach. I proposed it in a proposal submitted to the
USGS, and although the panel did not recommend funding, there was no objection
to the basic principles, which are rather simple. Because intensity measurements
are actually a proxy for  the ground motion during an earthquake, the effect of
the unconsolidated sediments of the Mississippi embayment can be inferred from
analysis of synthetic data. The idea is to generate synthetic seismograms for
embayment and hard-rock models, and to estimate magnitudes applying standard methods
to the synthetic data. A comparison of results for the two types of sites will
give us a good idea of the of the effect of the sediments on magnitude estimation.
Preliminary work I did seems to indicate that it may introduce a difference
in magnitude of about 0.5. Of course, more work is needed, but since I am currently
working on other projects, I hope somebody else may be interested in pursuing my
ideas.


2) I have not followed carefully the arguments on the probability of occurrence
of a magnitude 7.7 earthquake in the near future, but what I found striking is
the statement that there is a 7-10% probability that a magnitude 7.7 earthquake
can occur within any 50 year period. This was big news in the Memphis paper
in 2005 (Commercial Appeal, Dec. 11). If I understand this correctly, it means
that there is a significant probability that the strain accumulation required for
such an earthquake will be reached by 2055 at the latest. This is about 250 years
from the 1811-1812 earthquakes. If the recurrence rate is 500 years, my
interpretation is that a repeat of one of them requires a significant increase
in the rate of strain accumulation. This in turn will require a significant
acceleration in the physical process(es) responsible for the strain in the NMSZ.
So I wonder whether the 7-10% prediction is based on purely probabilistic
considerations, or whether it also uses other information, such as GPS. I would
appreciate it very much if someone could clarify this for me.

Jose Pujol
Professor
Dept. of Earth Sciences
The Univ. of Memphis.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geohazards.usgs.gov/pipermail/ceus-earthquake-hazards/attachments/20080219/85f92a33/attachment.html