[CEUS-earthquake-hazards] reply to Ellis Krinitzsky
jacob at ldeo.columbia.edu
jacob at ldeo.columbia.edu
Tue Feb 26 09:42:14 MST 2008
Dear Ellis:
Let's assume for the sake of the argument you were right (and I believe you are
not, since probability and uncertainty are intrinsicly connected through an
underlying distribution, whether it is sampled sufficiently or not, simply by
the fact that these distributions are linked to physical processes, and then we
make incomplete models of them, hopefully not just a few but several to explore
epistemic uncertainty as well); but lets assume you were right, then you made
an important point: building codes do NOT generally apply to critical
facilities. Critical facilities usually use special design methods (dynamic
analyses, etc) rather than prescriptive code designs. So I am not sure how your
very own argument advances THIS discussion which largely has to do with building
code applications of the seismic hazard mapping.
And if you were right on the critical facilities argument, then you have to ask
yourself why is EPRI and the nuclear industry INSISTING on the use of
probabilistic methods? One answer is probably: if you had to design a nuclear
power plant for all worst case scenarios (earthquake, tornado, flood,
terrorism, etc.) there would be no NPPs. Is this really what you are saying?
Are you also saying that all dams in the US or the world have been built to cope
with worst case scenarios? I have limited experience with dams. From my own
experience with the Tarbela Dam in Pakistan, designed by US engineering firms,
and one of the world's largest earth and rock-fill dams, across the Indus
River, is definitely NOT built to withstand the worst case earthquake one can
consider possible (although I would be at a loss how to define such a thing
other than probabilistically, when it comes right down to it, given how little
we know about the tectonics of the region !!)
Best
Klaus Jacob
=============
Quoting "Krinitzsky, Ellis L ERDC-GSL-MS Emeritus"
<Ellis.L.Krinitzsky at usace.army.mil>:
> Art,
>
> What I am saying is you can't account for uncertainty by probability, because
> you have too little data. It is easy by the deterministic method to get a
> worst case (mean plus sigma) and design safely. With your probability, you
> don't know what you have.
>
> I'd like to see you admit that one method is better than another for certain
> purposes.
>
> For structures that are critical (the consequences of failure are
> intolerable) you should use the deterministic method.
>
> Ellis
>
>
>
> -----Original Message-----
> From: ceus-earthquake-hazards-bounces at geohazards.usgs.gov
> [mailto:ceus-earthquake-hazards-bounces at geohazards.usgs.gov] On Behalf Of
> Arthur D Frankel
> Sent: Friday, February 22, 2008 12:28 PM
> To: ceus-earthquake-hazards at geohazards.usgs.gov
> Subject: [CEUS-earthquake-hazards] reply to Ellis Krinitzsky
>
>
> Ellis,
>
> I don't think the example you cite is applicable to the case of the logic
> tree we use for New Madrid magnitudes. In our logic tree we are trying to
> account for the uncertainty in the magnitude of characteristic (1811-12 type)
> earthquakes in the New Madrid source zone.
>
> In your example with two faults, a probabilistic hazard assessment would
> use the range of magnitudes specific for each fault. It does not average the
> magnitude between the two faults.
>
> In your earlier email, you expressed doubt about determining the rates of
> large earthquakes by extrapolating the rates of smaller earthquakes using a
> "b-line."
> This is not what is done for the New Madrid characteristic earthquakes
> (1811-12 type earthquakes). The average recurrence rate for the 1811-12
> type earthquakes is determined from the dating of sand blows (see Tuttle et
> al., 2002 BSSA), which shows that previous such events occurred around 1450
> and 900 A.D.
>
> -Art
>
>
>
> Art Frankel
> U.S. Geological Survey
> MS 966, Box 25046
> DFC
> Denver, CO 80225
> phone: 303-273-8556
> fax: 303-273-8600
> email: afrankel at usgs.gov
>
>
>
> "Krinitzsky, Ellis L ERDC-GSL-MS Emeritus"
> <Ellis.L.Krinitzsky at usace.army.mil>
>
> 02/22/2008 08:40 AM To
> "Arthur D Frankel" <afrankel at usgs.gov>, "Wang, Zhenming"
> <zmwang at email.uky.edu>
> cc
> <ceus-earthquake-hazards at geohazards.usgs.gov>
> Subject
> RE: [CEUS-earthquake-hazards] FW: reply to Joe Tomasello; buildings codes
> and earthquake hazard
>
>
>
>
>
>
> Art,
>
> I saw elsewhere you came to your values using a logic tree.
>
> Suppose you had two faults, one with a potential for M6, another for M8. If
> you average them you have an M7. If you designed for that M7 and you had an
> M8, you would be under designed. Yet the logic tree makes you do exactly
> that. Plus other illogical moves.
>
> You need to solve those problems first.
>
> Ellis
>
> -----Original Message-----
> From: ceus-earthquake-hazards-bounces at geohazards.usgs.gov
> [mailto:ceus-earthquake-hazards-bounces at geohazards.usgs.gov] On Behalf Of
> Arthur D Frankel
> Sent: Tuesday, February 19, 2008 10:01 PM
> To: Wang, Zhenming
> Cc: ceus-earthquake-hazards at geohazards.usgs.gov
> Subject: Re: [CEUS-earthquake-hazards] FW: reply to Joe Tomasello; buildings
> codes and earthquake hazard
>
> Zhenming,
>
> These two statements are not contradictory, when taken in the context that
> I wrote them.
>
> In the quote from my response to my comment, I was referring to the ground
> motions observed at any given location over time. At any particular site the
> ground motions with a 2% chance of being exceeded in 50 years, will occur, on
> average, once in 2500 years. As I said before, another way to express this is
> that these ground motions have a 1/2500 chance of being exceed each year.
>
> The point of my recent email is that each time an 1811-12 type earthquake
> occurs, there will be some locations that will experience the 2%/50 ground
> motions or larger, because of the spatial variability of ground motions. The
> set of sites that experience these higher ground motions will likely be
> different for each occurrence of this type of earthquake, because of the
> variability of ground motions from earthquake to earthquake. This apparent
> temporal variability of ground motions will occur even for successive
> earthquakes on the same fault, because of differences in rupture propagation
> and slip on the fault from earthquake to earthquake.
>
> As you design buildings to ground motions with lower probability levels, you
> are protecting a larger fraction of buildings from the range of ground
> motions expected during the next 1811-12 type earthquake.
>
> Art Frankel
> U.S. Geological Survey
> MS 966, Box 25046
> DFC
> Denver, CO 80225
> phone: 303-273-8556
> fax: 303-273-8600
> email: afrankel at usgs.gov
>
>
> -----ceus-earthquake-hazards-bounces at geohazards.usgs.gov wrote: -----
>
>
>
> To: "ceus-earthquake-hazards at geohazards.usgs.gov"
> <ceus-earthquake-hazards at geohazards.usgs.gov>
> From: "Wang, Zhenming" <zmwang at email.uky.edu>
> Sent by: ceus-earthquake-hazards-bounces at geohazards.usgs.gov
> Date: 02/19/2008 07:34AM
> Subject: [CEUS-earthquake-hazards] FW: reply to Joe
> Tomasello;
> buildings codes and earthquake hazard
>
>
>
> Here is another inconsistent statement on the national
> seismic hazard
> maps.
>
>
>
> " You have the mistaken notion that 2%/50 values are only
> observed
> once in 2500 years. This is flat out wrong. "
>
>
>
> In a response to our comment (Wang and others, 2005) that was
> published on Seismological Research Letter (Frankel, 2005), the
> interpretation was "the ground motion with 2% PE in 50 years is exceeded
> once, on average over 2,500 years, so that it has a 1/2500 annual probability
> of being exceeded."
>
>
>
>
>
>
>
> As demonstrated earlier, for a single M7.7 earthquake with
> 500 year
> recurrence interval in the New Madrid seismic zone, ground motion with
> 2,500-year return period (2% PE in 50 years) means there is about 20 percent
> probability that ground motion will be exceeded if the M7.7 earthquake
> occurs. In other words, if the ground motion with 2,500-year return period is
> selected for engineering design, we has a confidence level of 80% (not being
> exceeded) if the M7.7 earthquake occurs.
>
>
>
> Thanks.
>
>
>
> Zhenming
>
>
>
>
> ________________________________
>
>
> From: ceus-earthquake-hazards-bounces at geohazards.usgs.gov
> [mailto:ceus-earthquake-hazards-bounces at geohazards.usgs.gov] On Behalf Of
> Arthur D Frankel
> Sent: Friday, February 15, 2008 1:50 PM
> To: Joe Tomasello
> Cc: ceus-earthquake-hazards at geohazards.usgs.gov;
> mpetersen at usgs.gov
> Subject: Re: [CEUS-earthquake-hazards] reply to Joe
> Tomasello;
> buildings codes and earthquake hazard
>
>
>
>
> Joe,
>
> I feel I need to reply to you on the bulletin board, since
> you seem
> to want to keep this discussion going and you directly ask me questions. My
> apologies to Paul Segal and others who are fed up with this.
>
> I guess you missed one of the key points of my previous
> email. I
> said that the IBC values (2/3 times the ground motions with 2% probability of
> exceedance in 50 years [2%/50]) were probably experienced inMemphis during
> the 1811-12 earthquakes, based on intensity data from those earthquakes.
>
> You have this mistaken notion that IBC values (2/3 times
> the 2%/50
> motions) are only observed once in a millenium. You have the mistaken
> notion that 2%/50 values are only observed once in 2500 years. This is flat
> out wrong. Some locations will experience 2/3 times the 2%/50 ground
> motions during the next 1811-12 type earthquake. Some locations will
> experience the 2%/50 ground motions during the next 1811-12 type earthquake.
> This is a simple consequence of the observed variability of earthquake ground
> motions.
>
> As far as the magnitude issue you bring up, the USGS uses
> the
> range of magnitudes that various seismologists have determined for the
> 1811-12 earthquakes from the intensity data. The central values of these
> moment magnitude determinations range from 7.4-7.5 (Hough et al., 2000 in
> JGR) to 7.8 (Bakun and Hopper, 2004 in BSSA) to 8.0-8.1 (Johnston, 1996 in
> Geophysical Journal). In the national maps we use a logic tree to express
> this range, with a value of 7.7 given the highest weight. I used M7.7 in
> the scenario in my previous email, because it is in the center of the range
> of magnitudes determined for the 1811-12 earthquakes. I also gave results
> for a M7.4 earthquake.
>
> In the example in my previous email, I placed the
> scenario
> earthquake where the current seismicity trend for the New Madrid seismic zone
> is located. The closest distance to downtown Memphis is about 60 km.
>
> -Art
>
>
>
>
> Art Frankel
> U.S. Geological Survey
> MS 966, Box 25046
> DFC
> Denver , CO 80225
> phone: 303-273-8556
> fax: 303-273-8600
> email: afrankel at usgs.gov
>
>
>
> _______________________________________________
> CEUS-Earthquake-Hazards mailing list
> CEUS-Earthquake-Hazards at geohazards.usgs.gov
>
> https://geohazards.usgs.gov/mailman/listinfo/ceus-earthquake-hazards
>
>
>
> _______________________________________________
> CEUS-Earthquake-Hazards mailing list
> CEUS-Earthquake-Hazards at geohazards.usgs.gov
> https://geohazards.usgs.gov/mailman/listinfo/ceus-earthquake-hazards
>
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
More information about the CEUS-Earthquake-Hazards
mailing list