[CEUS-earthquake-hazards] Comment on China quake and hazards map
Julio J. Hernández
julher at cantv.net
Wed May 21 06:24:24 GMT 2008
Dear Zhenming,
considering your categoric response, I have been trying to understand your point of view.
I have reread several of your January and February emails, and the comments by
Schweig, Frankel, Hough, Tomasello and others.
I also read (Wang et al., EOS, 2003), the comment (Wang & Zhou, BSSA, 2007) on the
paper (Bommer & Abrahamson, BSSA, 2006) and their reply to you (BSSA, 2007).
I essentially agree with Bommer & Abrahamson and want to emphasize their last paragraph:
"The mathematical formulation of modern PSHA is correct. The key issues are the models used
for the seismic sources and the ground motion. The seismic-hazard analysis will benefit more from
focusing our energy on improving these models rather than trying to discredit a perfectly valid
approach to assessment of design ground motions."
Why cannot we mix temporal and spatial variabilities? Science is full with such combinations.
We can expect that the best explanation of earthquakes be a spatial-temporal process of the
evolution of the faults, the crust and the Earth.
A simple example concerning our issue. If we have a fault with sole characteristic M7 earthquakes
with recurrence interval = 250 years that lead to PGA ~ 0.4 g on average in a site, then
PGA ~ 0.4 g has a mean return period ~ 500 years. In fact, on average in half of cases
will be PGA > 0.4 g and in the other half will be PGA < 0.4 g. (exactly: > or =, etc).
In 2000 years we expect 8 M7 earthquakes and in 4 of them PGA >= 0.4 g.
Therefore, temporally speaking we have on average PGA >= 0.4 g every 500 years.
A spatial variability combines with a temporal recurrence and we obtain a temporal variability.
This a valid mathematical convolution of variables.
Moreover, the point is not the spatial or the temporal nature but occurrences (in space-time).
We combine simple occurrences (from different origin) and obtain compound occurrences.
But the point is to weigh up the Chinese maps. What is the explanation of PGA = 0.1 g in Chengdu?
If your supposition were right: "This incorrect treatment of ground-motion variability perhaps leads to
increased hazard estimates" (Wang & Zhou, BSSA, 2007) then the correct PGA < 0.1 g !!
Another possibility: they made an old-PSHA without variability of ground motion, similar to the
described ones in (Bommer & Abrahamson, BSSA, 2006); but for MRP= 475 years the
expected reduction of PGA is small (see Figure 4 of that paper). It doesn't explain so low PGA.
I think they used very low source parameters for the near fault. There are diverse possible reasons.
A standard-PSHA with a good seismogenic model should drive PGA >> 0.1 g.
This PSHA should estimate approximately the expected ground motion, never exactly
but taking into account the uncertainties.
Another option? If we use an intensity map with maximum recorded MMI, we still have
the problem of estimating the PGA for an intensity, a historical intensity!
Which are the uncertainties when translating MMI into PGA? Indeed larger.
Therefore, in future we must promote a good PSHA with good source models and updated
attenuation relationships, in order to China have modern maps.
Best regards, Julio J.
_____________________________________________________________________________
----- Original Message -----
From: Wang, Zhenming
To: Julio J. Hernández ; Central and Eastern U.S. Earthquake Hazards Listserve
Sent: Tuesday, May 20, 2008 3:36 PM
Subject: RE: [CEUS-earthquake-hazards] Comment on China quake and hazards map
Dear Julio,
Ground motion variability reflects spatial variation due to earthquake source and wave propagation (path), and is not a temporal measure. Using the ground motion variability in space to extrapolate ground motion occurrence in time is not appropriate. Temporal and spatial measurements of earthquake and its consequence (ground motion) at a site are two intrinsic parameters and should not be mixed one way or the other.
Thanks.
Zhenming
________________________________
From: Julio J. Hernández [julher at cantv.net]
Sent: Tuesday, May 20, 2008 2:39 PM
To: Central and Eastern U.S. Earthquake Hazards Listserve
Cc: Wang, Zhenming
Subject: Re: [CEUS-earthquake-hazards] Comment on China quake and hazards map
Dear Zhenming,
What I mean is that, for example, in an accepted variant of PSHA (Cornell, 1968), one assigns to each source (faults, ....) seismogenic parameters representative of its activity (b-value of Gutenberg-Richter's law, rate of activity for a minimum magnitude M0, maximum magnitude Mu, etc.). Given a fault that can lead to MMI = VIII in some site, those assigned parameters should not lead to PGA = 0.1 g in the same site for MRP = 475 years, unless that event is very rare for sure.
Due to uncertainties, it is not "correct" to assign to this fault so low parameters as if its long term activity were very low with certainty. In my opinion a good seismogenic model should not drive to that PGA in this case. Such a model must take into account geological data, seismological data and macroseismic data.
In several cases using only a kind of data leads to biased models. It is important to place emphasis in multidisciplinary studies.
Usually, when strange results are found the wrong thing is not the PSHA methodology but the input data.
On the other hand, the mean return period (MRP) for a value of PGA does not only depend on the MRP of the possible earthquakes. It also depends on the variability of the attenuation of ground motion. Even in an extremely deterministic scenario, with only a possible magnitude with a known return period in one fault, the PGA has many different possible values, with an expected value and a standard deviation. Obviously, it is possible to get a PGA for a given MRP while none earthquake has the same MRP. And if we have some earthquake with MRP < 475 years, we get a value of PGA with MRP ~ 475 years.
For example, regarding the strange PGA of the Chinese map, even with a low hazard given for a sole M6.5 earthquake (near the site) with MRP ~ 300 years, using the Abrahamson & Silva (1997) attenuation law, in rock at 10 km from rupture, we get PGA ~ 0.36 g for MRP ~ 475 years.
Best regards, Julio J.
_______________________________________________________
Julio J. Hernández
Consultant on Structural Engineering and Earthquake Engineering
Caracas, Venezuela
e-mail: julher at cantv.net<mailto:julher at cantv.net>
cellphone: +58-414-1293989
----- Original Message -----
From: Wang, Zhenming<mailto:zmwang at email.uky.edu>
To: Julio J. Hernández<mailto:julher at cantv.net> ; Central and Eastern U.S. Earthquake Hazards Listserve<mailto:ceus-earthquake-hazards at geohazards.usgs.gov>
Sent: Monday, May 19, 2008 5:02 PM
Subject: RE: [CEUS-earthquake-hazards] Comment on China quake and hazards map
Dear Julio,
There are enough studies and historical records in the area. The problem is not the seismogenic source model used, but PSHA. As you shown, recurrence interval of the large earthquake (like the May 12, 2008, event) is about 350-400 years. 500-year return period derived from PSHA do not really account for the ground motion from the large earthauke.
Thanks.
Zhenming
________________________________
From: Julio J. Hernández [julher at cantv.net]
Sent: Monday, May 19, 2008 2:23 PM
To: Central and Eastern U.S. Earthquake Hazards Listserve
Cc: Wang, Zhenming
Subject: Re: [CEUS-earthquake-hazards] Comment on China quake and hazards map
Zhenming,
First, thank you very much for the maps. (Also thanks to Rogers for the fault-quake maps).
I see the 1956 intensity map assigns a MMI ~ IX in the zone of 2008 earthquake.
Indeed, using this map (+ good structural designs) could save some life.
Regarding the PGA map, I think the problem is not the mean return period for designing,
but the seismogenic source model used for developing the map.
In the slides from Prof. Steve Gow one can see the 1713 earthquake on the fault.
I attach a small map with this earthquake and others (1713, 1748, 1792 and 1793 earthquakes);
data taken from Ganse & Nelson's catalog (BSSA, vol. 72, pp. 873-877, 1992).
According this catalog, the 1713 earthquake reached MMI=VIII and the 1748 earthquake MMI=VII.
Now, according to Wald et al., (EQS, Vol. 15, pp. 537-555, 1999) for a VIII earthq. one can
expect a PGA= 0.34 - 0.65 g, and for a VII earthquake PGA= 0.18 - 0.34 g. Both >> 0.1 g.
The 1713 earthq. happened 295 years ago. Of course, a difficulty is the unknown previous earthquake.
But, consideration of this earthquake must lead to expect a high PGA for a 475 years mean return period.
It is difficult to understand why a PSHA leads to ~0.1 g for this period.
For a site practically on the fault, which seismogenic parameters are supposed for obtaining a such PGA?
Because of these observations, I think the seismogenic source model used was inexplicably wrong.
Regards,
_______________________________________________________
Julio J. Hernández
Consultant on Structural Engineering and Earthquake Engineering
Caracas, Venezuela
e-mail: julher at cantv.net<mailto:julher at cantv.net<mailto:julher at cantv.net<mailto:julher at cantv.net>>
cellphone: +58-414-1293989
----- Original Message -----
From: Wang, Zhenming<mailto:zmwang at email.uky.edu>
To: jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu>> ; Central and EasternU.S. Earthquake Hazards Listserve<mailto:ceus-earthquake-hazards at geohazards.usgs.gov>
Cc: olboyd at usgs.gov<mailto:olboyd at usgs.gov<mailto:olboyd at usgs.gov<mailto:olboyd at usgs.gov>>
Sent: Sunday, May 18, 2008 7:39 AM
Subject: Re: [CEUS-earthquake-hazards] Comment on China quake and hazards map
Klaus,
I agree with you that "Hence this earthquake is VERY pertinent to the discussion of how to map seismic
hazards in regions with long recurrence periods for similar-sized earthquakes, New Madrid included."
Attached is the Chinese national seismic design map (1956 intensity map). This historical intensity map could save some life if it was used.
The problem may be the return period defined in PSHA. 500-year return period derived from a PSHA study is just a numerical number and is not equal to any recurrent interval of earthquake. Comparing return period of PSHA with recurrent interval of earthquake is misleading.
Thanks.
Zhenming
________________________________________
From: jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu<mailto:jacob at ldeo.columbia.edu>> [jacob at ldeo.columbia.edu]
Sent: Friday, May 16, 2008 10:43 PM
To: Central and Eastern U.S. Earthquake Hazards Listserve; Wang, Zhenming
Cc: ceus-earthquake-hazards at geohazards.usgs.gov<mailto:ceus-earthquake-hazards at geohazards.usgs.gov<mailto:ceus-earthquake-hazards at geohazards.usgs.gov<mailto:ceus-earthquake-hazards at geohazards.usgs.gov>>; olboyd at usgs.gov<mailto:olboyd at usgs.gov<mailto:olboyd at usgs.gov<mailto:olboyd at usgs.gov>>
Subject: Re: [CEUS-earthquake-hazards] Comment on China quake and hazards map
All:
Thanks Zhenming for the map.
>From what I can decipher from the map, the region of the M7.9 earthquake is
shown on the 10% in 50 years map as having PGAs between 10 and 20%g. Although I
have not heard any strong motion reports, I would guess that much of the
region, especially on the hanging wall side (the uplands, not in the basin)
may have seen PGA's of 0.5g and possibly larger.
The discrepancies between map values and likely real PGAs are most likely due to
the following:
>From what little I know about the region's geology and seismic history, there is
no earthquakes as large as this one in the historic record. Geological mapping
of faults has (see cooperative work between Burchfield's group at MIT
and many Chinese geologists) seem to NOT have found recently active SURFACE
faulting in the area, and what they found were seemingly older faults largely
with strike slip components (while this quake is largely a thrust with minor
strike slip).
This mapping may have missed that there could be a blind thrust with no surface
fault. But the topographic front looks formidable and needs relatively recent
thrusting/reverse faulting, blind or not blind.
The point is: if there was no large historic quake in the historic record, no
recognized fault with measurable slip rate, and low geodetic strain rates, then
the topography should have been a warning, albeit allowing for VERY LONG
RECURRENCE PERIODS OF SEVERAL 1000 YEARS for events on this fault or thrust
belt. If the recurrence period is this long, then it is hardly surprising that
on the hazards map for 10% in 50 years (average recurrence period 475 years)
this does zone does not show up very prominently.
This is exactly the reason why some time ago the US NEHRP hazard maps (i.e.
USGS maps) started to portray 2% in 50 years (2475 years recurrence period), to
catch regions like this with reasonably "safe" ( speak high) PGA values, or
sufficiently high spectral acceleration values for building code applications
(like in the CEUS).
Hence this earthquake is VERY pertinent to the discussion of how to map seismic
hazards in regions with long recurrence periods for similar-sized earthquakes,
New Madrid included.
Of course the tectonics is entirely different, but there are lessons to be
learned, and pertinent to the issues we all discussed with such passion in this
forum (but on different sides of the fence) a few month ago.
Of course there is the other issue about seismic building design and quality
control) or lack of both, but if there would have been full quality control,
the 10% in 50 year map PGA values don't provide sufficient protection.
And that is the lesson of this Eq. for the CEUS, and perhaps China will change
its code after this quake to longer recurrence periods, closer to what we have
for the US right now (2% in 50 y, or at least 2/3 of these hazard levels !!!!).
Best
Klaus Jacob
==================
Quoting "Wang, Zhenming" <zmwang at email.uky.edu<mailto:zmwang at email.uky.edu<mailto:zmwang at email.uky.edu<mailto:zmwang at email.uky.edu>>>:
> I would like to share with you all the Chinese National seismic design map
> (PGA with 10% PE in 50 years). This may explain one of the reasons that so
> many schools and hospitals collapsed in the Wenchuan earthquake.
>
> Thanks.
>
> ___________________________________
> Zhenming Wang, PhD, PE
> Head, Geologic Hazards Section
> Kentucky Geological Survey
> 228 Mining and Mineral Resources Building
> University of Kentucky
> Lexington, Kentucky 40506
> Phone:(859)257-5500x142
> Email: zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu<mailto:zmwang at uky.edu>>>
> Website:
> www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards<http://www.uky.edu/KGS/geologichazards>>>
> ____________________________________
>
>
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
________________________________
_______________________________________________
CEUS-Earthquake-Hazards mailing list
CEUS-Earthquake-Hazards at geohazards.usgs.gov<mailto:CEUS-Earthquake-Hazards at geohazards.usgs.gov<mailto:CEUS-Earthquake-Hazards at geohazards.usgs.gov<mailto:CEUS-Earthquake-Hazards at geohazards.usgs.gov>>
https://geohazards.usgs.gov/mailman/listinfo/ceus-earthquake-hazards
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://geohazards.usgs.gov/pipermail/ceus-earthquake-hazards/attachments/20080521/b0ea3e88/attachment-0001.html
More information about the CEUS-Earthquake-Hazards
mailing list