U.S. Department of the InteriorU.S. Geological Survey

CEUS Mean Earthquake Rates & Observations

Steve Harmsen, USGS

NRC-USGS Mmax Workshop

September 8-9, 2008

Figure A9–1

Basic Approach

Compare expected number of background-zone earthquakes in two sub-regions of CEUS with historical data. For expected rates, use 2008 USGS-NSHMP PSHA source model.

Comment on expected rates of M>6 background compared to historical record.

Comment on expected rates of M>7 background sources compared to paleoseismic evidence in those two sub-regions.

These are initial observations. This material is not abstracted from a journal article.

Figure A9–2

What is known about the fault?

Wide range of knowledge about faults Hazardous CEUS faults are Cheraw, Colorado and

Meers, Oklahoma (no location uncertainty) and NMSZ (with location uncert.). Earthquake size (Mchar) can be estimated from evidence.

Charleston, SC is a characteristic source zone. Size (Mchar) can be estimated from evidence.

Background sources often at locations with virtually no evidence for estimating fault size. Methodology assumes that long, capable faults are everywhere (close to every location in the CEUS and WUS).

Figure A9–3

Scope of Analysis

Background sources only No faults, fault zones, characteristic-earthquake

source zones are included in mean rate calculations, E[N].

Historical earthquakes used for tabulation of N on the basis of a declustered USGS catalog. Not explicitly considering range of uncertainty of historic earthquake magnitudes. Comment on the range of reported magnitudes.

Figure A9–4

CUS Region, Radius 970 kmNEUS Region, Radius 700 km

CUS region:Radius 970 km

NEUS region:Radius 700 km

Craton Mmax fault length 41 km

Margin Mmax faultLength 90 km

Figure A9–5

Figure A9–6A

NN

UA

L R

AT

E O

F E

XC

EE

DA

NC

E

MOMENT MAGNITUDE

How Many CEUS Eqs w/M>6 in Last 300 Years? Depends on who you ask. USGS ranked competing estimates,

NCEER estimate came out on top of the heap. Zero earthquakes with m or M>6 in CUS or NEUS circles,

according to declustered USGS catalog if you … Omit NMSZ mainshocks, foreshocks, and aftershocks of 1811-

1812 (handled by fault files, not background). Omit 1843 and 1895 NMSZ earthquakes (all have NCEER M <

5.5, but Johnston and some others give M>6 estimates). Omit Cape Ann November 1755 earthquake. NCEER M is 5.8 (mb

6+, Ebel) Omit Charleston, SC mainshock & aftershocks (outside the

study area). Also, handled by other files, not background. Omit St. Lawrence River-Charlevoix (M6+); 1935 Timiskaming

(M6.1, Bent; MMI VII); Grand Banks (1929, M7.2) (outside study area).

Omit 1897 Giles County, VA. Inferred M was 5.9. MMI VIII.

Figure A9–7

CUS Region, Radius 970 kmNEUS Region, Radius 700 km

CUS region:Radius 970 km

NEUS region:Radius 700 km

300 yearcompleteness

200 year completeness

Figure A9–8

Probability Statement About M>6 Historical Earthquakes

Let mu1 = Expected number of earthquakes having M>=6 in random 300 year period in NEUS circle. Mu1=2.85

Let mu2=Expected number of earthquakes with M>=6 in a random 200 year period in the CEUS circle. Mu2=2.72

Let mu = mu1 + mu2. Poisson Pr[0 M6+ | mu=5.57] = 0.0038 Pr[1 or less | mu=5.57] = 0.025

Figure A9–9

Is the significance test any good?

Only if observed N < 2. Some will claim that omitting Cape Ann,

Timiskaming, and Charlevoix is unfair. Some NMSZ “aftershocks” may be independent

sources. 1843 & 1895 M(NCEER)=5.4; other catalogs say M>=6 for these two earthquakes.

Completeness issues will be argued. Large % of these regions was raw frontier for substantial part of the time intervals.

Last 200 or 300 years may not be typical.

Figure A9–10

Even if test is good

Do not need to reduce MMax to get a non-significant test result

Could reduce predicted earthquake rates in the M6 to M7.5 range by a factor of 2 or so

2007 USGS PSHA in California did reduce background-source rates for M>6.5 sources to get better agreement with historical earthquake rate

Figure A9–11

M>7 Paleoearthquakes 6000 yrs

1 earthquake: Vincennes (M7+, Obermeier). 6100 years Let mu=expected number of M7+ background earthquakes in

the last 6000 years in either CUS or NEUS. Omit NMSZ mainshocks, Meers, OK, and South Carolina coastal plain, which are handled by other files.

From a previous slide, mu=12.5. Pr[1 or less M7+ source | mu=12.5] = 0.000 No CEUS NPP site applications indicate evidence for Holocene

earthquakes with M>=7 near the site. May want to include 1663 Charlevoix earthquake (M approx 7). Pr [2 or less | mu=12.5] = 0.0003 We need 6 or more M7+ paleoearthquakes to produce a non-

significant Poisson test when mu=12.5 (5 gives 0.015 p)

Figure A9–12

Summary

Initial look at these two CEUS regions shows that expected numbers of earthquakes of M>6 or M>7 seem to exceed observations by a significant amount.

Similar tests should be attempted after resolving some magnitude-uncertainty issues, trying better-researched completeness period estimates, and so on.

Figure A9–13