mapping earthquake hazard

Mapping Earthquake Mapping Earthquake Hazard in the UnitedHazard in the UnitedHazard in the United Hazard in the United

StatesStates

David M. BooreU.S. Geological Survey

A talk presented to the Geology Department, Oberlin College, Oberlin, Ohio, 9 March 2004College, Oberlin, Ohio, 9 March 2004

I b hIn both Canada and

the U.S., ,damaging

earthquakes areare

experienced over much of th tthe country,

making seismic hazard a

national issue

• From G. Atkinson

USGS National Seismic Hazard M 2002 U d tMaps: 2002 Update

A. Frankel, M. Petersen, C. Mueller, K Haller R Wheeler E LeyendeckerK. Haller, R. Wheeler, E. Leyendecker, R. Wesson, S. Harmsen, C. Cramer,

D Perkins and K RukstalesD. Perkins, and K. RukstalesU.S. Geological Survey

C lif i d d j i tl ithCalifornia maps produced jointly with California Geological Survey:

T Cao and W BryantT. Cao and W. Bryant

What’s Ahead?

• What is “earthquake (seismic) hazard’’?• Response spectrum: the measure of ground shaking

that is mapped • Mapping the hazard

– seismicity (with special attention to New Madrid)seismicity (with special attention to New Madrid)• where do earthquakes occur?• how often do they occur?• how large are they?

– ground motion• specify the ground shaking as a function of earthquake size

and distance from a siteti th h d l t b d– computing the hazard values to be mapped

– results• Paleoseismometry: precarious rocks

“Civilisation exists by geological consent, subject

h i h ito change without prior notice ”notice.

William Durant, historian

SEISMIC HAZARD -SEISMIC HAZARD -

the possibility of that consentthe possibility of that consent being withdrawn.g

SEISMIC HAZARDSEISMIC HAZARD is the possibility of

potentially destructiveearthquake effectsearthquake effects

occurring at a particular l ti ithi ifi dlocation within a specified

period.p

Earthquake EffectsFAULTINGSURFACE

RUPTURE TSUNAMISFAULTINGSURFACE RUPTURE TSUNAMISFAULTINGSURFACE RUPTURE TSUNAMIS

SEISMIC ENERGY RELEASE



LANDSLIDES LIQUEFACTIONLANDSLIDES LIQUEFACTIONLANDSLIDES LIQUEFACTION

STRONG GROUND MOTION

TOPOGRAPHIC RELIEF

SOIL DEPOSITSSTRONG GROUND MOTION

TOPOGRAPHIC RELIEF

SOIL DEPOSITSSTRONG GROUND MOTION

TOPOGRAPHIC RELIEF

SOIL DEPOSITS

AMPLIFIED GROUND SHAKING






RISK

HAZARD is not RISK

RISK =HAZARD * EXPOSURE * VULNERABILITYHAZARD EXPOSURE VULNERABILITY

The hazard is controlled by NatureThe hazard is controlled by Nature.

Vulnerability and Exposure are controlled by humans.

Seismic Risk Mitigation

HAZARD * EXPOSURE * VULNERABILITY COST

Seismic Risk Mitigation

HAZARD * EXPOSURE * VULNERABILITY = COST

AAssess

ControlReduce

Control

Balance

2003 San Simeon earthquake (M 6.5): 2 deaths in Paso Robles

0 480.48g

Damage in Paso Robles, CA, due to collapse of unreinforced masonry building (2 lives lost) during the 2003 San Simeon earthquake

•Contrast damage with that in Bam Iran (M 6 6)in Bam, Iran (M 6.6)•>30,000 deaths•Why so many deaths, compared to Paso Robles?

•Was ground motion higher than in Paso Robles? (0.98g pga in Bam; 0 48g 10 km fromBam; 0.48g 10 km from Paso Robles)•Was the construction less earthquake

?resistant?

0.98gg

Measures of ground motion forMeasures of ground motion for engineering purposes

• Peak motions (acceleration, velocity, displacement)

• Elastic response spectrap p

Elastic response spectraElastic response spectra

ÄuÄug

convert displacement spectrum into acceleration spectrum (multiply by (2π/T)2)

100

10

100

10

(cm

/s2 )

1

10

cem

ent

(cm

)

0 1

1

Acc

eler

atio

n

0.01

0.1

Rel

ativ

eD

ispl

ac

0 1 1 10 100

0.01

0.1

0 1 1 10 100

10-4

0.001

R

1999 Hector Mine Earthquake (M 7.1)

station 596 (r= 172 km), transverse component

0.1 1 10 100Period (sec)

0.1 1 10 100Period (sec)

pick off values of SA at 0.2 sec and 1 sec

100

10

100

10

(cm

/s2 )

1

10

cem

ent

(cm

)

0 1

1

Acc

eler

atio

n

0.2 sec 1 sec0.01

0.1

Rel

ativ

eD

ispl

ac

0 1 1 10 100

0.01

0.1

0 1 1 10 100

10-4

0.001

R



0.1 1 10 100Period (sec)

0.1 1 10 100Period (sec)

fit functions through values to form an approximate response spectrump p

100

10

100

SA = SA(0.2 s)

10

(cm

/s2 )

1

10

cem

ent

(cm

) SA = SA(1.0 s)/T

0 1

1

Acc

eler

atio

n

0.2 sec 1 sec0.01

0.1

Rel

ativ

eD

ispl

ac

0 1 1 10 100

0.01

0.1

0 1 1 10 100

10-4

0.001

R



similarity of SA(0.2) and SA(1.0) a coincidence here!

0.1 1 10 100Period (sec)

0.1 1 10 100Period (sec)

y ( ) ( )

U.S. National Seismic Hazard Map – 2002 Edition

Some Major Uses of the National SeismicSome Major Uses of the National Seismic Hazard Maps and Associated Products

• Building codes: International Building Code, International Residential Code, ASCE national design load standard, NEHRP ProvisionsD i f hi h b id d l dfill• Design of highway bridges, dams, landfills

• Loss estimation (e.g., HAZUS), earthquake iinsurance

• Emergency management, EQ scenarios

USGS Seismic Hazard Maps (1996 andUSGS Seismic Hazard Maps (1996 and update in 2002)

• Consensus of experts: regional workshops (CEUS 1995, 2000), external review panel, open review of interim maps on Internet

• Average hazard estimate, not worst case; d lt ti d ti di tiused alternative ground-motion prediction

equations and fault locations; uncertainty estimates published in 1997 2000 2001estimates published in 1997, 2000, 2001

Probabilistic Seismic Hazard Analysis (PSHA)

•• SeismicitySeismicity: for each spatial point, assign the probability of an earthquake with particular magnitude occurring each yearearthquake with particular magnitude occurring each year (consider all magnitudes in a range from small to large).

•• Ground motionGround motion: for a spatial point, compute the probability that a level of ground motion will be exceeded considering alla level of ground motion will be exceeded, considering all surrounding points as potential sources (each magnitude and distance can be thought of as a scenario).

Combine probabilities to obtain a frequency of exceedance for• Combine probabilities to obtain a frequency of exceedance for each scenario.

• Add frequencies of exceedance for a particular level of ground ( )motion (combining all scenarios). This gives the HAZARD HAZARD

CURVECURVE

S i i itS i i itSeismicitySeismicity

1. Identify the potential sources of future earthquakes

2. Estimate the maximum magnitude (Mmax) earthquake that ld ithi hcould occur within each source

3. Calculate the recurrence relationship that defines how frequently, on average, earthquakes of different magnitude q y g q goccur within each source.

Divide the US into WUS and CEUS

S A d f lt C i Pl iSan Andreas fault– Carrizo Plain(taken from a radio-controlled kite; see http://quake.usgs.gov/kap/carrizo/)

Faults (Fairview Valley, NV)

Note the flimsy cabin and stovepipe; does this say anything about the strength of ground shaking?

Intraplate Earthquakes

• The driving forces, and stress fields, that are characterized by intraplate earthquakes are difficult to characterize and vary widely

• One example mechanism for intraplate h k i i d i h l i learthquakes is stress associated with post-glacial

reboundSt t ti d k “f il d” ift• Stress concentrations and weak “failed” rifts another possibility

Specification of seismicityseismicity for the National Seismic Hazard MapsSeismic Hazard Maps

• 1 Use spatially-smoothed historic seismicity;1. Use spatially smoothed historic seismicity; assumes that moderate and large earthquakes will occur near previous M3+ events

• 2. Use large background zones based on broad geologic criteria; addresses non-stationary seismicity; quantifies hazard in areas with littleseismicity; quantifies hazard in areas with little historic seismicity but potential for damaging earthquakes

• 3. Use specific fault sources with recurrence rates determined from geologic slip rates, trenching st dies or paleoliq efaction datesstudies, or paleoliquefaction dates

Direct Inputs to Hazard Maps

• Earthquake catalogs (instrumental and historic)• Fault data (geologic slip rates, dates of past (g g p , p

events from trenching, fault geometry, etc.)• Effects of prehistoric earthquakes: p q

paleoliquefaction (New Madrid, Charleston, Wabash Valley), subsidence and uplift (Cascadia, Seattle flt)

• Geodetic data (NV-CA, Puget Lowland)

Note linear pattern of New Madrid

seismicity – but no f f l isurface faulting

found

New Madrid seismicity believed related to buried riftNew Madrid seismicity believed related to buried rift faults (under several km of overlying sediments)

The Smoking Guns for New Madrid EarthquakesEarthquakes

• 1811-12: three largest earthquakes felt as far away N E l d d i i i 9 10 ias New England, producing intensity 9-10 in

Memphis, very large liquefaction area• between 1300 and 1600 A D : sequence of three• between 1300 and 1600 A.D.: sequence of three

large earthquakes with similar liquefaction area as 1811-12 (Tuttle and Schweig)

• between 800 and 1000 A.D.: sequence of three large earthquakes with similar liquefaction area as 1811 12 (Tuttle and Schweig)1811-12 (Tuttle and Schweig)

• also: M6.6 earthquake in 1895 in Charleston, MO; M6 in 1843 in Marked Tree, AR; history of M5.1M6 in 1843 in Marked Tree, AR; history of M5.1 and smaller events since 1900

GroundGround--Motion Prediction EquationsMotion Prediction Equations

Gives mean and standard deviation of 1.0

response-spectrum ordinate (at a particular frequency)

f ti f kA

ccel

(g)

e:06

:53:

50

as a function of magnitude distance, site conditions, and perhaps other

0.1

rizon

talP

eak

ND

NR

.dra

w;

Dat

e:20

04-0

3-08

;Tim

e

perhaps other variables.

Larg

erH

o

1992 Landers M = 7 3

File

:C:\m

etu_

03\re

gres

s\B

JFLN

0.01

1992 Landers, M = 7.31994 NR, M=6.7 (reduced by RS-->SS factor)Boore et al., Strike Slip, M = 7.3, NEHRP Class D_+

10-1 1 101 102

Shortest Horiz. Dist. to Map View of Rupture Surface (km)

Deriving the Equations

• Regression analysis of observed data if have adequate observations (rare for most of the world).

• Regression analysis of simulated data for regions with inadequate data (making use of motions from

ll if il bl i dismaller events if available to constrain distance dependence of motions).H b id th d t i l ff t• Hybrid methods, capturing complex source effects from observed data and modifying for regional differencesdifferences.

8Western North America

Observed data adequate

6

7

entM

agni

tude

Obse ed data adequatefor regression exceptclose to large ‘quakes

5

6

Mom

e

Used by BJF93 for pga 3-09

-05;

Tim

e:14

:40:

01

1 10 100 10008

Eastern North America

_wna

_ena

_pga

.dra

w;

Dat

e:20

03

6

7

entM

agni

tude

File

:C:\m

etu_

03\re

gres

s\m

_d_

Observed data not adequate for regression,

5

6

Mom

e

AccelerographsS i hi St ti

use simulated data

1 10 100 1000Distance (km)

Seismographic Stations

Higher ground motions for given Magnitude, Distance for CEUS Earthquakes Compared with

WUS• Higher Q in crust: less attenuation with distance• Higher earthquake stress drop: more high-frequency

ground motion for specified moment magnitudeground motion for specified moment magnitude• Determined from instrumental analysis of small and

moderate events in CEUS and isoseismals of large ghistoric events

Distance-decay of regional shear waves determinedby Benz, Frankel, and Boore (1997)

Fits using magnitude-independent stress drop, omega –2 model

From Frankel (1994)

T = 0.2 sec

1

ion

(g)

M = 7.6

0.1

ectra

lAcc

eler

at

M = 5.6

0.01Spe

ENA: Atkinson and Boore, 1995ENA: Frankel et al., 1996WNA: Boore et al, 1997

10 20 30 100 2000.001

Distance (km)

How do we estimate ground motions for large earthquakes near New Madrid?

• use estimated magnitude to calculate ground motions f i d ti di ti tifrom various ground-motion prediction equations: stochastic models using source parameters and derived for small earthquakes; constant stress drop with magnitude model validated with felt area vs magnitudemagnitude model validated with felt area vs. magnitude data; in 2002 added two corner frequency model, hybrid extended-source model, and semi-empirical modelAtki d B (1998) d di ti ith• Atkinson and Boore (1998) compared predictions with regional ENAM data

• check with recorded ground motions of Bhuj, India earthquake

Hazard Methodology ExampleExample

a bEarthquake SourcesGround motion

r1

Hazard curvea

erat

ion

d1

d2

d3

d4

r1

r2

high seismicityM 7.6

k gr

ound

acc

elM7.6

r3

high seismicityzone

peak ground acceleration (pga)0.25gdistancepe

ak

0.5g

(AfterAnnual probability that earthquake occurs:

Source B 1/200=0.005

Source A Site

1/10=0.10

After Wang et al., 2003

2T=0.2 sec

1

POE = 0.16

POE = 0.50

"POE"= Probability of Exceeding the value of SA

0.1

0.2

SA

(g)

POE = 0.84

0

M = 5.5

1 2 10 20 1000.02

Distance (km).

2T=0.2 sec

1

POE = 0.16

POE = 0.50


0.44g

0.1

0.2

SA

(g)

POE = 0.840.26g

0.16g

0

M = 5.5

1 2 10 20 1000.02

Distance (km).

2T=0.2 sec

1

POE = 0.16

POE = 0.50


0.44g

0.1

0.2

SA

(g)

POE = 0.840.26g

0.16g

0

M = 5.5

"FOE"= Frequency of Exceedance (combining eq & sa prob)

For 0.44g, FOE = 0.10*0.16 = 0.016

1 2 10 20 1000.02

Distance (km).

2T=0.2 sec

1

POE = 0.16

POE = 0.50


0.44g

0.1

0.2

SA

(g)

POE = 0.840.26g

0.16g

0

M = 5.5


For 0.44g, FOE = 0.10*0.16 = 0.016For 0.26g, FOE = 0.10*0.50 = 0.050

1 2 10 20 1000.02

Distance (km).

2T=0.2 sec

1

POE = 0.16

POE = 0.50


0.44g

0.1

0.2

SA

(g)

POE = 0.840.26g

0.16g

0

M = 5.5


For 0.44g, FOE = 0.10*0.16 = 0.016For 0.26g, FOE = 0.10*0.50 = 0.050For 0.16g, FOE = 0.10*0.84 = 0.084

1 2 10 20 1000.02

Distance (km).

0.1

T = 0.2 sec

0.1

T = 0.2 sec

0.01ar

)Source A (M = 5.5, D = 10 km)

0.01ar

)Source A (M = 5.5, D = 10 km)

10-4

0.001

FOE

(per

yea

10-4

0.001

FOE

(per

yea

10-510-5

0 0.1 0.2 0.3 0.4 0.5 0.610-6

SA (g)0 0.1 0.2 0.3 0.4 0.5 0.6

10-6

SA (g)

0.1

T = 0.2 sec

Sum (Total Hazard)

0.01ar

)Source A (M = 5.5, D = 10 km)

10-4

0.001

FOE

(per

yea

Source B (M = 7.5, D = 50 km)

10-5

0 0.1 0.2 0.3 0.4 0.5 0.610-6

SA (g)

0.1

T = 1.0 sec

0.01ar

)

10-4

0.001

FOE

(per

yea

10-5

Source A (M = 5.5, D = 10 km)

Source B (M = 7 5 D = 50 km)

0 0.1 0.2 0.3 0.4 0.5 0.610-6

SA (g)

Source B (M 7.5, D 50 km)

0.1

T = 1.0 sec

0.01ar

)Sum (Total Hazard)

10-4

0.001

FOE

(per

yea

10-5

Source A (M = 5.5, D = 10 km)

Source B (M = 7 5 D = 50 km)

0 0.1 0.2 0.3 0.4 0.5 0.610-6

SA (g)

Source B (M 7.5, D 50 km)

0.1

T = 1.0 sec

0.01ar

)Sum (Total Hazard)

Specified FOE for hazard

10-4

0.001

FOE

(per

yea Specified FOE for hazard

10-5

Value of SA for map

0 0.1 0.2 0.3 0.4 0.5 0.610-6

SA (g)

Why the different sensitivity of T=1 s and T=0.2 s hazard to magnitude? Ground motionhazard to magnitude? Ground motion.

T = 1 0 sec T = 0 2 sec

1

g)

T = 1.0 sec T = 0.2 sec

0.1

Acc

eler

atio

n(g

M = 7.5M = 7.5

M = 5.5

0.01

Spe

ctra

lA

Western U S : Boore et al 1997

M = 5.5

Western U S : Boore et al 1997

10 20 30 1000.001

Distance (km)

Western U.S.: Boore et al, 1997

10 20 30 100Distance (km)

Western U.S.: Boore et al, 1997

1996 map (PGA with 2% PE in 50 yr)with seismicity and faultsw se s c y d u s

http://www.ohiodnr.com/OhioSeis/

Spectral response acceleration for 0.2 sec spectral ordinate

2% probability of exceedance in 50 yearsGround Motion

Maximum Considered Earthquake Ground Motion

%g30 - 4040 - 6060 - 8080 - 100100 - 125125 - 150150 - 175175 - 200175 - 200200 - 300> 300

constant value 150% g

0 50 100 150 20025Kilometers

Precarious Rocks

Jim Brune, U. Nevada Reno

Brune, 2000Near Antelope Valley fault, Walker, California (Brune, 2000)

Near Antelope Valley fault, Walker, California (Brune, 2000)

Q asi static toppling force F mg tanQuasi-static toppling force: F = mg tan α

Anooshehpoor et al., 2004

Preliminary Conclusions from Study ofPreliminary Conclusions from Study of Precarious Rocks

• Strong asymmetry in ground shaking from reverse faults (low on footwall side, high on hanging wall side)side)

• Ground motions for normal faults smaller than predicted by standard equationspredicted by standard equations

• Ground motions near San Andreas fault in S. California smaller than shown by hazard maps (!)

SUMMARYSUMMARY

• Defined hazard• Described response spectra • Basis for hazard maps: seismicity• Basis for hazard maps: ground motion• Mapping hazard

R l• Results• Paleoseismometry: precarious rocks

Stacy and Dave studying geology in the Dolomites, Italy, in January during Winter term

Brune, 2000

ENA: Atkinson and Boore, 1995ENA: Frankel et al., 1996

T = 1.0 sec

1

ion

(g)

,WNA: Boore et al, 1997

M = 7.6

0.1

ectra

lAcc

eler

at

M = 5.6

0.01Spe

10 20 30 100 2000.001

Distance (km)

ConclusionsConclusions

• USGS hazard maps are based on consensus of h d i fexperts; represent average hazard estimates from

alternative models; best maps for policy and design decisionsdesign decisions

• USGS hazard maps are derived from observations of past earthquakes in NM (1811-12, about 1500 and 900 A.D.), historical seismicity, geology, and models of ground motions for the region that have been validated with observed ground motions andbeen validated with observed ground motions and intensities

• Design maps need to have consistent rules for g pentire U.S.

GeologyGeologySeismic historySeismic history

EarthquakeEarthquakeL M d lL M d lSeismic historySeismic history

Earthquake SourcesEarthquake SourcesLoss ModelLoss Model

GroundGround--motion prediction equationsmotion prediction equationsand soil typeand soil typeOccurrenceOccurrence

ShakingShaking

Structural vulnerabilitiesStructural vulnerabilities

DamageDamage BuildingBuildingggInventoryInventory

VulnerabilityVulnerability$ Loss$ Loss

Hazard

Population

“Risk”

Seismic HazardSeismic Hazardshaking irrespective of consequence

Seismic RiskHazard * Exposure * VulnerabilityHazard * Exposure * Vulnerability

hazard * exposure * vulnerability = riskBaffin Island high low low

Vancouver high high high

Toronto low high moderate

TWO-FACTOR APPROACH TO CONSTRUCTING GROUND MOTION RESPONSE SPECTRAGROUND MOTION RESPONSE SPECTRA

GroundGround--Motion Prediction EquationsMotion Prediction Equations

Gives mean and standard deviation of response-deviation of responsespectrum ordinate (at a particular frequency) as a f ti f it dfunction of magnitude distance, site conditions, and perhaps other variables.

(E. Field)

1.0)

eak

Acc

el(g

)

06;T

ime:

14:2

6:36

0.1

Hor

izon

talP

\BJF

LND

NR

.dra

w;

Dat

e:20

03-0

9-0

Larg

er

1992 Landers, M = 7.31994 NR M=6 7 (reduced by RS >SS factor)

File

:C:\m

etu_

03\re

gres

s\

1 1 2

0.01

1994 NR, M=6.7 (reduced by RS-->SS factor)BJF, Strike Slip, M = 7.3, NEHRP Class DBJF, + - \5s

10-1 1 101 102

Shortest Horiz. Dist. to Map View of Rupture Surface (km)

8Western North America

Observed data adequate

6

7

ntM

agni

tude

Obse ed data adequatefor regression exceptclose to large ‘quakes

5

6

Mom

e

Used by BJF93 for pga -05;

Tim

e:14

:41:

02

1 10 100 1000

Used by BJF93 for pga

8World

na_b

jf_pe

er_p

ga.d

raw

;D

ate:

2003

-09

7

ntM

agni

tude

File

:C:\m

etu_

03\re

gres

s\m

_d_w

n

New recordings help fill in lack of data close to large ‘quakes (but can data be

5

6

Mom

en

PEER NGA (preliminary)

Includes1999 Chi-Chi (M 7.6)1999 Kocaeli (M 7.5)1978 Tabas (M 7.4)1986 Taiwan (M 7.3)1999 D (M 7 1)

q (used?)

1 10 100 1000Distance (km)

PEER NGA (preliminary) 1999 Duzce (M 7.1)

mapping earthquake hazard

Education