US – Japan Collaboration in
Ground Motion Simulation
Hiroshi Kawase
JSPS - Kyoto University
Paul Somerville and Jeff Bayless
SCEC - URS
Topics
• JSPS-SCEC comparison of strong motion
prediction techniques for disaster risk
evaluation
– Empirical Method
– Simulation Method
• Implementation of the Irikura Recipe for
Strong Motion Simulation on the SCEC
Broadband Platform
Comparison of Strong Motion Prediction
Techniques - Empirical Method
• We will select data for the commonly agreed
magnitude and distance ranges from the PEER NGA
database, and obtain average attenuation
characteristics for peak acceleration, peak velocity,
and response spectral acceleration using the method
of Sato (2010)
• The median ground motion levels and standard
deviations will be compared with the Japanese
results. The comparison will focus on event terms
and site amplification factors.
Comparison of Strong Motion Prediction
Techniques - Simulation Method
• We will select existing strong ground motion simulation
results for standard earthquake scenarios that have already
been done in Southern California and Japan.
• These may include the suites of broadband simulations that
we done for the Tall Buildings Initiative.
• We will exchange all the source parameters that were used in
generating the simulations.
• We will then repeat the simulations of the scenarios using the
other side’s source parameters in our simulation method
• These simulations will be done on the SCEC Broadband
Platform.
Comparison of Strong Motion Prediction
Techniques - Simulation Method (cont.)
• We will compare the average ground motion levels
and standard deviations of the simulated strong
motions obtained by each side, and with the
empirical ground motion models for each country
• If the comparisons reveal that significant bias exists
in the simulated ground motions, we will seek the
source of the bias and delineate the issues that we
must resolve for further improvement in strong
motion simulation techniques.
Implementation of the Irikura Recipe on
the SCEC Broadband Platform
• The Irikura Recipe is used in the seismic licensing of
nuclear power plants in Japan, and in the
development of scenario ground motion maps by the
Earthquake Research Committee.
• The following presentation is based on a
presentation prepared by Kojiro Irikura and Hiroe
Miyake
• Irikura, K. and H. Miyake (2011). Recipe for Predicting
Strong Ground Motion from Crustal Earthquake
Scenarios, Pure Appl. Geophys. 168 (2011), 85–104.
DOI 10.1007/s00024-010-0150-9.
Logistics for Implementation of the
Irikura Recipe on the SCEC Platform
• Download the code and its Japanese language
documentation from NIED (Fujiwara)
• Code is professionally written (Mitsubishi
Aerospace)
• Find a SCEC grad/postdoc to translate the
documentation into English
• SCEC – URS interaction on implementation
and testing
• URS verification against Japanese scenario
GMPE vs. Broadband Simulation for Scenario Earthquake:
Itoigawa-Shizuoka Tectonic-Line Earthquake
↓Ground Motion Prediction Equation
↑
Ground Motion Simulation by
the Hybrid Method (Irikura Recipe)
Earthquake Research Committee (2002)
Hybrid Method for Ground Motion Time History
stochastic Green’s function method (SGF)
Matching filter
crustal: ~ 1 s
subduction: > 2 s
Computational codes:
Long-period: Graves (1996), Aoi and Fujiwara (1999), Pitarka (1999), and etc.
Short-period by SGF: Kamae et al. (1991), Dan and Sato (1999)
Recipe for Strong Ground Motion Prediction
Outer Fault Parameters
Rupture area S is given.
Seismic moment Mo from the empirical relation of Mo-S.
Average static stress-drop Dsc from appropriate physical model
(e.g., circular crack model, tectonic loading model, etc.)
Inner Fault Parameters
Combined area of asperities Sa from the empirical relations of S-Sa
or Mo-Ao.
Stress drop on asperities Dsa based on the multiple asperity model.
Number of asperities from fault segments.
Average slip of asperities Da from dynamic simulations.
Effective stress for asperities sa and background area sb are given.
Slip velocity time function given as Kostrov-like function.
Extra Fault Parameters
Rupture nucleation and termination are related to fault geometry.
Irikura and Miyake (2001, 2011)
Source Characterization & Scaling
Mo-S: Outer Scaling (Fault Area)
Slip inversion
Outer Fault Parameters
S(=LW), Mo, Dsc
Mo-Sa: Inner Scaling (Asperity Area)
Inner Fault Parameters
Sa, Dsa , etc
Somerville et al. (1999)
Characterized Source Model
Miyake et al. (2003)
Characteristics of the Recipe
• Easy to link active fault studies.
• To adopt the characterized source model, not k-squared.
• To avoid the usage of random parameters, everybody can
reproduce ground motion time histories.
• To adopt the stochastic Green’s function method
providing ground motion time histories following the
omega-squared source model in a broadband period
range (e.g., 0.1-20 s).
• Well calibrated to match GMPEs for past earthquakes.
Empirical Relationships for the Recipe (1)
- Seismic Moment (Mo) vs. Total Rupture Area (S) -
inland crustal eq. (Wmax=20km)
subduction eq. (Wmax=100km)
Empirical Relationships for the Recipe (2-2)
- Mo vs. Acceleration Source Spectral Level (Ao) -
Theoretical relationship shows Aoa ∝ Mo1/3
A0  4   v r  s a 
a
Sa 
7
4
(from Madariaga, 1977; Boatwright, 1988)
2
vR 
const.
Empirical relationship shows Ao ∝ Mo1/3
S ∝ Mo2/3
M0
S 
Sa
Sa ∝ Mo2/3
Outer Fault Parameters
Parameters characterizing entire source area
Inland crustal earthquake
Step 1: Give total rupture area (S=LW)
Fault length (L) is related to grouping of active faults from geological and
geomophological survey.
Fault width (W) is related to thickness of seismogenic zones.
Step 2: Assume average static stress-drop (Dsc) on the fault
(about 2.3MPa from the empirical relationship by Somerville et al. (1999))
Step 3: Estimate total seismic moment (Mo)
M0 
16
from S and Dsc assuming
1 .5
7

a circular-crack model (Eshelby, 1957) for smaller L
or a loading model (Fujii and Matsu’ura, 2000) for larger L.
Ds c  S
1 .5
Inner Fault Parameters
Slip heterogeneity or roughness of faulting
Inland crustal earthquake
Step 4: Estimate combined area of asperities (Sa)
from empirical relation Sa-S
(Somerville et al., 1999; Irikura and Miyake, 2001, 2011 )
Sa/S = 0.22
Sa: combined area of asperities (inner)
S : total rupture area (outer)
Step 5: Estimate Stress Drop on Asperities (Dsa)
from multi-asperity model (Madariaga, 1979)
D s a  Ds c 
S
Sa
Dsa: stress drop on asperity (inner)
Dsc: average stress drop (outer)
Inner Fault Parameters
Slip heterogeneity or roughness of faulting
Inland crustal earthquake
Step 6: Estimate number of asperities (N): The asperities in the
entire fault rupture are related to the active-fault segments location
 from surface offsets measured along fault
Step 7: Estimate average slip on asperities (Da) based on Step 6 and
empirical relationships from dynamic simulations
(ex. N=1  Da/D=2.3, N=2  Da/D=2.0, N=3  Da/D=1.8)
reference: average Da/D = 2.0 (Somerville et al., 1999)
Extra Fault Parameters
Propagation pattern of rupture
Rupture starting point
Rupture propagation pattern
Rupture velocity
Inland crustal earthquakes
 Rupture nucleation and termination are
related to geomorphology of active
faults
Subduction-zone earthquakes
 Information from historical earthquakes
Nakata et al. (1998)