Need for NGA-East: EPRI CENA Ground Motion Report, Summary and Issues

advertisement
Need for NGA-East: EPRI CEUS
Ground Motion Report,
Summary and Issues
Norm Abrahamson
Mar 7, 2008
EPRI 2004 Approach
• Develop weights for existing ground motion
models
• Median and standard deviation are
separated
Regionalization
Followed EPRI (1993)
Model Clusters
•
•
•
•
1
2
3
4
Single Corner Stochastic
Double Corner Stochastic
Hybrid
Finite source / Greens Function
Models Grouped by Class
Cluster
Model Type
Models
1
Single Corner
Stochastic
Hwang and Huo (1997)
Silva et al (2002) - SC-CS
Silva et al (2002) - SC-CS-Sat
Silva et al (2002) - SC-VS
Toro et al (1997)
Frankel et al (1996)
2
Double Corner
Stochastic
Atkinson and Boore (1995)
Silva et al (2002) DC
Silva et al (2002) DC - Sat
3
Hybrid
Abrahamon & Silva (2002)
Atkinson (2001) & Sadigh et al (1997)
Campbell (2003)
4
Finite Source
/Greens Function
Somerville et al. (2001)
Candidate Models, Median M5
10 Hz
1 Hz
Candidate Models, Median M6
10 Hz
1 Hz
Candidate Models, Median M7
10 Hz
1 Hz
Candidate Models, Std Dev
1 Hz
10 Hz
Cluster Evaluation
• Consistency of the cluster median with
CEUS ground motion data
• Strength of the seismological principals
used in the model development
• Degree to which modeling of epistemic
uncertainty was considered in developing
individual ground motion models
EUS Data Used for Model
Evaluations
Consistency with CEUS Data
(Weights inversely proportional to variance)
Cluster
Relative Weight
Single Corner Stochastic
0.3639
Double Corner Stochastic
0.5869
Hybrid
0.0135
Finite Source /Greens
Function
0.0357
Cluster Weights Based on Seismological
Principles and Uncertainty Treatment
Seismological
Principles
Single
Corner
Double
Corner
Hybrid
GF
Medium
Medium
Medium
High
Explicit
modeling of
epistemic
0.40
0.20
0.50
0.00
Only
Parametric
0.60
0.80
0.50
1.00
No
consideration
0.0
0.0
0.0
0.0
Final Weight
0.245
0.221
0.257
0.277
Weights for Clusters
Importance Weights
0.25
Cluster
1
0.75
Consistency Seismological Composite
with data
/Uncertainty
Weight
0.3639
0.245
0.275
2
0.5869
0.221
0.312
3
0.0135
0.257
0.196
4
0.0357
0.277
0.217
Model Weights within a Cluster
Cluster
Model Type
Models
Weights
1
Single Corner
Stochastic
Hwang & Huo (1997)
Silva et al (2002) - SC-CS
Silva et al (2002) - SC-CS-Sat
Silva et al (2002) - SC-VS
Toro et al (1997)
Frankel et al (1996)
0.037
0.192
0.148
0.560
0.029
0.034
2
Double Corner
Stochastic
Atkinson & Boore (1995)
Silva et al (2002) DC
Silva et al (2002) DC - Sat
0.714
0.154
0.132
3
Hybrid
Abrahamon & Silva (2002)
Atkinson (2001) & Sadigh et al (1997)
Campbell (2003)
0.336
0.363
0.301
4
Finite Source
/Greens Function
Somerville et al. (2001)
1.0
Median
by
Cluster
PGA
Median
by
Cluster
T=1 sec
Cluster 1 Stress drops
Model
Median Stress Drop (Bars)
Frankel et al (1996)
150
Hwang and Huo (1997)
150
Silva (2002) Constant Ds
120
Silva (2002) Constant Ds
with saturation
Silva (2002) variable Ds
120
Toro et al (1997)
160 for M<=5.5,
70 for M8.5
120
Adjustments for Epicentral
Distance Based PSHA
• Some PSHA model earthquake as
Epicenters
• Need to allow models to use epicental
distance
• Two corrections needed
– Median RJB (Repi,M), Rrup (Repi,M)
– Increase in standard deviation
Logic Tree
Epistemic:
Stress-Drop
Median
(cluster 1)
Epistemic:
Path
(cluster 1)
Epistemic
Single
Corner
(cluster 1)
Epistemic
Double
Corner
(cluster 2)
Epistemic
Hybrid
(cluster 3)
Epistemic
GF
(cluster 4)
Issues
• Regionalization
– Same two regions as EPRI 1993?
– Rift vs non-rift
• Change in depth distribution
• If RJB is used, then are different models needed?
• Depends on scaling with depth
Regionalization
Regions Considered by EPRI 1993
Issues
• Distance Metric
– Do we need to consider epicenter distance
based models?
– RJB models may need to account for average
depth
Issues
• Comparisons with Data
– Hybrid models based on WUS models that
were not well constrained for M<5, but most
CEUS data is for M<5
Issues
• Finite Source /GF Simulations
– Based on a single model (Somerville et al 2001)
– MCEER project (1997) used multiple methods
and found a large range (factors of 2-3) in the
medians for the different simulation methods
Needs for NGA-East Approach
• Develop new models, with comprehensive
evaluations and reviews
– Revising weights for existing models is not going to
lead to significant improvements
• Where possible, collect/apply new data to
constrain ground motions
– Site conditions for GM stations
– Geological observations for GM constraints
– Ground motions from M>4 stable regions would-wide
• Coordinated application of different numerical
simulation methods
Download