RECENT DEVELOPMENTS IN STRONG MOTION SIMULATIONS FOR CEUS

1
RECENT DEVELOPMENTS IN STRONG MOTION SIMULATIONS
FOR CEUS

• Paul Somerville and Robert Graves URS Pasadena

MOTIVATION CEUS ground motion models are based
on simulations validated against sparse data
2
OUTLINE

• Strong ground motion simulation validation
• Seismic wave propagation
• Effect of crustal structure on ground motions
• Review of EPRI ground motion modeling
• Review of CEUS ground motion model (USGS), and
  modeling of NSN data (NRC)
• Improvement of simulations and models

3
NGA-E
NGA-East
4
Broadband Ground Motion Simulation Elastodynamic
Representation Theorem Ground motion U(t) can
be calculated from the convolution of the slip
time function D(t) on the fault with the Green's
function G(t) for the appropriate distance and
depth, integrated over the fault rupture
surface U(t) ? D(t) G(t) Combine long
period and short period simulations to generate
broadband time history
5
Validation of Ground Motion Simulations Against
Strong Motion Recordings

• Measure the difference between recorded and
  simulated ground motion (response spectral
  residuals)
• Bias Do the simulations systematically under-
  or over-predict the recorded ground motions?
• Standard Error What is the average difference
  between recorded and simulated values?

6
Validation against Arleta recording of Northridge
earthquake
Recording Simulations
Validation against a large set of Northridge
earthquake recordings
7
Validation against WUS earthquakes
8
Wave Propagation

• Critical reflections from the lower crust
• Contributions of different wave arrivals to
  ground motion attenuation

9
Critical Reflections from the Lower Crust
Direct, upgoing becomes weak with
increasing distance
Reflected, downgoing becomes strong beyond the
critical distance
30 km
40 km
10
Data
Simulation
Direct (S) fades as downgoing (ScS, SmS)
strengthens
Long duration due to scattering
This simulation has no scattering
11
Direct (upgoing) waves are strongest at close
distances but their amplitudes fade at larger
distances
Reflected (downgoing) waves are weak at close
distances but dominate at larger distances
(beyond about 50 km)
12
1988 Saguenay earthquake
13
Harvard data and simulation
Simulations work well at long periods
SmS etc.
sSms, SmSSmS
recorded
simulated
14
(No Transcript)
15
Contributions of Different Waves
Individual body waves
Combination
Complete simulation
16
Review of EPRI (1993) Ground Motion Modeling

• Survey of crustal structure in CEUS
• Simulation analysis of the effect of crustal
  structure on ground motion attenuation

17
Grenville
N. Appalachian
granite Conrad interface gabbro Moho interface
18
Crustal Structure Regions
Northern Appalachian
Grenville
19
Simulated attenuation in 16 different regions
20
Crustal Structure Regions
Northern Appalachian
Grenville
21
Effect of Crustal Structure
Eq Depth 7 km
Eq Depth 11 km
Eq Depth 15 km
Grenville - No Conrad Layer
N. Appalachians Conrad Layer
22
NRC Project (1998) Modeling Attenuation of NSN
Data in the Grenville Province (5 yrs of data)
o Data - Simulation -- Model
23
USGS Project (2001) CEUS Ground Motion Model
from Simulations

• Earthquake Source Scaling and Ground Motion
  Attenuation Relations in Central and Eastern
  North America, Somerville et al. (2001)
• Comparison with results obtained using the
  stochastic model Atkinson and Boore (1995) Toro
  et al. (1997)

24
Generating Ground Motion Models from Simulations

• Earthquake source scaling relations were derived
  from the source parameters of 3 eastern North
  American events (Hartzell, 1994)
• The source scaling relations were used to
  generate a large suite of ground motion
  simulations using a broadband Greens function
  method
• The simulated ground motions were used to
  generate a ground motion prediction model

25
Hartzell (1994) Rupture Models
26
Rupture Area Scaling
27
Scaling of Rise Time (slip duration on fault)
28
Source Scaling Relations in the CEUS

• Rupture area is smaller (0.4) than for
  tectonically active regions
• Average slip is 2.5 times larger than for
  tectonically active regions
• Rise time is 1.85 times larger than for
  tectonically active regions
• Slip velocity is 35 larger than for tectonically
  active regions (higher dynamic stress drop)

29
Comparison of Simulation-Based Ground Motion
Prediction Models
M 6.5
30
Ways to Improve CEUS Ground Motion Prediction
Models

• Use 15 years of NSN / ANSS data from CEUS to test
  earthquake source scaling and wave propagation
  models
• Apply broadband simulation methods, developed and
  validated in the WUS, to the CEUS environment
• Rigorously validate predictions against augmented
  CEUS data

31
Resources for Simulation

• SCEC Broadband Strong Motion Simulation Platform
  (Three Methods)
• USGS Golden Set of Four Strong Motion Simulation
  Methods (Steve Hartzell)

32
SCEC Kinematic Broadband Ground Motion Simulation
Platform

• Calculate Kinematic Broadband (0-10 Hz) Waveforms
  for Scenario Earthquakes
• Validation of simulations
• Verification of simulations
• Platform brings objectivity, transparency and
  repeatability to strong motion simulations

33
Low Frequency Simulation (lt 1 Hz) 1. 1D / 3D
GF 2. 3D AWM 3. Site Effects 4.

• Source Description
• CFM, ERF
• Mw, Dimension, Geometry

Kinematic Rupture Generator 1. Beroza 2.
Archuleta 3. Graves 4.
Combine Low High Frequency 1. Filters 2. Time
Shift 3.

• GF Libraries
• Site Lists
• Velocity Models

Standard Rupture Format
High Frequency Simulation (gt 1 Hz) 1. 1D / Ray
GF 2. Scattering 3. Site Effects 4.
SCEC Broadband Simulation Platform
Broadband Time Histories (0 10 Hz)
34
SUMMARY

• CEUS ground motion prediction models are based on
  simulations validated against sparse data
• We now have 15 years of high quality ground
  motion data from NSN / ANSS in the CEUS
• We have advanced broadband strong motion
  simulation procedures that have been validated in
  the WUS
• We have a platform for broadband strong motion
  simulation that brings objectivity, transparency
  and repeatability to strong motion simulations