Development of Procedures for the Rapid Estimation of Ground Shaking
Task 7: Ground Motion Estimates for Emergency Response

Objective
Develop methodology to automatically simulate strong ground motion
In sparse regional coverage
In near-real-time
With seamless shakemap product for areas with & without dense local strong motion instrument coverage

Task List
Finalize fault slip models & inverse methodology
Compare predicted to observed ground motions
Investigate site amplification factors
Ground motion simulation code improvements

Scope of the Problem
Uneven distribution of strong motion stations
Presently insufficient numbers of telemetered strong motion stations in northern California
Uneven distribution of seismic source regions
Regionally extensive support infrastructure

TriNet ShakeMaps for Landers and Northridge

Data Driven PGA Shakemap in Deteriorating Coverage

Inverse Methodology

Northridge Fault Slip Models

Slide 9

Slide 10

Predicted PGV (0.1-1.0 Hz)

Efforts to Improve Shakemaps
Application of site corrections
Application of hard rock and soft rock greenís functions
Integration of empirical attenuation relationships
Alternative methods of ground motion simulation

PGV Predicted to Observed Amplitude Ratios Relative to Site Class

Predicted to Observed PGV Ratios

Empirical Attenuation Model Shakemap

Implementation Plan
Automated finite source inversion to find:
Causative fault, fault slip dimension (approximately 15 minutes)
Fault slip distribution, rupture velocity and estimated dislocation rise time (approximately 30 minutes)
Preliminary line or course grid results used to generate a data/model shakemap using the Somerville et al. (1998) attenuation relationships
Approximately 15 minutes

Implementation Plan
Compute conservative shakemap (larger of the empirical or deterministic calculations)
Revised data/conservative-model shakemap
Approximately 45-60 minutes
Follow the data/model interpolation and site adjustment approach of the TriNet ShakeMap (Wald et al., 1999)

Combined Data/model Shakemaps (Northridge PGA)

Slide 19

Conclusions
Code developed for source inversion and ground motion simulation
Empirical site corrections not useful when applied to simulated ground motions
Alternative higher frequency simulation methods do improve shakemaps but not enough to offset additional computational cost

Conclusions
Site specific greenís functions improve the maps Ė application in other areas will require calibration
Directivity capable attenuation relationship improves shakemaps in several ways
Very rapid preliminary maps
Seamless base model for different levels of approximation
Base model for development of source specific conservative model
Reasonable estimates of peak and spectral values of ground motions

Recommendations
Develop model based site corrections using numerical methods
Extend the Somerville et al. (1997) attenuation relationship to include PGV and PGD