Engineering Perspective on Application of Simulated Ground Motions

Jonathan P. Stewart & Emel SeyhanUniversity of California, Los Angeles

Robert W. GravesUSGS Pasadena

Outline

• Need for verification, validation, and calibration (VVC)

• Validation procedures• Validation of ShakeOut motions• Calibration of high frequency IMs from Graves-

Pitarka hybrid procedure

Need for VVC

• Verification: consistency of motions across computational platforms for common conditions

Bielak et al., 2010

ShakeOut: example vel.(t)

Need for VVC

• Verification• Validation: results of analysis generally consistent

with recordings

Need for VVC

• Verification• Validation• Calibration: adjustment of parameters to achieve

desired attribute of simulated motions

Validation Procedures

• Waveform comparison using earthquake data

Graves and Pitarka, 2010

Validation Procedures

• Waveform comparison using earthquake data

• IM comparison to earthquake data

Closest Distance (km)

Graves and Pitarka, 2010

Validation Procedures

• Waveform comparison using earthquake data

• IM comparison to earthquake data

• IM comparison to GMPEs

Frankel, 2009

ShakeOut Validation

• Motions from hybrid procedure

ShakeOut Validation

• Motions from hybrid procedure

• Analysis of residuals– 4 NGA GMPEs– R, Vs30, Z for site i

ShakeOut Validation

• Motions from hybrid procedure

• Analysis of residuals• Too-fast distance

attenuation

SO

PH

R

R

Star et al., 20xx

ShakeOut Validation

• Motions from hybrid procedure

• Analysis of residuals• Too-fast distance

attenuation• Too-low intra-event

standard deviation

Star et al., 20xx

Calibration

• Adjustment to high frequency component of hybrid procedure

Calibration

• Adjustment to high frequency component of hybrid procedure

• Increase Q to slow distance attenuation

Seyhan et al., 20xx

Calibration

• Adjustment to high frequency component of hybrid procedure

• Increase Q to slow distance attenuation

Seyhan et al., 20xx

Calibration

• Adjustment to high frequency component of hybrid procedure

• Increase Q to slow distance attenuation

• Randomize Fourier amplitude spectra

Seyhan et al., 20xx

Ar fAfA exp

sA = f(M)

Calibration

• Adjustment to high frequency component of hybrid procedure

• Increase Q to slow distance attenuation

• Randomize Fourier amplitude spectra

Seyhan et al., 20xx

Conclusions

• Engineers want to use simulated motions• Needed for conditions having sparse data• Verification – Validation – Calibration– Each step documented– Avoid circularity in use of data