SCEC Award Number 16117 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Development of a broadband 3D pseudo-dynamic rupture generator for geometrically complex faults: Part 2
Investigator(s)
Name Organization
Kim Olsen San Diego State University
Other Participants PhD candiate William Savran
SCEC Priorities 6e, 6b, 6c SCEC Groups GMSV, CME, GMP
Report Due Date 03/15/2017 Date Report Submitted 06/05/2017
Project Abstract
 Spontaneous rupture simulations using geometrically-rough faults have been shown to produce realistic far-field spectra and comparable fits with GMPEs, but they are too computationally demanding for use with physics-based probabilistic seismic hazard analysis efforts such as the Southern California Earthquake Center CyberShake or Broadband Platforms. Here, we present our implementation of a kinematic rupture generator that mimics (at least in a statistical sense) the processes of rough-fault spontaneous rupture models that are responsible for generating realistic broadband sources. To this end, we analyze ~100 dynamic rupture simulations on strike-slip faults ranging from Mw 6.4 - 7.2. We find that our dynamic simulations follow empirical scaling relationships for inter-plate strike-slip events, and provide source spectra comparable with an ω-2 source model. To define our kinematic source model, we use an exponential source-time function parameterized in terms of slip, peak slip velocity, and rupture velocity. These parameters are represented by random fields whose distributional parameters are estimated from the dynamic ensembles. Our method uses sequential Gaussian co-simulation to generate the random spatial fields using two-point statistics defined by a linear model of coregionalization. We incorporate a nested-model of linearly independent variogram basis functions to capture correlations at all scale lengths between all source parameters and the initial friction on the fault. Additionally, we introduce the variability observed in one-point statistics by defining a multivariate Gaussian random variable model. We show that ground motions calculated by the kinematic model have amplitudes and spectral content comparable with those of the dynamic rupture simulations.
Intellectual Merit SCEC aims to advance seismic hazard analysis to higher frequencies which requires transparent and efficient methods to generate kinematic source functions. Most current kinematic rupture generators have not been tested at the higher frequencies. The rupture generator here includes the energy at the higher frequencies from data-constrained geometrical roughness of the fault.
Broader Impacts Immediate applications of the proposed kinematic rupture generator include the SDSU Broadband Platform (BBP) method (currently sharing the Graves and Pitarka rupture generator), CyberShake and UCERF3. The project includes training of William Savran and support for his doctoral in the Joint Doctoral Program between SDSU and UCSD.
Exemplary Figure Combine Figs 2a and 5. Fig. 2a: Stochastic source model for a Mw 6.9 strike-slip earthquake scenario generated by our KRG. (a) Final slip (Δu (m)) with rupture time contours plotted in light gray, with the resulting surface-slip plotted on the top axes. (b) Peak-slip-velocity ( (m/s)) and (c) normalized rupture-velocity (, where is the local shear-wave velocity. Fig. 5: Normalized Fourier acceleration spectra averaged over all stations on 500m x 500m grid spanning entire free-surface. We apply a Butterworth lowpass filter to the seismograms with corner frequency =7.5 Hz, and each spectrum is normalized to unit RMS amplitude between 2.5 – 5.0 Hz before stacking. Fault-parallel corresponds to the Vx-component and fault-normal corresponds to the Vy-component of ground motion. Credit: W.H.Savran.