Verification and Validation of High-Frequency (fmax = 5 Hz) Ground Motion Simulations of the 2014 M 5.1 La Habra, California, earthquake

Ricardo Taborda, Kim B. Olsen, Robert W. Graves, Fabio Silva, Naeem Khoshnevis, William H. Savran, Daniel Roten, Zheqiang Shi, Christine A. Goulet, Jacobo Bielak, Philip J. Maechling, Yifeng Cui, & Thomas H. Jordan

Submitted September 9, 2016, SCEC Contribution #6475, 2016 SCEC Annual Meeting Poster #272 (PDF)

The Southern California Earthquake Center (SCEC) High-F project seeks to advance physics-based, deterministic earthquake simulation with the long-term objective of improving ground motion prediction and seismic hazard models. An important aspect of this involves the verification of models and simulation methods, and the validation of synthetics with respect to data. The present study describes a recent concerted effort to verify different simulation methods, and attempts to reproduce the ground motions from the 2014 M 5.1 La Habra, California, earthquake in the greater Los Angeles region over a simulation domain of 180 km x 135 km, and 62 km in depth. The simulations are done using three high-performance computing codes; two of them implement the finite difference method and the third one implements a finite element approach. The models are tailored to satisfy a maximum frequency of 5 Hz and a minimum shear wave velocity of 500 m/s. We rely on the latest version of the community velocity model CVM-S4.26.M01, and use a point source and two finite source models. The point source model is defined by a mechanism derived from strong-motion data and a slip-time signal obtained from a dynamic rough-fault rupture model. The two finite source models come from (i) an independent source inversion study, and (ii) a simulation done with a kinematic rupture generator also used in the SCEC Broadband Platform. At the verification stage, we compared synthetics from the three codes for the point source model, using the 3D regional structure and frequency independent attenuation (Q) models. The comparisons between the three codes exhibit very good agreement. For the validation stage, we compared seismograms collected at 300+ recording stations from different regional strong-motion seismic monitoring networks with synthetics obtained for the three different source models. Initial results indicate that extended source models, even for a moderate-size earthquake like the one considered here, tend to lead to better fits with data. Current and future efforts concentrate on defining the best possible source model based on goodness-of-fit comparison metrics, and testing the influence of other parameters such as a frequency dependent Q model and small scale heterogeneities. Simulations have been carried out on NCSA Blue Waters and on OLCF Titan.

Key Words
Ground motion simulation, High-performance computing, Verification, Validation, La Habra earthquake

Taborda, R., Olsen, K. B., Graves, R. W., Silva, F., Khoshnevis, N., Savran, W. H., Roten, D., Shi, Z., Goulet, C. A., Bielak, J., Maechling, P. J., Cui, Y., & Jordan, T. H. (2016, 09). Verification and Validation of High-Frequency (fmax = 5 Hz) Ground Motion Simulations of the 2014 M 5.1 La Habra, California, earthquake. Poster Presentation at 2016 SCEC Annual Meeting.

Related Projects & Working Groups
Ground Motion Prediction (GMP)