Performance of Physics-based Deterministic Ground Motion Simulations: Building Confidence in Using Broad-Band Synthetic Ground Motion in Engineering Applications

Recent improvements of large-scale ground motion simulations resulting from physics-based rupture and wave propagation models and accessible high-performance computing have made possible the potential use of synthetic ground motion in engineering applications. They provide scientists and engineers with new data that can yield new insight on the characteristics of ground motions and the variability of the infrastructure response. Such simulations can also make a significant contribution to the reduction of uncertainty in ground motion models (GMMs) that are being developed for large earthquakes at near-fault distances where ground motion variability is not fully captured by current sparsely recorded data.

An important aspect of the development and calibration of regional physics-based simulation platforms is the validation of their methodology and synthetic ground motion. Independent criteria that involve direct comparisons with recorded earthquakes, comparisons with exiting region-specific ground motion models for scenario earthquakes, and comparison with recorded buildings response should be requirements for trusted validation analysis.

In an effort to build confidence in simulated ground motion we compiled published results of validation analysis performed by several modeling teams and analyzed the general performance of their physics-based ground motion simulations. We focused on broad-band simulations that use a deterministic approach in computing ground motion time histories for crustal earthquakes. The analysis includes simulation results of selected recorded and scenario earthquakes with the magnitude ranging from 5.4 to 7.5.   The goals of our study are to demonstrate that current physics-based kinematic rupture models can produce ground motions that agree with observed ones and empirical estimates, and that well-constrained reginal velocity models are capable of producing the expected wave scattering affecting ground motion variability and amplitude at local and regional distances. Satisfying these goals provides confidence in the predictive capabilities of the simulation platforms and the quality of synthetic ground motion in various engineering applications, including development of non-ergodic GMMs and building response analysis.