Joyner Lecture: Simulating Realistic Earthquake Ground Motions
Rob Graves, USGS Pasadena
Wednesday, July 17, 2019 at 10:30 AM
- Building 3, Rambo Auditorium
- Tom Hanks
Ground motion simulations are an important resource for augmenting recorded motions, assessing impacts for scenario earthquakes, and exploring parametric sensitivity. Providing confidence that simulations are realistic requires demonstrating that they not only reproduce motions from past earthquakes, but also that they can predict motions for future events. Ideally, simulations should capture effects due to complexities in the rupture process, as well as effects due to large-scale (e.g., basins) and small-scale (e.g., scattering or site effects) variations in the seismic velocity structure. Accounting for these features is challenging due to uncertainty in the expected ranges of the required parameters. Furthermore, adding increased detail to the model space increases the computational requirements. Simulation approaches have traditionally been classified as “deterministic” or “stochastic” depending on the level of complexity used to describe earthquake rupture and wave propagation effects. Because of knowledge limitations and computational cost, the deterministic approach is typically employed for lower frequencies (< 1 Hz) and the stochastic approach at higher frequencies (> 1Hz). This distinction naturally leads to a hybrid approach, where the separate low- and high-frequency responses are combined to produce a broadband response. In my talk, I will describe the features behind the hybrid simulation approach along with examples of its application to model recorded earthquake ground motions. I will also describe refinements to the deterministic approach that extend its range of applicability to higher frequencies. Finally, I will summarize those cases where simulations are most beneficial and explore some frontiers scientists are facing in ground motion simulations.