Toward Hybrid Broadband Ground Motion Simulation Validation for Mw>3.5 New Zealand Earthquakes

Robin L. Lee, Brendon A. Bradley, & Xavier Bellagamba

Submitted August 15, 2018, SCEC Contribution #8610, 2018 SCEC Annual Meeting Poster #016

Recent advances in computational performance and physics-based ground motion simulation methodologies are making physics-based seismic hazard analysis a reality. Extensive validation of physics-based ground motion simulations against observations are needed to both understand their predictive performance (and hence their degree-of-belief weighting in a logic tree for prediction), and efficiently identify areas for their iterative improvement. Recent ground motion simulation validation efforts in New Zealand have mostly been focused on the Canterbury region where the 2010-2011 Canterbury Earthquake Sequence occurred, and we have begun to extend our focus to New Zealand in its entirety. This poster presents preliminary ground motion simulation validation results considering small-to-moderate magnitude (3.5<Mw<5.0) earthquake events across New Zealand, with clusters of earthquakes generally located along the tectonic plate boundary. Source characteristics are obtained from centroid moment tensor solutions provided by GeoNet and a unified New Zealand Velocity Model provides crustal seismic velocities. A total of 638 earthquake ruptures, modelled as point sources, are considered with over 5000 quality-assured ground motions recorded at 382 recording stations. In order to scale our analyses to a large number of ground motions we have also developed a neural network for ground motion quality classification, and introduced event-specific computational mesh determination to optimize total compute core hour requirements. The performance of the simulations, and conventional empirical ground motion models (for benchmarking purposes), are quantified using a mixed-effects analysis framework where the biases and uncertainties associated with the simulation methodology and systematic source, site and paths effects are explicitly determined.

Key Words
ground motion simulation, validation, New Zealand, small magnitude

Citation
Lee, R. L., Bradley, B. A., & Bellagamba, X. (2018, 08). Toward Hybrid Broadband Ground Motion Simulation Validation for Mw>3.5 New Zealand Earthquakes. Poster Presentation at 2018 SCEC Annual Meeting.


Related Projects & Working Groups
Ground Motions