Mike Dupuis

High-quality earthquake ground-motion records are required for various applications in engineering and seismology; however, quality assessment of ground-motion records is time-consuming if done manually and poorly handled by automation with conventional mathematical functions. Machine learning is well suited to this problem, and a supervised deep-learning-based model was developed to estimate the quality of all types of ground-motion records through training on 1096 example records from earthquakes in New Zealand, which is an active tectonic environment with crustal and subduction earthquakes. The model estimates a quality and minimum usable frequency for each record component and can handle one-, two-, or three-component records. The estimations were found to match manually labeled test data well, and the model was able to accurately replicate manual quality classifications from other published studies based on the requirements of three different engineering applications. The component-level quality and minimum usable frequency estimations provide flexibility to assess record quality based on diverse requirements and make the model useful for a range of potential applications. We apply the model to enable automated record classification for 43,398 ground motions from GeoNet as part of the development of a new curated ground-motion database for New Zealand.

High-quality large-magnitude subduction earthquake ground-motion records are needed as inputs for response history analysis of dams in regions adjacent to the Cascadia Subduction Zone. Quality assessment of candidate ground-motion records is time consuming if done manually and poorly handled by automation with conventional mathematical functions; therefore, a supervised deep-learning-based model was developed in a previous study to estimate the quality and minimum usable frequency of ground-motion records through training on 1,096 records from earthquakes in New Zealand, which is an active tectonic environment with crustal and subduction earthquakes. In that study, the model was found to perform well for small-to-moderate magnitude earthquake records from active shallow crustal, subduction slab, and subduction interface earthquakes; however, the model’s performance for large-magnitude earthquake records was not investigated. In this study, we evaluate the performance of the model for assessment of records from the 2010 M8.8 Maule and 2011 M9.1 Tohoku subduction interface earthquakes. We utilize high-quality processed subduction records and then superimpose various amplitudes of artificial background noise to degrade quality and then apply the model quality for quality classification. Eleven high-quality ground-motion records were selected based on the model results and then linearly scaled to a target spectral acceleration for a hypothetical site in British Columbia to produce a suite of large-magnitude subduction interface earthquake ground-motions suitable for structural response history analysis.

Estimating ground motion from large-magnitude subduction earthquakes is critical for assessing seismic risk for dams near subduction zones. This study presents physics-based hybrid broadband simulations for 50 Mw8.7 rupture scenarios on the Hikurangi Subduction Zone in Aotearoa New Zealand, using subduction-specific parameter models validated with subduction interface events. Ground motions were simulated at 54 dam sites to support response history analysis and infrastructure risk assessment. The simulations reveal strong and spatially variable shaking, with characteristically long durations and high Arias intensities. Ground-motion intensity was found to be sensitive to rupture characteristics, including hypocentre location, subevent locations, rupture velocity, and stress parameter. These results provide critical insights for seismic hazard analysis, seismic structural response assessment, and resilience planning for dam infrastructure exposed to potential megathrust earthquakes.

Physics-based simulation of subduction earthquake ground motions remains less comprehensively validated than for shallow crustal earthquakes, despite subduction events contributing significantly to global seismic hazard. In this study, subduction-specific simulation models were developed and validated for small-magnitude (Mw3.5–5) interface and slab earthquakes using hybrid broadband ground-motion simulation. Simulation models were constrained by (1) global empirical ground-motion models, (2) a global database of finite-fault rupture models, and (3) global ground-motion simulation studies. The simulation models include subduction source-specific representations for stress parameter and rupture velocity, with a significant depth dependence of the stress parameter for slab earthquakes. Volcanic backarc effects on anelastic path attenuation are also included through path attenuation-based scaling of the rock quality factors in the simulations. The simulations leverage models for site and basin effects which have been validated using crustal earthquakes, for which there are many recorded ground motions, and the subduction-specific model modification mainly affect the high-frequency component of the ground-motion simulations above 1Hz. Simulation predictions were validated against a compiled dataset of observed subduction earthquake ground-motion records in New Zealand. The subduction-specific modifications significantly improve predictive performance compared to the use of simulation parameter values for active shallow crustal earthquakes and provide comparable accuracy to prior simulations for crustal earthquakes in New Zealand. Despite these promising results, further studies are needed to address prediction residuals at low frequencies (f ≤ 1 Hz), as well as extending validation to larger magnitude earthquakes.

Earthquake ground motions recorded at rock sites can be used to reduce seismic hazard uncertainty for dams and other critical infrastructure on rock; however, few ground motion records exist from known rock sites because most seismic stations are on soil or lack measured shear wave velocities. In this study, a dataset of California ground motions is examined to identify additional instrumented “Very Dense Soil” and “Soft Rock” sites which may actually be “Rock”. This approach offers a cost- and time-efficient alternative to installing new seismic stations. To identify possible rock sites, partially-crossed linear mixed-effects regression is used to compute site residuals from empirical ground-motion model predictions and recorded ground motions. Based on these site residuals and other site characteristics, machine learning is then applied to estimate shear wave velocities at sites without measured velocities. Using these estimated shear wave velocities, a ranked list of candidate rock sites is developed. Geophysical testing is proposed at a selection of these sites to verify the high estimated shear wave velocities, with a focus on sites with many recorded ground motions, as an efficient means to expand the catalogue of rock ground motions.