CSZ Full-Margin Megathrust Earthquake Scenarios, in 3D Broadband Ground Shaking Simulations of Cascadia Subduction Zone Earthquake Scenarios

This dataset contains 3D broadband ground motion simulations for 36 earthquake scenarios for full-margin megathrust earthquakes in the Cascadia Subduction Zone (CSZ), expanding upon the M9 Project database (Frankel et al., 2018; Wirth et al., 2018) by including a wider range of earthquake magnitudes and slip distributions. We generate full-margin CSZ megathrust earthquakes using a logic tree approach. Our logic tree includes the downdip extent of slip, the magnitude, the updip extent of slip, and the slip distribution. Event magnitude is calculated using magnitude-area relationships. Outside of the logic tree, we vary the hypocenter location randomly within the northern, central, and southern regions and the location of high frequency generating areas within the downdip region of the subduction zone. Compared to M9 project scenarios, we include variable updip extents of slip, allowing rupture on both the megathrust and secondary splay faults (frontal thrusts), and variable event magnitudes.

Ground motions are simulated using the same approach and many of the same input parameters as the M9 Project ground motions. We utilize a hybrid approach, where long period (<1 Hz) ground motions are calculated deterministically in a 3D numerical simulation including 3D seismic velocity structure (Stephenson et al., 2017) and short period ground motions (1-10Hz) are calculated stochastically using the summation of point-sources that rupture within the high-frequency regions (i.e. subevents) along the fault, specifying a source magnitude, a 1D velocity model based on a profile within the Puget Lowlands, a stress drop (200 bar), and a regional attenuation factor, using the open-source code SMSIM (Boore, 1983, 1996). These are combined using a matched filter at 1 Hz. Broadband waveforms are resolved to 10 Hz, deterministic waveforms are resolved to 1 Hz, both use a minimum Vs = 600 m/s. The deterministic simulations are run using the open-source code SPECFEM3D (Komatitsch & Tromp, 2002b, 2002a,https://github.com/SPECFEM/specfem3d) and take between 25-40 hrs on 1152 cores using Cascade Lake Nodes (CLX) on the Frontera HPC at the Texas Advanced Computing Center (TACC). Using identical source scenarios, we verified SPECFEM3D against the numerical code used in the M9 Project, finding a comparable match, with slight variations due to differences in interpolation and meshing methods.