Synthetic Seismograms from Physics-based Modeling of Heterogeneous Rupture for Large Subduction Earthquakes

Large subduction earthquakes with magnitudes (Mw) greater than 8.0 are devastating events. Such large earthquakes remain poorly recorded due to their infrequent occurrence. This lack of observational data limits our ability to study rupture dynamics and accurately predict future broadband ground motion. To address these limitations, physics-based modeling has emerged as a powerful approach for understanding the rupture dynamics of large subduction earthquakes and associated ground motions. In this study, we analyze kinematic rupture processes and their influence on synthetic seismogram simulations. We revisit the rupture characteristics and ground motion variability for a large megathrust earthquake in Central Chile, the Illapel Mw 8.3 (2015). We use the data derived from the Bayesian inversion framework presented by Caballero et al. (2023) as input for the forward modeling of ground motion. To capture the finite source effects of a heterogeneous slip distribution, we discretize each sub-fault into point sources limited by a separation dependent on the maximum frequency resolution. With this in mind, we interpolate the seismic moment and define the rupture propagation across the rupture plane. We implement two different codes to compute the resulting ground motions: Axitra (Cotton & Coutant, 1997) and Pyrocko-GF (Heimann et al., 2019). Both codes employ distinct methods for Green’s functions computation and source representation, allowing a comparative analysis of their capabilities in reproducing strong motion. Pyrocko-GF efficiently handles low-frequency simulations with pre-computed Green’s functions, while Axitra provides broadband synthetic seismograms up to 20 Hz. However, with Pyrocko-GF it is also possible to reach high frequencies by adding Green´s functions to its FOMOSTO program. The synthetic seismograms were compared against strong-motion data, focusing on stations at a maximum of 5° of the hypocenter. Key parameters such as peak ground displacement, waveform similarity, and spectral content were analyzed. Additionally, we evaluated the impact of different source time functions on predictions. Our results provide insights into the importance of incorporating heterogeneous rupture scenarios for large earthquakes, as well as the challenges of modeling high-frequency ground motions using different numerical approaches. It is foreseen that our methodology and results will be used in a full physics-based seismic-tsunami hazard assessment for Central Chile.