Advancements in multi-rupture time-dependent seismic hazard modeling, including fault interaction

Abstract

Several recent earthquake events (e.g., 2008 moment-magnitude (MW) 8.0 Wenchuan, China; 2016 MW 7.8 Kaikōura earthquake, New Zealand; 2019 MW 6.4–7.1 Ridgecrest sequence, USA) have emphasized the need to explicitly account for fault sources in probabilistic seismic hazard analysis (PSHA). Fault-based PSHA currently involves a number of significant but necessary modeling assumptions that mainly relate to fault segmentation, multi-segment event occurrence, long-term fault interaction, and time-dependent/independent earthquake recurrence. Each of these issues is typically investigated in isolation, neglecting the implications of their dependencies. This study offers a review of the current literature on fault-based PSHA, unifying state-of-the-art advances in the field within a single harmonized framework. The framework specifically incorporates some underlying methodologies of the latest Uniform California Earthquake Rupture Forecast (UCERF3; Field et al., 2014), providing a comprehensive means of relaxing fault segmentation, accounting for multi-segment ruptures in a standardized way, interpreting available fault data (e.g., slip rates and paleoseismic data) consistently, and inferring time-dependent probabilities of mainshock occurrence. The proposed framework also explicitly accounts for fault-interaction triggering between major known faults, using the approach outlined by Mignan et al. (2016) and Toda et al. (1998). A simple case study is established to demonstrate the framework's capabilities and limitations, involving a holistic investigation of the aforementioned modeling assumptions' effect on the seismic hazard estimates. The main findings of this study are (1) the ground-motion amplitude estimates can change significantly (for certain return periods) depending on the segmentation assumptions used (e.g., strict segmentation or relaxed segmentation, excluding multi-segment ruptures); (2) considering an ensemble of faults with a time-dependent occurrence model changes the shape of the hazard curve with respect to the time-independent assumption; (3) faults with the largest contribution to the hazard can differ between the time-dependent and time-independent cases; and (4) accounting for fault interaction may change the hazard estimates with respect to those obtained using classic time-dependent analysis (for which fault interaction is neglected). The framework provides a clear means of leveraging paleoseismic campaigns and slip rate data collections to potentially better constrain hazard estimates.