Abstract
Physics-based dynamic rupture models capture the variability of earthquake slip in space and time and can account for the structural complexity inherent to subduction zones. Here we link tsunami generation, propagation, and coastal inundation with 3D earthquake dynamic rupture (DR) models initialized using a 2D seismo-thermo-mechanical geodynamic (SC) model simulating both subduction dynamics and seismic cycles. We analyze a total of 15 subduction-initialized 3D dynamic rupture-tsunami scenarios in which the tsunami source arises from the time-dependent co-seismic seafloor displacements with flat bathymetry and inundation on a linearly sloping beach. We first vary the location of the hypocenter to generate 12 distinct unilateral and bilateral propagating earthquake scenarios. Large-scale fault topography leads to localized up- or downdip propagating supershear rupture depending on hypocentral depth. Albeit dynamic earthquakes differ (rupture speed, peak slip-rate, fault slip, bimaterial effects), the effects of hypocentral depth (25–40 km) on tsunami dynamics are negligible. Lateral hypocenter variations lead to small effects such as delayed wave arrival of up to 100 s and differences in tsunami amplitude of up to 0.4 m at the coast. We next analyse inundation on a coastline with complex topo-bathymetry which increases tsunami wave amplitudes up to ≈1.5 m compared to a linearly sloping beach. Motivated by structural heterogeneity in subduction zones, we analyse a scenario with increased Poisson's ratio of ν = 0.3 which results in close to double the amount of shallow fault slip, ≈1.5 m higher vertical seafloor displacement, and a difference of up to ≈1.5 m in coastal tsunami amplitudes. Lastly, we model a dynamic rupture “tsunami earthquake” with low rupture velocity and low peak slip rates but twice as high tsunami potential energy. We triple fracture energy which again doubles the amount of shallow fault slip, but also causes a 2 m higher vertical seafloor uplift and the highest coastal tsunami amplitude (≈7.5 m) and inundation area compared to all other scenarios. Our mechanically consistent analysis for a generic megathrust setting can provide building blocks toward using physics-based dynamic rupture modeling in Probabilistic Tsunami Hazard Analysis.
Highlights
Earthquake sources are governed by highly non-linear multiphysics and multi-scale processes leading to large variability in dynamic and kinematic properties such as rupture speed, slip rate, energy radiation, and slip distribution (e.g., Oglesby et al, 2000; Kaneko et al, 2008; Gabriel et al, 2012; Bao et al, 2019; Ulrich et al, 2019a; Gabriel et al, 2020)
In this study we introduce a number of important differences to previous work. 2D linking including approximations to match seismic cycle (SC) and DR fracture energy during slip events leads to differences in slip magnitude between the SC and DR modeling and large magnitudes and high rupture speed in dynamic rupture scenarios
In our 12 scenarios listed in Table 1, different hypocenter depths lead to pronounced differences in dynamic rupture propagation
Summary
Earthquake sources are governed by highly non-linear multiphysics and multi-scale processes leading to large variability in dynamic and kinematic properties such as rupture speed, slip rate, energy radiation, and slip distribution (e.g., Oglesby et al, 2000; Kaneko et al, 2008; Gabriel et al, 2012; Bao et al, 2019; Ulrich et al, 2019a; Gabriel et al, 2020). Such variability may impact the generation, propagation, and inundation of earthquake-generated tsunami or secondary tsunami generation mechanisms such as triggered landslides (e.g., Sepúlveda et al, 2020). Models of tsunami generated by large earthquakes can routinely and quickly use kinematic finite fault models constrained by inversion of seismic, geodetic, and other geophysical data (Geist and Yoshioka, 1996; Ji et al, 2002; Babeyko et al, 2010; Maeda et al, 2013; Allgeyer and Cummins, 2014; Mai and Thingbaijam, 2014; Bletery et al, 2016; Jamelot et al, 2019), but are challenged by the inherent non-uniqueness of kinematic source models (Mai et al, 2016)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
