Fast and slow intraplate ruptures during the 19 October 2020 magnitude 7.6 Shumagin earthquake

Tsunamis pose a significant hazard, particularly those generated by slow ruptures on shallow subduction zone faults. The July 22, 2020 M_w 7.8 Simeonof megathrust earthquake in the Alaska subduction zone, while significant, produced a relatively small local tsunami (approximately 30 cm). However, its October 19, 2020 M_w 7.6 aftershock, despite being smaller and exhibiting a predominantly strike-slip mechanism based on initial seismic wave analysis, generated a surprisingly large tsunami. This discrepancy motivated a detailed investigation into the aftershock's source characteristics using seismic, geodetic, and tsunami data. The study focuses on understanding the source mechanism responsible for the unexpectedly large tsunami generated by the aftershock. This is crucial for improving tsunami hazard assessments, as the conventional models may not accurately predict the tsunami potential of such complex events. The research aims to unravel the source complexities, including the interplay of fast and slow ruptures on multiple faults, and their contribution to tsunami generation. Understanding this phenomenon is paramount for refining seismic hazard models and ultimately safeguarding coastal communities from future tsunamis.

Previous research has established that large tsunamigenic earthquakes typically occur in subduction zones and involve thrust faulting on the plate boundary. The 2020 Shumagin mainshock, while large, exhibited energy trapping on the continental shelf, limiting far-field tsunami amplitudes. Large intraplate ruptures seaward of megathrust events are usually associated with normal faulting, contrasting with the aftershock's unusual strike-slip faulting. Lateral gradients in megathrust coupling, ranging from strongly coupled to weakly coupled zones, have been observed and may influence the stress state, leading to complex rupture behavior. Previous studies have highlighted the importance of seafloor deformation extending seaward of the shelf break in enhancing tsunami excitation. This is especially critical when coupled with unfavorable faulting geometries, such as strike-slip faulting, which is less efficient in generating vertical seafloor displacements compared to thrust events. Existing tsunami models often rely on simple representations of the earthquake rupture, which may fail to capture the complexity and result in underestimation of the potential tsunami hazard.

The study used a multi-faceted approach, integrating seismic, geodetic, and tsunami data. Teleseismic P and SH waveforms, regional broadband and strong-motion recordings, and regional GNSS high-rate time series and static offsets were used to constrain a finite-fault model. Initially, a single-fault model was attempted. The authors then expanded to explore a two-fault model to account for a non-double-couple component observed in the long-period moment tensor. The model involved inverting slip distributions on two faults—one intraslab strike-slip fault and another fault in the upper plate, its exact geometry remaining somewhat uncertain—using simulated annealing algorithms. The positions, orientations, and slip distributions on these faults were adjusted until a good fit to the seismic and geodetic data was achieved. To model the tsunami waveforms, the authors utilized the NEOWAVE non-hydrostatic code, a model suitable for resolving the dynamic processes on steep continental slopes. The seafloor motions from the two-fault model were used as input for the tsunami simulation. However, this initial two-fault model failed to adequately match the observed tsunami waveforms, indicating a missing component in the source description. To address this, a simplified dipole source was initially introduced as a parametric function to represent potential additional seafloor deformation. This allowed for a systematic search of parameters like the spatial extent and timing of the seafloor uplift to best match the tsunami observations. The parameters were adjusted to find the most plausible configuration that accurately replicated the observed tsunami waveforms. The success of the dipole model then guided the exploration of more physically realistic fault dislocation models, such as slow thrust slip on the shallow megathrust or a splay fault. These models were assessed for their ability to match both the tsunami data and the high-rate GNSS data. The authors tested various orientations, depths, and dimensions of the additional fault, adjusting parameters to achieve a balance between matching the tsunami data and avoiding the contradiction with the GNSS observations. Long-period spectral analysis using global seismic recordings was conducted to confirm the consistency of the proposed model and compare it with the long-period moment tensor solutions. Coulomb stress changes were calculated to investigate whether the fast rupture could have triggered the slow slip event. The model also utilized a 1-D layered velocity model for Green's function calculations and a high-resolution bathymetry model for tsunami modeling.

The study's key findings highlight the unprecedented complexity of the M_w 7.6 aftershock source. The two-fault fast-slip model, while successfully matching seismic and geodetic data, significantly underpredicted the observed tsunami amplitudes. This led to the discovery of a previously unidentified slow-slip event. A third fault, almost perpendicular to the trench and dipping westward, was introduced in the model. This fault involved slow thrust slip (lasting more than 5 minutes) near the continental slope, producing the dominant component of the observed tsunami. The total seismic moment of the aftershock was divided into components of 2.5 × 10²⁰ Nm (M_w 7.5) for the intraslab rupture, 0.29 × 10²⁰ Nm (M_w 7.0) for the upper plate rupture, and 1.8 × 10²⁰ Nm for the slow-slip event. The slow slip caused a significant seafloor uplift near the continental slope, generating long-period tsunami waves (15-45 min). These waves propagated across the North Pacific, contributing significantly to the large tsunami signals observed in Hawaii. Importantly, the slow-slip event had minimal expression in seismic or geodetic observations, emphasizing the limitations of relying solely on seismic or geodetic data in tsunami hazard assessment. Analysis of Coulomb stress changes suggests that the fast slip events may have statically triggered the slow slip. The study concludes that the significant tsunami generation was a result of a complex interplay between fast and slow slip on multiple faults. The spatial offset and timing differences between the two resulted in destructive interference of wave systems, except in regions close to the Shumagin Islands, where arrivals were aligned.

The findings significantly advance the understanding of tsunami earthquake generation. The unexpected dominance of the slow-slip component, with minimal seismic expression, highlights the importance of considering such events in hazard assessments. The occurrence of slow thrusting on a near-trench perpendicular fault adds a new dimension to tsunami hazard assessment. While the precise geometry and triggering mechanism of the slow slip remain uncertain, the study strongly suggests that the fast ruptures likely triggered the slow-slip event, either dynamically or statically. The model presented offers a viable explanation for the observed tsunami amplitudes and their contrast with the relatively small tsunami from the larger M_w 7.8 mainshock. The interaction between the fast and slow slip components, resulting in constructive and destructive interference patterns across the Pacific, is critical in understanding the variability of tsunami amplitudes at different locations. The long-period tsunami waves produced by the slow-slip component are particularly important considering resonance effects on insular shelves, like those of the Hawaiian Islands. This study's implication on tsunami hazard assessment and the need for integrated analysis using various datasets will improve future predictions of tsunami potential.

The 19 October 2020 M_w 7.6 Shumagin earthquake demonstrated an unprecedented combination of fast and slow slip on multiple faults, resulting in unexpectedly large tsunami generation. The slow-slip component, while significant in tsunami generation, had minimal expression in seismic and geodetic data, highlighting the limitations of relying solely on those data for tsunami hazard assessment. The study emphasizes the need for integrated analyses combining seismic, geodetic, and tsunami observations to accurately model complex rupture processes and to improve tsunami hazard assessments. Future research should focus on obtaining higher-resolution bathymetric and reflection seismic data to improve imaging of the shallow structures and the potential slow slip fault. Campaign seafloor geodetic observations could provide further constraints on the process, including contributions from the mainshock and fast-slip components.

The study acknowledges several limitations. The precise geometry and depth of the upper-plate fault responsible for the fast rupture remain uncertain, although the authors' preferred model provides a good fit to the data. While the dipole model helped identify the general location and timing of the additional tsunami source, it's a simplified representation and may not entirely capture the true physical process. Obtaining a unique solution for the slow-slip source is challenging due to the limited data available and the complex interplay between the different rupture components. The study is dependent on the accuracy of the bathymetric models used in the tsunami simulations.