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

The 22 July 2020 Mw 7.8 Simeonof megathrust earthquake beneath the Shumagin Islands generated relatively small far-field tsunamis due to energy trapping on the continental shelf. On 19 October 2020, a Mw 7.6 aftershock with an oblique strike-slip mechanism occurred seaward of the up-dip edge of the mainshock rupture and produced unexpectedly large tsunami signals at DART buoys and tide gauges in Alaska and Hawaii, despite being smaller in magnitude and having an intrinsically less tsunamigenic faulting style. The event’s aftershocks span depths of ~5–40 km, straddling the megathrust, with substantial upper plate activity. Typical outer-rise intraplate events involve trench-parallel normal faulting, so the observed ~50° east-dipping strike-slip mechanism indicates unusual stresses possibly related to along-strike gradients in megathrust coupling between strongly coupled Semidi segments and the weakly coupled Shumagin segment. The study aims to resolve the source processes responsible for the unexpectedly large tsunami by integrating seismic, geodetic, and tsunami observations.

Prior studies documented the 22 July 2020 Mw 7.8 Simeonof event, including partial rupture of a weakly coupled megathrust and tsunami characteristics influenced by shelf trapping and de-shoaling. Typical intraplate ruptures seaward of megathrusts are normal-faulting and trench-parallel, contrasting with the October aftershock’s strike-slip geometry. Reflection seismology has imaged a shallow north-dipping normal fault in the upper plate beneath the shelf. Coupling variations and asperity behavior along the Alaska-Aleutian subduction zone have been reported, including transitions from locked to creeping behavior and persistent asperities. Modeling approaches that combine seismic and geodetic data to invert finite faults and to simulate near-field tsunamis (e.g., NEOWAVE) are established and have been used in comparable events (e.g., 2011 Tohoku, 2021 Chignik). The literature on tsunami generation by horizontal seafloor displacement and on parametric dipole models for slumps informs the search for additional sources beyond fast coseismic slip.

- Data collection and preprocessing: 62 teleseismic P and 50 SH broadband recordings (30°–90°), 9 regional broadband and 6 local strong-motion stations (<700 km), 8 GNSS coseismic static offsets, and 8 high-rate (hr) GNSS time series (stations including AC12, AC28). Instrument responses removed; seismic data filtered 1–300 s (teleseismic) and 5–100 s (regional); resampled at 0.2 s; manual picking of first arrivals; 180 s inversion window for joint inversion. - Finite-fault inversion: Non-linear simulated annealing joint inversion of teleseismic body waves, regional waveforms, GNSS statics, and hr-GNSS time series. Initial single-plane model (strike 350°, dip 50°E) then a preferred two-fault model to match non-double-couple radiation. Subfaults 5 km × 5 km; rake constrained to right-lateral strike-slip on intraslab fault and pure dip-slip on secondary fault. Slip 0–8 m, rise/fall intervals 0.6–6.0 s (total 1.2–12 s), rupture velocity 0.5–3.0 km/s. Green’s functions computed with a 1-D layered velocity model; equal weighting for GNSS statics and seismic waveforms. - Tsunami modeling: NEOWAVE non-hydrostatic model with nested spherical grids across the North Pacific. Grid resolutions: Level-1 at 2 arc-min (~3700 m), Level-2 at 24–30 arc-sec (~740–925 m), Level-3 at 6 arc-sec (~185 m), Levels 4–5 at 0.3–0.4 arc-sec (~9–12 m) for harbors (Hilo, Kahului, Sand Point, King Cove); Manning n=0.025. Bathymetry/DEM from GEBCO, multibeam, LiDAR, and NCEI datasets. Seafloor deformation from elastic half-space (Okada) and horizontal seafloor motion accounted for. - Dipole parameterization search: A two-lobed surface dipole (uplift/subsidence) used as a flexible kinematic proxy to localize additional tsunami source near the shelf break, exploring onset time (lag), location/orientation along bathymetry contours, size, and amplitude through forward modeling to fit DART waveforms (sensitive to phase and period). Conducted 160 realizations; preferred dipole: depressed patch ~25 × 16.7 km, ~2 m depth, uplifted counterpart scaled by α=1.21, initiated ~4 min after fast rupture, located ~0.2° WSW of epicenter with ~20 km positional uncertainty along the shelf break. - Physical slow-slip source testing: Explored slow rupture on (i) shallow megathrust, (ii) trench-parallel upper plate splay, and (iii) trench-perpendicular upper plate thrust geometries. Ensured minimal seismic/hr-GNSS signals by using long source durations (>5 min) and delayed onset. Models matching tsunami but violating hr-GNSS at AC12/AC28 were rejected. - Preferred three-fault model: Two fast-slip faults (intraslab strike-slip and upper-plate oblique-normal) plus a delayed slow-slip upper-plate thrust on a fault striking ~190°, dipping 30°W, located below continental slope near shelf break, with L=W=20 km, slip ~15 m, onset ~30 s after initiation, duration ~300 s. Tsunami predictions validated at DARTs and tide gauges without violating seismic or hr-GNSS time histories. - Long-period spectral analysis: Fundamental mode Rayleigh and Love wave spectra at 256 s corrected to source using PREM; compared four-lobed radiation patterns to two-fault model predictions; slow-slip component yields weak long-period amplitudes and large phase shifts, explaining lack of clear seismic signature. - Coulomb stress modeling: ΔCFS = Δτ + μ′ΔσN with μ′=0.4 computed on the preferred slow-slip fault (Strike 190°, dip 30°, rake 90°) due to both fast-slip faults (tested for alternative shallow fault dips). Increases up to ~0.5 MPa at 3–13 km depth suggest static triggering viability. - Validation: Comprehensive comparison to DART (46402, 46403, 46409, 46410, 46414, 46415) and tide gauges (Sand Point, King Cove, Hilo, Kahului). Assessed spectral energy (periods ~15–45 min) and arrival-time adjustments noted for plotting alignment only; ensured hr-GNSS shows no deformation 30–330 s post fast-slip consistent with slow-slip duration.

- Fast-slip rupture complexity: The Mw 7.6 aftershock involved two fast ruptures completing within ~40 s at 2–3 km/s: (1) a dominant intraslab strike-slip fault (strike ~350°, dip ~50°E) with seismic moment Mo ≈ 2.5 × 10^20 Nm (Mw ~7.5); (2) a smaller upper-plate oblique-normal fault below the shelf near the shelf break with Mo ≈ 0.29 × 10^20 Nm (Mw ~7.0). The composite moment tensor explains the observed non-double-couple radiation and fits seismic and geodetic (static offsets and hr-GNSS) observations. - Tsunami mismatch with fast-slip alone: Despite fitting seismic/geodetic data and initial DART pulse timing and first peak/shoulder, the two-fault fast-slip model underestimates the main second peak and subsequent trough at multiple DARTs (46402, 46414, 46409, 46415), underpredicts Sand Point amplitudes (with ~14 min early modeled arrival), and shows poor agreement at Hawaii tide gauges (Hilo, Kahului). Spectral energy at ~20–45 min periods is underestimated, indicating a missing delayed source. - Additional delayed source required: Forward modeling with a kinematic dipole indicates a seafloor deformation localized across the shelf break, with uplift on the continental slope and drawdown near/landward of the shelf break, delayed by ~4–5 min relative to initial fast rupture. Preferred dipole implies ~±2 m vertical deflections over ~20–25 km scales with ~20 km positional uncertainty along the shelf break. - Successful physical slow-slip model: A slow thrust-slip rupture on an upper-plate fault striking ~190° (near-perpendicular to trench), dipping ~30°W, located near the continental slope/shelf break, starting ~30 s after earthquake initiation and lasting >5 min (~300 s), with ~15 m slip over 20 × 20 km (Mo ≈ 1.8 × 10^20 Nm), reproduces DART and tide-gauge waveforms (including 15–45 min periods and Hawaii resonance) without detectable seismic or hr-GNSS signals between 30–330 s. Alternative slow-slip on shallow megathrust or trench-parallel splay faults can match tsunami but violate hr-GNSS, thus rejected. - Interference and far-field amplification: Spatial offset, onset delay (~30 s), and long duration (~300 s) between fast and slow components cause phase lags and destructive interference generally, but constructive interference around the Shumagin Islands; slow-slip near the shelf break radiates larger basin-wide tsunami than the deeper mainshock. Wave periods (15–45 min) align with Hawaiian shelf resonances, explaining large amplitudes in Hawaii. - Triggering: Coulomb stress increases up to ~0.5 MPa on the slow-slip fault region from fast-slip ruptures support static triggering; dynamic stresses likely also significant. - Unprecedented source complexity: Coseismic ruptures on both sides of the megathrust (intraslab and upper plate) plus a delayed, tsunamigenic, slow upper-plate thrust fault nearly trench-perpendicular constitute an unprecedented combination explaining the unexpectedly large tsunami from a predominantly strike-slip Mw 7.6 event.

The study resolves the paradox of a smaller, mainly strike-slip Mw 7.6 aftershock generating a larger tsunami than the preceding Mw 7.8 thrust mainshock. Fast-slip ruptures above and below the plate interface alone cannot account for the observed DART and tide-gauge signals; a delayed slow-slip thrust within the upper plate near the shelf break, striking nearly perpendicular to the trench, provides the missing source of long-period tsunami energy. The slow-slip’s long duration and location suppress its seismic and onshore geodetic signatures while efficiently uplifting the continental slope and producing drawdown across the shelf break, shaping the spectral content and phasing of the tsunami. Phase interactions between fast- and slow-generated waves produce destructive interference in many directions but constructive interference locally and amplify signals at periods resonant with Hawaiian insular shelves, explaining unexpectedly large amplitudes in Hawaii. The results underscore how along-strike variations in megathrust coupling and complex upper-plate structure can foster compound intraplate faulting and trigger slow-slip. This highlights a previously underappreciated tsunami hazard mechanism: tsunamigenic slow-slip on non-splay, trench-perpendicular upper-plate faults near the shelf break, potentially producing substantial far-field tsunamis even when seismic and onshore geodetic signals are muted.

By integrating seismic, geodetic, and tsunami observations with joint finite-fault inversions and hydrodynamic modeling, the study demonstrates that the 19 October 2020 Mw 7.6 Shumagin aftershock comprised: (i) fast intraslab strike-slip rupture; (ii) fast upper-plate oblique-normal rupture; and (iii) a delayed (>5 min), slow upper-plate thrust rupture striking nearly perpendicular to the trench near the shelf break. This three-fault compound source reconciles the large basin-wide tsunami, including amplified signals in Hawaii, with minimal seismic/geodetic expression of the slow component. The work identifies an unprecedented intraplate rupture configuration and emphasizes that slow-slip on non-splay upper-plate faults can pose significant tsunami hazards. Future work should focus on high-resolution 3D seismic imaging of the shallow prism to identify potential trench-perpendicular faults, expanded seafloor geodesy to resolve shallow deformation, improved bathymetric monitoring to detect mass-wasting, and refined stress-change analyses to understand triggering mechanisms and coupling gradients.

- Non-uniqueness: While the preferred three-fault model fits all datasets, the solution is not unique; alternate shallow fast-fault geometries with the same slow-slip component can also fit tsunami data. - Geometric uncertainty: The slow-slip fault strike and dip are constrained within roughly ±15°, with ~20 km absolute position uncertainty along the shelf break. - Observational gaps: Onshore geodetic and InSAR cannot resolve the proposed slow-slip; lack of pre/post-event high-resolution bathymetry and dense reflection profiles limits independent confirmation of the slow-slip fault geometry. - Rejected alternatives: Slow slip on the shallow megathrust or trench-parallel splay faults can match tsunami but contradict hr-GNSS observations; slump interpretations are geometrically and geodetically challenging, though not entirely excluded without dedicated high-resolution seafloor surveys. - Seismic detectability: The long duration and location of the slow-slip reduce seismic spectral amplitudes and alter phasing, hindering definitive detection in long-period surface waves.