Geological evidence for repeated slip-to-the-trench style megathrust earthquakes at the Japan Trench Open Access

The large amount (>40 m) of shallow slip documented on the megathrust fault during the 2011 CE M_w 9.1 Tōhoku-oki earthquake contributed to the unexpectedly large tsunami that devasted the Japan coastline (Ide et al., 2011; Lay, 2018; Kodaira et al., 2021). This first documentation of coseismic slip propagation along the décollement up to the trench axis revolutionized understanding of the seismogenic portion of subduction zones (e.g., Chester et al., 2013). At present, it is unclear whether the observed “slip-to-the-trench” type behavior was anomalous or whether it could be characteristic of megathrust earthquakes at the Japan Trench and other subduction zones. Establishing if, and how often, slip-to-the-trench occurred during past megathrust earthquakes is therefore key for diagnosing future hazard.

Several data sets offer information on past megathrust earthquakes at the Japan Trench. Perhaps most comprehensively, sedimentary records of tsunami deposits (e.g., Goto et al., 2019; Sawai, 2020) and submarine seismoturbidites (e.g., Ikehara et al., 2016; Usami et al., 2018; Kioka et al., 2019; Strasser at al., 2024) constrain the age of several earthquakes attributed to slip on the megathrust fault throughout the Holocene. This includes the 869 CE Jogan earthquake, which may have been a predecessor of the Tōhoku-oki earthquake in terms of its size (Fig. 1A; Sugawara et al., 2012; Namegaya and Satake, 2014). However, because these paleoseismic event deposits represent only indirect or “off-fault” evidence of past earthquakes, it remains challenging to constrain the up-dip limit of rupture, i.e., whether the shallowest part of the megathrust fault hosted coseismic slip. Alternatively, “on-fault” evidence from thermal biomarkers in rock samples recovered from the décollement at International Ocean Discovery Program (IODP) Site C0019 (located at ~38°N in the area of highest slip in 2011; Fig. 1A) confirm that coseismic slip on the shallow megathrust fault has occurred repeatedly (Rabinowitz et al., 2020), but the timing of these events is unknown. The inability of these data sets to conclusively link shallow slip to past megathrust earthquakes means the recurrence of slip-to-the-trench behavior is currently unconstrained.

Data sets that definitively document slip to the trench for the Tōhoku-oki earthquake include differential bathymetry, seismic data, sediment cores, and observations during submersible dives that depict seafloor uplift, thrust-up structures, and the actual fault scarp in the trench (Fujiwara et al., 2011; Kodaira et al., 2012; Ueda et al., 2023). Pre- and post-earthquake seismic reflection profiles through a trench-fill basin immediately seaward of the zone of highest slip (near IODP Site C0019; Fig. 1A) show kilometer-scale coseismic slumping of the prism toe facilitated by slip on the shallow megathrust, causing imbrication of almost the entire trench-fill sediment sequence (Kodaira et al., 2012; Strasser et al., 2013; Fig 1B). In this regard, subsurface imaging of the coseismically deformed trench-fill stratigraphy uniquely captures geologic evidence for slip-to-the-trench behavior during the Tōhoku-oki earthquake.

If older earthquakes generated similar coseismic deformation to trench-fill sediments, evidence of slip-to-the-trench behavior may be recorded in the deeper basin stratigraphy or in other basins along strike. Hence, this study aims to link (1) interpretations of deformed trench-fill stratigraphy imaged in subsurface data with (2) event stratigraphy from seismoturbidites in sediment cores as a means to explore if and how often shallow slip might have been associated with past megathrust earthquakes at the Japan Trench.

Selecting an appropriate basin in which to study past coseismic deformation is complicated because a number of factors influence whether evidence is recorded and preserved by trench-fill sediments. Schottenfels et al. (2024) showed how subducting horst-graben structures define (1) the position of the décollement as it approaches basins at the trench axis and, therefore, (2) how slip can propagate into trench-fill sediments. For example, at the trench basin near Site C0019 (Fig. 1A), the alignment of the overriding prism toe with the subducting graben wall creates a large stepdown of the décollement, which was key in promoting gravitational slumping during the Tōhoku-oki earthquake (Fig. 1B; Strasser et al., 2013; Ueda et al., 2023). So, targeting trench basins with similar geometry may increase the chances of detecting evidence of past coseismic deformation recorded in the trench-fill stratigraphy.

However, the scale of deformation observed at the trench basin near Site C0019 (within the 2011 high-slip zone; Fig. 1A) presents challenges for studying older earthquake records. If the entire trench-fill sequence were imbricated by coseismic slumping in each earthquake, evidence of older deformation would be overprinted. Following the assumption that past megathrust earthquakes at the Japan Trench occurred in a similar area as the Tōhoku-oki event (Ikehara et al., 2016; Usami et al., 2018), basins located at the edges of the 2011 high-slip zone may offer an optimal balance between coseismic deformation that is (1) large enough to be resolved in subsurface imaging data, and (2) small enough that sedimentation between successive events allows the burial and stratigraphic preservation of deformation signatures. Based on this, we identify “Basin C2” (as in Strasser et al., 2023), located at 38.75°N just north of the 2011 high-slip zone, as an ideal study site (Fig. 1A). Here, the position of the currently subducting graben relative to the overriding prism toe is in a state facilitating locally enhanced slip at the trench axis (Fig. 1C; Ueda et al., 2023), but differential bathymetry indicates minimal seafloor uplift in 2011 (Fig. 1A; Kodaira et al., 2021; Zhang et al., 2023).

Nakamura et al. (2023) displayed the subsurface stratigraphy of Basin C2 in high-resolution multichannel seismic (HRMCS) line HDMY093, perpendicular to the trench. We reprocessed HDMY093 by applying deghosting, dephasing, and corrections for airgun bubble effects, followed by surface-consistent deconvolution, pre-stack Kirchhoff time migration using a refined velocity model, and post-migration noise attenuation. With this, we gained better vertical resolution (in the order of ~5 m), required to interpret subtle details indicating deformed stratigraphy in the shallow trench fill (see Figs. S3–S4 in the Supplemental Material¹).

In previous work, Kioka et al. (2019) used trench-parallel hydroacoustic sub-bottom profiler (SBP) data with decimeter vertical resolution to map “event beds” in the upper ~40 m of trench-fill basin stratigraphy. Event beds are characterized by acoustically transparent units that pond above prominent high-amplitude reflections. We mapped the same SBP-scale event beds onto an intersecting trench-perpendicular SBP line colocated with our reprocessed HRMCS profile across Basin C2 (Fig. 2D).

Strasser et al. (2024) recently linked the SBP-scale event beds mapped by Kioka et al. (2019) to >0.5-m-thick, homogenous facies overlying the coarse, sandy bases of seismoturbidites in ~40 m sediment cores collected during IODP Expedition 386. Note that the seismoturbidite associated with the Tōhoku-oki earthquake is too thin to be resolved by HRMCS and SBP data in Basin C2. In the longest sediment core (M0083F; Fig. 1C; see Fig. S5 for more detail), the uppermost SBP-scale seismoturbidites correspond to the historical 1454 CE Kyotoku and 869 CE Jogan earthquakes (older seismoturbidites have unconstrained ages), providing chronostratigraphic tie points for interpreting the trench-fill sequence (Strasser et al., 2024).

To test for stratigraphic correlation between seismoturbidites and potential coseismic deformation signatures in the trench-fill sequence, we examined the HRMCS, SBP, and sediment core data sets at the same vertical scale (Figs. S1–S2). Good stratigraphic agreement allowed us to tie the base of seismoturbidites traced by highest-amplitude reflections in the SBP data with prominent high-amplitude reflections in the HRMCS data. This correlation constrained the P-wave velocity for time-to-depth conversion in the trench fill to be ~1600 m/s (as in Strasser et al., 2024), allowing measurement of subsurface features in HRMCS data.

At the landward (western) edge of Basin C2, the trench-fill sequence exhibits parallel horizontal reflections that dip gently westward along distinct horizons A, B, C, and D (at ~1.0°, 1.4°, 1.5°, and 1.8°, respectively; Fig. 2A). We interpret these horizons as basal detachment surfaces, from which westward-dipping (~30°) thrust faults branch upward to form distinct, small imbricate thrust wedges (~10–35 m height, 350–750 m length) that taper eastward. At their proximal (western) end, the high-amplitude detachment surfaces connect downward toward the décollement, i.e., the main megathrust fault zone (“i” in Fig. 2A). This suggests that the interpreted frontally confined compression of trench-fill sediments is linked to lateral (west-to-east) propagation of the frontal prism. The detachment surfaces also connect upward to eastward-dipping listric reflections, interpreted as sliding surfaces of gravitational slumps (“ii” in Fig. 2A). The cross-cutting relationships of imbricate thrust wedges A, B, C, and D are therefore consistent with episodic deformation of trench-fill sediments and paleo-seafloor uplift caused by displacement along the shallow megathrust fault and locally enhanced by gravitational instability at the prism toe (cf. Ueda et al., 2023).

SBP data along the same transect through Basin C2 supports the interpreted deformation at the landward basin edge, where a distinct convex-upward structure (~4 m high) scatters the acoustic signal between ~10.01 and 10.04 s two-way traveltime (Fig. 2B). The easternmost edge of the scattered feature cuts the parallel reflections of undeformed trench-fill stratigraphy along a westward-dipping interface (Fig. 2B), which directly corresponds to the distal-most thrust fault at the snout of imbricate thrust wedge A in the HRMCS data (Fig. 2A). The acoustically transparent unit mapped in blue in Figure 2 corresponds to the seismoturbidite from the 869 CE Jogan earthquake (Strasser et al., 2024) and directly onlaps the top of up-thrusted sediments in imbricate thrust wedge A (Fig. 2B). High-amplitude reflections below the Jogan seismoturbidite are parallel and dip slightly toward the thrust fault, whereas high-amplitude reflections above the Jogan seismoturbidite extend almost horizontally over the up-thrusted sediments (Fig. 2B). This sequence means the formation of imbricate thrust wedge A must have occurred immediately before or during the Jogan earthquake in order for the uplifted paleo-seafloor to be draped by the several-meters-thick Jogan seismoturbidite.

In our correlation of data sets in Basin C2, we show that the formation of imbricate thrust wedge A is associated with displacement along the décollement and that this occurred immediately before the deposition of the Jogan seismoturbidite (Fig. 2). Connection to a listric interface at the prism toe (“ii” in Fig. 2A) indicates cogenetic slumping, which likely contributed to localized compression of poorly consolidated trench-fill sediments and uplift of the paleo-seafloor. Together, our observations document strong evidence that the shallow megathrust fault hosted coseismic slip to the trench during the Jogan earthquake, resulting in the deformation of trench-fill sediments in Basin C2. This conclusion reinforces the Jogan earthquake as the closest historical analogue to the Tōhoku-oki earthquake and supports tsunami deposit–based inundation models that estimate similar magnitudes for both events (Sugawara et al., 2012; Namegaya and Satake, 2014).

Moreover, our findings contribute important advances in understanding the up-dip termination of shallow slip for the Jogan earthquake, which until now has been unresolved. Identifying slip to the trench in Basin C2 indicates rupture extended farther north and closer to the trench than was (1) previously estimated for the Jogan earthquake, and (2) documented for the Tōhoku-oki earthquake (Fig. 1A). This implies that the area of shallow slip in great subduction earthquakes can be spatially and temporally variable (Philibosian and Meltzner, 2020). These insights confirm our unique core-to-seismic correlation approach as a valuable method to constrain the distribution of shallow slip in past earthquakes and highlight its potential for identifying similar behavior at other subduction zones.

Our documentation of slip to the trench during the Jogan earthquake bolsters other data sets that hypothesize shallow slip on the Japan Trench megathrust fault has occurred in the past (Rabinowitz et al., 2020; Kodaira et al., 2020). Although the 40 m length of the M0083F core (Strasser et al., 2024) currently limits our conclusive chronostratigraphic correlation to the Jogan event only, the record of trench-fill deformation in the HRMCS data at Basin C2 can be extended much further. In Figure 3, we identify at least five older stratigraphic intervals with similar imbricate thrust wedges that are onlapped by acoustically transparent ponding units consistent with the characteristics of seismoturbidites. Based on repetition of the same sequence of coeval imbricate thrust faulting, paleo-seafloor uplift, and onlapping of thick turbidites, we infer the deeper intervals also represent shallow coseismic slip propagation to the trench during past megathrust earthquakes. Applying our core-to-seismic methods to the full ~120 m trench-fill sequence at Basin C2 could therefore unlock a much longer record of slip-to-the-trench style ruptures—the results of which would be beneficial not only for informing tsunami and earthquake hazard at the Japan Trench but also for advancing global understanding of spatiotemporal patterns of megathrust earthquake behavior.

This study used data from the International Ocean Discovery Program (IODP). The authors thank the science party and crew of IODP Expedition 386 and several other cruises facilitated by the Japan Agency for Marine-Earth Science and Technology, M. Ortler for assistance with maps, as well as G. Moore and C. Regalla for constructive comments that improved this paper. Research was funded by the Austrian Science Fund (FWF) (grant https://doi.org/10.55776/P36809 to Strasser) and Japan Society for the Promotion of Science (grant 23K22586 to Ikehara).