Complex tsunamigenic near-trench seafloor deformation during the 2011 Tohoku-Oki earthquake

The 2011 Tohoku-Oki (Mw 9.1) earthquake exhibited unprecedented shallow rupture behavior, but near-trench coseismic slip remains poorly resolved due to limited offshore geodetic data near the trench. Existing rupture models based on seismic, geodetic, and tsunami data disagree on whether maximum slip peaked at the trench axis or down-dip and on the northern extent of large near-trench slip, which is critical for explaining the devastating Sanriku tsunami. Inversions using seismic/geodetic data tend to limit large slip south of ~39°N, while tsunami-informed inversions allow extensions northward. The scarcity of near-field observations (few GPS-acoustic stations >50 km from trench; one OBP with large horizontal uncertainty) hampers resolution. Differential bathymetry can capture offshore deformation but historically has low horizontal resolution, complicating interpretation where slip is highly heterogeneous. This study aims to improve horizontal resolution of differential bathymetry to clarify near-trench slip behavior, assess off-fault deformation in the frontal prism, and delineate the northern extent of trench-breaching slip.

Prior studies using differential bathymetry revealed large slips approaching the trench and uplift on the outer landward slope, challenging the concept of an aseismic frontal wedge. However, previous bathymetric matching used large data windows to reduce noise, yielding horizontal resolutions of tens of kilometers and assuming spatial homogeneity—an approach that can bias results where slip is highly variable. Multidisciplinary inversions provided contrasting shallow slip distributions: some place peak slip at the trench axis, others down-dip; and estimates of northern slip extent vary, with tsunami-based inversions often suggesting extension beyond 39°N. Nearby tracks (e.g., MY101, MY102) have shown trenchward-increasing slip, whereas other data suggest variability along strike. There is debate on whether near-trench slip direction north of 39°N rotates to accommodate trench geometry. The role of off-fault processes (inelastic wedge deformation, submarine slope failures, and potential splay faulting) has been proposed to influence vertical seafloor deformation and tsunamigenesis. This study builds on these findings by enhancing the resolution and uncertainty characterization of differential bathymetry.

Data: Multibeam bathymetry collected by R/Vs Natsushima, Kaiyo, Yokosuka, Mirai, and Kairei (2001–2012) from JAMSTEC. Foreshock/afterslip contributions were estimated to be ≤1–2 m and assumed negligible relative to coseismic signals. Data were manually edited (removing line transitions, invalid points), median-filtered for outliers, projected to Gauss plane coordinates, and gridded at 20 m to preserve short-wavelength morphology. Data correction and registration: Following Kodaira et al., pre- and post-event datasets were geo-coordinated assuming minimal change on the outer rise. Horizontal offsets were estimated via correlation matching over the seaward outer-rise overlap, removed, then vertical offsets were estimated as the median depth difference over the seaward slope and removed. Landward slope coseismic displacements were referenced relative to the outer-rise. Segmentation and windowing: Profiles were segmented with centers every 5 km along dip. Track 1 used 5 km screen windows (also 10 km for verification). Track 2 used 6 km windows (10 km for verification) due to sparser data. To reduce SVP/tide/ship-loading biases, the median depth within each window was subtracted so matching relied on contours. Data with grazing angles >40° were excluded. Horizontal displacement estimation: Tracks were subdivided along trench-normal direction; pre/post-event bathymetry within each window were matched independently by correlation. For track 1, displacement estimates were projected to the dip-slip direction (slip direction constrained by GPSA). For track 2, estimates were projected east–west due to uncertain slip azimuth north of 39°N. Reduced window sizes increased spatial resolution while acknowledging local heterogeneity; an iterative framework handled athwart bias and outliers. Uncertainty estimation: Uncertainties were derived using variance of unit weight based on the seaward slope, where true coseismic slip is negligible. Fine-scale matching on the seaward slope produced a sample to compute error variances; uncertainties for each estimate scale with a weight term. Uncertainty is larger in the north–south component than east–west (due to trench-parallel morphology) and larger on the seaward slope than landward, with typical horizontal uncertainty on the order of ~20 m. Vertical displacement estimation: Depth differences were binned at 500 m along dip; medians computed per bin to form vertical deformation profiles. For track 1, depth-dependent bias in multibeam measurements produced a slight slope in differences; a robust regression against depth was used to remove this bias. The typical uncertainty of vertical estimates is within ~2 m. Tsunami inversion profiles at 20 km and 40 km resolution were used for comparison. Quality control and interpretation: Outer-rise estimates served as noise benchmarks. Large near-trench horizontal estimates on the seaward slope were treated cautiously due to smooth morphology and temporal changes. Spatial correlations with structural features (backstop interface outcrop, trench-parallel ridge, potential splay fault) and geomorphic signatures (landslide deposits, subsidence) were examined to interpret off-fault processes.

- Track 1 (∼38.2°N, main rupture region): - Two landward-slope zones with distinct behavior. - Zone I (∼26 km up-dip span): Strong trenchward decay of slip; horizontal displacement ~50 m at ~40 km from trench, decreasing to ~20 m near zone’s east end. Vertical uplift >10 m at the western end, decreasing toward the trench. Interpreted as velocity-strengthening shallow fault behavior; contrasts with nearby MY101/MY102 tracks that showed trenchward-increasing slip. - Zone II: Lower horizontal displacement with slight seaward taper; vertical uplift trend increases trenchward. The mismatch between horizontal and vertical trends indicates off-fault processes. The inversion of uplift trend coincides with the backstop interface outcrop, consistent with inelastic deformation of unconsolidated frontal prism sediments that efficiently convert rupture energy to uplift while reducing shallow megathrust slip. A pronounced uplift peak (>15 m) at a trench-parallel ridge is inconsistent with the estimated horizontal slip (~22 m vs. ~50 m needed for such uplift at ~5° dip), suggesting contributions from submarine slope failure (landslide) and possible activation of a landward-dipping ancillary splay fault beneath the ridge. Near the trench axis, horizontal estimates increase abruptly but are statistically uncertain; vertical deformation decreases sharply at the trench axis, with seaward-slope subsidence and trench-axis shallowing attributed to mass wasting. - Outer-rise estimates fluctuate around zero, informing realistic uncertainty; despite uncertainties, a robust trenchward decrease in horizontal displacement on the landward slope is established. - Track 2 (∼39.05°N, northern region): - Landward slope (~17 km) exhibits coherent trenchward motion with maximum east–west horizontal displacement >20 m near the trench; statistically significant given uncertainties. - Demonstrates that large trench-breaching slip extended immediately north of 39°N; further north, slip decays rapidly with a strong meridional gradient (~20 m over ~10 km), delineating the northern boundary of the main rupture. - Vertical uplift peaks immediately landward of the trench axis; negative correlation between uplift and horizontal displacement trends again indicates dominant off-fault deformation in the frontal prism. - Integrative implications: - At ~38.2°N, relatively low slip proximal to the trench supports models placing the maximum slip down-dip and aligns with sea-surface displacement from tsunami inversion and multidisciplinary slip models. - The uplift trend inversion near the backstop interface and similar patterns on track 2 support pervasive inelastic deformation of the frontal prism during the earthquake, which enhances tsunamigenic uplift while reducing shallow slip and absorbing high-frequency energy. - Strong along-strike heterogeneity is evident: co-existence of velocity-strengthening and weakening behaviors in a compact near-trench region; northern terminus of trench-breaching slip aligns with distribution of smectite-rich pelagic clay on the incoming plate, suggesting material control. - Methodologically, enhanced differential bathymetry achieved <10 km horizontal resolution with ~20 m horizontal accuracy, enabling finer-scale mapping of near-trench deformation.

The study addresses the poorly constrained near-trench slip behavior of the 2011 Tohoku-Oki earthquake by improving the spatial resolution of differential bathymetry. The findings show that in the central corridor near ~38.2°N, shallow fault behavior includes velocity-strengthening segments where slip decays toward the trench, contrasting with adjacent areas where slip peaks at the trench. This supports source models with maximum slip down-dip of the trench and is consistent with independent tsunami and geodetic inferences. The observed inversion of the vertical uplift trend near the backstop interface, together with a trenchward-tapering horizontal displacement, cannot be explained by elastic fault slip alone. The data implicate inelastic deformation of the unconsolidated frontal prism as a key process, with localized contributions from mass wasting and possible splay faulting. This mechanism efficiently transfers rupture energy into seafloor uplift while diminishing shallow slip, thereby enhancing tsunami generation. North of 39°N, track 2 demonstrates that large near-trench slip extended just beyond 39°N but decayed rapidly further north, reconciling conflicting geodetic and tsunami-based inversions and clarifying the source region relevant to the large Sanriku run-up. The similarity of horizontal/vertical mismatches across both tracks suggests that inelastic wedge deformation may be widespread during such events. These results highlight the strong along-strike heterogeneity of shallow megathrust friction and the importance of material properties (e.g., smectite-rich pelagic clays) in controlling trench-breaching slip. Overall, the study underscores that elastic-only models oversimplify outer wedge deformation and that improved near-trench mapping is essential for accurate tsunami hazard assessment.

This work develops and applies an enhanced differential bathymetry approach that increases horizontal resolution (<10 km) while maintaining moderate accuracy (~20 m), enabling detailed mapping of near-trench coseismic deformation during the 2011 Tohoku-Oki earthquake. Key contributions include: (1) identification of a velocity-strengthening shallow fault segment in the main rupture region where slip decreases toward the trench; (2) recognition that off-fault inelastic deformation within the frontal prism, along with localized mass wasting and possible splay faulting, predominantly controls near-trench uplift and tsunamigenesis; and (3) delineation of the northern extent of trench-breaching slip immediately north of 39°N, beyond which slip decays rapidly. These results reveal pronounced spatial heterogeneity of shallow rupture behavior and emphasize the role of wedge material properties in governing slip to the trench. Future work should expand repeated bathymetric surveys to improve coverage, integrate bathymetry with near-field seafloor geodesy (GPSA, OBP) and seismic imaging, refine models of inelastic wedge processes, and apply these methods to other subduction margins for improved tsunami hazard evaluation.

- Near-field geodetic observations remain sparse near the trench; GPSA sites are >50 km from the trench and the OBP gauge has large horizontal positioning uncertainty (~20 m). - Differential bathymetry matching has inherently large horizontal uncertainties (~20 m), larger on the smoother seaward slope and for north–south components; abrupt large near-trench horizontal estimates on the seaward slope are statistically insignificant due to smooth morphology and temporal bathymetric variations. - Horizontal displacement directions are poorly constrained in some regions (e.g., north of 39°N), requiring projection assumptions (dip-slip for track 1, east–west for track 2). - Vertical estimates require correction for depth-dependent biases; residual artefacts from data quality, matching residuals, or earthquake-triggered slope failures may affect local features. - Registration assumes minimal outer-rise change; unmodeled foreshock/afterslip (estimated ≤1–2 m) could slightly bias differential results. - Localized contributions from landslides or splay-fault activation complicate attribution of uplift sources in specific segments.