The 2011 Tohoku-Oki earthquake (Mw 9.1) was one of the most well-recorded earthquakes, revealing unprecedented shallow rupture behaviors that significantly advanced our understanding of megathrust rupture dynamics and tsunamigenesis. However, critical details about coseismic seafloor slip in the near-trench region, where ruptures are highly tsunamigenic, remain unclear. Existing multidisciplinary observations have led to numerous rupture models, but these vary significantly in their spatial extent and distribution patterns near the trench, reflecting fundamentally different fault mechanisms. While a large slip approaching the trench is established, there's no consensus on whether the maximum slip occurred directly at the trench axis or further down-dip. Similarly, whether this large, trench-peaking slip is a common feature or reflects site-specific geology remains unresolved. The northern extent of the large near-trench slip, crucial for understanding the devastating tsunami along the Sanriku coast, is also debated, with some models suggesting it extends north of 39°N, while others do not. This discrepancy stems largely from the scarcity of offshore geodetic observations, particularly near the trench, with limited GPS-acoustic (GPSA) stations and ocean-bottom pressure (OBP) gauges providing quantitative data. Differential bathymetry offers a unique alternative, but previous studies have suffered from low horizontal resolution, blurring crucial morphological features. This study aims to improve the horizontal resolution of differential bathymetry estimates to analyze coseismic slip behavior near and across the trench at a finer scale, addressing the existing uncertainties and providing a more refined understanding of the near-trench tsunamigenic processes.

Numerous studies have attempted to model the 2011 Tohoku-Oki earthquake rupture using various datasets including seismic, geodetic, and tsunami data (Sun et al., 2017; Lay, 2018; Wang et al., 2018; Lay et al., 2011; Yue & Lay, 2011; Romano et al., 2014; Minson et al., 2014; Jiang & Simons, 2016; Wang & Kinoshita, 2013; Uchida & Bürgmann, 2021). These studies have yielded varying results regarding the spatial extent and distribution of coseismic slip, particularly near the trench. Ozawa et al. (2011) showed evidence for large coseismic and postseismic slip, while Satake et al. (2013) inferred coseismic slip distribution from tsunami waveform data. Hasegawa & Yoshida (2015) reviewed preceding seismic activity and slow slip events, and Yamazaki et al. (2018) presented a self-consistent fault slip model. However, inconsistencies remain, especially concerning the location of maximum slip and the northern limit of large near-trench slip (Simons et al., 2011). The limited availability of near-trench geodetic observations, such as GPSA and OBP data (Kido et al., 2011; Sato et al., 2011; Ito et al., 2011), has hindered a comprehensive understanding of the near-trench rupture process. Previous studies utilizing differential bathymetry (Fujiwara et al., 2011; Kodaira et al., 2022) have shown promise but have been limited by the low resolution of horizontal displacement estimates (Fujiwara et al., 2021; Kodaira et al., 2020; Maksymowicz et al., 2017). This limitation has led to uncertainties in the interpretation of near-trench coseismic deformation and its contribution to tsunami generation.

This study utilizes differential bathymetry data from multiple research vessels (Natsushima, Kaiyo, Yokosuka, Mirai, and Kairei) collected between 2001 and 2012. The data underwent preprocessing steps including manual editing, outlier filtering (median filter), and projection onto the Gauss plane coordinate system, before being gridded into 20 m intervals. Data correction was performed following Kodaira et al. (2012), initially geo-coordinating the pre- and post-event datasets based on the assumption of minimal bathymetry change in the outer-rise region. Horizontal and vertical offsets were estimated through correlation matching and median depth difference calculations, respectively. To improve resolution, the profiles were segmented into blocks with 5 km (Track 1) or 6 km (Track 2) center spacing along the dip direction. A sliding window approach was used for bathymetry correlation matching, with window sizes of 5 km and 10 km (Track 1) and 6 km and 10 km (Track 2). A median depth subtraction within each window was implemented to remove absolute depth information and focus on contour information, and data with grazing angles >40° were excluded. Horizontal displacement amplitude was calculated by projecting the bathymetry matching results onto the slip-dip direction (Track 1) or the east-west direction (Track 2). Vertical displacement was determined by binning depth difference values along the dip-slip direction at a 500 m window scale, computing the median of each bin, and accounting for a depth-dependent bias through robust fitting. Uncertainty estimations were performed for both horizontal and vertical displacements. The Matlab codes used for correlation matching are available at [https://codeocean.com/capsule/5251089/tree](https://codeocean.com/capsule/5251089/tree).

Analysis of Track 1, located in the main rupture region, revealed a complex along-track variation of seafloor deformation. The landward slope was subdivided into two zones: Zone I, characterized by a trenchward decay of slip amplitude, indicating velocity-strengthening fault friction; and Zone II, showing a relatively low horizontal displacement with a trenchward increase in vertical uplift. This mismatch between horizontal and vertical deformation in Zone II suggests inelastic deformation of unconsolidated sediments in the frontal prism as a primary factor. Submarine slope failure and splay fault activation were also identified as contributing factors to the observed uplift. Track 2, located north of the main rupture region, showed coherent trenchward movement of the frontal prism with a maximum horizontal displacement exceeding 20 m, providing clear evidence that large coseismic slip extends beyond 39°N. The seafloor uplift pattern in Track 2 is negatively correlated with the horizontal displacement trend, further supporting the importance of off-fault inelastic deformation. The study's improved resolution reveals a velocity-strengthening behavior near the trench in the central corridor of the main rupture region, contrasting with the coseismic weakening observed in nearby areas. The inversion of the seafloor uplift trend in the frontal prism, coinciding with the backstop interface outcrop, strongly suggests the dominant role of inelastic deformation. This inelastic deformation pattern closely resembles that derived from tsunami inversion results, highlighting its relevance to tsunami excitation. The study also delineated the northern boundary of the main rupture extension, showing a sharp decrease in slip northward of 39°N. The uncertainty analysis, based on increased sample sizes on the seaward slope, suggests that the trenchward decrease in horizontal displacement amplitude in the landward slope is a robust feature.

The findings address the unresolved issues concerning the near-trench rupture behavior of the 2011 Tohoku-Oki earthquake by providing high-resolution spatial information on coseismic seafloor deformation. The velocity-strengthening behavior observed in Track 1 contrasts with previous models suggesting consistent trenchward slip increase, supporting models where maximum slip occurs down-dip from the trench. The significant off-fault inelastic deformation revealed in both Track 1 and Track 2, particularly its correlation with the backstop interface outcrop, provides a plausible explanation for the observed mismatch between horizontal and vertical displacements and its direct relevance to tsunami generation. The delineation of the northern boundary of the large coseismic slip contributes significantly to understanding the tsunami's impact on the Sanriku coast. The strong spatial heterogeneity in shallow rupture behavior observed across different locations highlights the complexity of the rupture process and the inadequacy of relying solely on limited near-field geodetic observations. The dominant role of inelastic deformation in the frontal prism, particularly in generating tsunami, challenges the traditional elastic-dominant model and suggests its importance in tsunami potential evaluation worldwide.

This study significantly advances our understanding of near-trench coseismic deformation during the 2011 Tohoku-Oki earthquake through high-resolution differential bathymetry analysis. The findings highlight the importance of inelastic deformation in the frontal prism, its close connection to tsunami generation, and the spatial heterogeneity of shallow rupture behavior. The improved resolution obtained using the refined methodology allows for a more accurate characterization of the complex interactions between fault slip and off-fault deformation. Future research should focus on further refining the models of inelastic deformation and incorporating them into more comprehensive tsunami generation models. Additionally, expanding the application of this refined methodology to other subduction zones would be highly beneficial.

The study relies on bathymetry data from multiple sources collected over a period of time. Although the impact of foreshocks and afterslips was considered minimal, their influence cannot be completely ruled out. The inherent uncertainties associated with bathymetry matching, particularly in the seaward slope, could affect the precision of displacement estimates. The interpretation of off-fault deformation relies on several assumptions and may be affected by other unmodeled processes. The study focuses on two specific tracks and may not fully capture the overall complexity of near-trench deformation along the entire rupture zone.