We utilize the slowness-enhanced back projection (SEBP) and joint finite fault inversion (FFI), which combines body waves, surface waves, and 3D ground displacements to image the rupture process and slip distribution of the Mw7.4 Maduo earthquake. The results indicate a 160-km-long bilateral rupture occurring on a north-dipping fault. The WNW branch ruptures a length of ~75 km at 2.7 km/s, while the ESE branch ruptures a length of ~85 km at 3 km/s. Most slip concentrates above 10 km depth, with several major slip patches up to 5.7 m located on the east segment of the main fault and on the bifurcated branching fault. The comparison between SEBP and FFI shows the consistency between large slip pulses and high-frequency sources. We observe up to 3 m slip with large uncertainty in depth beyond 20 km, which is abnormal since it requires rupture penetrating into the ductile layers. We propose that the apparent deep slip could be an artificial compensation to match the observed moment, which indicates that the crust in the source region is more rigid than the current understanding. Or the deep creeping fault is turned into seismic ones by the strain localization and dynamic weakening. The stress analysis on the forks of the fault demonstrates that the branching behavior on the eastern fork could be well explained by the pre-stress inclination, rupture speed, and branching angle.

We examine the source parameters of four Mw≥7.0 intraslab earthquakes that occurred near the Tohoku coast over the past two decades: 2003, 2011, 2021, and 2022. By analyzing the finite fault slip histories constrained by inland strong motion observations, we found that these earthquakes occurred within the upper plane of the subducted Pacific Plate due to downdip compression caused by plate unbending. These earthquakes have a more compact fault area and higher stress drop compared to shallow crustal earthquakes. Additionally, intraslab earthquakes have much slower relative rupture velocity than shallow crustal earthquakes. Good spatial correlations between the static stress drop and slip rate are found, which may suggest the compatibility between dynamic stress drop and static stress drop. The rupture area, average slip, asperity area, average static stress drop over the entire fault, and asperities are consistent with the reported scaling relationship for global intraslab earthquakes within a similar depth range. Using plate unbending, we found the recurrence intervals of these intraslab earthquakes are around 600 years, which is comparable with that of the 2011 Tohoku earthquake. A visual spatial-correlation between the locations of these earthquakes and seismicity in the lower plane is reported. These findings provide insights into the tectonic background and source parameters of intraslab earthquakes in the Tohoku region and contribute to better seismic hazard assessment.