Through a series of iterative simulations of the dual source tsunami, we found the SMF-triggering time delay that led to the best match with the observed arrival time of higher-frequency waves at the GPS and DART buoys (Fig. 2) to be ti = 135 s. This timing is consistent with the 110 s delay obtained from tsunami
Beta-Sheet Breaker (e.g., Takagawa and Tomita, 2012). It can be explained by the finite time of shear wave propagation from the earthquake hypocenter to the toe of the accretionary prism ( Yue and Lay, 2011), and loss of sediment shear strength from liquefaction along the basal décollement that is normal in SMF. SMF motion terminates at t = ti + tf = 198 s (3.3 min) from the time of earthquake rupture; this total time is identical to the delay independently found from the simplified travel time analysis of Fig. 3. As seen in Fig. 7b, there is downward seafloor displacement at the rear of the SMF and upward at the front, as expected for a rotational failure and confirmed by the post-earthquake bathymetry of the SMF source used ( Fig. 5). The free surface elevations of the SMF generated wave at t = 300 s are in the range − 10 to 16 m ( Fig. 10 and Fig. 11). As expected from similar earlier work (e.g., Day et al., 2005 and Tappin et al., 200
cool , at
pollen grains and later times, the tsunami wave train generated by the SMF located north of the main earthquake rupture, is more directional than waves generated by the earthquake source with, despite the cylindrical pattern of generated waves, higher amplitude waves being generated near the SMF azimuthal direction of motion (axis). The SMF generated waves also are shorter, higher-frequency (and thus more dispersive) waves than those generated to the south by the earthquake ( Fig. 2, Fig. 10 and Fig. 11). The results of the dual source tsunami are presented as two videos: an overall view of the tsunami and a close-up in the region near the Sanriku Coast (for explanation of video see caption for Fig. 11).
