On the Possibility of Slip-to-trench Rupture in Cascadia Megathrust Earthquakes

Tsunamis may be generated by subduction zone earthquakes in four ways: (1) elastic seafloor deformation of the upper plate induced by a buried rupture, (2) enhanced seafloor uplift due to splay faulting, (3) seaward motion of the sloping seafloor due to slip-to-trench rupture, and (4) activation of multiple thrusts and back-thrusts due to sudden shortening of the frontal accretionary prism. Cascadia megathrust rupture models previously developed for tsunami hazard assessment include the first two scenarios. The 2011 Mw 9.0 Tohoku-oki earthquake which exhibited dramatic coseismic slip at the trench raised a new question. Can the shallowest portion of the Cascadia megathrust also slip to trench in great earthquakes as in the Tohoku-oki earthquake or would it normally resist coseimic rupture but creep aseismically after the earthquake as in the 2005 Mw 8.7 Nias earthquake? We reanalyzed seismic images from marine multichannel seismic surveys conducted in 1985 and 1989 with a new focus on the accretionary wedge deformation front. The incoming plate at Cascadia is blanketed by ~3km sediment near the deformation front. Off Vancouver Island, deformation style varies along the subduction margin. In a southern portion there are multiple thrusts dipping landward. Half-way north the vergence changes to dominantly back-thrusting. Farther north, in the Explorer segment, both seaward and landward vergent thrusts are present. This is in sharp contrast to the sediment-starved Japan trench where one continuous decollement extends all the way to the trench, a structure style that facilitates slip-to-trench rupture. Given the complex structure at Cascadia's deformation front, slip-to-trench rupture appears to be much less likely. Scenarios (1) and (4) may be the more likely source scenarios for Cascadia. However, for tsunami hazard assessment, all the rupture scenarios should be considered even for the low probability slip-to-trench rupture scenario involving frontal thrusts with structural details very different from the Japan trench. Therefore, we developed a suit of rupture models using a 3D dislocation model based on the latest structural knowledge, rupture mechanics and observations of recent large tsunamigenic earthquakes such as those at the Sumatra, Chile, and Japan subduction zones.