T43A-2959
Numerical simulation of earthquake rupture sequences on the Manila thrust fault: Effects of seamount subduction

Thursday, 17 December 2015
Poster Hall (Moscone South)
Hongyu Yu1, Yajing Liu2, Jieyuan Ning3, Changrong He4 and Lei Zhang4, (1)ITAG Institute of Theoretical and Applied Geophysics, Peking University, Beijing, China, (2)McGill University, Montreal, QC, Canada, (3)Peking University, School of Earth and Space Sciences, Beijing, China, (4)Institute of Geology, China Earthquake Administration, Beijing, China
Abstract:
The Manila subduction zone is located at the convergent boundary between the Philippine Sea Plate and the Sunda/Eurasian Plate from offshore Taiwan to northern Luzon of Philippines, where only infrequent M7 earthquakes were observed in modern seismological instrumentation history. The lack of great events (M8+) indicates the subduction fault is either aseismically slipping or is accumulating strain energy toward rapid release in a great earthquake. Here we conduct numerical simulations of earthquake rupture sequences in the framework of rate-state-friction along the 15-19.5ºN segment of the 3D plate boundary with subducted seamounts. Rate-state frictional properties are constrained by laboratory friction experiments conducted on IODP Expedition 349, South China Sea (SCS), drilling samples from the basaltic basement rock under 100ºC – 600ºC, effective normal stress of 50 MPa and pore pressure of 100 MPa.

During the modeled 2000-year period, the maximum magnitude of earthquakes is Mw7. Each sequence repeats every ~200 years and is consisted of three sub-events, event 1 (Mw7) that can overcome the barrier, where dip angle changes most rapidly along the strike, to rupture the entire fault. Events 2 (Mw 6.4) and 3 (Mw 5.7) are of smaller magnitudes and result in north-south segmented rupture pattern. We further quantify the potential of earthquake nucleation by the S-ratio (lower S ratio means the initial stress is closer to peak strength, hence more likely to nucleate an earthquake). The subducted seamount shows higher S-ratios than its surroundings mostly, implying an unlikely nucleate area. Our results are qualitatively similar to 2D subduction earthquake modeling by Herrendörfer et al. (2015, 2-3 events per supercycle and median long-term S is 0.5-1).

Finally, we plan to use our coseismic rupture model results as inputs for a tsunami propagation model in SCS. Compared to the kinematic seafloor deformation input, our physics-based earthquake source model and its ability to explore a broad range of parameters, including the location and shape of subducted seamounts, will help improve earthquake and tsunami hazards assessment and mitigation in the populated regions around the SCS.