print page print page

Dynamic rupture simulation of the 2011 Tohoku earthquake

On March 11th 2011, a violent Mw=9 earthquake stroke Japan causing 28000 victims, including casualties and missing. This earthquake triggered a devastating tsunami causing severe damage in cities and nuclear facilities along the Japanese coast. The amount of data recorded from this earthquake is exceptional, with thousands of broadband, strong motion and continuous GPS sensors located all over Japan. This provides a great opportunity for seismologist and engineers to deeply investigate the rupture process to better understand the physics of this type of earthquakes and their associated effects like tsunamis.

Based on the compilation of seismological, geodetic, bathymetric and tsunami observations as well as source inversion and back-projection studies of the 2011 Mw 9.0 Tohoku earthquake show that this earthquake is mainly characterized by three main features:

  1. unusual large slip over 50m;
  2. complex rupture patterns with multiple rupture fronts and slip reactivation;
  3. distinct depth dependent frequency radiation patterns, in which the shallower part of the fault dominate the low frequency radiation and the deep part dominates the high frequency radiation.

In order to explain such observations we examine the complex rupture process of this event by means of dynamic rupture simulations. We employ the 3D spectral element code SPECFEM3D-SESAME, in which we recently implemented the capability of solving for dynamic fault rupture. Through the usage of a full-featured software toolkit CUBIT for robust unstructured mesh generation, our model can incorporate a non-planar geometry of the megathrust interface. Figure(a) shows the geometry of Japanese subuction zone done with CUBIT and Figure(b) shows a view of the megathurst interface and the bottom surface meshed with linear coarsening in the west-east direction.

Using this unstructured mesh we set up a model including a linear slip-weakening friction law in which frictional strength drops initially a certain value, but then drops down again if the fault is subjected to large slips. This frictional law has been proposed by Kanamori and Heaton (2000), who claim that earthquakes under extremely large slip undergo fault melting, pressurization, lubrication or other thermal weakening mechanisms that reduces further the frictional strength to lower levels. Additionally we add to this model large asperity patches of radius between 30 to 50 km in the intermediate and shallower part of the slab, and small patches of asperities at the bottom of the slab to account for the strong high frequency radiation. The very shallow part of the fault is considered as a stable zone that operates during rupture with an enhanced energy absorption mechanism. We model this zone by assuming negative stress drop and large critical slip distance.

Using the above model we reproduce the main patterns of 2011 Tohoku earthquake, generating tremendous final slip bigger than 50m in the central part of the slab and about 30m in the sallow part close to the trench, see figure(c). This huge slip close to the trench may be one of the key features to the generation of such a big tsunami followed by this earthquake. Finally, our simulations show sharp peak slip velocities at the bottom of the slab, see figure (d), which is in agreement with high frequency energy radiation observed from back-projection studies, Meng et. al. (2011).

typo3 by akea
Open Source Web Design - original design by tri-star web design