[S-SS32_1AM2] Fault Rheology and Earthquake Dynamics

[SSS32-10] Dynamic backthrust branching: role of barriers, and implications

*Shiqing XU¹, Eiichi FUKUYAMA¹, Yehuda BEN-ZION², Jean-paul AMPUERO³ (1.NIED, 2.University of Southern California, 3.California Institute of Technology)

Keywords:earthquake branching, friction of fault zones, fault barrier

Increasing evidence indicates that backthrusts may become active during or after megathrust ruptures in subduction zones, such as in Chile and Sumatra (Melnick et al., 2012; Singh et al., 2011). Previous studies of relevant mechanisms mainly focused on the interaction between forethrusts and the megathrust. Here we investigate through dynamic rupture simulations how backthrusts may be activated by megathrust ruptures in subduction zone environments. Assuming a single backthrust branch that is backward inclined to the compressional side of a continuous main fault, our results show that (1) fast speed and long propagation distance of the main rupture favor the activation of backthrust; (2) the outward propagation of the activated branch rupture interacts with the main fault mainly in the backward direction, while the tapered slip towards the branch end at the junction affects the main rupture behavior around the junction. We further assume an effective barrier for the main fault at the junction, motivated by the previous studies that barriers of various types (e.g. sharp fault bend, fault end, and transition region with increased basal friction) can also generate backthrusts during the long-term quasi-static process. Compared to the case without barrier, one prominent effect of the barrier is to arrest or delay the forward propagation of the main rupture, such that a resultant backward stress lobe as discussed in Xu and Ben-Zion (2013) can load the backthrust branch over a considerable time. This is particularly important for rupture activation along relatively immature backthrusts within sediments, where the nucleation time leading to the spontaneous propagation phase could be long, due to the large effective Dc, low frictional strength drop, or surface roughness. Indeed, our additional results confirm that the barrier model, although not always necessary, is more favorable for the activation of backthrusts with increased dynamic friction. Our study has several implications: (i) it agrees with the quasi-static model based on the critical taper theory and limit analysis (Cubas et al., 2013) that an increase of basal friction towards the toe may statistically favor the activation of backthrusts near the up-dip limit of megathrust ruptures; (ii) there are also possibilities that backthrusts can still be activated by a propagating rupture, therefore the dynamic critical taper theory developed by Wang and Hu (2006) needs to be improved. In fact, not only the region near the up-dip limit of the seismogenic zone can be pushed into a critical state, successive region around the propagating rupture front within the seismogenic zone can also be temporarily stressed to failure and may even sustain a failure propagation along preexisting branches; (iii) it provides a specific example of compressional-side antithetic branching that can support the early speculation of fault behavior at junctions (King, 1986; Andrews, 1989).