Estimation of the Breakdown Slip from Strong-Motion Seismograms: Insights from Numerical Experiments

Fukuyama and Mikumo (2007) presented a method to extract informa- tion about the friction controlling earthquake rupture directly from near-field seismo- grams recorded out to several kilometers from a fault. This method computes a parameter (D 00) approximating the breakdown slip (Dc) from the rake-parallel dis- placement at the time of the peak particle velocity. The validation of the method was based on simple 2D steady-state Green's functions, an approach lacking sufficient physics of the rupture process to support their conclusions. Here, we use 3D simula- tions of subshear strike-slip spontaneous rupture propagation to demonstrate that D 00 is almost always controlled by rupture and wave propagation effects, rather than the fault friction process. Only if rupture reaches the Earth's surface and within a short distance from the fault (Rc) is it possible to extract information about Dc from strong-motion data, due to the fast decay with distance from the fault of the seismic energy related to the stress breakdown process. Beyond Rc, D 00 is controlled by the dynamic stress drop from the earthquake rupture without information about Dc. Rc is comparable to the length of the fault cohesive zone where the breakdown process takes place during rupture and approximately equal to 80% of the wavelength associated with the break- down frequency, defined as the reciprocal of the time span required by the stress to drop to the dynamic level. Our findings suggest that the Dc estimate by Fukuyama and Mikumo (2007) for the 2000 Mw 6.6 Tottori earthquake is unrelated to the breakdown slip. Moreover, the very narrow widths for Rc obtained from our simulations are in agreement with previous studies challenging the resolution of bandlimited strong- motion data to estimate Dc from kinematic-rupture models (Guatteri and Spudich, 2000; Spudich and Guatteri, 2004).