TITLE:

NONLINEAR INVERSION FOR EARTHQUAKE RUPTURE PARAMETERS (No. 40)

DATE:

Friday, April 2nd, 2004

TIME:

3:30 PM

LOCATION:

GMCS 214

SPEAKER:

Kim Bak Olsen, Department of Geological Science, San Diego State University

ABSTRACT:

Seismologists have attempted to unravel the rupture history of large earthquakes for decades. Such information is critical to understand how earthquakes initiate, propagate and arrest. The conventional procedure is to match recorded and synthetic accelerograms solving for the slip or sliprate on the fault However, such kinematic rupture inversion is only based on physical plausible considerations, and does not solve the mechanical problem, and the results may be non-causal and biased by unphysical constraints on the rupture velocity and adopted source-time function. A more physically correct model would estimate the dynamics of spontaneous rupture propagation, i.e., the stress and friction parameters. Spontaneous rupture propagation assumes that the rupture process of an earthquake is a stress relaxation process. Rupture starts when the surrounding stress exceeds the frictional strength on the fault.

We have developed a systematic nonlinear inversion method for estimating rupture propagation and the underlying dynamic parameters for large historical earthquakes. The rupture modeling is carried out using a three-dimensional finite-difference method, and the inversion is implemented by a neighbourhood algorithm, minimizing the misfit between computed and observed near-fault seismograms. We test the method by estimating the stress drop within 32 regions on the causative fault for the 2000 magnitude 6.6 Western Tottori, Japan, earthquake. While the dynamic models show both similarities and differences with the conventional kinematic models, our method provides an ensemble of physically-correct models with plausible rupture propagation for the earthquake.

HOST:

Steve Day

DOWNLOAD: