A Physics-Based Simulation of the 2003 Tokachi-oki, Japan, Earthquake to Predict Strong Ground Motions

Abstract

We demonstrate the usefulness of physics-based integrated simulation of subduction earthquakes for making strong-motion predictions in a deterministic way. We estimated strong ground motions caused by a 2003 Tokachi-oki-like earthquake, taking into account the stress accumulation due to tectonic loading, dynamic rupture propagation due to the failure of asperities, and generation of strong ground motions. Then we compared the synthetic seismograms with those observed during the 2003 Tokachi-oki earthquake ( M 8.0) to confirm that simple but physically reasonable modeling can obtain a reasonable estimate of ground motions of large earthquakes in view of peak ground velocities, duration, and velocity response spectra. It should be emphasized that this procedure is quite important to constructing a physics-based earthquake generation model that would be available for the prediction of strong ground motions and mitigation of disasters due to future large earthquakes. And this feature will not be achieved by traditional experience-based kinematic modeling approaches. Of course, this simulation includes quite a few uncertain parameters, but the parameter search experiment itself will help us construct useful scenarios of possible disasters we will suffer in the future. Hypocenter location is one of the most uncertain parameters in this simulation. We investigated the effect of uncertainty of hypocenter location on the computed strong motions quantitatively. The result shows that the variation in response spectra of ground velocities is within a factor of 2 at 3.3 sec and a factor of 4 at 15 sec, which is not serious compared to the local site amplification effects.