Doctoral Dissertation 2007

Study on Strong Motion Generation Based on Detailed Analysis of Earthquake Source Rupture Process

Kimiyuki Asano

To illuminate the strong ground motion generation process during large earthquakes, reliable source modeling is quite necessary. The detailed source rupture process relating to the strong motion generation is investigated in two approaches. This thesis addresses the stability and reliability of solutions obtained by kinematic source inversions using theoretical Green's functions, the characterization of static stress drop on asperities of inland crustal earthquakes, and the estimation of final slip and peak slip velocity distributions by a new source inversion method that involves the empirical Green's function method using broadband strong ground motion data.

The first approach is the estimation of the kinematic source rupture process of the 2004 Chuetsu, mid Niigata prefecture, Japan, earthquake (MW 6.6) by applying the kinematic linear waveform inversion method to dense observed strong ground motion data with well-calibrated Green's functions. A set of adequate velocity structure models to calculate the Green's functions for each station is estimated by forward waveform modeling of small events' records. The spatial extent and location of asperities are precisely obtained by the source inversion. Two tests are demonstrated to examine the stability of the obtained solution. The first test examines the effect of the number of stations on the variation in the slip at each subfault. Convergence of the solution and decrease in its standard deviation at each subfault are observed, and the optimum number of stations is estimated. The other test examines the effect of wave types (P or S-wave) used in the inversion on its solution. This effect is negligible if the number of stations is sufficient and the velocity structure model at each strong motion station is carefully calibrated.

The characterization of static stress drop distributions is investigated for inland crustal earthquakes based on kinematic source models, which have been obtained from waveform inversions of strong motion data. The static stress changes of these earthquakes are calculated from the final slip distributions of the kinematic source inversion results. It is one of applications of kinematic source models to physics of earthquake. The result indicates that static stress drop on asperity depends on its depth. The surface breaking asperity appears to have smaller stress drop than the buried asperities. This information is also useful for understanding the physical nature of asperity.

As for the second approach, a new source inversion method to estimate slip, peak slip velocity, and rupture time distributions is proposed based on the empirical Green's function technique. This method jointly utilizes velocity waveforms in the low-frequency range (0.2-1 Hz) and acceleration envelopes in the high-frequency range (1-10 Hz). The proposed method is applied to the 2003 off Miyagi prefecture, Japan, earthquake (MW 7.0). The source parameters of the small event, whose ground motions are used as the empirical Green's functions, are obtained by a circular crack source model. The result of the joint inversion shows that the strong motions of the mainshock are mainly radiated from two asperities. The peak slip velocity is large around the hypocenter and near the northern edge of an asperity whereas its final slip distribution has the maximum close to the center of the asperity. The nature of the strong motion generation area is thought of as the asperity including such a spatial heterogeneity inside it. The high-frequency seismic waves are radiated not only from the edge of the asperities but from the whole area of asperities.

For quantitative evaluation of strong ground motions from an earthquake, effects of the faulting type, depth, and tectonic environment on its asperities should be included in source modeling as well as physical process of seismic wave radiations inside asperities. The evaluation of these effects is required to be based on detailed analyses of observations for individual earthquakes both in low-frequency and in high-frequency ranges. This thesis demonstrates a practical way to obtain them.