Some aspects of earthquake seismology: slip partitioning along major convergent plate boundaries; composite source model for estimation of strong motion; and nonlinear soil response modeling

Abstract

Studies on three aspects of earthquake seismology were conducted. Firstly, differences between observed earthquake slip vectors and those predicted from global plate motion models along major convergent plate boundaries were investigated using the Harvard Moment Tensor Catalog as the principal data base. Discrepancies in the rates and geometry of relative plate motion along subduction zones are generally attributed to aseismic slip and slip partitioning and they provide important insight into the limitations of the validity of plate tectonic theory in specific regions. It was found that subduction zones characterized by back-arc spreading tend to show the greatest degree of slip partitioning, while subduction zones without back-arc spreading show less slip partitioning, and the partitioning is accommodated by strike-slip motion, back-arc extension, or a combination of both. Secondly, a composite source model for estimation of strong ground motions was developed. The composite source model method is easy to implement and all the parameters in the model are constrained by physical phenomena. The resulting synthetic accelerations, velocities, and displacements have realistic appearance and fit the statistical properties of the observed seismograms reasonably well. The success in using the composite source model as a source description for generating realistic seismograms suggests that there might be some kinship between the actual earthquake source and a fractal distribution of random asperities. Thus, the composite source model method shows great promise not only for computing synthetic strong ground motions at a specific site but also it may enhance understanding of actual earthquake sources. Finally, nonlinear soil effects on strong ground motion was examined by numerical modeling. The numerical model successfully produces the effects that are usually cited as evidence of nonlinearity: decreased spectral ratios of surface-to-input motion near the dominant frequency of soil; decreased statistical uncertainty in prediction of peak acceleration; and increased effective period of surface motion. The numerical modeling also indicates that for uphole/downhole experiments, the nonlinear soil effect can be detected by three frequency bands in the spectral ratios: unaffected at low frequencies, decreased at intermediate frequencies, and increased at the highest frequencies.