Forward modeling and inversion of near-source earthquake ground motion

A method to calculate the response on the surface of a multilayered half-space for a fault of finite width and infinite length is presented. The model involves a piecewise-rectilinear and continuous rupture front propagating at a constant rupture velocity along the length of a fault of arbitrary dip angle. The motion produced by this steady state dislocation model corresponds to the passage of the rupture front phase, which is a predominant phase in the near-source region away from the ends of a finite fault. A series of validation test, using both three- and two-dimensional kinematic fault models, and a limited set of parametric studies clarifying the mechanisms involved in the generation of high amplitudes, are presented. It is shown that the distribution of peak horizontal velocities in the near-source region calculated by use of this model compare favorably with the regression results of Joyner and Boore (1981). The steady state model gives an efficient way to synthesize a ubiquitous, intermediate frequency, high-amplitude pulse observed in many near-source records. It is employed as a forward model for waveform inversion of near-source data from the 1966 Parkfield, 1979 Imperial Valley, and 1980 Mexicali Valley earthquakes. The inferred shapes for the rupture front and for the distribution of slip as a function of depth are consistent with previous results obtained by use of more general models. The results obtained show that a strong velocity pulse observed in the near-source region can be modeled as the passage of the rupture front phase and that the supershear propagation of the rupture front in the sedimentary layers of the medium provides a mechanism for the generation of the observed large amplitudes.; Subsequently, a method is presented for obtaining a unique temporal and spatial slip distribution on a prescribed fault plane from inversion of earthquake ground motions. The inverse problem is formulated in the frequency domain where the spatial distribution of slip is found for a set of frequency values. The time dependence of slip is then recovered by Fourier synthesis. A new norm minimization condition is presented, based upon the scalar wave equation. (Abstract shortened with permission of author.)...