SIMULTANEOUS INVERSION OF THE THREE-DIMENSIONAL VELOCITY STRUCTURE AND MICROEARTHQUAKE HYPOCENTERS IN THE COSO GEOTHERMAL AREA AND THE GEYSERS - CLEAR LAKE GEOTHERMAL AREA, CALIFORNIA

This dissertation simultaneously treats the inversion of microearthquake arrival-time data to infer three-dimensional velocity structure and microearthquake hypocenters. It is divided into two main parts. The first part presents a numerical inversion scheme which is adapted from nonlinear inversion techniques. The second part presents the application of the method to the investigation of the layer velocity model and three-dimensional velocity structure of two geothermal areas. The arrival times of the P-phase and the S-phase are used in these inversions.; Earthquake hypocenters are structurally constrained. Therefore, the determination of the crustal velocity structure using earthquake sources of unknown focal depth, distributed irregularly over a tectonic area, is frequently unstable or converges poorly during the inversion of the shallow crustal structure. A partitioned-constraint algorithm which differs from the conventional Baysian type approach is adopted to limit the perturbation of the model parameter during each iteration of the calculation. The advantage of this constraint is the use of contextual information at each step to reduce local ambiguities and to simultaneously predict the model parameters. The low-velocity layer is also successfully recovered in the synthetic inversion study.; Among the different joint inversions carried out in this study are (1) the simultaneous inversion of the P-wave and S-wave arrival-time data to infer the velocity model and hypocenters, (2) the joint inversion of arrival-time data for several different initial velocity models and hypocenters, and (3) the joint inversion of arrival-time data for different initial velocity models and a single hypocenter models. An alternative approach to the inversion, which solves for the slowness instead of the velocity, is adopted since the perturbation of the slowness can be more easily constrained than perturbations of velocity. The three-dimensional inversion uses the average velocity model obtained from the layer-velocity inversion as an initial model.; In the application, this method is used to study the velocity structure of two geothermal areas: (1) The Geysers-Clear Lake geothermal area in northern California and (2) the Coso Hot Springs geothermal area in southeastern California. The procedure developed and carried out here is (1) inversion of the S-wave and P-wave arrival-time data for a layered velocity model using homogeneous station corrections, (2) simultaneous inversion of the same data for both a heterogeneous three-dimensional velocity structure for P- and S-waves and the hypocenters of all events using homogeneous station corrections, and (3) calculation of lateral variation of Poisson's ratio within layers.