Source parameter estimation and related uncertainties of small earthquakes in southern California

Understanding the physics of earthquake rupture is critical to providing accurate estimates of seismic hazard and for effectively mitigating these hazards. Matching physical models to seismic data in order to better understand the earthquake rupture processes requires accurate and precise estimates of earthquake source properties. Measuring source properties such as rupture size and stress drop must include accounting for the effects of seismic wave propagation and making proper assumptions about the rupture process. This thesis focuses on the methods of estimating source properties of small earthquakes and on the application of these methods to earthquakes in two distinctly different seismogenic regions of California. In the San Jacinto Fault Zone, earthquakes recorded by a small aperture array allow quantification of source parameter uncertainties using empirical Green's functions and frequency-domain techniques. These uncertainties are frequently overlooked in source parameter estimation, and this study constrains them to ̃30% of estimate values. A non-parametric time- domain method using a set of empirical Green's functions is described and applied to a series of example earthquakes. This approach minimizes the assumptions regarding the rupture process and can be used to study less simple ruptures. Correcting for the effects of seismic wave propagation is an important aspect of techniques used in source parameter estimation, and the conditions necessary to effectively use nearby earthquakes as path corrections are tested and quantified. At the San Andreas Fault near Parkfield, the high degree of waveform similarity among closely spaced earthquakes is used to apply spatially averaged propagation path corrections and search for rupture directivity effects. This analysis shows that the population of small earthquakes in this region does not have a consistent unilateral rupture direction, but 70% of M>3 earthquakes exhibit characteristics of southeast-directed rupture. Computational models featuring a fault interface separating two materials for Parkfield-like conditions agree with the preferential southeast-directed rupture and present potential implications for earthquake source physics. Combined, these studies of earthquake source parameter estimation can be used to improve future source parameter estimates, offer appropriate metrics for establishing uncertainty bounds, and contribute to the study of earthquake source physic