Near - Source Imaging of Seismic Rupture

The nature of failure in earthquakes is one of the most fundamental questions in geophysics. We approach this problem using near-source seismic recordings of strongground motion to image the rupture process of earthquakes. We apply the technique to three well-recorded earthquakes in California and interpret the results in terms of earthquake rupture dynamics. To calculate theoretical seismograms we use a high-frequency, near-source approximation that allows the use of geometrical ray theory at distances greater than a wavelength. This approach has several advantages. It is several orders of magnitude faster than techniques that use complete Green's functions. This increase in speed allows us to parameterize the rupture process differently than most previous studies. Ray theory also allows us to include the effects of smooth, laterally-varying structure on near-source seismograms. We have developed the capability to do this using dynamic ray tracing and have applied it to a simple, idealized, fault-zone model. We parameterize the seismic source as a distributed displacement discontinuity across a planar fault that slips only once, when the rupture front passes, with a spatially varying rupture time and slip intensity. This parameterization results in a nonlinear relation between the observed seismograms and the faulting model. We perform a linearization of the problem and solve the nonlinear inverse problem by iteratively perturbing an assumed starting model. The inversion is performed using a regularized, back-projection technique. The resolution of the method is analyzed using sensitivity tests. The technique is applied to the 1966 Parkfield earthquake, the 1979 Imperial Valley earthquake, and the 1984 Morgan Hill earthquake. In each of these earthquakes there is evidence for strong heterogeneity in the slip amplitude. In the case of the Imperial Valley and Morgan Hill earthquakes, where the data are sufficient to distinguish between variations in slip amplitude and variations in rupture time, there is evidence for strong heterogeneity in the rupture velocity as well as the slip amplitude. The observed heterogeneity in the rupture process is attributable to heterogeities in the stress and strength. In the case of the Morgan Hill and Parkfield earthquakes there are features evident in the surface trace of the fault and in the distribution of seismicity with depth that are plausibly related to the large-scale heterogeneities in the rupture process. In the case of the Imperial valley earthquake the relation between large-scale heterogeneity and fault zone structure is less clear. 3 We also find indirect evidence for complexity of the faulting process at much smaller length-scales. Our estimates of the shear fracture energy in these earthquakes ranges from 520 x 105 J/m2. This is several orders of magnitude higher than found for laboratory samples of a simple fracture. The greatly increased surface area of a complex earthquake rupture may account for the high apparent shear fracture energies that we find. Thesis Committee: Dr. Dr. Dr. Dr. Dr. Thomas H. Jordan Theodore R. Madden M. Nafi Toksiz Vernon Cormier Teng-fong Wong Thesis Supervisor University of Connecticut at Storrs State University of New York at Stonybrook