<p> The rapid modeling and detection of earthquakes and transient deformation is a problem of extreme societal importance for earthquake early warning and rapid hazard response. To date, GPS data is not used in earthquake early warning or rapid source modeling even in Japan or California where the most extensive geophysical networks exist. This dissertation focuses on creating algorithms for automated modeling of earthquakes and transient slip events using GPS data in the western United States and Japan. First, I focus on the creation and use of high-rate GPS and combined seismogeodetic data for applications in earthquake early warning and rapid slip inversions. Leveraging data from earthquakes in Japan and southern California, I demonstrate that an accurate magnitude estimate can be made within seconds using P wave displacement scaling, and that a heterogeneous static slip model can be generated within 2-3 minutes. The preliminary source characterization is sufficiently robust to independently confirm the extent of fault slip used for rapid assessment of strong ground motions and improved tsunami warning in subduction zone environments. Secondly, I investigate the automated detection of transient slow slip events in Cascadia using daily positional estimates from GPS. Proper geodetic characterization of transient deformation is necessary for studies of regional interseismic, coseismic and postseismic tectonics, and miscalculations can affect our understanding of the regional stress field. I utilize the relative strength index (RSI) from financial forecasting to create a complete record of slow slip from continuous GPS stations in the Cascadia subduction zone between 1996 and 2012. I create a complete history of slow slip across the Cascadia subduction zone, fully characterizing the timing, progression, and magnitude of events. Finally, using a combination of continuous and campaign GPS measurements, I characterize the amount of extension, shear and subsidence in the Salton Trough, one of the most complex zone of active faulting and seismicity in California. I show the implications that faulting in the Salton Trough has for the evolution of the Brawley Seismic Zone, and more importantly, the southern San Andreas fault.</p>

<p> Long-offset, large-angle reflections have great potential for both both velocity and density inversion. Prestack migration has angle-dependent wavelet stretch effects, which lowers the image resolution at large reflection angles. Current stretch correction filters operate on migrated images. We develop a new stretch-free imaging condition, which does a shrink-and-shift operation on the extracted propagation wavelet after extrapolation, but before the imaging condition is applied. Most existing amplitude versus angle methods do modeling with Zoeppritz approximations, which lose accuracy at large angles. We develop a new inversion method using the "exact" Zoeppritz equation, which is able to invert up to 4 elastic parameters, comparing to the conventional 2- or 3-term inversion. Near- and post-critical reflections have phase shifts as well as amplitude increases. We propose using phase versus angle data (combined with the amplitude data) to do elastic inversion. Phase is less affected than amplitudes by the transmission losses, which makes it more accurate for a target with many overburden layers. The plane-wave reflection coefficients given by the Zoeppritz equation are not applicable near the critical angle in time-space domain, and the modeling of accurate spherical-wave reflection coefficients is expensive. The tau-p transform decomposes spherical waves into plane waves, and thus provides a way of applying the Zoeppritz equation to wide-angle reflections. We test the accuracy of forward modeling and develop a new target-oriented amplitude and phase versus angle inversion algorithm using a tau-p transform and ray tracing, which applies to laterally heterogeneous models. The ray tracing links the reflection angle at the target reflector and the apparent slowness at the receiver, which enables extracting the amplitude and phase versus angle data in the tau-p domain. We suggest using precritical amplitudes and postcritical phases in the tau-p domain for inversion, which uses the Zoeppritz equation to do efficient forward modeling, and can estimate the elastic parameters beneath the target reflector.</p>

The structure of the Earth's lower mantle has key implications to the distribution of potential mantle reservoirs and the dynamics of mantle convection. To characterize the nature of mantle heterogeneity, three teleseismic studies are combined in this thesis to expand the observational knowledge of middle mantle and lower mantle features. Array stacking techniques (linear, phase weight, Nth root, common-conversion point) are applied to P-wave coda data from West Pacific and South American subduction zones to search for middle mantle discontinuities and scatterers using S to P converted waveforms. In the West Pacific, the mid-mantle reflectors associated with a high velocity anomaly (HVA) are observed dipping towards the western edge of the study region at depths from 930km to 1080km. In contrast, scatterers beneath the middle mantle of South America occur in three depth families at &sim;830-950km, &sim;1020-1150km, and &sim;1450-1550km, primarily sample the edge of HVA's or occur in regions where no velocity anomaly is present with a distinct geometry. Where mineral phase changes within the slab are a viable solution for scatterers/reflectors associated with HVA's, scatterers without geometry occurring in unperturbed mantle as observed in South America are likely due to small scale heterogeneity in the lower mantle. The final study relates to searching for the root of a deep plume source beneath the Galapagos Islands using analysis of SKS and SKKS waveforms. 248 Residual differential travel times of SKKS-SKS are measured and combined with a statistical analysis to reveal a region of positive residual differential travel times in the northeast portion of the sampled region. First order modeling of three potential low velocity structures, an ultra low velocity zone (ULVZ), a plume conduit, and a D" structure, are used as a diagnostic tool. We find that the most likely cause for the observed SKKS-SKS travel time residuals is a low velocity structure within the D" layer. To determine the presence of lower mantle anisotropy, amplitude ratios of the radial and transverse component of SKS and SKKS waveform are calculated and compared. Regions with significant differences in anisotropy are interpreted as the edge of the flow field of a hypothetical mantle upwelling.

Depth-averaged horizontal strain rates in oceanic lithosphere due to thermal contraction are determined. Calculated strain rates range from &ap;10 -2 Myr-1 (near the mid-ocean ridge) to &ap;10 -5 Myr-1 (for the oldest oceanic lithosphere). The average thermal contractional strain rate in oceanic lithosphere is &ap;10 -4 Myr-1. Newly created lithosphere is displaced toward old ocean basins at a rate that is 1.35% of the half-rate of seafloor spreading, giving displacement rates of 0.1 to 1.1 km Myr-1. The bias in plate displacement rates estimated from marine magnetic anomalies, expressed as a percentage of the full spreading rate, is 0.60% or 0.85% depending on the age of the magnetic isochron used to estimate current plate velocity. The displacement rate due to thermal contraction parallel to a mid-ocean ridge could be as large as &ap;10 mm/yr. Strain rates due to thermo-elastic stresses are an order of magnitude smaller than the strain rate calculated when these stresses are neglected.

Early thermal history models predicted a high Urey ratio (>0.7), in contrast to more recent geochemical estimates of 0.15-0.6. To potentially help resolve this discrepancy, we explore the effects of adding continental growth to thermal history models. The addition of continents has two effects, continents sequester radioactive material out of the mantle, and continents insulate the mantle. We incorporated these coupled effects into thermal history models to constrain the continental growth scenarios that can satisfy observational constraints for present heat flow, mantle potential temperature, and Urey ratios. To further constrain the allowable class of models we also included a cooling core in the thermal history calculation. This provides two added constraints; a core heat flux >15 mW/m2 to generate a geodynamo and an inner core size consistent with present observations. These constraints limited the class of allowable continental growth models to those with a component of progressive growth.

The crustal and lithospheric structure of the northern South America plate boundary with the southeast Caribbean has been the focus of many studies. In this region, westward subduction of (Atlantic) oceanic South America transitions to east-west transform between continental South America and the Caribbean plate. Previous models invoke a poorly-constrained component of north-south convergence between the Caribbean and continental South America, predicting that the westward subduction transitions to northwest-dipping subduction beneath the Serrania del Interior. These models predict that continental crust extends north of the Venezuela coast beneath the Leeward Antilles remnant arc islands, and that the Leeward Antilles are accreting onto South America. The results presented in this dissertation determine instead that the dextral strike-slip system along the Venezuelan coast cuts near-vertically through the crust and offsets the Moho. The strike-slip system fundamentally defines the plate boundary, deriving from a shear tear through the entire lithosphere that is actively propagating north of the Paria peninsula. This shear tear detaches subducting oceanic crust from buoyant continental crust along the weakened, former passive margin. Thrust faults flanking the strike-slip system to the north and south dip systematically toward the plate boundary. These faults have been previously interpreted as delineating a 300 km-wide diffuse plate boundary zone, caused by oblique convergence partitioned into orthogonal thrust and strike-slip displacements. Instead, these faults are driven largely by vertical rather than horizontal tectonics, and are the result of the geodynamic response to the shear tear.

This thesis is a case study concerning the application of reflection seismic methods to a very specific instance: mapping an extremely shallow underground feature in the framework of groundwater contamination and remediation ongoing at Operable Unit 2, Hill Air Force Base (UT). The target is the bottom of a paleo-channel eroded in a clay layer at a depth of 10-15 m from the topography and filled with a mix of sands, clays and gravels, which have different compaction and water saturation. The channel acts as contaminant trap for dense non-aqueous phase liquids, therefore a detailed map of its geometry, lateral boundaries and depth is crucial to the remediation effort. The clay-rich sediments that cover much of the study area preclude any effective use of the most commonly utilized ground-penetrating radar. Here I present the results of the processing of a 3-D reflection dataset acquired in 2000 over the buried channel, showing that it is possible to image and delineate the geometry and depth of the very shallow structure using reflection seismic techniques. The resulting contour map of the buried channel can be directly used to plan injection/extraction well placements in the ongoing ground water remediation program. The outcome is in good accordance with the available well data in the deepest, critical part of the buried paleo-channel, where the contaminant is assumed to be concentrated. The map of the channel obtained from the reflection seismic data, although taking advantage of the available borehole information eliminates the need for many wells. In addition, I designed and performed a 2-D elastic-wave seismic survey at the same site in 2000, using three-component sources and receivers. The processing effort resulted in horizontally polarized shear-wave and P-wave stack sections across and along the channel. Despite the less efficient seismic source, the S-wave images are superior to the correspondent P-wave images. Finally, after inverting the first arrival traveltime of both P- and S-wave data, I obtained maps of the VP/VS ratio and equivalent Poisson's ratio along the acquisition lines. These maps highlight areas with different sand and shale content, allowing distinguishing lithology changes in the subsurface.

Orientations of vertical fractures (i.e., transversely isotropic with axis of symmetry horizontal) can be detected using the Amplitude-Versus-Offset (AVO) response for the reflected p-waves. A MATLAB program that calculates the AVO response at a plane interface between any two transversely isotropic media is presented. The program agrees with the Zoeppritz equations when both media are isotropic.

Most current interpretations of lower crustal seismic reflectivity suggest the existence of fine-scale ($\approx$100 m thick) layered structures at depth. Typical common-midpoint (CMP) stacked images of deep structures are, however, noisy and show discontinuous reflections characterized by numerous subhorizontal segments. Present interpretation techniques cannot definitively resolve whether such a reflection response is attributable to heterogeneity at the target or whether the seismic image is distorted by propagation effects and contaminated by noise. The quantitative assessment of lateral heterogeneity in the deep crust is fundamental to understanding mechanisms of crustal formation and evolution. Here, crustal heterogeneity is represented by velocity structures that vary randomly in two dimensions, with a correlation distance comparable to the dominant wavelength of the seismic source. Synthetic CMP seismic data are computed for various models using a fourth-order finite-difference solution to the acoustic wave equation. A conventional data processing sequence produces CMP stacked sections with greater lateral continuity than is present at the target and an overall appearance comparable to that of field-recorded data. Lateral coherence is quantified using the spectral coefficient of coherence, applied to trace pairs having various spatial separations. Increased continuity in the CMP stack is attributable to the dip-filtering action of stacking and can be compensated by the application of migration before stack or equivalent processes. The reflection response observed in common-shot trace gathers shows amplitude and lateral coherence increasing with offset, an effect attributable to the source-receiver geometry. Observed wavefield coherence is related to the correlation properties of the target through a convolutional expression. A field data example from the Black Forest, Germany shows that enhanced coherence can be expected in wide-angle field experiments and that a model having two-dimensional heterogeneity matches field data previously interpreted in terms of extreme fine-scale layering. A densely-sampled field data set from the Basin and Range Province, Nevada shows increased coherence for P$\sb{\rm m}$P at large offsets. Observed lateral coherence values are lower than predicted for scattering purely in the Moho transition zone; coherence levels can be modeled by including scattered energy from inhomogeneities above the target zone.

Rice conducted an onshore-offshore seismic survey across the central California transform margin. Refraction velocity modeling of shot and receiver gathers has generated two models with similar near surface features. Model synthetics produce excellent fits to first arrivals and good matches for later arrivals in the data. CMP stacking of nearer offset traces imaged an event interpreted as reflections from the top of a dipping lower crustal layer. Model raytracing indicates layer flattening landward and seaward of the coast. High amplitude late arrivals seen at long offset on receiver gathers are modeled by imbricating the lower crust beneath the coast. This feature is interpreted as a result of overthrusting of the continental and Farallon plates onto the Pacific plate. The models differ in the middle crust, where one includes a laterally discontinuous low velocity zone. No direct indication of this LVZ exists in the data or in offshore reflection profiles.