<p dir="ltr">Studying changes in seismic velocities beneath two separate volcanic systems in the Aleutian arc. Focusing on eruptive behavior, this research delineates subsurface changes through the lens of changes in seismic velocity.</p>

<p dir="ltr">The Alpine Fault, located in the South Island, New Zealand, is a subject of intense geological study due to its potential to trigger large earthquakes. It encompasses a complex system with the interplay of mechanics, thermodynamics, and fluid. Gaining insights into these systems not only enhances our understanding of the fault but also holds the potential to guide risk mitigation efforts.</p><p dir="ltr">The damage extent and fracture networks within the metamorphic rock mass adjacent to the fault can be effectively characterized by seismic anisotropy, an elastic property of rock, where seismic waves travel at different speeds with variation directions. This thesis presents a comprehensive exploration of seismic anisotropy in the hanging wall immediately adjacent to the principal slip zone of the Alpine Fault in New Zealand. Leveraging the borehole seismic data from a unique scientific drilling project and advanced numerical modeling techniques, the ultimate goal is to invert and parameterize the bulk seismic anisotropy.</p><p dir="ltr">Motivated by these challenges, the thesis undertakes several key initiatives: The first effort focuses on gaining a comprehensive understanding of an innovative method for seismic measurement: Distributed Acoustic Sensing (DAS) – examining its operational principles, factors influencing observed wavelets, and how it contrasts with traditional point sensors for accurate interpretation. Subsequently, the research introduces the implementation of an open-source seismic wave solver designed for modeling elastic wave propagation in complicated anisotropic media. This solver is further optimized for computational efficiency with its performance rigorously benchmarked.</p><p dir="ltr">With this preparedness, the inversion is further facilitated by high-performance computing (HPC) and a deep-learning algorithm specifically designed for automatically picking transit times. The inverted bulk elastic constants, compared to the intact rock, reveal 28% to 35% reductions in qP-wave velocity, characterizing the damage due to mesoscale fracture. Further analysis sheds light on the existence of orthogonal fracture sets and an intricate geometrical arrangement that agree with the previous borehole image log. This represents an advancement in our ability to characterize and understand the geologic processes with seismic anisotropy.</p>

<p dir="ltr">The Cascadia subduction zone features many along-strike variations in geophysical signatures that appear independent of properties in the subducting Juan de Fuca plate. Past studies have hypothesized that controls on these variations, namely subcretion, seem linked to overriding plate characteristics but may be influenced by characteristics of the downgoing slab as well. Nowhere is this more apparent than in southern Cascadia, which features the highest seismogenesis, broadest forearc topography, and lowest Bouguer gravity along the Cascadia margin. Additionally, the northward migration of deformation related to the San Andreas fault’s evolution and potential subslab buoyancies introduce further complexities making it difficult to parse contributions of tectonic processes to individual geophysical observations. To better understand contributions from Cascadia subduction and San Andreas evolution on tectonic processes, 60 Magseis Fairview nodal seismometers were deployed in southern Cascadia (Klamath Mountains) between April and May of 2020. We perform ambient noise tomography using Rayleigh and Love waves to constrain radial anisotropy and reveal seismic characteristics in the forearc. We find low VSV (<3.4 km/s) in the lower crust of the forearc consistent with previous studies. This is paired with high (>10%) positive radial anisotropy suggesting these materials are dominated by (sub)horizontal fabrics. We also observe relatively high VSV and VSH and negative radial anisotropy (~ -10%) in the upper crust of the forearc to ~10 km depth. These results suggest that the upper crust, which is dominated by the Klamath terrane, is characterized by (sub-vertical) deformational fabrics, likely related to brittle deformation superimposed on the accretionary history of the Klamath terrane, while the lower crust shows fabrics consistent with what would be expected due to basal accretion of oceanic crust (e.g, sedimentary rocks with or without basaltic slivers). The correlation of positive radial anisotropy with low shear-wave velocities (~3.4 km/s), low Bouguer gravity, high conductivity, and high rates of seismogenic activity (LFEs, tremor distribution, and episodic slow slip events) suggest that this basally accreted material may be infiltrated by fluids derived from the downgoing oceanic lithosphere.</p>

<p dir="ltr">As cameras increasingly populate urban environments for surveillance, the threat of data breaches and losses escalates as well. The rapid advancements in generative Artificial Intelligence have greatly simplified the replication of individuals’ appearances from video footage. This capability poses a grave risk as malicious entities can exploit it for various nefarious purposes, including identity theft and tracking individuals’ daily activities to facilitate theft or burglary.</p><p dir="ltr">To reduce reliance on video surveillance systems, this study introduces Urban Seismic Event Detection (USED), a deep learning-based technique aimed at extracting information about urban seismic events. Our approach involves synthesizing training data through a small batch of manually labelled field data. Additionally, we explore the utilization of unlabeled field data in training through semi-supervised learning, with the implementation of a mean-teacher approach. We also introduce pre-processing and post-processing techniques tailored to seismic data. Subsequently, we evaluate the trained models using synthetic, real, and unlabeled data and compare the results with recent statistical methods. Finally, we discuss the insights gained and the limitations encountered in our approach, while also proposing potential avenues for future research.</p>