Explore the vast range of titles available.

Africa, a continent straddling multiple plates, exhibits diverse tectonospheric behavior, including intracontinental instabilities, underscoring the need for a comprehensive assessment of key geodynamic controls. While extensive work has advanced knowledge of its tectonic makeup and of many active deformations, many aspects of its contemporaneous mechanical response remain unexplained. Case in point: over the past 3 decades of seismicity in West Africa, seismological evaluations of driving mechanisms have been lacking. Moreover, other essential components of geodynamic analyses, such as architectural models, seem inadequate for accurate complementary stress prediction, which are often employed as a remedy for sparse earthquake analyses. This is, in part, due to the low resolution of seismic discontinuity catalogs that can inform layered contrasts in the seismic tomography model. In this dissertation, I address the incompleteness of the seismological constraints of Africa’s lithospheric dynamics, focusing on (1) the outstandingly understudied regions (e.g., West Africa), and (2) the improvement of seismic-discontinuities-imaging, allowing for a clearer explanation of their characteristics and the roots of their existence within the mantle. In the first endeavor, I scrutinized the seismic records of all low- to intermediate-magnitude earthquakes to compute focal mechanisms and subsequently invert for local and regional stress. These analyses offer unprecedented insight into the mode of faulting and the stress distribution across West Africa. The major findings include: (1) A Laterally variable stress field with different drivers between the passive margins and the continental interior, (2) the prominent role of oceanic transform structure in the stress-transfer, (3) the importance of first-order stress within the continental interior, albeit second-order stress perturbation, and outstanding anthropogenic influence in Mauritania. In the second endeavor, I revisit the seismic mantle discontinuities using P-to-S receiver functions, processed with a sparsity-promoting Radon transform algorithm that eliminates reverberations imprint. The new set of mantle images reveals the depth of seismic discontinuities with improved certainty, and shows (1) 4 layering signatures within the upper mantle, which bear thermal, compositional, or grain-scale influences; (2) low-velocity layers above and below the mantle-transition zone, closely related to plume influences and limitedly to subduction influences, all invoking accumulations of partial-melt.

Recent advances in machine learning have enhanced the capability of multi-attribute analysis to resolve relationships within high-dimensional data, enabling more sophisticated reservoir characterization. However, the ability of multi-attribute analysis to produce geologically meaningful results remains strongly controlled by the diversity, relevance, and spatial extent of the input data. Seismic amplitude data is inherently band-limited and does not adequately capture low-frequency components, which limits its ability to represent subsurface variability. Inversion products derived from low-frequency models provide a complementary source of information to address this limitation.This study investigates the effects of incorporating pre-stack simultaneous AVO inversion products into multi-attribute analysis. To evaluate the impact of inversion products, a comparative framework based on two input datasets is established. The first scenario utilizes attributes derived solely from seismic amplitude data, whereas the second incorporates P- and S-impedance volumes into the attribute set. The Pohokura Field in the Taranaki Basin is used as a case study, where multi-attribute analysis is performed for both scenarios using Principal Component Analysis (PCA) and K-means clustering. The results indicate that the inclusion of inversion products leads to a shift in data representation from a structure primarily governed by seismic amplitude response to one that is more consistent with subsurface geology and controlled by elastic properties. This transition improves the delineation of reservoir and seal units and enhances the definition of reservoir boundaries, lithological trends, heterogeneity zones, structural discontinuities, and facies distributions. Sand-prone intervals that are not clearly resolved in the attribute-only scenario become detectable when inversion products are included. Furthermore, the inclusion of inversion products results in reduced noise and improved vertical and lateral resolution, which enhances interpretability and enables a more reliable characterization of the complex Mangahewa reservoir. The achievements gained by incorporating inversion products into multi-attribute analysis highlight the importance of enriching the frequency content of the input data. In this regard, multi-attribute analysis should be regarded not simply as a data integration approach, but as a design-oriented process in which the selection and organization of the input dataset fundamentally control the quality of the results.

Machine learning (ML) techniques are increasingly adopted in seismic interpretation workflows, yet their rapid integration has outpaced the development of frameworks for evaluating whether their outputs are geologically meaningful. Too often, ML outputs are treated as geologic answers rather than as one step in a broader interpretation process, optimizing for statistical performance rather than geologic validity.This dissertation develops and demonstrates a practical framework for integrating unsupervised ML into seismic facies interpretation within a geologically informed context, using a deep-water channel system in the Taranaki Basin, New Zealand as the primary study area. Key findings demonstrate that dimensionality reduction (DR) improves both visual interpretability and unsupervised clustering performance across 55 model configurations — but that statistical performance alone is insufficient for model selection. The best-performing statistical model is not necessarily the most geologically meaningful one. A robust geologic foundation, including systematic seismic visualization, understanding of depositional processes, and stratigraphic context, must precede and inform ML evaluation.The central contribution of this work is a transferable framework that guides practitioners through attribute selection, DR technique selection, model evaluation, and geologic validation. The framework offers a disciplined counterbalance to the pressure of rapid ML adoption: the algorithm answers what you ask, not what you mean. The goal is not to replace geologic reasoning with computation, but to demonstrate how ML, when implemented thoughtfully within a geologically informed workflow, can enhance interpretive resolution and increase confidence in geologic models.

Subduction zones are complex, self-organized systems in which plate buoyancy, mantle rheology, metamorphic processes, and geometry interact to control slab deformation, trench motion, and mantle flow. Understanding how these processes couple across spatial scales and modeling frameworks remains a central challenge in geodynamics. This dissertation investigates the physical controls on subduction dynamics using a suite of two- and three-dimensional free-subduction models that progressively incorporate physically based rheology, compositionally dependent phase transitions, and geometric effects.The first part of this work focuses on the plate boundary shear zone, which transitions from a brittle fault near the surface to a ductile shear zone at depth. Rather than prescribing a weak layer with an arbitrary cutoff depth, the shear zone viscosity is linked to the basalt–eclogite phase transition, providing a physical mechanism for the emergence of a globally consistent mantle decoupling depth. This formulation naturally leads to self-sustained, one-sided subduction and captures feedbacks between slab deformation, shallow slab dip, trench retreat, and slab thermal structure. The results further suggest that slab-top melting localized near the mantle decoupling depth may be a common feature of mature, water-saturated subduction zones.The second part of the dissertation examines how slab age, slab width, and dimensionality influence modes of slab deformation. By accounting for realistic mechanical and rheological conditions governing slab buoyancy, strength, and deformation, this study systematically compares two-dimensional and three-dimensional free-subduction models. The results show that slab age influences deformation style while also modulating the effects of slab width on trench curvature, and that the stage of subduction can obscure direct correlations between slab age and deformation mode.The final part of this dissertation investigates the influence of model geometry by developing a unified three-dimensional free-subduction framework that enables direct comparison between Cartesian and spherical geometries with adjustable domain size. Earth’s sphericity is shown to enhance along-strike deformation while suppressing down-dip buckling. A shallow lower boundary modifies internal slab deformation and mantle return flow by acting as a vertical constraint on slab descent and mantle circulation. In contrast, nearby side boundaries modulate mantle convection patterns in a manner analogous to interactions between adjacent slabs, highlighting the importance of lateral confinement and slab–slab interactions in governing subduction dynamics.Together, these studies emphasize the importance of physically based shear zone implementations and consistent geometric treatment when modeling subduction systems. In addition, the unified two- and three-dimensional framework provides guidance for interpreting model results in a consistent physical context.

In seismic multi-attribute analysis, unsupervised machine learning helps identify patterns inherent in the data across the multiple attribute dimensions. It summarizes the information intuitively for appraisal of the underlying geological objects responsible for the different seismic signal expressions. Conventionally, though, unsupervised machine learning products—in the form of volumes and maps—serves mainly as leads for the seismic interpreter, while applying further seismic attribute theory for interpretation still requires going back to the original input attribute volumes. This points to a deeper challenge: unsupervised methods inherently lack an interpretability layer, leaving the model's internal reasoning opaque to the interpreter.Explainability frameworks such as SHapley Additive exPlanations (SHAP) have gained traction in supervised learning, where a well-defined output metric—typically a prediction probability or classification score—enables the attribution of input feature contributions. Without an equivalent metric, unsupervised models have largely been outside the scope of SHAP-based interpretability. To bridge this gap, this study utilizes probability distributions from a Gaussian Mixture Model (GMM) as a proxy metric for cluster membership confidence, effectively applying SHAP to an unsupervised multi-attribute machine learning workflow.This integration enables analysis into how individual seismic attributes contribute to each cluster assignment. Which means each appraised cluster is then accompanied by geophysically meaningful information that highlights the key signal expressions and attributes characterizing it. This added interpretability works to equip the geoscientist with data-driven quantitative means to support their interpretation, while simultaneously grounding the machine learning results back to seismic attribute theory.In this study, we found that the SHAP graphical descriptions align with seismic attribute characterization of geological targets in the Taranaki Basin, showing how the cluster is expressed in each attribute space. The graphical descriptions enhance the role of the unsupervised machine learning workflows in seismic multi-attribute analysis and provide intuition-driven mechanisms for quality control and trust in machine learning products.

Glaciohydraulic tremor is the seismic energy generated by turbulent water flow through the subglacial environment, characterized by frequencies between 1.5 and 10 Hz. Tremor is known to serve as a reliable proxy for variations in glacier runoff, with theoretical power-law relationships relating tremor amplitude and subglacial discharge under distinct hydrological regimes. However, the quantitative relationship between tremor and discharge remains poorly constrained. Existing theory lacks rigorous testing across multiple glacial systems, and no existing model has the capacity to directly predict discharge quantities from seismic observations. This research integrates stream gauge records, passive seismic observations, meteorological data, and remote sensing imagery with hydrological modeling and statistical analysis to quantify this relationship across four alpine glaciers in Alaska and the French Alps. A degree-day melt model accounts for snow and ice melt, combined with precipitation contributions across surrounding watersheds, and is smoothed using exponential convolution and calibrated by least squares fitting to downstream gauge observations. I use the ratio between modeled discharges at the source and gauge to scale gauge observations and estimate the discharge at tremor source locations beneath the glacier. Comparison between source discharge and tremor amplitude corrected for attenuation across all sites yields a best-fit empirical relationship of Q=(5.50±1.20 )×10^8 V^(1.615±0.015), achieving R^2 of 0.739 with 95% of predictions falling within a factor of 2.89 of the corresponding estimated discharge. This result closely agrees with the theoretically derived relationship fitted with a fixed exponent giving V1.6 for conditions expected to govern fluctuations in discharge during peak melt season and exhibits identical statistical performance across this dataset. Model analysis reveals this relationship is highly sensitive to individual glacial environments. Cross-validation testing indicates that theoretical predictions with the fixed exponent of 1.6 perform best across diverse systems and broadly predict discharge within a factor of five. Unconstrained, best-fit relationships, where both the scaling factors and exponents are free parameters, vary considerably across testing iterations, likely reflecting differences in subglacial conditions and discharge magnitudes of the training dataset. The model demonstrates greatest reliability in well-developed, channelized drainage systems in smaller alpine environments. Ultimately, this work establishes the first multi-glacier empirical validation of the tremor-discharge power relationship, advancing seismic methods as a quantitative tool for hydrological monitoring of glacial systems where conventional gauging is impossible.

This study examines the patterns of seismicity in western Nepal within the central Himalayan seismic gap and determines whether current earthquake occurrence represents short-term stress release through clustered seismicity or long-term strain accumulation reflected by independent background earthquakes. A unified earthquake catalog was developed by combining regional and global earthquake catalogs and homogenizing magnitudes to moment magnitude (Mw). Completeness analysis based on the Stepp method indicated that this catalog is reliable for Mw ≥ 4.8. A nearest neighbor declustering algorithm has been implemented to identify clustered earthquakes from background seismicity due to long-term tectonic loading. Frequency-magnitude analysis shows b-values consistent with active tectonic environments. Coulomb stress analysis for the Mw 7.8 Gorkha earthquake in 2015 has been performed to assess its impact on western Nepal. The results show that background seismicity dominates and that static stress changes due to this event are small, indicating continued strain accumulation and ongoing seismic hazard in this area.

As seismic attributes were first incorporated into academic and industry workflows, fault detection and visualization became a primary objective. Faults and their associated damage zones exert significant control on subsurface fluid flow, seal integrity, and reservoir compartmentalization. While attributes such as coherence have proven effective for delineating fault geometries, quantitative workflows for constraining fault damage zone widths from seismic data remain underdeveloped.This study investigates the use of coherence and rejected noise derived from structure-oriented filtering (SOF) to characterize fault damage zones within the Bacalhau 3D seismic dataset in the post-salt interval of the Santos Basin. Coherence consistently delineated fault geometries and produced damage zone width estimates comparable to field-based outcrop measurements, reproducing the near 1:1 linear scaling relationship between fault slip and damage zone width documented in field studies. Rejected noise frequently mimicked seismic stratigraphy, reducing its reliability for estimating fault zone widths. However, noise aligned with interpreted faults was morphologically distinct and enhanced continuity along small-offset faults that do not register a coherence anomaly. At times, rejected noise also modeled a substantially wider damage zone than coherence. In the pre-salt carbonates of the Bacalhau field, well data indicated that coherence primarily responds to the most intensely fractured intervals, whereas rejected noise is more sensitive to mechanical and petrophysical transitions, capturing shifts from tight rock to porous and/or fractured rock not resolved by coherence. These results demonstrate that coherence and SOF-derived noise are complementary attributes, and when integrated with well data, can provide a more robust framework for seismic characterization of fault damage zones.

With increasing computational capabilities, seismology has taken aim at numerical modelling of seismic wavefields at increasingly-shorter wavelengths, enabling the mapping of Earth's interior structure at unprecedented resolution. There is, however, still substantial value in long-period wavefields, on the period of 40 seconds and above. At these periods long-wavelength averages of structure may be observed and, as the influence of Earth's self-gravitation becomes non-negligible, sensitivity to density structure is also present. In addition, recent observations of transient earthquake-induced gravity signals have created a need to simulate wave propagation on realistic, self-gravitating Earth models for application to earthquake early-warning. This thesis may be divided into two overarching parts that both consider numerical methods for modelling long-period seismic wavefields, on complex Earth models. In the first, the development and application of a spectral-infinite-element method is described. This method enables simulation of self-gravitating wave-propagation on complex, 3D Earth models. The approach is then applied to so-called `prompt elasto-gravitational signals' which are simulated on a 3D Earth model for the first time. The second section describes a spectral-element approach to traditional normal-mode splitting calculations. This new method enables the rapid construction of splitting matrices for arbitrary parametrisations of material heterogeneity. Following benchmarks demonstrating its utility, this splitting code is integrated into a Markov chain Monte Carlo framework for the joint inversion of body-wave and normal-mode data, with a focus on Earth's inner core anisotropy.