Mantle circulation models are a modified class of mantle convection simulations assimilating recent plate motions as the surface velocity boundary condition. In this thesis, I present a suite of mantle circulation models assimilating the past 300 million years of tectonic history. By comparing model predictions of present day mantle temperature anomalies to mantle structure imaged by seismic tomography one can better understand the physical properties of Earth’s mantle. Given a mantle model with realistic physical properties, plate reconstructions can also be tested. Mantle viscosity is the most significant property affecting mantle circulation models. For subducted slabs to sink to depths predicted by tomography studies a lower mantle viscosity increase of around thirty times is required. For models with a factor of ten increase slabs do not remain at mid-mantle depths for long enough, while a factor of one hundred increase causes slab sinking rates too slow to match imaged tomographic anomalies. An endothermic phase changes could potentially layer mantle convection into two independent layers. In models assimilating plate motions, no model containing an endothermic phase change reaches a fully layered state, even with unrealistically large, negative Clapeyron slopes. The onset of plate tectonics could potentially break down a two-layered mantle into a partially layered state, similar to the present day mantle. Predictions of mantle heterogeneity from high-resolution, global mantle circulation models match well with complex mantle structure imaged by seismic tomography in the Tethys region. These models indicate that a more complicated history of subduction during the closure of the Neotethys Ocean is required to match the imaged mantle structure. Subduction is required in two locations, one at the Eurasian margin and a second behind a back-arc ocean opening in the Neotethys Ocean. Simultaneous subduction at both plate boundaries appears not to be necessary. Global mantle circulation models estimate long-wavelength dynamic topography with amplitudes of up to five kilometres. The largest amplitude signal of dynamic topography is at plate boundaries, suggesting that near surface density variations in the mantle contribute significantly to the dynamic topography signal. The five-kilometre amplitude of topography is larger than predicted elsewhere and is explained by the inclusion of near surface density variations, commonly ignored by other global calculations of dynamic topography. If dynamic topography is defined as ‘any topography arising from flow within Earth’s mantle’ then near surface density variations are significant to the dynamic signal. Predictions of dynamic topography from mantle circulation models reveal a dichotomy between continental and oceanic regions. Oceanic crust is a part of the mantle convection system and so predicted topography for ocean regions matches well with the expected depth versus age curve for oceanic crust. Continental regions are significantly subsided relative to oceans in the dynamic signal, suggesting that isostatic effects mask continental dynamic topography. When predictions of dynamic topography are corrected for isostatic effects and crustal thickness, an accurate estimate of Earth’s observed topography is generated. This work contributes to an on going debate on the nature of dynamic topography on global and regional scales.

The Platreef, located in the northern limb of the Bushveld Complex in South Africa, is a world-class Ni-Cu-PGE deposit. The complexity of the deposit has meant that despite the numerous studies, developing an accepted genetic model to account for the variations observed has been difficult. While some authors have suggested that it is part of the Upper Critical Zone, correlating it to the Merensky Reef, others have suggested that the Platreef is unrelated to mineralisation found elsewhere in the Bushveld Complex. The model tested is the multiple staging chamber model developed by McDonald and Holwell, that proposes that the parental magma was upgraded in PGE (plus Ni and Cu) prior to emplacement. Key to testing this model has been the analysis of immiscible sulphide inclusions trapped within chromite grains, believed to represent the early parental magma. Analysis has shown that they contain high PGE tenors, significant semi-metal (Bi, Te and As) content and the low S/Se ratios of the inclusions suggest a mantle source. Interaction of the sulphide liquid with multiple batches of magma in the staging chamber is proposed to have enabled enrichment to occur prior to emplacement through a process known as multi-stage dissolution upgrading. The analysis of chromite grains from the three study farms has shown that the variation in chromite composition is dependent on host lithology and the location of the sample along strike of the Platreef. Some correlation can be made with chromites from the UG2 but Platreef chromites cannot be directly correlated to those from the Merensky Reef. Investigation of PGE concentrations within the BMS from Zwartfontein has shown a strong association between PGE and BMS and that the distribution of PGE is consistent with fractional crystallisation of a sulphide liquid. The PGM study has shown that variation along strike and down dip of the Platreef is not strictly controlled by footwall lithologies as previously proposed. Variation is suggested to be the result of differing temperatures and ƒO2 conditions due to the proximity around proposed feeder zones. In order to further test the staging chamber model, S isotope analysis should be carried out on the sulphide inclusions to ascertain if a magmatic signature is present. In addition, further support to the model may be achieved by examining other Lower Zone bodies for chalcophile element depletion.

The morphologies of pyroclasts in basaltic and silicic ash deposits and tuffs in volcanic sequences have been used by previous authors to determine eruption processes in proximal settings commonly from modern eruption deposits. However, little research has been conducted on the finest tephra fraction, ash and fine-ash, in distal settings. The study of tephra in distal localities, >100 km from source, may provide important information regarding the dispersal of ash from highly explosive large-scale eruptions, for which proximal deposits are no longer present. The grain morphologies (blocky, bubble-wall, irregular and rod-like) and vesicularity of pyroclasts can be used to determine eruption processes and distinguish between ‘dry’ magmatic and ‘wet’ hydrovolcanic eruption-styles. This research is mainly concerned with the characteristics of pyroclasts in basaltic ash/tuff likely formed by highly explosive eruptions, and whether the explosivity of these eruptions was due to magma-water interactions and hydrovolcanism. This study includes a re-examination, and detailed petrographical analysis, of tuffs in the Early Eocene Balder Formation, North Sea Basin, and Harwich Formation, SE England, where the source of tephra produced by explosive volcanism and the extent of basaltic ash dispersal have been contested. Previously, these basaltic tuffs have been attributed to large-scale hydrovolcanic eruptions that occurred during initial sea-floor spreading in the NE Atlantic during the separation of Greenland from NW Europe. These tuffs are compared and contrasted with the basaltic Saksunarvatn Ash (10.3 ka BP) of Grímsvötn Volcano, Iceland, and various non-lava lithologies of the Miocene Columbia River Flood Basalt Province, USA. The primary characteristics of pyroclasts may be modified by abrasion and alteration, particularly during additional transport and deposition. Therefore, the features created directly by the fragmentation and quenching of pyroclasts need to be distinguished from features created by the reworking of this material in the sedimentary environment. It is therefore essential to determine the characteristics of pyroclastic and volcaniclastic materials, which can be achieved by comparing data on the overall deposit characteristics, grain-types, grain-size distribution and grain-morphologies. For example, pyroclastic deposits of ash-grade are dominated by up to 100% igneous/volcanic grain-types, particularly glassy pyroclasts; in thin-section the pyroclasts have smooth-edges and may have undergone limited alteration in modern deposits. In contrast, volcaniclastic deposits may contain a wide range of clast-types, including a greater number of non-volcanic grains, and sedimentary structures such as cross-stratification; in thin-section, reworked pyroclasts of ash-grade (volcaniclastic grains) often have jagged-edges, and varying levels of alteration. This study builds upon the observations and naming schemes of previous authors and proposes a scheme applicable to distal deposits of ash and fine-ash, of both basaltic and silicic composition, including highly altered material.

The Central Arran Igneous Complex - a system of intrusive and pyroclastic volcanic rocks - is one of the least understood onshore 'central complexes' of the British Palaeogene Igneous Province (BPIP). The BPIP is part of the wider North Atlantic Igneous Province, formed during opening of the North Atlantic Ocean and the arrival of the Iceland plume at the base of the lithosphere. Despite being known about for over a century, the magmatic and volcanic processes that formed the Central Arran Igneous Complex remain poorly understood. The Central Arran Igneous Complex comprises a number of granitic and dioritic intrusions, a caldera-fill succession of pyroclastic and sedimentary rocks intruded by a dolerite sill, and a series of dykes ranging in composition from picrite to pitchstone. This study uses a combination of field mapping, whole-rock elemental geochemistry, radiogenic isotope geochemistry, and U-Pb zircon geochronology to determine how these units and their magmas formed, how they relate to one another, and their relationship with other igneous rocks throughout the BPIP. Detailed study of the intra-caldera stratigraphy allows a prolonged volcanic history to be pieced together, with pyroclastic units separated by erosional surfaces indicating periods of volcanic quiescence and sedimentary processes. Following the initial period of caldera collapse, which was accompanied by highly explosive eruptions, volcanism was generally less explosive and formed abundant high-grade and lava-like ignimbrites. The majority of magmas were derived from a mantle source compositionally transitional between N-MORB and the Iceland plume. Geochemical differences between the mafic units are derived from different degrees of melting, from different source regions. All magmas show some degree of crustal contamination by various units, and their isotope geochemistry can be used to constrain the poorly understood crustal architecture in the region. U-Pb zircon geochronology shows that magmatism in central Arran took place over a very short space of time. The difference in interpreted 206Pb/238U ages between the oldest (Allt Ruadh Member ignimbrites; 58.92 ± 0.19 Ma) and the youngest (granitic Glenloig Hybrids; 58.71 ± 0.07 Ma) dated units is ≤ 470 kyr.

This work discusses the dynamic development of hydraulic fractures, their evolution and the resulting seismicity during fluid injection in a coupled numerical model. The model describes coupling between a solid that can fracture dynamically and a compressible fluid that can push back at the rock and open fractures. With a series of numerical simulations it is shown how the fracture pattern and seismicity change depending on changes in depth, injection rate, Young's modulus and breaking strength. Simulations indicate that the Young's modulus has the largest influence on the fracture dynamics and also the related seismicity. Simulations of rocks with a Young's modulus smaller than 10 GPa show dominant mode I failure and a growth of fracture aperture with a decrease in Young's modulus. Simulations of rocks with a Young's modulus higher than 10 GPa show fractures with a constant aperture and fracture growth that is mainly governed by a growth in crack length and an increasing amount of mode II failure. This change in fracture geometry evolution has an effect on the observed seismicity. Rocks with a Young's modulus of 10 GPa have the smallest moment magnitude while both decrease and increase of Young's modulus value contribute to a growth of the seismic moment magnitude. The signal is further altered by non-linear change in dip and tensile angle depending on the Young's modulus value. It is proposed that two distinct failure regimes are observed in the simulations. Below 10 GPa a fracture propagates through growth in aperture, this causes the fracture tip to be under constant extension. For rocks above 10 GPa, the aperture is small and the fracture is under compression. In this case fracture growth is driven by stress intensification at the crack tip, which causes fracture opening to have greater proportion of mode II compared to mode I. To suppliment the observations made from numerical simulations, laboratory experiments with air injection into vertically orientated Hele-Shaw cell were carried out. Strain analysis of the recorded experiments showed stress regimes that are very similar to the ones observed during numerical simulations with soft rocks. In both cases negative strain fields could be observed in front of the fracture tip. This indicates that fracture propagation for soft materials is driven by tensile failure and walls being pushed apart. Further analysis on fracture propagation mechanisms and solid media response were carried out. These results are applicable to the prediction of fracture dynamics and seismicity during fluid injection, especially since we see a transition from one failure regime to another at around 10 GPa, a Young's modulus that lies in the middle of possible values for natural shale rocks.

This thesis uses high-quality 3D seismic data from the Broad Fourteens Basin (Southern North Sea), Espírito Santo Basin (SE Brazil), and Taranaki Basin (New Zealand) to characterise the evolution of geological structures related to differential compaction and subsidence; also known as subtle hydrocarbon traps. Each chapter tackles deformation over a different geological feature, spanning from salt-withdrawal basins, to submarine channel complexes and associated mass-transport deposits. These chapters subsequently discuss the impact the results have on the hydrocarbon industry. Included in these discussions are the importance of subtle traps on carbon capture and storage, local sealing potential, and reservoir distribution. The Broad Fourteens Basin dataset was used to investigate concentric faults associated with salt withdrawal from below Triassic units. Throw-depth and throw-distance plots helped to understand the growth histories of the concentric faults. It was shown that these faults formed as a result of the bending of strata due to differential subsidence during salt withdrawal. Slip tendency analyses assessed the likelihood for faults to reactivate and transmit fluids whenever pore fluid pressure is increased. This approach simulated a typical profile during carbon capture and storage. It was shown that concentric faults will reactivate if pore fluid pressures are increased above 30 MPa at the relevant sub-surface depths, leaking fluids (including stored CO2) past regional seal intervals in the basin. Data from the Espírito Santo Basin were first used to assess the timing and magnitude of differential compaction over a submarine channel complex. Thickness-relief models helped quantify both the variations in thickness in overburden strata. Smaller channels associated with downslope knickpoints were located within the channel complex. Differential compaction over channels produced four-way dip closures, as coarse-grained sediments were deposited at the knickpoint base. These provide adequate structural traps after early burial. The Espírito Santo Basin 3D survey was used in a third chapter to assess how differential compaction affected sediment distribution over a mass-transport deposit. As large remnant and rafted blocks entrained within the MTD were buried, differential compaction produced anticlines over them. This created a rugged seafloor and the topographic highs confined sediment moving downslope, allowing it to pond in discrete depocentres. Results from the data analysis chapters were compared with compaction-related structures documented in the published literature. A novel classification for subtle structural traps associated with differential compaction was produced, separating each feature into one of four types; Type A: folds over tectonic structures >2 km wide; Type B: folds over sedimentary packages, typically elongate, ~500 m to 5 km wide; Type C: folds over topographic features that are 20 m to 2 km wide; Type D: folds over sub-seismic/outcrop features no larger than 20 m. The results of the classification can be used as a first assessment when recognising a compaction-related fold and to rapidly assess its evolution and effectiveness as a subtle hydrocarbon trap.

Predicting the geometry and continuity of clastic units within lava-dominated sequences in volcanic margin settings is problematic as they are typically laterally discontinuous, relatively thin and often poorly imaged in the subsurface. Although such sequences are well known, detailed studies of their lithofacies architecture are rare and poorly constrained, which can result in major challenges to hydrocarbon exploration in these settings. As the demand for hydrocarbons increases, exploration is being focussed on more challenging stratigraphic and tectonic settings, such as volcanic margins. Consequently, it is necessary we enhance our understanding of how petroleum systems interact with volcanic-prone sequences, in order to maximise recovery of hydrocarbons. Using a field analogue in conjunction with remote sensing datasets is fundamental to understanding these complicated systems. This study utilises the Hreppar Formation (HF) in SW Iceland as an analogue to understand elements of petroleum systems to reduce the challenges and risk associated with hydrocarbon exploration within volcanic-dominated basins. The HF at Flúðir, comprises basaltic lavas and interbasaltic sedimentary rocks of Plio-Pleistocene age (3.3-0.7Ma) and provides excellent 3D exposure. This study provides a comprehensive evaluation of the geology in this area and its relevance and importance to hydrocarbon exploration. Detailed field mapping and graphic logging have been combined with field panoramas and photogrammetry to characterise the sequence in detail and to identify the lateral (dis)connectivity of the clastic units, the main lithofacies, the different facies architectures, structural elements and drainage pathways within the HF. The detailed field data presented here are generally all below seismic resolution. In an offshore setting, with currently available technology, it is highly unlikely this level of detail can be captured using remote sensing tools alone. The advantage of using a field analogue such as that of the HF is the level of detail which can be captured. This enables the gap in scale, between field-scale and seismic¬/well-scale to be bridged. It allows models to be ground truthed, which reduces uncertainty and risk, essential to hydrocarbon exploration. This research identifies complex interaction between volcanic, glacial and fluvial systems, underpinned by a strong tectonic influence. >60% of the HF is dominated by sub-aerial basaltic lavas and predicting where lithofacies occur in these types of environments is challenging, however, through initial quantitative analysis of volcanic and sedimentary units in the HF, basic prediction in similar settings is possible. The field data collected in the HF can inform every stage of the development of a hydrocarbon field in a volcanic margin, from determining the architecture of a potential reservoir to defining the main structures and potential fluid pathways as well as deciding how to produce the field.

Layered intrusions offer an exceptional opportunity to study magmatic processes operating in the upper crust, including the concentration of world class precious-metal deposits. Despite the importance for understanding igneous processes and mineral resources, the fundamental construction mechanism of these intrusions is not well understood. The Rum Layered Suite (RLS) in NW Scotland represents an archetypal open-system layered intrusion that formed as part of the North Atlantic Igneous Province ~60 Ma. The peridotite cumulates within the RLS are poorly studied, where formation is typically attributed to the early stages of batch-fractionation from picritic and basaltic magma. This study focuses on the peridotite cumulates from Unit 10 in the Eastern Layered Intrusion (ELI) and Western Layered Intrusion (WLI). The cumulates contain abundant harrisite, an unusual skeletal-olivine bearing peridotite that crystallised in situ. Harrisite layers are used to represent picrite sills, with features of harrisite in Unit 10 and WLI, inconsistent with processes occurring on the magma chamber floor, also highlighting multiple replenishment events. The unusual olivine morphology in harrisites is controlled by the crystal mush temperature on intrusion, enhancing undercooling (i.e., chilling). Picrite emplacement caused a reaction with intercumulus plagioclase, forming Cr-spinel seams in situ alongside harrisite. The seam formed in discrete intra-mush melt channels where spinel chemistry, texture, and precious-metal enrichment is controlled by melt flux. Lateral magma intrusion and melt migration was facilitated by slow gravitational collapse of the crystal mush, producing planar shear zones and hot tears in a high-crystallinity framework. Repeated sill emplacement caused widespread dissolution to pre-existing cumulate and recrystallisation of primary mineralogy in situ by infiltrating melt. The evidence presented here is used to argue against traditional magma chamber models of layered intrusion growth, where the RLS instead represents a sill complex, consisting of numerous sill-like replenishment events into pre-existing feldspathic cumulate. The upward growth of layered sequences typically attributed to relative age is refuted as sill emplacement is random. As such, the structural position of cumulate sequences in layered intrusions, especially for structurally low ultramafic cumulates, cannot be used as an indicator of relative age. The sill complex model for layered intrusion growth is not restricted to the RLS, where evidence presented here supports sill emplacement mechanisms even in the largest layered intrusions.

The modes in which faults can propagate and grow through subsurface rocks and strata are key to the establishment of fluid paths in sedimentary basins; faults are potential conduits for fluid in some regions, at the same time they are associated with fault-related traps in others. The classical fault propagation models addressed in the published literature have so-far considered isolated, linkage (lateral-tip linkage and dip-linkage), constant-length, and coherent models. However, the propagation histories of faults in regions dominated by salt tectonics are scarcely documented; rather, the existing fault propagation models lack critical thinking when applied to crestal faults, particularly due to the limited resolution of imaged strata in most publications, and the relatively small size of crestal faults (length < 2.3 km, maximum throw < 50 m). With the increasing use of high-resolution seismic data in recent decades, it is now possible to undertake research into the evolution of both crestal faults and fluid flow paths in regions dominated by salt tectonics. In parallel, the uniqueness of crestal faults in terms of their scales has brought up important questions about how data resolution and scale variance influence many a fault analysis, and the current fault propagation models, when based on seismic and outcrop information. This research uses high resolution seismic data from the Espírito Santo Basin, offshore SE Brazil, to investigate the growth histories of crestal faults, fluid flow in an area of significant salt tectonics, and how crestal faults are associated with traps in supra- salt successions. To answer the question, in a second stage, of how scale variance can influence the analysis of faults’ propagation histories, data from Somerset (Bristol Channel) and the Ierapetra Basin (Crete) were collected in the field to broaden the database in this thesis from the larger, rift-basin scale to the seismic and sub-seismic scales. Segment linkage is predominant in areas where crestal faults grow. Interpreted crestal faults in SE Brazil propagated vertically and horizontally. Horizontal propagation was often hindered by natural barriers such as an accommodation zone (Chapter 4), or oblique transfer zones (Chapter 5), onto which faults terminate. Vertical propagation stopped when the fault meets the sea floor or when vertical propagation was accommodated by blind faults or larger (adjacent) faults showing relatively large displacements. Hence, this thesis shows that the propagation of crestal faults does not follow a ‘coherent growth model’. Rather, the geometry and history of propagation of discrete faults segments are not comparable. In SE Brazil, large fault segments propagated to link with non-reactivated small fault segments on the crest of the salt ridge, and can show later ‘blind’ propagation towards the surface. In terms of how scale variance can potentially (negatively) influence fault growth models interpreted on seismic data and in the field, a new quantitative method and two new parameters (sampling interval and module error) are introduced in this thesis for faults of multiple scales - from a few meters to 10s of kilometers. Sampling interval has a significant influence on the interpretation of fault growth histories. By changing one’s sampling interval: 1) the interpretation of fault geometries is significantly changed; 2) maximum fault throw values are underestimated; 3) fault segments are underrepresented; 4) the geometry of fault linkage zones is changed; 5) the width of fault linkage zones is underestimated; 6) fault interaction zones are lost. Using the SE Brazil seismic data, the accuracy of Throw-Distance plots was shown to be quantitatively lost when sampling intervals were larger than 37.5 m (every 3 shot-points) for the ‘unique’ crestal fault families in this thesis. However, this thesis demonstrates that sampling intervals adopted by interpreters should differ depending on the resolution of seismic data used, and the total length of investigated structures. A practical sampling interval/fault length ratio is therefore proposed in this work to address the caveats behind using variable (and indiscriminate) sampling intervals when analysing faults. Supra-salt sequences capable of promoting episodic fluid flow in regions of salt tectonics are of vital economic importance. Following on the two latter themes (crestal faulting and fault scaling), the thesis addressed the episodic fluid flow documented in the Espírito Santo Basin in a third stage. The results of this section are proposed as a case study for supra-salt sequences. In detail, seal failure is systematically recorded in the study area, and is interpreted to have contributed to most of the supra-salt fluid flow events investigated in SE Brazil. Six types of traps are therefore widely identified in supra-salt successions of the Espírito Santo Basin – all forming examples of trapping geometries in sedimentary basins associated with salt tectonics. Regardless of a thermogenic or diagenetic origin for fluid off Espírito Santo, the results in this thesis demonstrate important (and focused) fluid flow above salt giants when, at least, two critical conditions are observed: 1) a certain thickness of overburden strata is deposited on top of the salt structures, 2) the generation of highly developed (i.e. large) crestal fault systems is observed over these same salt structures. It is therefore postulated that, if overburden strata is thinner than a certain value, or pressure imposed by growing salt increases significantly, active salt intrusion occurring together with fluid flow will replace more focused fluid flow features in salt giants.

The formation of layered intrusions remains one of the most important, yet unresolved issues in the study of mafic-ultramafic systems, although they are of major economic significance, hosting more than 80 % of the world's platinum-group element (PGE) resource. In many layered intrusions, PGE mineralisation is associated with stratigraphic intervals that are characterised by pronounced igneous layering. Thus, the origin of layering and the emplacement mechanism of individual layers are closely related to the formation of PGE deposits. In this study, field-based and analytical evidence from the Bushveld Complex of South Africa and the Monchegorsk Complex in Russia is used to gain a better understanding of the small- and large-scale processes associated with the emplacement of layered intrusions. Detailed examination of drill core and field exposures suggest that sill-like intrusions of crystal mushes play an important role in the formation of layered intrusions, especially in the lower ultramafic portions of large complexes. In contrast, the in situ Sr isotope analysis of plagioclase from the upper portion of the Bushveld Complex indicates that the more mafic portions may also crystallise in situ from crystal-poor magmas, which can also undergo mixing. Moreover, mineralogical and microtextural work based on high-resolution elemental mapping highlights the importance of melt migration at different stages of cumulate solidification as a consequence of displacement by convecting interstitial liquids and compaction. Further, broadly stratiform PGE mineralisation in the Monchegorsk Complex cannot always be explained by a classic PGE reef model, in which the mineralised horizon marks the transition from sulfide-undersaturated to sulfide-saturated conditions. It is more likely that preformed sulfides were entrained in crystal mushes and emplaced into a semi-consolidated cumulate pile at different levels of the layered intrusion. Ultimately, thermal modelling shows that a multi-stage emplacement history of a complex should be regarded as highly prospective with respect to PGE-Ni-Cu mineralisation.