This thesis develops a chronology of jökulhlaup (glacier burst flood) activity from Mýrdalsjökull in southern Iceland. New geornorphological, sedimentological and tephrochronological data has identified 15 flow events during the Holocene. The majority of these were hyperconcentrated flow events originating from, or close to, the northwest area of the ice cap and are associated with subglacial volcanism. One flood originated in the Vei«!»ivötn area and on 3 occasions flooding from Katla may have been accompanied by floods from Eyjafjallajökull. A further two events relate to re­mobilisation of thick airfall tephra deposits. Silicic pumice found on the sandur and close to the ice margin indicates that the Markarfljót acted as a terrestrial transport route for pumice found along North Atlantic coasts, and was possibly a route for silicic Katla jökulhlaups. Additionally, flood routes and glacial landforms show that Entujökull reached a maximum Holocene extent in the mid-Holocene, extending farther downvalley than during the Little Ice Age. In prehistory, floods were directed to both the south and west of Mýrdalsjökull. Similarly timed jökulhlaups took these paths when floods also flowed from Eyjafjöll into the Markarfljót. This suggests that concurrent routing of floods to the south and west of Mýrdalsjökull is related to synchronous volcanic activity in Katla and Eyjafjallajökull. Since the 10^th Century most Katla floods have been routed to the south east, possibly reflecting changes in intra-caldera eruption sites or subglacial topographic change associated with the Eldgjá eruption in c.935 AD, as suggested by Larsen (2000). The environmental impacts of these floods were significant. Late prehistoric and early historic floods had a major role in shaping the landscape faced by the earliest Norse colonisers of the region. Future flooding could pose a distinct hazard to farmland on the sandur plain downvalley and to popular tourist areas in North Þórsmörk.

This thesis explores the influence that meltwater exerts on styles of ice-marginal sedimentation, using past and present examples from Iceland. The study glaciers display marked contrasts in form, size and composition of moraines which are unlikely to reflect differences in rates of subglacial erosion. This is because the study glaciers occupy a similar climate, show similar relief, sit above similar bedrock, and are inferred to flow at similar speeds. The observed variation in moraine properties must reflect some other process which intervenes to modify sediment transport relationships prior to the arrival of debris at the ice edge. I argue that this key factor which controls sediment transport - and, as a result, the potential to form moraines - is the behaviour of subglacial meltwater flows. Studies of the sediment load of its outlet river show that Sólheimajökull is a highly erosive glacier, yet the quantity of debris carried by the <I>ice</I> is extremely small. Consequently, present-day moraine formation is extremely limited. This can best be explained as the product of an aggressive subglacial drainage network which captures and evacuates the bulk of debris generated by subglacial erosion. This state of high efficiency subglacial flushing is likely to dominate the sediment budget of many temperate glaciers. Whereas the present-day margin of Sólheimajökull is debris-poor, the present-day margins of Gígjökull and Steinholtsjökull are debris-rich. This debris consists of two major populations: 1) rounded closets set in a sorted coarse sand and gravel matrix, derived from a series of englacial debris bands, and 2) sub-angular clasts in a poorly-sorted matrix, derived from unusually thick sequences of basal ice.

The study of reservoir seal formation characteristics is vital to the success of carbon sequestration projects. The unique properties of these formations allows for the safe long-term storage of carbon dioxide. These intrinsic properties also give rise to numerous experiment complexities outside of the realm of traditional core characterization techniques. Samples were obtained to represent the main classes of cap rocks; shales from both a quarry in the UK and a Spanish carbon storage pilot site, anhydrite from UK extraction mines, and a evaporite sample from a reservoir located in the Middle East. An apparatus has been constructed capable of measuring the permeability and capillary threshold pressure of reservoir cap rocks. The pressure decay technique was used to measure the permeability relationship of clay-rich and evaporite samples with varying applied stresses was measured. Unique trends are observed for each geologic sample exhibiting minimums in permeability. The initial reduction of permeability as effective pressure was increased was due compaction and the subsequent increase at high stresses was due to the opening of micro-fractures. The capillary threshold pressures of each sample were determined using three different techniques. A novel technique takes advantage of the pressure decay permeability measurements technique in quantifying extremely small fluid volumes during initial sample drainage. Capillary threshold pressures were shown to also be dependant on applied system stress. The capillary threshold pressure was observed to decrease dramatically following the increase in permeability with further increasing effective pressure. Imaging capabilities were also explored, ranging from core scale to nanometre scale techniques. Computerized micro-tomography was used in plug sample evaluation, and in the observation of fractured system behaviour under varying stress. Scanning electron microscopy paired with focused ion beam milling was used to extract the 3D pore space of the ceramic allowing for permeability estimates from numerical simulations.

Natural fractures are ubiquitous in crustal rocks and often dominate the bulk properties of geological formations. The development of numerical tools to model the geometry, geomechanics and fluid flow behaviour of natural fracture networks is a challenging issue which is relevant to many rock engineering applications. The thesis first presents a study of the statistics and tectonism of a multiscale fracture system in limestone, from which the complexity of natural fractures is illustrated with respect to hierarchical topologies and underlying mechanisms. To simulate the geomechanical behaviour of rock masses embedded with natural fractures, the finite-discrete element method (FEMDEM) is integrated with a joint constitutive model (JCM) to solve the solid mechanics problems of such intricate discontinuity systems explicitly represented by discrete fracture network (DFN) models. This computational formulation can calculate the stress/strain fields of the rock matrix, capture the mechanical interactions of discrete rock blocks, characterise the non-linear deformation of rough fractures and mimic the propagation of new cracks driven by stress concentrations. The developed simulation tool is used to derive the aperture distribution of various fracture networks under different geomechanical conditions, based on which the stress-dependent fluid flow is further analysed. A novel upscaling approach to fracture network models is developed to evaluate the scaling of the equivalent permeability of fractured rocks under in-situ stresses. The combined JCM-FEMDEM model is further applied to simulate the progressive rock mass failure around an underground excavation in a crystalline rock with pre-existing discontinuities. The scope of this thesis covers the scenarios of both two-dimensional (2D) and three-dimensional (3D) fracture networks with pre-existing natural fractures and stress-induced new cracks. The research findings demonstrate the importance of integrating explicit DFN representations and conducting geomechanical computations for more meaningful assessments of the hydromechanical behaviour of naturally fractured rocks.

This thesis develops a three-dimensional time-dependent numerical model that dynamically simulates the evolution of the flow and thermal regime of the Antarctic Ice Sheet in response to climatic changes. As one of the most important heat sinks and water reservoirs in the Earth’s energy and water systems, the behaviour of the Antarctic Ice Sheet strongly influences the natural environment on the Earth. This thesis first reviews the factors that affect the thermal and dynamical processes in the Antarctic Ice Sheet system and quantifies their functions in the system by setting up their mathematical relationships with other factors. The universal conservation equations of mass and energy serve as the governing equations of the ice sheet model. Different boundary and initial conditions for solving the governing equation set represent different scenarios of the ice sheet system. They are crucial components of the ice sheet model and are analysed at every interface between the Antarctic Ice Sheet system and the outside environment. The ADI scheme is used to compute the ice sheet model in one-dimensional, two-dimensional, and finally three-dimensional systems in order to highlight the role of the various factors affecting the behaviour of the ice sheet. The sensitivities of the various factors are tested in these idealised ice sheet systems. The parameters employed to model the Antarctic Ice Sheet in the three-dimensional system are then calibrated against the field data. By reconstructing the thermal and dynamical regimes of the Antarctic Ice Sheet under present-day and glacial maximum climatic conditions, the ice sheet model simulates the behaviour of the Antarctic Ice Sheet under changing climate. The modelling results support the idea of a stable East Antarctic Ice Sheet and a comparatively dynamic West Antarctic Ice Sheet.

This study analyses surveys carried out on the Isle of Lewis between 1993 and 1997, by the author. Surveys were targeted across representative areas of terrain to the west and east of the south of Loch Roag, and also at its southern, inland, end. Later an additional area was examined, the Isle of Bernera, located in the centre of Loch Roag, and much of the coastline of Lewis was also surveyed during a coastal erosion assessment commissioned by Historic Scotland, one of the important discoveries of which was a large number of promontory enclosures distributed all round the coast of the island. These latter surveys were used in this study to provide a wider base, and a test, for the results of the initial analysis. The aim of the study was to examine the viability of using archaeological data derived primarily from survey, supported with limited excavation, to model site and settlement distribution and landscape development. The extensive data sets were drawn not only from the author’ own surveys but also from the Royal Commission on Ancient and Historic Monuments of Scotland (National Monuments Record of Scotland and the First Edition Survey Project), and from the survey carried out by Ian Armit in the Bhaltos area in the early 1990s. Data was processed using GIS software (IDRISI and MAPINFO) to allow the location of individual classes of monument to be compared to the prevailing terrain types on the island. Monument type locations were also compared to each other and to place-names, and to the geomorphology. The resultant data allowed a model to be formulated for each monument type, that describes not only its relationship to terrain types but also data such as prominence in the landscape, association or otherwise with other monument classes, place-names and frequency. Following the initial formulation of the model, its predications were tested against the results of the survey of Great and Little Bernera.

Meteorological data from one automatic weather station installed on two different glaciers at 3335 m and 4667 m in the Chilean Central Andes demonstrate that the relative importance of the terms in the energy balance equation shows substantial differnces from that of mid-latitude Alpine glaciers, with turbulent heat transfer accounting for 10 to 20% of the net energy balance. Intense evaporation, together with effective radiative cooling, dew point well below freezing and intense solar radiation favour differential ablation of the snow surface, which results in the formation of snow penitentes, these unique forms result in a further modification of the energy balance. Stable conditions and a clear atmosphere permit effective incorporation of fully distributed solar radiative models with good results. Melt rates are compared with theoretical models, showing that while evaporation is enhanced in the Andes, surface morphology decreases total ablation. Modelling work suggests that penitentes enhance conservation of snow cover and the consequences of their loss, due to changes in the initial climatic conditions, might be increased ablation over the whole season, decreased glacier mass balance and faster depletion of water resources. To implement distributed energy balance models both at the basin scale and at the microtopography scale, two additional tools where developed. One is a set of algorithms for the computation of terrain parameters from digital elevation models (DEMs) suitable for rough topography. These algorithms, together with those developed for shading and insolation, permit the treatment of overhanging surfaces, such as the penitentes. Another tool is a flexible and inexpensive remote sensing tool for albedo estimation using conventional terrestrial photography. The technique consists in georeferencing oblique photographs to a digital elevation model (DEM), defining a mapping function between the pixel of the image and the corresponding cell on the DEM. Once the image is georeferenced, the reflectance values recorded by the film or digital camera are corrected for topographic and atmospheric influences and for the effect of the photographic process. By comparing these to a surface of known albedo, the spatial distribution of albedos is calculated. The validation of the technique on Alpine terrain shows good agreement with measured values.

This thesis presents the first dynamic imaging of fluid/rock reaction using X-ray microtomography (μ-CT) and focuses on three series of experiments: (1) imaging a homogenous carbonate during dissolution using a laboratory scanner; (2) imaging heterogeneous carbonates at multiple flow rates using a synchrotron pink beam; (3) imaging the same rocks using a laboratory scanner at multiple reactive conditions incorporating effluent analysis. First the in situ reservoir condition imaging apparatus was adapted to image Ketton carbonate dynamically using a laboratory μ-CT scanner. 10 images were acquired over 2½ hours. Porosity and surface area were measured from the images and permeability and connectivity were calculated using flow models. Ketton dissolved uniformly at these conditions although the effective reaction rate (reff) was 16 times lower than those measured in batch reactor experiments with no transport limitations. Second the experimental apparatus was used with fast synchrotron-based μ-CT to image two more complex carbonates, Estaillades and Portland Basebed at two different flow conditions. ~100 images were taken over 2 hours, which captured the complexity of dissolution. It was found that the type of dissolution is both pore structure and flow rate dependent. A new type of dissolution, channelling, is observed which has a reff up to 100 times lower than batch rates. Third, effluent analysis was incorporated into the experimental apparatus. All three rocks were imaged again at two separate reactive conditions. The reff was between 10 and 100 times lower than the batch rates, with the lowest rates in samples with the most channelized flow, confirming that transport limitations are the dominant mechanism in determining reff at the fluid/solid boundary. Effluent analysis confirmed that using the in situ, rather than the injected pH, to determine reff is valid in the uniform regime, but overestimates reff with channelling by an order of magnitude.

Barrier dunes represent potentially long-term, but complex, archives of coastal evolution. The examples occupying the Wilderness embayment, on the southern Cape coast of South Africa, form a regionally unique system of three shore-parallel barriers reaching up to 200 m in height and extending up to ~32 km alongshore. This research combines chronological and sediment provenance analyses to reconstruct the emplacement and evolution of the Wilderness barrier dunes through the Mid- to Late-Quaternary. Thirty-six new luminescence ages collected from ten sites across the three Wilderness barriers are presented, and are combined with a compilation of dates from the literature to produce a high-resolution chronology of barrier accumulation. The record spans at least the last two glacial-interglacial cycles, with notable phases between 245-217 ka, 155-143 ka, 128-121 ka, 91-86 ka and post-6 ka. Analysis of trace element geochemistry, heavy minerals, particle size, carbonate content and offshore topographic evidence all combine to indicate the provenance of the barrier sands has remained constant throughout their formation, and must involve marine transport pathways. The hypothesis that barrier accumulation at Wilderness during periods of low sea level was sustained by terrestrial aeolian activity is thus disproven, and evidence for a regional pre-MIS 5 marine transgression is provided. The terrestrially derived fraction of the barrier sands predominantly comprises quartzitic material derived from Table Mountain Group (TMG) rocks, most likely sourced from the Gouritz River ~75 km west of Wilderness. In addition to sediment from the TMG, the barrier sands also contain contributions of material derived from local geology, of material recycled from previous generations of aeolianite, and of authigenic marine sediment. The extensive coversand deposits inland of the Wilderness embayment, dated to >1.6 Ma using isothermal thermoluminescence, are demonstrated not to have made any significant input of sediment to the barriers. The Wilderness barriers record a complex history of erosion, as well as deposition through the Mid- to Late-Quaternary, and the preserved record clearly reflects the influence of local nearshore bathymetry on the rate of sea-level regression. The importance of previous generations of aeolianite in both fixing the position of subsequent depositional episodes, and protecting them from erosion, is also evident. The barriers exhibit similar behaviour to deposits on tectonically stable coastlines elsewhere, and contrast with the more complete and widely spaced barrier records present on uplifting coasts.

Groundwater is widely developed for town water supplies in weathered crystalline rocks in sub-Saharan Africa but the sustainability of this abstraction is unknown. Groundwater flow and storage in aquifers underlying two towns in central (Wobulenzi) and southwestern (Rukungiri) Uganda are assessed using environmental tracers and aquifer responses to hydraulic stress. Stable isotope ratios ( H: H, O: O) in precipitation and groundwater, the timely response of groundwater levels to bimodal rainfall, and short groundwater residence times of less than 22 years and big proportions of modern groundwater (5 to 100 %), derived from reconstructed atmospheric inputs of H and anthopogenic gases (CFC-113, CFC-12, CFC-11), clearly indicate active rainfall-fed, groundwater recharge. Diagnostic plots (s versus t/r , log-log, log-linear, derivative, flow dimension) of drawdown responses are used to inform conceptual models of groundwater flow. In Wobulenzi, linear flow through individual bedrock fractures at early pumping times (between 600 and 1500 minutes) is succeeded by radial flow through interconnected fractures that induce vertical flow in a thick weathered (regolith) aquifer at late pumping times (approximately 1800 minutes). Groundwater abstraction from bedrock fractures of 12 m h" per borehole, has not significantly affected groundwater storage over the last 9 years due to vertical leakage from overlying weathered aquifers that is commensurate to rainfall-fed recharge. In Rukungiri, a highly productive aquifer comprising coarse-grained, fluvial sediments, is identified in palaeochannels of former westerly flowing river networks. Fluvial sediments can feature significant thicknesses in palaeochannels of major river networks truncated by Miocene to Pleistocene rifting but subsequent erosion in the intra-arch basin, draining to the downfaulted rift floor significantly constrains the extent and thickness of the aquifers and, hence, the sustainability of groundwater abstraction. Depletion of groundwater storage over the last 8 years as a result of abstraction (12 m h" per borehole), is indicated by water-level declines of 2.5 m a". The identification of the palaeochannel aquifer provides new insight into the understanding of the relationship between the geomorphology and hydrogeology of deeply weathered environments and a new target for groundwater development in the humid tropics.