Subglacial hydrology is known to influence the flow of ice. However, difficulty in accessing the base of large ice sheets has made determining the interaction between ice streams, basal sediment and water difficult to discern. The aim of this thesis is to determine the influence of subglacial hydrology on the flow of the West Antarctic ice streams. This is achieved through development of a numerical flowline model, the Hydrology, Ice and Till (HIT) model. Ice thermodynamics are coupled to a till layer of Coulomb plastic rheology. The porosity of the till changes with basal melt and freeze and can be augmented by water transported through a subglacial conduit system. Water availability strongly affects ice flow, as till porosity influences the till failure strength and thereby the basal resistance of the ice. The model was developed in four stages and a number of sensitivity tests were performed. It was then applied to Kamb Ice Stream (Ice Stream C) and Whillians Ice Stream (Ice Stream B), West Antarctica. Results confirm that ice streams are capable of oscillating between fast and slow velocity states. Cycles are generated at the grounding line of an ice stream and the speed of the transition from slow to fast flow is governed by water availability. The period of oscillation of the cycles for the West Antarctic ice streams was found to be several hundred years, which is in line with observations of stagnation and reactivation of these ice streams. This shows that subglacial hydrology has a role in modulating the flow variability of ice streams and that rather large changes in the flow of the West Antarctic ice streams are likely to occur this century.

Ice stream research has gained momentum in the last decade due to the increased need to understand ice sheet dynamics and instability and, by extension, the role ice streams have on climate change and sea level rise. Although significant progress has been made recently in understanding the role of ice streams in ice sheet dynamics, much remains to be done for documenting and understanding paleo-ice stream records. This is especially true for terrestrial paleo-ice streams like those that operated in the southwest Laurentide Ice Sheet. In previous studies evidence was shown for at least two large paleo-ice stream systems in southern Saskatchewan and a model was proposed involving major glacial dynamics shifts during the Late Wisconsinan linked to ice streaming in the Prairies. The goal of this research is to further characterize drift provenance and to verify sediment-landform assemblages that are central to the proposed reconstruction. This is done through spatial analysis of sediment and landform characteristics (e.g. compositional data, till fabric, landform identification). Provenance is also investigated using radiometric data and 40Ar/39Ar dating of hornblende grains. All of the data are geo-referenced in GIS to examine the spatial relationships. Results show spatially consistent patterns that fit within the structure of the proposed ice stream model. Ten assemblages were delineated, some showing landform evidence of southwestern and southeastern flow, while others show an older western signature through compositional data. The ice flow dynamics were characterized by a combination of landform evidence and compositional data, where three distinct ice flow phases can be seen. Of these three flow phases, two can be characterized by paleo-ice streams. The older south westward Maskwa system flowed against the regional slope, creating a large area of mega-scale glacial lineations, as well as transverse ridges, and was bound on either side by hummocky terrain. The preservation of the southwest Maskwa system was due to the abrupt shift to the southeastward Buffalo system. The Buffalo system captured the subglacial water from the Maskwa causing its shutdown, which fed the James Lobe until it collapsed. The Buffalo paleo-ice stream was the youngest and least stable of the two systems, as shown in the cross-cutting landform evidence. This approach combines multiple methods of analysis to go beyond the geomorphologic evidence to test the main underlying assumptions of paleo-ice stream landsystem models. This is critical if we are to understand the processes involved in the formation of paleo-ice streams and to reconstruct their evolution. Further characterization of the paleo-ice stream systems in the Prairies is critical to improve our understanding of how large ice sheets, like the Laurentide Ice Sheet, evolved and eventually collapsed.

Ice stream research has gained momentum in the last decade due to the increased need to understand ice sheet dynamics and instability and, by extension, the role ice streams have on climate change and sea level rise. Although significant progress has been made recently in understanding the role of ice streams in ice sheet dynamics, much remains to be done for documenting and understanding paleo-ice stream records. This is especially true for terrestrial paleo-ice streams like those that operated in the southwest Laurentide Ice Sheet. In previous studies evidence was shown for at least two large paleo-ice stream systems in southern Saskatchewan and a model was proposed involving major glacial dynamics shifts during the Late Wisconsinan linked to ice streaming in the Prairies. The goal of this research is to further characterize drift provenance and to verify sediment-landform assemblages that are central to the proposed reconstruction. This is done through spatial analysis of sediment and landform characteristics (e.g. compositional data, till fabric, landform identification). Provenance is also investigated using radiometric data and 40Ar/39Ar dating of hornblende grains. All of the data are geo-referenced in GIS to examine the spatial relationships. Results show spatially consistent patterns that fit within the structure of the proposed ice stream model. Ten assemblages were delineated, some showing landform evidence of southwestern and southeastern flow, while others show an older western signature through compositional data. The ice flow dynamics were characterized by a combination of landform evidence and compositional data, where three distinct ice flow phases can be seen. Of these three flow phases, two can be characterized by paleo-ice streams. The older south westward Maskwa system flowed against the regional slope, creating a large area of mega-scale glacial lineations, as well as transverse ridges, and was bound on either side by hummocky terrain. The preservation of the southwest Maskwa system was due to the abrupt shift to the southeastward Buffalo system. The Buffalo system captured the subglacial water from the Maskwa causing its shutdown, which fed the James Lobe until it collapsed. The Buffalo paleo-ice stream was the youngest and least stable of the two systems, as shown in the cross-cutting landform evidence. This approach combines multiple methods of analysis to go beyond the geomorphologic evidence to test the main underlying assumptions of paleo-ice stream landsystem models. This is critical if we are to understand the processes involved in the formation of paleo-ice streams and to reconstruct their evolution. Further characterization of the paleo-ice stream systems in the Prairies is critical to improve our understanding of how large ice sheets, like the Laurentide Ice Sheet, evolved and eventually collapsed.

<p>This thesis presents a palaeoglaciological study aimed to determine the location, geometry and temporal evolution of palaeo-ice streams of the north-easternmost Laurentide Ice Sheet. The work was accomplished through the geomorphological interpretation of satellite imagery over 3.19 x 10⁶ km² of the Canadian Arctic, using a glaciological inversion scheme. Ice streams were active in this region during most of the time between the Last Glacial Maximum and the last deglaciation. A web of ice streams and inter-ice stream areas existed. Three major ice stream networks are identified: the M'Clintock Channel, Gulf of Boothia – Lancaster Sound and Hudson Strait. The M'Clintock Channel bears the most complex landform record, comprising three generations of palaeo-ice streams. Their location was weakly controlled by the subglacial topography and their geometry was determined by frozen-bed portions of the ice sheet, thus providing evidence for pure ice streams in the Laurentide Ice Sheet. In contrast, the more pronounced relief of the Gulf of Boothia – Lancaster Sound corridor supported topographically controlled ice streams. The landform record on emerged land along Hudson Strait is insufficient to support the existence of ice streams. It is therefore proposed that ice streams were constrained within the deep parts of the strait while flanked by cold-based zones on the margins. Small transient ice streams on Baffin and Prince of Wales islands drained local remnant ice caps during the collapse of the ice sheet. Analysis of the controls on the location and flow of palaeo-ice streams suggests that the interaction between the subglacial topography and thermal state of the substrate plays a more fundamental role than the geology. It is concluded that the behaviour of ice streams cannot be explained in terms of environmental controls alone, but the complex dynamics of ice stream shear margins and onset zones must be considered.</p>

This thesis presents a palaeoglaciological study aimed to determine the location, geometry and temporal evolution of palaeo-ice streams of the north-easternmost Laurentide Ice Sheet. The work was accomplished through the geomorphological interpretation of satellite imagery over 3.19 x 106 km2 of the Canadian Arctic, using a glaciological inversion scheme. Ice streams were active in this region during most of the time between the Last Glacial Maximum and the last deglaciation. A web of ice streams and inter-ice stream areas existed. Three major ice stream networks are identified: the M'Clintock Channel, Gulf of Boothia – Lancaster Sound and Hudson Strait. The M'Clintock Channel bears the most complex landform record, comprising three generations of palaeo-ice streams. Their location was weakly controlled by the subglacial topography and their geometry was determined by frozen-bed portions of the ice sheet, thus providing evidence for pure ice streams in the Laurentide Ice Sheet. In contrast, the more pronounced relief of the Gulf of Boothia – Lancaster Sound corridor supported topographically controlled ice streams. The landform record on emerged land along Hudson Strait is insufficient to support the existence of ice streams. It is therefore proposed that ice streams were constrained within the deep parts of the strait while flanked by cold-based zones on the margins. Small transient ice streams on Baffin and Prince of Wales islands drained local remnant ice caps during the collapse of the ice sheet. Analysis of the controls on the location and flow of palaeo-ice streams suggests that the interaction between the subglacial topography and thermal state of the substrate plays a more fundamental role than the geology. It is concluded that the behaviour of ice streams cannot be explained in terms of environmental controls alone, but the complex dynamics of ice stream shear margins and onset zones must be considered.

The lateral "corners" where Kamb and Whillans Ice Streams (KIS and WIS) discharge into the Ross Ice Shelf share common geometries and ice mechanical settings. At both corners of the now-stagnant KIS outlet, shear margins of apparently different ages confine regions with a relatively flat, smooth surface expression. These features are called the "Duckfoot" on the northern, right-lateral side and the "Goosefoot" on the other. It has been suggested, on evidence found in ice internal layers, that the flat ice terrains on KIS were afloat in the recent past, at a time when the ice stream grounding line was upstream of its present location. The overdeepening in the bed just upstream of the KIS grounding line supports this view of the past geometry. The right-lateral margin at the outlet of the currently active WIS, the location of Subglacial Lake Englehardt (SLE), appears to have many similarities with the right lateral margin of KIS, though with a less developed looking inboard margin. This paper presents a mechanical analysis using surface and bed topography and velocity datasets comparing the Duckfoot flat ice terrain with the terrain around Subglacial Lake Englehardt. At both locations mechanical thinning along shear margins and lows in the bed topography redirects basal water routing towards the features. Here, I consider the history of these features and their role in ice stream variability by comparison of the relict and modern features and via numerical modeling of ice shelf grounding and ungrounding in response to variations in ice flow. We propose two scenarios for the development of flat ice terrains/subglacial lakes at the outlets of ice streams. In the first, development of a lake in the hydraulic potential low along a shear margin forces a margin jump as shearing develops along the inboard shore of the margin lake. This thesis presents evidence for an inboard (relative to the main outboard shear margin) zone of shear along the inboard shoreline of SLE, suggesting that subglacial lakes along shear margins are capable of facilitating shear margin jumps. In the second, grounding line advance around a relative low in the bed, creating adjacent margins along the lakeshores, forms a remnant lake. Discerning which of these scenarios is appropriate at the KIS outlet has implications for understanding the history of the ice stream grounding line. An ice flow model is used to place these local conditions in a regional context by studying the effect of internal perturbations, such as ice rise stagnation or inward margin jumps, on grounding line position. Bathymetry is important in determining ice stream flow in the ways that might not be otherwise realized in 1-D flow model studies. In the numerical modeling experiments, grounding line advance across the KIS outlet is mediated by the overdeepening in the bed and proceeds not in the direction of ice flow but transverse to flow. This finding adds complexity to both a flowline view of grounding line migration and the theory that grounding lines are unstable in the presence of inward sloping bed topography.

Cette thèse concerne la modélisation des calottes de glace qui ont couvert l'hémisphère nord durant les dernières périodes glaciaires. Des améliorations de la physique ont été effectéees sur le modèle de calotte de glace Antarctique du LGGE, nommé GRISLI, afin de rendre ce modèle portable sur l'hémisphère nord. Une nouvelle méthode pour déterminer la position du front des ice shelves (plates-formes de glace flottante) a été mise au point, des conditions aux limites spécifiques au front des ice shelves ont été ajoutées. Un nouveau schéma pour le drainage de l'eau sous-glaciaire et un critère basé sur la pression de l'eau sous-glaciaire ont permis de localiser les ice streams (ou fleuves de glace) de façon bien plus réaliste qu'auparavant. Ce nouveau modèle est appliqué aux calottes de l'hémisphère nord et simule leur évolution lors du dernier cycle glaciaire-interglaciaire en comparant l'impact relatif du bilan de masse en surface et de la dynamique. L'évolution de la calotte eurasienne aurout de l'évènement à 90 000 ans (BP) a été détaillée pour étudier l'impact des lacs proglaciaires et le rôle des ice shelves pendant l'avancée et le retrait sur les mers de Barents et de Kara.

Indiana University-Purdue University Indianapolis (IUPUI) / In January 2007, 26 samples of till from 6 different moraines along the Reedy Glacier, East Antarctica were collected with the goal of differentiating between these samples and till collected from the base of the Whillans, Kamb, and Bindschadler Ice Streams of West Antarctica. The ability to differentiate between East and West Antarctic ice will allow us to constrain ice flow into the central Ross Sea during the Last Glacial Maximum (LGM), which has implications for more accurate reconstructions of the Ross Ice Sheet and its behavior. Moraines sampled from the head of Reedy Glacier give insight to the geology beneath the EAIS, and may be representative of what the glacier is eroding from its bed. Samples along the trunk of the glacier capture representative rock types eroded along the length of Reedy Glacier. At each moraine 3 replicate sub-sites were selected for collection to represent the diversity of material within each moraine. Comparisons are based on the composition of pebbles, particle size distributions, and sand petrography. Analysis of the pebble fraction shows that each sub-site contains similar rock types, however, the concentration of each rock type varies as much as 25-35%. Similar variation is also seen within the sub-site sand fraction. Both the pebble and sand fraction reflect the mapped bedrock geology. The dominant pebble types are coarse-grained felsic and intermediate igneous rocks, as well as quartzite. Similarly felsic igneous grains, quartzite, quartz, and feldspar characterize the sand fraction. Particle size analysis shows that v Reedy Glacier till averages 85% sand. The subglacial West Antarctic samples contain approximately 30% sand, and equal amounts of silt and clay, approximately 35% each. An observation of the sand fraction from beneath the West Antarctic Ice Streams shows composition similar to tills from Reedy Glacier. However, tills from the base of the West Antarctic Ice Streams contain up to 75% polymict grains, and in contrast, these grains are absent in the tills from Reedy Glacier. These sand-sized polymict grains dominate material from the base of Whillans and Bindschadler Ice Streams, whereas material from the base of Kamb Ice Stream contains grains of felsic igneous, quartz, feldspar, and few to no polymict grains. In addition to the polymict grains, the sand fraction in the ice stream cores contains trace fragments of sedimentary, and volcanic rocks, both of which are absent from the Reedy Glacier sand fraction. However, polymict grains are believed to represent a process occurring beneath the ice sheet, rather than indicate provenance. It is difficult to differentiate between the two tills, as both contain high concentrations of felsic-intermediate igneous lithics, quartz, and feldspar. The central Ross Sea contains sediment similar in rock type and mineralogy as seen within sediments from both Reedy Glacier, and the base of the ice streams of West Antarctica.

Sveriges geomorfologi har till största del bildats under den senaste glaciationens erosion och depositions processer. Vid denna glaciations avsmältning rann smältvattnet bort från glaciären i isälvarna genom tunnlar under inlandsisen. Från dessa isälvar bildas åsar när materialet som transporterats med glaciären förs med i det turbulenta flödet i isälven innan det avsätts och bygger upp dessa långsträckta terrängformer. Isälvarnas position under inlandsisen bestäms av hydrauliska potentialen vilken bestämmer rörelseriktningen för vatten under glaciären. I vissa områden kommer den ha en lägre potential och andra en högre potential beroende på isens tjocklek och formen av den underliggande terrängen. Där den hydrauliska potentialen konvergerar kommer isälvar bildas av de stora mängder smältvatten som transporteras bort. Syftet med detta projekt var att bekräfta åsbildningen i Svealand genom att göra en modellering av den hydrauliska potentialen under inlandsisen i ArcGIS. Två modelleringsalgoritmer för flödesriktning kallade D8 och D-infinity, jämfördes också för att avföra vilken av dessa som bäst modellerar smältvattnets flödesvägar under inlandsisen. Tidsintervallen 11 000, 12 000 och 13 000 år sedan valdes för denna undersökning då inlandsisen då gick från att ligga strax söder om Svealand till mellersta Svealand. Mellan 11 000 och 10 000 år sedan avsmälte glaciären väldigt hastigt och retirerade till Kaledoniderna i norra Sverige och används därför ej i denna undersökning.   I ArcGIS beräknades hydropotentialen för de olika tidpunkterna varefter sänkor i rastret fylldes upp. Flödesriktningen och flödesackumulationen beräknades för både D8 och D-infinity. Från flödesackumulationen togs isälvarna fram genom en omklassning av rastret. För att avgöra vilken flödesriktningsalgoritm som modellerade smältvattnets flödesvägar under inlandsisen bäst användes ett verktyg som summerade antalet pixlar av isälvar som låg under polygoner över nutida åsar i Svealand. Från detta beräknades en procentskillnad mellan D8 och D-infinity för att avgöra vilken av dem som stämde bäst överens med de nutida åsarna. En karta för varje tidsintervall som undersöktes samanställdes med de modellerade isälvarna och de nutida åsarna i Svealand för att visuellt avgöra om det gick urskilja en kronologisk bildningsföljd av åsarna.  D-infinity beräknades modellera isälvarna upp till 2,5 procentenheter bättre än D8. Skillnaden mellan modelleringsalgoritmerna minskade i takt med att glaciärtäckningen av Svealand minskade. I kartorna som sammanställdes gick en viss kronologisk trend att se. Vissa åsar som när de låg långt från inlandsisens kant blev inte modellerade som isälvar förrän inlandsisens kant kom närmare dem. Vissa modellerade isälvar låg inte direkt på de nutida åsarna utan lite till sidan av dem vilket skulle kunna bero på att den rumsliga upplösningen som modelleringen gjordes i var för grov eller att fler faktorer behöver tas med i modelleringen. Till exempel skulle snävare tidsintervall kunnat användas i modelleringen för att bekräfta den kronologiska bildningsföljden av åsar. Transmissivitetens koppling till den hydrauliska potentialen och avståndet mellan isälvar hade möjligtvis också förbättrat modelleringen av isälvarnas rumsliga position. / Sweden's geomorphology has largely been formed during the recent glaciation erosion deposition processes of recent glaciation. During the deglaciation, the meltwater ran away from the glacier in the channel ice streams through tunnels under the ice sheet from which eskers were formed. The position of the glaciers in the ice sheet is determined by the hydraulic potential, which determines the direction of movement of water below the glacier. In some areas, it will have a lower potential and others a higher potential depending on the thickness of the ice and the shape of the underlying terrain. In places where the hydraulic potential converges, ice rivers will be formed by the large amounts of melt water transported away. The purpose of this project was to confirm the esker formation in Svealand by modelling the hydraulic potential during the glaciation in ArcGIS. Two flow direction modelling algorithms, D8 and D-infinity, were also compared to determine which of these best models the meltwater flow paths under the ice sheet. The years 11,000, 12,000 and 13,000 years ago were chosen for this study as the ice sheet then went from lying just south of Svealand to central Svealand. Between 11,000 and 10,000 years ago, the glacier melted very rapidly and retreated to the Caledonids in northern Sweden and is therefore not used in this study. In ArcGIS, the hydraulic potential for the different time intervals was calculated, after which sinks in the grid were filled up, the flow direction and the flow accumulation were calculated for both D8 and D-infinity. From the flow accumulation, the channel ice streams were generated by a reclassification of the grid. To determine which flow direction algorithm best modelled the meltwater flow paths under the ice sheet, a tool was used that summed the number of pixels of channel ice stream that were below polygons over current eskers in Svealand. From this, a percentage between D8 and D-infinity was calculated to determine which of them best matched the current eskers. A map for each assessed year was compiled with the modelled channel ice streams, ice sheet edge and the current eskers in Svealand to visually determine whether a chronological sequence of the eskers could be discerned. D-infinity was estimated to model the channel ice streams up to 2.5 percentage points better than D8. The difference between the modelling algorithms decreased as the glacier coverage of Svealand decreased. In the maps that were compiled, a chronological trend could be inferred to a certain point. Some eskers that, when they were far from the edge of the ice sheet, were not modelled as channel ice streams until the edge of the ice sheet were much closer to them. The modelling could be improved by increasing the resolution in which the modelling was made as it might have been too coarse or that more factors need to be included in the modelling. For example, narrower time intervals could be used in the modelling to confirm the chronological sequence of ridges. The connection between the transmissivity and the hydraulic potential and the distance between ice rivers may have also improved the modelling of the spatial position of the ice rivers.

The West Antarctic Ice Sheet is drained primarily by five major ice streams, which together control the volume of ice discharged into the ocean across the grounding line. The grounding line of Kamb Ice Stream (KIS) is unusual because the ice stream upstream of it is stagnant. Here, a set of surface features--shore-parallel, long wavelength, low amplitude undulations--found only at that grounding line are examined and found to be "pinch and swell" features formed by an instability in the viscous deformation of the ice. When a relatively competent layer is surrounded by lower strength materials, particular wavelength features within the layer may be amplified under certain layer thickness and strain rate conditions. The undulations at KIS grounding line are possible due to the relatively large strain rates and particular ice thickness at that location. Several data sets are used to characterize the surface features. High resolution surface profiles are created using kinematic GPS carried on board a sled that was used to tow ice penetrating radar equipment. The radar data are used to examine the relationship between surface shape and basal crevasses. Additional surface profiles are created using ICESat laser altimeter observations. Repeat GPS surveys of a strain grid across the grounding line yields strain rate information. Analysis of repeat observations over tidal cycles and multi-day intervals shows that the features are not standing or traveling waves. Together, these observations are then used to evaluate the contributions of elastic and viscous deformation of the ice in creating the grounding line undulations.