This thesis is concerned with soil erosion from hillslopes in semi-arid Spain. It approaches the problem f rom three related angles which are reflected in its structure: Part 1: The literature is evaluated to determine the current conceptual understanding of the processes of soil erosion, and the organisation of those concepts and evidence from field and laboratory studies into predictive models of the major processes of erosion - rainfall dynamics, rainsplash, infiltration and runoff generation, runoff routing and wash erosion mechanics. Part 2: A field study of the major factors of erosion and their results is conducted over two field seasons for four experimental hilislope sites in southern Spain, and provides detailed information on the spatial and temporal patterns of erosion and their controls. Four main contextual variables are analysed - topography, vegetation cover, surface sediments (soil and coarse fraction) and micro-topographic roughness. The results from four process experiments are presented - infiltration from ponded and simulated rainfall experiments; runoff generation and soil loss from fifty microplot rainfall simulations; natural storm runoff and erosion losses from sub-site plots situated at the midpoint and base of each site; and seasonal and annual patterns of ground surface change derived from soil pin data. Part 3: Concern is with the development of selected erosion process simulation models. These include - the stochastic generation of a temporally nested rainfall series (yield, intensity, duration of wet and dry periods) for the annual/seasonal distribution of daily rainfall, and, subject to threshold criteria, the hourly pattern of storms within a day; an infiltration-runoff generation model based on a modified Green and Ampt (1911) infiltration equation, which predicts time to runoff, volume of pre-saturation soil storage, and runoff volumes. The latter are routed across plots according to an algorithm which determines potential flow lines in three dimensional space. A finite difference kinematic flow model using the Manning flow equation is developed and calibrated from microplot data; finally an erosion model developed from the prototype of Meyer and Wischmeier (1969) predicts spatial and temporal variability of erosion from hillslopes. Major changes to the original model include three-dimensional space and contour curvature, dynamic modelling of flow generation and routing, and physically based detachment and transport mechanics of overland flow. The thesis is divided into two volumes - Volume I comprises the text arranged into three parts. Volume II, for ease of reference, presents supporting and background material, which may be read in parallel with Volume I, and includes Appendices of raw data, results of data analysis, computer programs and simulation results; a list of symbols by chapter; a set of photographic plates, and a full bibliography.

Mountainous regions are ubiquitously dissected by river networks. These networks are the main drivers by which climate and tectonic signals are transmitted to the rest of the landscape, and control the response timescale of the landscape to these external forcings. Furthermore, river systems set the downslope boundary conditions for hillslope sediment transport, which controls landscape denudation. Therefore, understanding the controls on the organisation and structure of river networks in upland landscapes is an important goal in Earth surface processes research. The recent introduction of high-resolution topographic data, such as airborne lidar data, has revolutionised our ability to extract information from the topography, providing new opportunities for linking geomorphic process with landscape form. This thesis is focused on developing techniques for analysing high-resolution topographic data to quantify and understand controls on the structure of fiuvial systems in upland landscapes. Firstly, I develop and test new algorithms for objective feature extraction from lidar-derived digital elevation models (DEMs). I present a new method for identifying the upstream extent of channel processes by identifying scaling breaks in river long profiles. I then compare this new method to three existing methods of channel extraction, using field-mapped channel heads from four field sites in the US. I find that the new method presented here, along with another method of identifying channels based on valley geometry, most accurately reproduces the measured channel heads in all four field sites. I then present a new method for identifying floodplains and fiuvial terraces from DEMs based on two thresholds: local gradient, and elevation compared to the nearest channel. These thresholds are calculated statistically from the DEM using quantile-quantile plots and do not need to be set manually for each landscape in question. I test this new method against field-mapped floodplain initiation points, published flood hazard maps, and digitised terrace surfaces from eight field sites in both the US and the UK. This method provides a new tool for rapidly and objectively identifying floodplain and terrace features on a landscape scale, with applications including flood risk mapping, landscape evolution modelling, and quantification of sediment storage and routing. Finally, I apply these new algorithms to examine the density of channel networks across a range of mountainous landscapes, and explore implications for fluvial incision models. I compare the relationship between drainage density (Dd) and erosion rate (E) using both analytical solutions and numerical modelling, and find that varying the channel slope exponent (n) in detachment-limited fluvial incision models controls the relationship between Dd and E. Following on from this, I quantify Dd for five field sites throughout the US. For two of these field sites I compare Dd to cosmogenic radionuclide (CRN)-derived erosion rates, and for each site I use mean hilltop curvature as a proxy for erosion rate where CRN-derived erosion rates are not available. I find that there is a significant positive relationship between Dd, E, and hilltop curvature across four out of the five field sites. In contrast to assumptions made in many studies of fluvial incision, this positive relationship suggests that the channel slope exponent n is greater than unity for each of these landscapes, with fundamental implications for both landscape evolution and sediment transport.

Debris flows can be considered the most significant geological hazard in areas of high relief in Scotland having impacted upon slope foot infrastructure several times in recent years. The potency of this geohazard is anticipated to increase over the coming decades due to a climatologically enforced upturn in debris flow frequency. In thisresearch material and topographic controls on debris flow activity are investigated using a combination of field and laboratory based analysis of debris flows at six study sites across upland Scotland. Centrifuge modelling is also used to simulate theinitiation of debris flows in soils with varying particle size distributions.Spatial densities of debris flow measured in the field indicate that hillslopes underlain by sandstone and granitic bedrocks, which tend to be mantled by coarser sand rich soils, have a greater frequency of flows than those underlain by schist andextrusive lava bedrocks. Higher debris flow densities on slopes underlain by sandstone and granite lithologies are facilitated by high permeability in overlying regolith matrixes allowing more rapid increase in pore water pressures duringrainstorms although this is likely to be further influenced by packing and organic content. Centrifuge modelling of hillslope debris flows also demonstrate that sandier soils are generally geotechnically more susceptible to slope failure.The susceptibility of a hillslope to debris flow is strongly influenced by slope geometry and morphology. Hillslopes with persistently steep slopes and a high incidence of concavities, gullies and couloirs are topographically more predisposed todebris flow activity due to greater shear stresses and morphologically controlled, gravity induced concentrations of hillslope hydrology. The majority of material in channelised debris flows is entrained during the gully propagation stage of the massmovement. Consequently, such events can be considered accumulative channelised debris flows. Longer and steeper gullies with greater sediment capacities are more likely to yield larger flow mass movements. Coupling between open hillslopes andbedrock gullies is shown to be an essential component for conceptualisation of the debris flow geohazard.Due to the role they play in amplifying debris flow magnitude, hazard management should be focussed around bedrock gullies and stream channels. Highesthazard rankings should be assigned to slope foot infrastructure in proximity to gullied stream channels with high sediment capacities and long, steep profiles conducive to large accumulative channelised debris flows. To avoid detrimental aesthetic impact, hazard management should be strongly geared towards utilisation of lower impactexposure reduction techniques and less visually intrusive engineering approaches such as increasing culvert capacity to accommodate debris flows. During realignment or the planning of future transport infrastructure, culverts with capacities significantly exceeding those required for purely hydrodynamic considerations should be placed straight on to stream channels avoiding proximal gully bends.

This thesis includes two different studies in an attempt to investigate and better understand the key characteristics of landscape evolution. In the first study, the rate of surface particle creep was investigated through the use of Terrestrial lidar at an archaeological site in Grand Canyon National Park. The second study developed ways to quantify metrics of the Colorado River drainage and reports the role of bedrock strength in the irregular profile of the trunk Colorado River drainage. Archaeological sites along the Colorado River corridor in Grand Canyon National Park are eroding due to a variety of surficial processes. The nature of surface particle creep is difficult to quantify and managers of this sensitive landscape wish to know the rates of erosion in order to make timely decisions regarding preservation. In the first study, two scans of a single convex hillslope were collected over the span of 12 months through the use of a ground-based lidar instrument. The scans were used to track the movement of rock clasts. This study, with a relatively small data set, did not show the expected positive relations of creep rate to slope or clast size, but did not preclude the existence of these relations either. The remarkably irregular long profile of the Colorado River has inspired several questions about the role of knickpoint recession, tectonics, and bedrock in the landscape evolution of Grand Canyon and the region. Bedrock resistance to erosion has a fundamental role in controlling topography and surface processes. In this second study, a data set of bedrock strength data was compiled and presented, providing relations of bedrock strength to hydraulicdriving forces of the trunk Colorado River drainage. Results indicate that rock strength and topographic metrics are strongly correlated in the middle to lower reaches of the plateau drainage. In the upper reaches of the drainage, intact-rock strength values are ~25% higher without a matching increase in stream power. As more tensile strength samples are analyzed and appropriately scaled with respect to fracturing and shale content, we believe we will see a clearer and more consistent pattern in the upper reaches.

La morphologie de l’île tropicale volcanique de la Réunion (Océan Indien) est caractérisée par des vallées fortement incisées et des structures pluri-kilométriques particulières de forme circulaire appelées « cirques ». Ces morphologies sont présentes à la fois sur les flancs du volcan éteint, le Piton des Neiges, et sur les flancs du volcan en activité, le Piton de la Fournaise. Elles témoignent des processus d’érosion et de transport sédimentaire intenses qui s’y produisent. L’intensité de ces processus est liée à la fois au relief prononcé et aux évènements météorologiques extrêmes, les cyclones, durant lesquels de fortes pluies provoquent un ruissèlement important sur les versants et les crues des rivières. Ces crues, d’un débit qui peut atteindre plusieurs milliers de mètres cubes d’eau par seconde, sont responsables d’un transport important des sédiments vers la côte. Dans ce travail, nous avons testé des solutions permettant de mesurer des volumes sédimentaires érodés et transportés par l’utilisation de données de télédétection multi-sources. En effet l’intensité des crues ne permet pas une mesure précise des volumes sédimentaires transportés par les rivières durant les cyclones, les appareils de mesures classiques ne résistant pas à ces évènements très énergétiques. Nous avons donc développé des méthodologies d’acquisition et de traitement de la topographie par LiDAR terrestre et par imagerie optique acquises depuis un ULM. Ces deux techniques ont été mises en oeuvre sur le terrain en mai 2014 et en mai 2015. De plus, afin d’étendre la période temporelle des observations nous avons valorisé les données d’archives d’imagerie aérienne qui existent depuis 1949 sur l’île. Nous avons calculé des MNT diachroniques, et nous avons développé une méthodologie de corrélation d’images SPOT-5 en contexte tropical fortement végétalisé.Ces travaux ouvrent de nouvelles perspectives pour explorer le potentiel de la télédétection afin d’affiner les connaissances sur les volumes de sédiments érodés et transportés en contexte volcanique tropical. / The morphology of the tropical volcanic island of la Réunion (Indian Ocean) is characterized by deeply incised canyons and plurikilometric structures with a circular shape called cirques. These morphologies are present both on the slope of the extinct volcano, the Piton des Neiges, and the active volcano, the Piton de la Fournaise, and are the consequence of intense sedimentary erosion and transport processes. The intensity of these processes is linked to the pronounced surface relief and to extreme meteorological events, such as cyclones, during which heavy rainfall causes significant runoff on hillslopes and subsequent river floods. During extreme floods, flows can reach several thousands of cubic meters per second, and is the primary mechanism responsible for the transport of sediment to the seashore.The intensity of extreme floods makes precise measurements of the sedimentary volumes transported by rivers during cyclones difficult because typical in-situ measuring devices are not able to withstand such energetic events. In this work, novel solutions to measure eroded and transported sedimentary material using multi-source remote-sensing data are evaluated. Methods were developed to acquire and process topographic data using terrestrial LiDAR and optical imagery, with photographs acquired from a microlight. These techniques were implemented on the island in May 2014 and May 2015. Furthermore, in order to extend the temporal observation period, archived aerial images acquired since 1949 over the island were used. Diachronic DEMs were processed and a methodology for SPOT-5 image correlation in a densely vegetated tropical context was developed.This work opens new possibilities to explore the potential of remote-sensing techniques in order to better understand the erosion and transport of sediment volumes in this tropical volcanic context.

Hillslopes usually have high heterogeneity in terms of landscape processes. Interactions occur between geology, geomorphological processes, and vegetation distribution on a hillslope. This study was undertaken to assess the processes and interactions of geology, regolith production, geomorphological processes, channel formation and how these are influenced by the vegetation on a portion of Mariepskop. Mariepskop forms part of the Drakensberg Escarpment, but is a separate hillslope within the Mpumalanga Province. A north-eastern portion of the Mariepskop forms the study site, with a drainage line located within the site. Deciduous bush covers most of the study site, and grassland patches occur on the southern parts of the study site. Quartz-feldspar-biotite gneiss dominates most of the area with the cliff and higher parts consisting of feldspar-rich schist. Three site visits were undertaken where bedrock geology, weathering, soil formation, erosion, mass movement processes and the drainage channel were assessed. Maps of these processes as well as slope profiling and plan forms were compiled. According to the results, Mariepskop shows heterogeneous processes both laterally and vertically, with various degrees of interactions taking place. Underlying geology, mass movements on higher altitudes, and soil creep on lower altitudes occur on both the northern and southern parts. Processes mainly occurring on the northern part are rockfall from drainage channel incision, weathering, rill erosion and fluvial erosion within the drainage channel. Main processes on the southern part are mass movement in term form of slumping/debris flow, and erosion, in particular rainsplash and overland flow. Soil is deeper on northern part than on southern part of the study site. Geomorphological processes interact with the vegetation distribution over the study area. Grassland patches on the southern part of the study site are mainly due to slumping/debris flow, rainsplash erosion, convexity of the plan form (therefore no valleys) and oxidic soils occurrence. Similar geomorphological processes will probably influence grassland patches over the rest of Mariepskop. / Dissertation (MSc)--University of Pretoria, 2013. / gm2014 / Geography, Geoinformatics and Meteorology / unrestricted

In situ measurements of soil nitrogen dynamics is a potential method for evaluating the health of constructed wetlands following oil sands mining. The objective of this study is to measure and compare the soil nitrogen availability of a reclaimed fen (Sandhill fen) with a nutrient-rich reference fen (Poplar fen) and a nutrient-poor reference fen (Pauciflora fen) in the Athabasca oil sands region of northern Alberta. Total Nitrogen (TN), Nitrate (NO3-) and Ammonium (NH4+) supply rates were determined along wetland hillslope transects using Western Ag Innovations Plant Root Simulator (PRSTM) probes at all three sites in 2014. Net N mineralization, net nitrification and net ammonification were determined simultaneously using the buried polyethylene bag sampling method. Overall, TN supply rates were greatest at the poor fen and least at the constructed Sandhill fen. In contrast, mineralization was greatest at the rich fen but again least at the Sandhill fen. Mineralization at the Sandhill fen was controlled evenly by ammonification and nitrification, whereas the two natural sites were controlled by ammonification. Relatively low N supply rates and mineralization at the Sandhill fen were likely due to lower soil organic matter and limited soil moisture in these newly constructed substrates. Spatial differences along the hillslopes also varied among sites. The Sandhill fen had higher TN supply rates at the upslope positions but no significant differences in net N mineralization rates along the hillslopes. The rich fen also had higher TN supply rates at the upslope but greatest mineralization rates downslope. These results highlight the importance of N storage and transport processes and offer insight into the N status of a constructed fen. / Thesis / Master of Science (MSc)