Although levee deposits make up a significant part of modern and ancient deep-marine slope systems, details of their internal lithological composition and stratal architecture remain poorly documented. At the Castle Creek study area, strata of the Neoproterozoic Isaac Formation (Windermere Supergroup) crop out superbly in a kilometre-scale section through a sinuous deep-water channel-levee system (ICC3). Levee deposits near the outer bend of the channel consist of sandstone-rich (sandstone-to-mudstone ratio of 68:42), medium- to thick-bedded turbidites interstratified with thinly-bedded turbidites. Structureless sandstone (Ta), planar laminated sandstone (Tb), non-climbing ripple cross-stratified sandstone (Tc) and massive and laminated siltstone (Td) are common. Thick beds generally thicken and then thin and fine laterally over about 300 m. Thin-bedded strata, in contrast, thin and fine negligibly over similar distances. In the distal part of the outer-bend levee (up to 700 m laterally away from the channel) strata consist predominantly of thin-bedded Tcd turbidites with a much lower sandstone-to-mudstone ratio (35:65). On the opposite side of the channel, inner-bend levee deposits are mudstone-rich, locally as low as 15:85, and consist mostly of thin-bedded, Tcd turbidites, although thicker-bedded, Ta-d turbidites are more common in the lower part of the section. Lateral thinning and fining of beds is more rapid than their outer-bend counterpart. Levee deposits of ICC3 comprise three stacked decametre-scale upward-thinning and -fining successions. Each is interpreted to record a depositional history consisting of lateral channel migration, levee deposition, channel filling, and distal levee deposition. During the early stage of increasing levee relief it is proposed that the termini of individual beds progressively backstep towards the channel margin resulting in an overall lateral thinning of the stratal profile. This interpretation notably contrasts the common assumption that levee morphology is the result of the vertical stacking of beds that dip. In addition to field studies, laboratory experiments were conducted to determine the depositional threshold of non-climbing ripple cross-stratification, which is common in levee strata of ICC3. It was determined that non-climbing ripples form when bed aggradation rates are less than 0.015 cm/sec, and most probably in flows made up of poorly sorted sediment.

At the Castle Creek study area, levee deposits are well-exposed over an area of ~2.6 km wide and ~90 m thick. This provides an opportunity to describe their lateral and vertical lithological changes, and accordingly details about their reservoir geometry and stratal continuity. Here, levee deposits are divided vertically into packages, each consisting of a sand-rich lower part overlain sharply by a mud-rich upper part. Each lower part displays a consistent thickening then thinning trend laterally away from its genetically related channel. The characteristics of these packages suggest that they were controlled by recurring changes in the structure of channellized flows, which in turn was controlled by grain size and grain sorting. This ultimately was controlled by short-term changes in relative sea level. Moreover, some mud- and sand-rich strata are rich in residual carbon suggesting that mid-fan levees can serve as source rocks for hydrocarbon generation, and also reservoirs.

Indiana University-Purdue University Indianapolis (IUPUI) / Floods are one of the worst disasters in the United States. Each year, the government allocates a tremendous amount of manpower and money on flood prevention initiatives. As the first defense line, levees provide protection from temporary flooding (Makhdoom, 2013). These embankments are broadly classified according to the areas they protect, which could either be urban or agricultural levees within floodplains. In the U.S., most of the levees are handled by government agencies such as the U.S. Army Corps of Engineers and the Federal Emergency Management Services. On the other hand, non-levee embankments created by individual farmers (Olson & Morton, 2013) or naturally formed levee-like structures may not be in the government database. The initial purpose of this research was to assist Polis center on the “Mapping of Non-Levee Embankments in the Indiana” project. The non-levee embankments are not certified or engineered levee-like structures. They, therefore, impose lateral constraints on flood flows, reducing the floodplain storage capacity and increasing the flood velocity. These non-levee embankments can cause stream erosion and downstream flooding. Therefore, it is important to know the locations of these features. The first part of the proposed method adapted the Empirical Bayesian theorem and the low pass filter techniques to extract elevated linear features from LiDAR elevation data – Digital Elevation Model (DEM). The second part of the proposed methods combined the Minimum Bounding Circle (MBC) method and the Youden Index to locate the optimal threshold value that can be used to determine whether the extracted features are levee-like structures. The focus of this study is not only limited to artificial levee-like structures, but also takes the natural levees, or any potential levee-like structures into account because this study assumes all embankments play important roles during flood events.

In this dissertation, three studies were done for radar and hyperspectral remote sensing applications using statistical techniques. The first study investigated a relationship between synthetic aperture radar backscatter and in situ soil properties for levee monitoring. A series of statistical analyses were performed to investigate potential correlations between three independent polarization channels of radar backscatter and various soil properties. The results showed a weak but considerable correlation between the cross-polarized (HV) radar backscatter coefficients and several soil properties. The second study performed effective statistical feature extraction for levee slide classification. Images about a levee are often very large, and it is difficult to monitor levee conditions quickly because of high computational cost and large memory requirement. Therefore, a time-efficient method to monitor levee conditions is necessary. The traditional support vector machine (SVM) did not work well on original radar images with three bands, requiring extraction of discriminative features. Gray level co-occurrence matrix is a powerful method to extract textural information from grey-scale images, but it may not be practical for a big data in terms of calculation time. In this study, very efficient feature extraction methods with spatial filtering were used, including a weighted average filter and a majority filter in conjunction with a nonlinear band normalization process. Feature extraction with these filters, along with normalized bands, yielded comparable results to gray level co-occurrence matrix with a much lower computational cost. The third study focused on the case when only a small number of ground truth labels were available for hyperspectral image classification. To overcome the difficulty of not having enough training samples, a semisupervised method was proposed. The main idea was to expand ground truth using a relationship between labeled and unlabeled data. A fast self-training algorithm was developed in this study. Reliable unlabeled samples were chosen based on SVM output with majority voting or weighted majority voting, and added to labeled data to build a better SVM classifier. The results showed that majority voting and weighted majority voting could effectively select reliable unlabeled data, and weighted majority voting yielded better performance than majority voting.

No description available.

Floodplain is an important location for the economic and social development of society throughout history, although it afflicted by different disasters like floods and bank erosion from time to time. Population dynamics and distribution trends have important effects on the landscape and society. Population dynamics in floodplains influenced mostly by the flood events and related human activities to protect themselves and the properties. Thus, floodplain population dynamics is vital to understand human-nature relationships, disaster risk assessment, flood forecasting and flood damage costs in the floodplains. Economic losses and fatalities due to flood events increase with increased flood frequency and flood magnitude. Extreme hydrological changes are responsible for such changes in the flood events. Population growth, urbanization and developmental activities worsened the situation further in the floodplain areas. Learning effect, i.e. flood memory is noticeable after the flood which diminishes with time, however, levee, which is a popular flood protection measure, causes further reduction of learning effect. Researchers suggest that levees create a false sense of safety and can influence the mobilization of floodplain population. Several pieces of research indicate that along with flood events and intervention measures, inundation depth, durations, flood type (coastal/ riverine), economic, technological and political characteristics of the societies influence flood memory, hence, influence floodplain population dynamics. However, long term floodplain population dynamics and how it is related to different flood-related, i.e. hydrological and sociological variables are yet to be confirmed. Flood is one of the major natural disasters in the United States. About 10% of the population lives in the 100-year and 500-year floodplain of the United States. This study carried out to investigate the long-term trend of floodplain population dynamics in the USA and how the dynamics changes from country to state to county level considering Arkansas State and Sebastian County to observe the difference. Also, the relationships of floodplain population dynamics with hydrological and sociological variables such as flood events, damage, intervention measures and poverty assessed in the national, state and county level. Then the dynamics of the national level, the state (Arkansas) level and the county (Sebastian) level were compared to identify changes in dynamics from macro to micro-scale. Floodplain population dynamics evaluated in terms of population growth rate and the ratio of floodplain population to the total population from 1790 – 2010. The median centroids of the population growth rate from 1790 – 2010 were estimated to observe the changes in the central tendency of the population growth rate in the total area and in the floodplain overtime to locate the typical place of floodplain population growth. High-resolution raster image (1 km x 1 km) of historical population data and a high-resolution raster image (250 m x 250 m) of floodplain data were analysed using ArcGIS to evaluate floodplain population dynamics. Historical data regarding flood events, damage, structural and non- structural flood intervention measures and poverty from available databases were extracted, processed and analysed to establish the relationship among population dynamics, hydrological and social variables. Finally, the population dynamics of the three levels were compared statistically and discussed for changes. It is found that the central tendency of the total population growth rate and floodplain population growth rate of the USA shifts from east to west direction. The central tendency of the overall population growth rate located near the lower Mississippi river basin in the 1800s. It is consistent near the geographical centroid of the USA in Kansas since 1900. For the floodplain population, the central tendency of the growth rate stayed consistent in Arkansas from 1910 – 2010, i.e. in the lower Mississippi river basin. This location indicates the floodplains in Arkansas was preferable for floodplain population settlement for many decades. Change in floodplain population proportion of the whole country USA from 1790 – 2010 shows that the proportion of floodplain population is more than 80% during the initial years of settlement. Then it reduced to less than 40% over time. At the state level – Arkansas, the proportion of floodplain population is also more than 80% during initial settlement and reduced to less than 60% over time for counties with high percentage floodplain area. Flood events, structural and non-structural protection measures influenced floodplain population dynamics over time and learning effect, i.e. social memory of flood found was seven years. At the county level- Sebastian county, Arkansas, the proportion of floodplain population varies significantly with a variance of 0.04 and a standard deviation of 0.19. The ratio of floodplain population in Sebastian was 29% in 1820, which increased and decreased over the years and about 70% in 2010. In addition to flood events, different intervention measures, the floodplain population dynamics in the Sebastian county are influenced by the social, economic and political phenomenon, examining which is out of the scope of this study. The learning effect of flood seems to be eight years in the case of Sebastian county. The population dynamics influenced by a different social, geographical, economic, political and hydrological phenomenon which makes it a complicated system to comprehend within the present scope of the study. The result shows that the floodplain population dynamics can vary for different geographical scale. Therefore, the geographic extent should also need to consider while using population dynamics in flood disaster management and mitigation strategy development, policy formulation and implementation. / Flodslätter är viktiga platser för samhällets ekonomiska samt sociala utveckling och har varit det sedan är lång tid tillbaka, även om de från och till har drabbats av olika katastrofer såsom översvämning och erosion. Översvämning är en av de största naturliga katastrofer som sker i USA. Ungefär 10 % av befolkningen bor i 100- och 500-års flodslätter som emellertid förändras på grund av olika anledningar, inklusive översvämningar. I denna studie undersöks den tillfälliga relationen mellan översvämning, relaterad skada, olika ingripande åtgärder och fattigdomsnivå med en befolkningsdynamik i flodslättsområden. Vidare sker denna undersökning i tre olika nivåer – nationellt: USA som ett helt land, stats-nivå: Arkansas och län: Sebastian County, Arkansas där de olika dynamikerna i de tre nivåerna jämförs. På den nationella nivån utvärderades befolkningsdynamiken med avseende på flodslätten till den totala populationen och befolkningstillväxt. Resultatet visar att befolkningsdynamiken förändras beroende på om den betraktas ur ett nationellt, statligt eller läns perspektiv. Ur ett nationellt perspektiv flyttar flodslättsbefolkningar till olika områden med tiden. Ur ett statligt perspektiv är flodslättsbefolkningar känsliga mot översvämningar, skada och olika ingripande åtgärder. På länsnivå, utöver översvämningsrelaterade variabler, är flodslättsbefolkningar känsliga mot andra socio-ekonomiska faktorer som inte behöver vara relaterade till översvämningar. Resultatet indikerar således att översvämningar inte alltid behöver vara den faktorn som styr befolkningsdynamiken i flodslättsområden på den lokala nivån samt beslutet att flytta till en flodslätt påverkas av olika socio- ekonomiska faktorer, inklusive översvämning. På den nationella nivån kan däremot både påverkan av översvämningar och relaterade variabler på befolkningsdynamiken i flodslättsområden observeras.

Increasing interest, from the U.S. Army Corps of Engineers (USACE) and other agencies, in using deep-mixing methods (DMM) to improve the stability of levees constructed on soft ground is driven by the need to reduce levee footprints and environmental impacts and to allow for more rapid construction. Suitable methods for analysis and design of these systems are needed to ensure that the DMM technology is properly applied. DMM shear walls oriented perpendicular to the levee alignment are an effective arrangement for supporting unbalanced lateral loads. Shear walls constructed by overlapping individual DMM columns installed with single-axis or multiple axis equipment include vertical joints caused by the reduced width of the wall at the overlap between adjacent columns. These joints can be made weaker by misalignment during construction, which reduces the efficiency of the overlap. Depending on the prevalence and strength of these joints, complex failure mechanisms, such as racking due to slipping along vertical joints between adjacent installations in the shear walls, can occur. Ordinary limit equilibrium analyses only account for a composite shearing failure mode; whereas, numerical stress-strain analyses can account for other failure modes. Five case studies provided by the USACE were analyzed to evaluate the behavior of levee and floodwall systems founded on soft ground stabilized with DMM shear walls. These identified and illustrated potential failure mechanisms of these types of systems. Two-dimensional numerical stability and settlement analyses were performed for the case studies using the FLAC computer program. The key findings and conclusions for the individual case studies were assessed and integrated into general conclusions about design of deep-mixing support for levees and floodwalls. One of the significant findings from this research was to identify the potential for a partial depth racking failure, which can control design when the DMM shear walls are socketted into a relatively strong bearing layer. The potential for partial depth racking failure is not discussed in the literature and represents a new failure mode identified by this research. This discovery also highlights the importance of adapting suitable methods for analysis and design of these systems to address all potential failure modes. / Ph. D.

Infrastructure plays an instrumental role in the shaping of the landscape across many scales and is a critical human component within the landscape, yet these systems have tended to ignore the function of appearance and aesthetics in their design. Consequently, the relationship between our infrastructure, the environment, and us has become increasingly opaque. The majority of the vast infrastructure systems that weave throughout the landscape promote a mono-functional agenda which is relegated to the background of our everyday experiences. By investigating the traditional methods of designing infrastructure, we can begin to understand how to integrate aesthetics into the design of infrastructure. This is explored through one of the largest infrastructure systems in the United States; flood control. Flood control infrastructure in is an extensive system that has formed a protective barrier between human and natural processes for over 200 years. Its largest component, the levee, is an elegantly simple structure that contains many layers of significant cultural and historic aesthetic narratives. This thesis focuses on the levee as an infrastructure that mediates between natural processes and human development and studies how it can perform aesthetically to convey new meaning and value. What is the potential of the levee to become expressive in our lives, and be designed in such a way to move us? This new infrastructural paradigm explores the implications of utilizing aesthetics as an expressive and significant function of levee design that can inform and inspire the public and define a new dialogue between man, nature, and technology. / Master of Landscape Architecture

Les digues de protection contre les inondations sont d’une importance stratégique incontournable pour la sécurité des personnes et des biens. Une meilleure connaissance de leur fiabilité structurelle constitue un enjeu majeur pour les ingénieurs et pour les gestionnaires de ces ouvrages afin de répondre aux réglementations en vigueur et potentiellement les faire évoluer. La réglementation en France relative à la sûreté des ouvrages hydrauliques exige la réalisation d’études de dangers, au sein desquelles les démarches probabilistes d’évaluation de la sécurité prennent une place de plus en plus importante. De par la complexité de leurs mécanismes de rupture et l’absence de condition d’état-limite précis pour certains mécanismes de défaillance tels que l’érosion interne, la surverse ou l’affouillement, le calcul d’une probabilité de défaillance par des approches quantitatives demeure à ce jour impossible. Les digues induisent des problématiques particulières pour l’évaluation de leur fiabilité structurelle où l’intervention d’experts s’avère nécessaire et centrale. Ceux-ci doivent procéder à l’interprétation des données disponibles et la prise en compte de leurs incertitudes, pour pouvoir ensuite évaluer la fiabilité structurelle de digues en termes de probabilité de défaillance. L’objectif de la thèse est l’élaboration d’une démarche complète d’aide à l’évaluation probabiliste de la fiabilité structurelle des digues à partir du jugement expert. Une démarche scientifiquement justifiée pour le recueil et l’exploitation des évaluations expertes, incertaines mais quantitatives, de la fiabilité structurelle des ouvrages, sous la forme d’une probabilité de défaillance ou d’un coefficient de sécurité, assorti(e) d’une marge d’incertitude probabiliste. Afin de répondre à cet objectif, deux démarches ont été développée, « EiCAD » et « EiDA », toutes reposant (dans des ordres différents) sur les phases suivantes :- Une phase d’élicitation individuelle des avis experts (Ei) permettant le recueil des évaluations expertes probabilistes par la construction d’un formulaire de questionnement ;- Une phase de calibration (C) permettant de pondérer les évaluations expertes en fonction des degrés de précision et de justesse ;- Une phase d’agrégation (A) permettant une prise en compte simultanée de plusieurs évaluations expertes probabilistes ;- Une phase de débiaisage (D) permettant de traiter les biais susceptibles d’entacher les évaluations expertes probabilistes. Les démarches développées ont été mises en œuvre sur des cas de digues du Drac, dans l’agglomération grenobloise, pour évaluer la probabilité de défaillance, par jugement expert, vis-à-vis de mécanisme de rupture par glissement et par érosion interne / River levees as protective measures against flooding are a matter of utmost strategic importance for the security of people and property. A better knowledge of their structural reliability is a major challenge for engineers and managers of these structures in order to meet current regulations and potentially develop them. In France, the regulations relating to the safety of hydraulic structures require the realization of hazard studies, in which probabilistic safety evaluation procedures take on a more and more important role. Due to the complexity of their failure mechanisms and the lack of a specific condition of limit-state for some failure mechanisms such as internal erosion and overtopping or scour, calculating a probability of failure by quantitative approaches remains impossible to this day. Levees induce specific problems in evaluating their structural reliability where expert intervention is necessary and pivotal. They must interpret the available data and take into account the uncertainties in their analysis, in order to evaluate the structural reliability of levees in terms of probability of failure. The aim of the thesis is to develop and elaborate a complete approach for supporting probabilistic evaluation of structural reliability of levees based on expert judgments. It is designed as a scientifically justified approach to collect and use uncertain but quantitative expert evaluations of structural reliability, in the form of a failure probability or a safety factor, accompanied by a margin of probabilistic uncertainty. To meet this goal, two approaches have been developed, "EiCAD" and "EiDA" composed of the following phases (in different orders):- individual elicitation phase of expert judgment (Ei) allowing the collection of probabilistic expert evaluations by a constricted questionnaire ;- calibration phase (C) to weigh the probabilistic expert evaluations elicited depending on the degrees of precision and accuracy that can be provided;- aggregation phase (A) during which the probabilistic expert evaluations elicited by several experts are taken into account ;- debiasing phase (D) during which biases that may affect the probabilistic expert evaluations are removed. The developed approaches have been implemented for the cases of Drac levees, located in the Grenoble agglomeration, to assess the probability of failure, by expert judgment, with regard to sliding and internal erosion mechanism of failure

Les réservoirs clastiques sont constitués pour la plupart d'un agencement géométrique de plusieurs structures. Le but de ces travaux est de fournir une représentation volumétrique de ces différents objets afin de pouvoir reproduire la grande hétérogénéité des dépôts sédimentaires. Nous proposons une paramétrisation tridimensionnelle compacte des objets sédimentaires qui permet de représenter des géométries variées et qui fournit un espace curvilinéaire pour la modélisation des hétérogénéités internes des structures sédimentaires. Les modèles de corps géologiques sont définis par une représentation par frontières, chacune des frontières étant construite par une surface paramétrique déformable. La formulation mathématique utilisée pour construire les espaces paramétriques sont les B-Splines rationnelles non uniformes dites NURBS (pour Non Uniform Rational B-Splines). Chaque forme élémentaire est contrôlée par des règles de déformations afin de maintenir la géométrie et la cohérence des objets durant l'édition. Les modèles de structures sédimentaires sont appliqués à la simulation de chenaux et des structures qui y sont liées telles que les barres d'accrétion latérale. Ils servent alors de support pour réaliser des simulations de propriétés pétrophysiques qui suivent l'espace paramétrique particulier de chaque objet. Les modèles d'objets sédimentaires sont également appliqués dans le cadre de l'interprétation semi-automatique d'images géologiques en adaptant des méthodes classiques d'extraction de formes. Cette approche est appliquée sur des images satellites de chenaux alluviaux. Des résultats préliminaires sur des données de sismiques 3D sont également présentés / Most clastic deposits consist in geometric arrangements of several structures. The purpose of this work is to provide a volumetric representation of these objects in order to reproduce the high heterogeneity of the sedimentary deposits. These models are constructed to be flexible and compact in order to model the sedimentary formations at different scales. The sedimentary models are defined by a boundary representation ; each boundary is defined by a deformable parametric surface. The proposed parametrization is the Non Uniform Rational B-Spline (NURBS). Each elementary shape is controlled by deformation rules and has connection constraints with associated objects, in order to maintain the geometry and the consistency through editing. Sedimentary structure models are applied to channel simulation and to the construction of the related structures such as point bars. These structures are then used as framework for petrophysical property simulations. The models are also applied to semi-automatically interpret geological images by adapting classical shape extraction methods. This approach is applied on satellite pictures showing alluvial channels and some preliminary results on 3D seismic time slices are also presented