京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第22049号 / 工博第4630号 / 新制||工||1722(附属図書館) / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 中川 一, 教授 藤田 正治, 准教授 川池 健司 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Storm surge, the pileup of seawater occurring as a result of high surface stresses and strong currents generated by extreme storm events such as hurricanes, is known to cause greater loss of life than these storms' associated winds. For example, inland flooding from the storm surge along the Gulf Coast during Hurricane Katrina killed hundreds of people. Previous storms produced even larger death tolls. Simultaneously, dune, barrier island, and channel erosion taking place during a hurricane leads to the removal of major flow controls, which significantly affects inland inundation. Also, excessive sea bed scouring around pilings can compromise the structural integrity of bridges, levees, piers, and buildings. Modeling these processes requires tightly coupling a bed morphology equation to the shallow water equations (SWE). Discontinuous Galerkin finite element methods (DGFEMs) are a natural choice for modeling this coupled system, given the need to solve these problems on large, complicated, unstructured computational meshes, as well as the desire to implement hp-adaptivity for capturing the dynamic features of the solution. Comprehensive modeling of these processes in the coastal zone presents several challenges and open questions. Most existing hydrodynamic models use a fixed-bed approach; the bottom is not allowed to evolve in response to the fluid motion. With respect to movable-bed models, there is no single, generally accepted mathematical model in use. Numerical challenges include coupling models of processes that exhibit disparate time scales during fair weather, but possibly similar time scales during intense storms. The main goals of this dissertation include implementing a robust, efficient, tightly-coupled morphological model using the local discontinuous Galerkin (LDG) method within the existing Advanced Circulation (ADCIRC) modeling framework, performing systematic code and model verification (using test cases with known solutions, proven convergence rates, or well-documented physical behavior), analyzing the stability and accuracy of the implemented numerical scheme by way of a priori error estimates, and ultimately laying some of the necessary groundwork needed to simultaneously model storm surges and bed morphodynamics during extreme storm events. / text

Cette thèse, à dominante expérimentale, tente de décrire, de comprendre et de quantifier le transport sédimentaire pour un système torrentiel caractérisé par une forte pente et des conditions hydrauliques faibles. Une plateforme expérimentale a été mise en place et instrumentée. Elle permet de contrôler quatre paramètres du système : la pente du canal, le débit hydraulique, le débit solide et la granulométrie. La durée des expériences est suffisamment longue pour couvrir l'ensemble des fluctuations gouvernant le transport. L'étude de l'évolution d'un même mélange granulaire soumis à un forçage hydraulique fort, (Shield = 1.37 Shield critique) ou à un forçage hydraulique faible (Shield = 0.94 Shield critique) a permis de mettre en évidence, au sein des structures sédimentaires observées, deux classes de comportement différentes. Pour l'expérience à Faible Transport, le lit est globalement pavé au cours de la durée de l'expérience qui est de 110 heures. Des structures de type step-pools sont présentes et très stables, et contrôlent la dynamique de transport du système à court terme et à long terme. A court terme, le transport solide n'est pas uniforme mais s'effectue à travers des processus transitoires d'érosion et de dépôt « de proche en proche », associés respectivement à la destruction et à la formation de step-pools dans le système. A long terme, les step-pools conduisent à un pavage uniforme sur l'ensemble du canal. Des destructions périodiques de ce pavage de surface sont également observées et sont associées à une capacité de transport pour le système pouvant atteindre jusqu'à 15 fois le débit solide moyen. Pour l'expérience à Fort Transport, d'une durée de 92 heures, la morphologie du lit fluctue entre deux états extrêmes. Un lit plat dont la surface est formée de particules fines, à forte capacité de transport et un lit pavé à forte pente et à faible capacité de transport. Les step-pools ne sont, ici, que des états éphémères associés à la phase de transition du système qui sont rapidement détruits par les nappes de charriage. Ces dernières sont le mode principal de transport pour le système et expliquent les évolutions cycliques mesurées à court et à long terme. Enfin, nous avons étudié les conséquences des phénomènes décrits ci-dessus en conditions hors équilibre à forte pente. Nous avons pour cela réalisé deux expériences : une première expérience où nous étudions le cas d'un système en phase d'aggradation et une seconde où nous prenons le cas d'un système contrôlé par deux seuils de correction torrentielle. La première a nécessité une longue période afin que le système atteigne une condition d'équilibre dynamique : plus que 210 heures. L'équilibre local du système est atteint progressivement de l'amont vers l'aval, les parties les plus proches de l'alimentation atteignant en premiers l'état d'équilibre. De plus, pendant l'aggradation les fluctuations du système autour de la moyenne sont inférieures à celles mesurées pour un système à l'équilibre. La seconde expérience a montré que l'aménagement d'un système torrentiel avec des seuils peut avoir un impact non négligeable sur la quantité maximum de matériau transporté au cours d'un seul événement. Néanmoins, l'intensité maximale instantanée du transport et le comportement global du système à long terme ne sont en rien affectés par la présence de seuils. / This experimental thesis aims to describe, understand and quantify the sediment dynamics of a torrential system, characterized by steep slope and low hydraulic conditions. An experimental platform has been developed and instrumented for monitoring the channel slope, the water discharge, the feed solid discharge and the bed grain size distribution. The duration of the experiments were sufficiently long to cover all the fluctuations governing the bedload transport phenomenon. Two flow conditions were considered, in order to observe the bed dynamic under different forcing. A strong hydraulic forcing capable to move the bed pavement (Shield parameter = 1.37 Critical Shield parameter) was compared to a low hydraulic forcing supposed to not disturb the bed pavement (Shield Parameter = 0.94 Critical Shield Parameter). We measured for each case the sediment dynamic and associated bedforms. For the “Low Transport” experiment the bed was generally paved over all the duration of the experiment (110 hours). Bedforms similar to step-pools developed and were present for all the duration of the run. These macroforms seems to control the transport dynamic of the system in the short and long term. In the short term, bedload transport was not uniform. Instead, transient processes of erosion / deposition associated with the formation / destruction of step-pools were observed. In the long term, step-pools lead to the formation of a uniform paved bed on the whole channel length. Periodic destruction of bed pavement was also observed and associated with a transport capacity that could reach 15 times the value of average measured bedload transport. For the “High Transport” experiment (92 hours) the channel morphology fluctuates between two extremes. A flat bed characterised by highly mobile fine sediments and high transport capacity, and a steep paved bed corresponding to a low transport capacity. For this experiment step-pools were just a transitory state associated with the transition between the two extremes and they are quickly destroyed by the bedload sheets propagation. Bedload sheets were the main transport mode for the system. Their formation and propagation can explain the cyclical evolution of the system in the short and long term. We finally studied the consequences of the observed phenomena for non-equilibrium systems on steep slopes. We conducted two experiments: one in order to reproduce an aggradational system and the second to reproduce the dynamic of a system controlled by two check dams. A very long duration was necessary for reaching the dynamic equilibrium condition with the aggradational system (more than 210 hours). The local equilibrium was achieved gradually from upstream to downstream, with the flume sections closest to the feeding device attaining first the dynamic equilibrium. Experiments with check dams showed that this kind of slope correction can have a significant impact on the maximum amount of material transported during a single event. However, the maximum instantaneous transport and the long term sediment behaviour of the system are not affected by the presence of the check dams.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第19292号 / 工博第4089号 / 新制||工||1630(附属図書館) / 32294 / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 角 哲也, 准教授 竹門 康弘, 准教授 KANTOUSH Sameh Ahmed / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

Made available in DSpace on 2017-07-10T17:30:55Z (GMT). No. of bitstreams: 1 Andrey Luis Binda.pdf: 3846554 bytes, checksum: bff1c6b39e4b9f07d3cbffa410b26168 (MD5) Previous issue date: 2009-07-24 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Trees, twigs and leaves fallen inside of the channel, form a significant component of the fluvial environment and natural occurrence in fluvial systems that cut forested areas. All these vegetal materials that are found inside of the channel are called wood debris. The wood debris play a key function in the ecology and geomorphology of channels, being able to affect the processes of the channels in an ample reach: since the hydraulic roughness, retention and sized grain of bed sediments, until the creation of steps and pools, being able to the times to control the standard of the channel and distance and/variation of fluvial features that are basic for the formation of aquatic habitat. In this context, the present research had as objective, to verify the influence of accumulations of wood debris on the morphology and sedimentology of stream bed in the river Guabiroba, Guarapuava, Center-South of the Paraná. The used techniques had included: installation of permanent transversal profiles (to the upstream and downstream of the accumulations) and sections of batimetric measurement, survey of the longitudinal profile and grain sized characterization of bed sediments. The measurement was carried throughout 15 months (December of 2007 the March of 2009) in four sections located in the main course of the river. The rainfall during the measurement allowed to particularizing the occurrence of only one period with rains above average (3 - 4ª Campaign). The results indicate the occurrence of sedimentation in the second and fifth measurement and processes of erosion in the third and fourth measurement. These processes erosive-depositional can in part be attributed to the pluviometric regime, with the periods of high entailed concentration of rains to the erosion of the stream bed and periods of related scarce and not concentrated rains with the deposition of sediments. The analysis of the data allows to establish some consideration: 1) predominance of processes of erosion in the sections of batimetric measurement; 2) bigger rate deposition in downstream cross-sectional profiles of the accumulations; 3) predominance of coarser sediments (gravel medium - sand very coarse) to upstream and finer sediments (gravel fine - sand medium) to downstream; 4) bigger diversity of depths in the immediacy of the accumulations; 5) establishment of riffles by the forced deposition of sediments and pools for the scour of the stream bed. / Árvores, galhos e folhas caídos dentro do canal formam um significativo componente do ambiente fluvial e de ocorrência natural em sistemas fluviais que cortam áreas florestadas. Toda essa carga de material de origem vegetal que se encontra dentro do canal é chamada de detritos lenhosos. Os detritos lenhosos desempenham função chave na ecologia e geomorfologia de canais, podendo afetar os processos do canal em um amplo alcance: desde a rugosidade hidráulica, retenção e grânulometria dos sedimentos de leito, até a criação de degraus e depressões, podendo, às vezes, controlar o padrão do canal e a distância/variação de feições fluviais que são fundamentais para a formação de habitat aquáticos. Num contexto desses, a presente pesquisa teve como objetivo verificar a influência de acumulações de detritos lenhosos sobre a morfologia e a sedimentologia de leito no Rio Guabiroba, município de Guarapuava, mesorregião Centro-Sul do Paraná. As técnicas utilizadas incluíram: instalação de perfis transversais permanentes (a montante e a jusante das acumulações) e seções de monitoramento batimétrico, levantamento do perfil longitudinal e caracterização grânulométrica de sedimentos do fundo. O monitoramento foi realizado ao longo de 15 meses (dezembro de 2007 a março de 2009) em quatro pontos localizados em dois trechos no curso principal do rio. A pluviosidade durante o monitoramento permitiu individualizar a ocorrência de apenas um período com chuvas acima da média (3ª a 4ª campanha). Os resultados indicam a ocorrência de sedimentação na segunda e quinta campanhas e processos de erosão na terceira e quarta campanhas. Esses processos erosivo-deposicionais podem, em parte, ser atribuídos ao regime pluviométrico, com os períodos de alta concentração de chuvas vinculados à erosão do leito e períodos de chuvas escassas e não concentradas relacionados com a deposição de sedimentos. A análise dos dados permite estabelecer algumas considerações: 1) predominância de processos de erosão nas seções de monitoramento batimétrico (Blocos 3D); 2) maior taxa de deposição nos perfis transversais situados a jusante das acumulações; 3) predomínio de sedimentos mais grossos (seixo médio - areia muito grossa) a montante da acumulação do que a jusante (seixo fino - areia média); 4) maior diversidade de profundidades nas imediações das acumulações; 5) estabelecimento de soleiras pela deposição forçada de sedimentos e depressões pela escavação do leito.

Made available in DSpace on 2017-05-12T14:42:26Z (GMT). No. of bitstreams: 1 Andrey Luis Binda.pdf: 3846554 bytes, checksum: bff1c6b39e4b9f07d3cbffa410b26168 (MD5) Previous issue date: 2009-07-24 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Trees, twigs and leaves fallen inside of the channel, form a significant component of the fluvial environment and natural occurrence in fluvial systems that cut forested areas. All these vegetal materials that are found inside of the channel are called wood debris. The wood debris play a key function in the ecology and geomorphology of channels, being able to affect the processes of the channels in an ample reach: since the hydraulic roughness, retention and sized grain of bed sediments, until the creation of steps and pools, being able to the times to control the standard of the channel and distance and/variation of fluvial features that are basic for the formation of aquatic habitat. In this context, the present research had as objective, to verify the influence of accumulations of wood debris on the morphology and sedimentology of stream bed in the river Guabiroba, Guarapuava, Center-South of the Paraná. The used techniques had included: installation of permanent transversal profiles (to the upstream and downstream of the accumulations) and sections of batimetric measurement, survey of the longitudinal profile and grain sized characterization of bed sediments. The measurement was carried throughout 15 months (December of 2007 the March of 2009) in four sections located in the main course of the river. The rainfall during the measurement allowed to particularizing the occurrence of only one period with rains above average (3 - 4ª Campaign). The results indicate the occurrence of sedimentation in the second and fifth measurement and processes of erosion in the third and fourth measurement. These processes erosive-depositional can in part be attributed to the pluviometric regime, with the periods of high entailed concentration of rains to the erosion of the stream bed and periods of related scarce and not concentrated rains with the deposition of sediments. The analysis of the data allows to establish some consideration: 1) predominance of processes of erosion in the sections of batimetric measurement; 2) bigger rate deposition in downstream cross-sectional profiles of the accumulations; 3) predominance of coarser sediments (gravel medium - sand very coarse) to upstream and finer sediments (gravel fine - sand medium) to downstream; 4) bigger diversity of depths in the immediacy of the accumulations; 5) establishment of riffles by the forced deposition of sediments and pools for the scour of the stream bed. / Árvores, galhos e folhas caídos dentro do canal formam um significativo componente do ambiente fluvial e de ocorrência natural em sistemas fluviais que cortam áreas florestadas. Toda essa carga de material de origem vegetal que se encontra dentro do canal é chamada de detritos lenhosos. Os detritos lenhosos desempenham função chave na ecologia e geomorfologia de canais, podendo afetar os processos do canal em um amplo alcance: desde a rugosidade hidráulica, retenção e grânulometria dos sedimentos de leito, até a criação de degraus e depressões, podendo, às vezes, controlar o padrão do canal e a distância/variação de feições fluviais que são fundamentais para a formação de habitat aquáticos. Num contexto desses, a presente pesquisa teve como objetivo verificar a influência de acumulações de detritos lenhosos sobre a morfologia e a sedimentologia de leito no Rio Guabiroba, município de Guarapuava, mesorregião Centro-Sul do Paraná. As técnicas utilizadas incluíram: instalação de perfis transversais permanentes (a montante e a jusante das acumulações) e seções de monitoramento batimétrico, levantamento do perfil longitudinal e caracterização grânulométrica de sedimentos do fundo. O monitoramento foi realizado ao longo de 15 meses (dezembro de 2007 a março de 2009) em quatro pontos localizados em dois trechos no curso principal do rio. A pluviosidade durante o monitoramento permitiu individualizar a ocorrência de apenas um período com chuvas acima da média (3ª a 4ª campanha). Os resultados indicam a ocorrência de sedimentação na segunda e quinta campanhas e processos de erosão na terceira e quarta campanhas. Esses processos erosivo-deposicionais podem, em parte, ser atribuídos ao regime pluviométrico, com os períodos de alta concentração de chuvas vinculados à erosão do leito e períodos de chuvas escassas e não concentradas relacionados com a deposição de sedimentos. A análise dos dados permite estabelecer algumas considerações: 1) predominância de processos de erosão nas seções de monitoramento batimétrico (Blocos 3D); 2) maior taxa de deposição nos perfis transversais situados a jusante das acumulações; 3) predomínio de sedimentos mais grossos (seixo médio - areia muito grossa) a montante da acumulação do que a jusante (seixo fino - areia média); 4) maior diversidade de profundidades nas imediações das acumulações; 5) estabelecimento de soleiras pela deposição forçada de sedimentos e depressões pela escavação do leito.

L'objectif ultime en géomorphologie fluviale est d'expliquer les formes des cours d'eau et leur évolution temporelle et spatiale. La multiplication des études nous a mené à la réalisation que les systèmes géomorphologiques sont complexes. Les formes observées sont plus que la somme des processus individuels qui les régissent en raison d’interactions et de rétroactions non-linéaires à de multiples échelles spatiales et temporelles. Dans ce contexte, le but général de la thèse est de proposer et de tester de nouvelles avenues de recherche afin de mieux appréhender la complexité des dynamiques fluviales en utilisant des approches méthodologiques et analytiques mettant l’accent sur les interactions entre l’écoulement, le transport de sédiments en charge fond et la morphologie du lit en rivière graveleuse. Cette orientation découle du constat que les paradigmes actuels en géomorphologie fluviale n’arrivent pas à expliquer adéquatement la variabilité naturelle du transport en charge de fond ainsi que des formes du lit qui en résultent. Cinq pistes de réflexion sont développées sous forme d’articles basés sur des études de cas : 1. L'intégration des échelles de variation de l'écoulement permet d’insérer la notion de structures turbulentes dans des pulsations de plus grande échelle et d'améliorer la compréhension de la variabilité du transport de sédiments. 2. La quantification des taux de changement de l’écoulement (accélération /décélération) au cours d’une crue permet d’expliquer la variabilité des flux de transport en charge fond autant que la magnitude de l’écoulement. 3. L’utilisation de techniques de mesures complémentaires révèle une nouvelle dynamique du lit des rivières graveleuses, la dilatation et la contraction du lit suite à une crue. 4. La remise en cause du fait généralement accepté que le transport en charge de fond est corrélé positivement à l'intensité des modifications morphologiques en raison d’un problème associé aux échelles différentes des processus en cause. 5. L’approche systémique des dynamiques fluviales par l’utilisation d’analyses multivariées permet d’appréhender la complexité des dynamiques de rétroactions linéaires et non-linéaires dans l’évolution d’un chenal et d’illustrer l’importance de l’historique récent des changements géomorphologiques en réponse aux crues. Cette thèse se veut une avancée conceptuelle issue d'une profonde réflexion sur les approches classiques que l'on utilise en géomorphologie fluviale depuis plusieurs décennies. Elle est basée sur un jeu de données unique récolté lors du suivi intensif de 21 évènements de crue dans un petit cours d’eau à lit de graviers, le ruisseau Béard (Québec). Le protocole expérimental axé sur la simultanéité des mesures de l’écoulement, de la morphologie du lit et du transport de sédiments en charge de fond a permis de centrer la recherche directement sur les interactions entre les processus plutôt que sur les processus individuels, une approche rarement utilisée en géomorphologie fluviale. Chacun des chapitres illustre un nouveau concept ou une nouvelle approche permettant de résoudre certaines des impasses rencontrées actuellement en géomorphologie fluviale. Ces travaux ont des implications importantes pour la compréhension de la dynamique des lits de rivières et des habitats fluviaux et servent de point de départ pour de nouveaux développements. / The overarching objective in fluvial geomorphology is to explain river morphology and its temporal and spatial evolution. The multiplication of studies led to the realization that geomorphic systems are complex and difficult to understand. The observed river morphology is more than the sum of the individual processes which govern them because of the superimposition of process interactions and non-linear feedback loops at multiple scales. In this context, the goal of this thesis is to explore new avenues of research in order to comprehend more accurately the complexity of gravel-bed river dynamics by using methodological and analytical approaches focusing on the interactions between the flow, transport of sediments as bedload and the bed morphology. This orientation of the thesis is in line with the fact that current paradigms in fluvial geomorphology have not succeeded in explaining adequately the natural variability of the sediment transport and its consequences on the river bedforms. Five avenues are developed in the form of articles based on case studies: 1. The study of the scales of variation of the flow allows us to integrate the turbulent flow structures into flow pulsations at larger scales and to potentially improve our understanding comprehension of bedload sediment transport variability. 2. The quantification of the rates of change of the flow magnitude (acceleration /deceleration) during a flood explains as much the variability of bedload transport fluxes as the flow intensity itself. 3. The use of complementary measurement techniques of bedload processes can reveal new river bed dynamics in gravel-bed rivers: the dilation and contraction of the bed resulting from a flood. 4. The investigation of the generally accepted fact that bedload sediment transport is correlated with the intensity of the morphological changes reveals that the two processes do not relate directly due to their different characteristic scales. 5. The systemic approach of river dynamics based on the use of multivariate analytical techniques is better suited for the study of complex linear and non-linear interactions and feedbacks in the evolution of a river channel. This analysis has shown the importance of the recent history of the morphological changes in response to a flood. The conceptual orientation of this thesis results from a deep reflection on the traditional approaches previously used for several decades in fluvial geomorphology. It is based on a single very extensive data set collected during 21 flood events in a small gravel-bed river, Béard creek (Quebec). The experimental protocol aimed at the simultaneity of processes measurements in order to focus the analysis on the interactions between the flow, the channel morphology and the bedload sediment transport rather than on the individual processes, an approach seldom used in fluvial geomorphology. Each chapter reports on a new concept or a new approach making it possible to resolve some of the issues met in fluvial geomorphology. This work has important implications for the understanding of river bed dynamics and fluvial habitats and is a starting point for new developments.

L'objectif ultime en géomorphologie fluviale est d'expliquer les formes des cours d'eau et leur évolution temporelle et spatiale. La multiplication des études nous a mené à la réalisation que les systèmes géomorphologiques sont complexes. Les formes observées sont plus que la somme des processus individuels qui les régissent en raison d’interactions et de rétroactions non-linéaires à de multiples échelles spatiales et temporelles. Dans ce contexte, le but général de la thèse est de proposer et de tester de nouvelles avenues de recherche afin de mieux appréhender la complexité des dynamiques fluviales en utilisant des approches méthodologiques et analytiques mettant l’accent sur les interactions entre l’écoulement, le transport de sédiments en charge fond et la morphologie du lit en rivière graveleuse. Cette orientation découle du constat que les paradigmes actuels en géomorphologie fluviale n’arrivent pas à expliquer adéquatement la variabilité naturelle du transport en charge de fond ainsi que des formes du lit qui en résultent. Cinq pistes de réflexion sont développées sous forme d’articles basés sur des études de cas : 1. L'intégration des échelles de variation de l'écoulement permet d’insérer la notion de structures turbulentes dans des pulsations de plus grande échelle et d'améliorer la compréhension de la variabilité du transport de sédiments. 2. La quantification des taux de changement de l’écoulement (accélération /décélération) au cours d’une crue permet d’expliquer la variabilité des flux de transport en charge fond autant que la magnitude de l’écoulement. 3. L’utilisation de techniques de mesures complémentaires révèle une nouvelle dynamique du lit des rivières graveleuses, la dilatation et la contraction du lit suite à une crue. 4. La remise en cause du fait généralement accepté que le transport en charge de fond est corrélé positivement à l'intensité des modifications morphologiques en raison d’un problème associé aux échelles différentes des processus en cause. 5. L’approche systémique des dynamiques fluviales par l’utilisation d’analyses multivariées permet d’appréhender la complexité des dynamiques de rétroactions linéaires et non-linéaires dans l’évolution d’un chenal et d’illustrer l’importance de l’historique récent des changements géomorphologiques en réponse aux crues. Cette thèse se veut une avancée conceptuelle issue d'une profonde réflexion sur les approches classiques que l'on utilise en géomorphologie fluviale depuis plusieurs décennies. Elle est basée sur un jeu de données unique récolté lors du suivi intensif de 21 évènements de crue dans un petit cours d’eau à lit de graviers, le ruisseau Béard (Québec). Le protocole expérimental axé sur la simultanéité des mesures de l’écoulement, de la morphologie du lit et du transport de sédiments en charge de fond a permis de centrer la recherche directement sur les interactions entre les processus plutôt que sur les processus individuels, une approche rarement utilisée en géomorphologie fluviale. Chacun des chapitres illustre un nouveau concept ou une nouvelle approche permettant de résoudre certaines des impasses rencontrées actuellement en géomorphologie fluviale. Ces travaux ont des implications importantes pour la compréhension de la dynamique des lits de rivières et des habitats fluviaux et servent de point de départ pour de nouveaux développements. / The overarching objective in fluvial geomorphology is to explain river morphology and its temporal and spatial evolution. The multiplication of studies led to the realization that geomorphic systems are complex and difficult to understand. The observed river morphology is more than the sum of the individual processes which govern them because of the superimposition of process interactions and non-linear feedback loops at multiple scales. In this context, the goal of this thesis is to explore new avenues of research in order to comprehend more accurately the complexity of gravel-bed river dynamics by using methodological and analytical approaches focusing on the interactions between the flow, transport of sediments as bedload and the bed morphology. This orientation of the thesis is in line with the fact that current paradigms in fluvial geomorphology have not succeeded in explaining adequately the natural variability of the sediment transport and its consequences on the river bedforms. Five avenues are developed in the form of articles based on case studies: 1. The study of the scales of variation of the flow allows us to integrate the turbulent flow structures into flow pulsations at larger scales and to potentially improve our understanding comprehension of bedload sediment transport variability. 2. The quantification of the rates of change of the flow magnitude (acceleration /deceleration) during a flood explains as much the variability of bedload transport fluxes as the flow intensity itself. 3. The use of complementary measurement techniques of bedload processes can reveal new river bed dynamics in gravel-bed rivers: the dilation and contraction of the bed resulting from a flood. 4. The investigation of the generally accepted fact that bedload sediment transport is correlated with the intensity of the morphological changes reveals that the two processes do not relate directly due to their different characteristic scales. 5. The systemic approach of river dynamics based on the use of multivariate analytical techniques is better suited for the study of complex linear and non-linear interactions and feedbacks in the evolution of a river channel. This analysis has shown the importance of the recent history of the morphological changes in response to a flood. The conceptual orientation of this thesis results from a deep reflection on the traditional approaches previously used for several decades in fluvial geomorphology. It is based on a single very extensive data set collected during 21 flood events in a small gravel-bed river, Béard creek (Quebec). The experimental protocol aimed at the simultaneity of processes measurements in order to focus the analysis on the interactions between the flow, the channel morphology and the bedload sediment transport rather than on the individual processes, an approach seldom used in fluvial geomorphology. Each chapter reports on a new concept or a new approach making it possible to resolve some of the issues met in fluvial geomorphology. This work has important implications for the understanding of river bed dynamics and fluvial habitats and is a starting point for new developments.