In November, 2006, a storm generated a minimum of 34 cm of precipitation in six days, triggering debris flows in many of the drainages on all sides of Mount Hood, Oregon. Of the eleven drainages surveyed, seven experienced debris flows; these include the White River, Salmon River, Clark Creek, Newton Creek, Eliot Creek, Ladd Creek and Sandy River basins. Flows in the White River, Eliot Creek, and Newton Creek, caused major damage to bridges and roadways. Initiation elevations averaged around 1,860 meters. Initiation zone material was predominantly sand (45-82%) with gravel (15-49%) and had few fines (3-5%). Four debris flows were triggered by landslides caused by undercutting of the river banks. Three developed through coalescence of multiple small debris flows within major channels and were termed "headless debris flows". Physical and morphological characterization of source areas was used to assess factors controlling debris flow initiation. Although findings indicate that all major drainages on Mount Hood are capable of producing debris flows, drainages with direct connection to a glacier, low percentages of vegetation, and moderate gradients in the upper basin were the most susceptible. Among basins not having debris flows, neither the Zigzag River nor Polallie Creek have a direct connection to a glacier, And the Muddy Fork and the Coe both have high percentages of vegetated slopes. The material in the upper basin of the Muddy Fork is predominately rock making initiation there weathering-limited. Additionally, the Muddy Fork and the Zigzag have two of the steepest gradients on the mountain. This pattern suggests that material there is regularly transported downstream through normal fluvial processes rather than building up to be catastrophically removed through debris flow processes.

Debris flows caused by heavy rains occurred in November of 2006 on several Cascade volcanoes. Mt. Adams experienced debris flows in seven of eighteen drainages including Adams Creek, Big Muddy Creek, Lewis Creek, Little Muddy Creek, Muddy Fork, Rusk Creek and Salt Creek. Six debris flows occurred on the northeast side of the mountain. A landslide initiated one debris flow, three were initiated by heavy water flow and in channel landslides, and three were initiated by a coalescence of eroded channels (headless debris flows). Four pre-2006 debris flows were found in the Cascade Creek, Crofton Creek, Hellroaring Creek and Morrison Creek drainages. Every 2006 debris flow initiated in Quaternary glacial drift. Attributes of the drainages were investigated to determine differences between drainages with debris flows and those without. The upper basins of drainages with debris flows averaged 37% glacial coverage, 29% bedrock and 35% unconsolidated material. The upper basins of drainages without debris flows without averaged 12% glacial coverage, 63% bedrock, and 25% unconsolidated material. All of the drainages with debris flows were directly connected to a glacier, opposed to only 36% of the drainages without debris flows. Drainages with debris flows averaged 18% slopes above 33°, 10% vegetation, a gradient of 0.38, a Melton's Ruggedness Number of 0.62, an average annual rainfall of 2.16 m, and -52% glacier lost between 1904-2006. The upper basins of drainages without debris flows averaged 11% slopes above 33°, 18% vegetation, a gradient of 0.31, a MRN of 0.58, an average annual rainfall of 2.38 m, and -41% glacier lost between 1904-2006. A multiple logistic regression was performed to determine factors with highest influence on predicting the probability of a debris flow. Influencing factors were percent glacial coverage and average annual rainfall. They predicted the 2006 debris flows with an 89% accuracy rate. This model was used to produce a debris flow hazard map. Due to the number of Cascade volcanoes that experienced debris flows as a result of the November 2006 storm, data of this type could be combined from multiple mountains to construct a general Cascade Mountain debris flow hazard model.

No description available.

Non-equilibrium systems have no set method of analysis, and a wide array of dynamics can be present in such systems. In this work we present three very different non-equilibrium models, inspired by biological systems and phenomena, that we analyze through computational means to showcase both the range of dynamics encompassed by these systems, as well as various techniques used to analyze them. The first system we model is a surface plasmon resonance (SPR) cell, a device used to determine the binding rates between various species of chemicals. We simulate the SPR cell and compare these computational results with a mean-field approximation, and find that such a simplification fails for a wide range of reaction rates that have been observed between different species of chemicals. Specifically, the mean-field approximation places limits on the possible resolution of the measured rates, and such an analysis fails to capture very fast dynamics between chemicals. The second system we analyzed is an avalanching neural network that models cascading neural activity seen in monkeys, rats, and humans. We used a model devised by Lombardi, Herrmann, de Arcangelis et al. to simulate this system and characterized its behavior as the fraction of inhibitory neurons was changed. At low fractions of inhibitory neurons we observed epileptic-like behavior in the system, as well as extended tails in the avalanche strength and duration distributions, which dominate the system in this regime. We also observed how the connectivity of these networks evolved under the effects of different inhibitory fractions, and found the high fractions of inhibitory neurons cause networks to evolve more sparsely, while networks with low fractions maintain their initial connectivity. We demonstrated two strategies to control the extreme avalanches present at low inhibitory fractions through either the random or targeted disabling of neurons. The final system we present is a sparsely encoding convolutional neural network, a computational system inspired by the human visual cortex that has been engineered to reconstruct images inputted into the network using a series of "patterns" learned from previous images as basis elements. The network attempts to do so "sparsely," so that the fewest number of neurons are used. Such systems are often used for denoising tasks, where noisy or fragmented images are reconstructed. We observed a minimum in this denoising error as the fraction of active neurons was varied, and observed the depth and location of this minimum to obey finite-size scaling laws that suggest the system is undergoing a second-order phase transition. We can use these finite-size scaling relations to further optimize this system by tuning it to the critical point for any given system size. / Doctor of Philosophy / Non-equilibrium systems have no set method of analysis, and a wide array of dynamics can be present in such systems. In this work we present three very different non-equilibrium models, inspired by biological systems and phenomena, that we analyze through computational means to showcase both the range of dynamics encompassed by these systems, as well as various techniques used to analyze them. The first system we model is a surface plasmon resonance (SPR) cell, a device used to determine the binding rates between various species of chemicals. We simulate the SPR cell and compare these computational results with a mean-field approximation, and find that such a simplification fails for a wide range of reaction rates that have been observed between different species of chemicals. Specifically, the mean-field approximation places limits on the possible resolution of the measured rates, and such an analysis fails to capture very fast dynamics between chemicals. The second system we analyzed is an avalanching neural network that models cascading neural activity seen in monkeys, rats, and humans. We used a model devised by Lombardi, Herrmann, de Arcangelis et al. to simulate this system and characterized its behavior as the fraction of inhibitory neurons was changed. At low fractions of inhibitory neurons we observed epileptic-like behavior in the system, as well as extended tails in the avalanche strength and duration distributions, which dominate the system in this regime. We also observed how the connectivity of these networks evolved under the effects of different inhibitory fractions, and found the high fractions of inhibitory neurons cause networks to evolve more sparsely, while networks with low fractions maintain their initial connectivity. We demonstrated two strategies to control the extreme avalanches present at low inhibitory fractions through either the random or targeted disabling of neurons. The final system we present is a sparsely encoding convolutional neural network, a computational system inspired by the human visual cortex that has been engineered to reconstruct images inputted into the network using a series of “patterns” learned from previous images as basis elements. The network attempts to do so “sparsely,” so that the fewest number of neurons are used. Such systems are often used for denoising tasks, where noisy or fragmented images are reconstructed. We observed a minimum in this denoising error as the fraction of active neurons was varied, and observed the depth and location of this minimum to obey finite-size scaling laws that suggest the system is undergoing a second-order phase transition. We can use these finite-size scaling relations to further optimize this system by tuning it to the critical point for any given system size.

A capacidade de um sistema sensorial detectar estímulos eficientemente é tradicionalmente dimensionada pela faixa dinâmica, que é simplesmente uma medida da extensão do intervalo de intensidades de estímulo para as quais a rede é suficientemente sensível. Muitas vezes, sistemas biológicos exibem largas faixas dinâmicas, que abrangem diversas ordens de magnitude. A compreensão desse fenômeno não é trivial, haja vista que todos os neurônios apresentam janelas de sensibilidade muito estreitas. Tentativas de explicação baseadas em argumentos de recrutamento sequencial dos neurônios sensoriais, com efeitos essencialmente aditivos, simplesmente não são realísticas, haja vista que seria preciso que os limiares de ativação das unidades também apresentassem um escalonamento por várias ordens de magnitude, para cobrir a faixa dinâmica empiricamente observada em nível macroscópico. Notavelmente, o modelo Kinouchi-Copelli (KC), que carrega o nome de seus idealizadores, mostrou que aquele comportamento pode ser um efeito coletivo (não aditivo) do conjunto de neurônios sensoriais. O modelo KC é uma rede de unidades excitáveis com dinâmicas estocásticas e acoplados segundo uma topologia de grafo aleatório. Kinouchi e Copelli mostraram que a taxa espontânea de disparo dos neurônios (ou atividade média) sinaliza uma transição de fase fora do equilíbrio do tipo ordem-desordem, e que exatamente no ponto crítico desta transição (em termos de um parâmetro ligado às características estruturais da rede) a sensibilidade a estímulos externos é máxima, ou seja, a faixa dinâmica exibe uma otimização crítica. Neste trabalho, investigamos como o ponto crítico depende da topologia, utilizando os modelos mais comuns das chamadas redes complexas. Além disso, estudamos computacionalmente os padrões de atividade (avalanches neuronais) exibidos pelo modelo, com especial atenção às mudanças qualitativas de comportamento devido às mudanças de topologia. Comentaremos também a relação desses resultados com experimentos recentes de monitoramento de dinâmicas neurais. / The capacity of a sensory system in efficiently detecting stimuli is usually given by the dynamic range, a simple measure of the range of stimulus intensity over which the network is sensible enough. Many times biological systems exhibit large dynamic ranges, covering many orders of magnitude. There is no easy explanation for that, since individual neurons present very short dynamic ranges isolatedly. Arguments based on sequential recruitment are doomed to failure since the corresponding arrangement of the limiar thresholds of the units over many orders of magnitude is unrealistic. Notably the so-called Kinouchi-Copelli (KC) model strongly suggested that large dynamic range should be a collective effect of the sensory neurons. The KC model is a network of stochastic excitable elements coupled as a random graph. KC showed the spontaneous activity of the network signals an order-disorder nonequilibrium phase transition and that the dynamic range exhibits an optimum precisely at the critical point (in terms of a control parameter related to structural properties of the network). In this work, we investigate how the critical point depends on the topology, considering the alternatives among the standard complex networks. We also study the burts of activity (neuronal avalanches) exhibited by the model, focusing on the qualitative changes due to alternative topologies. Finally we comment on possible connections among our results and recent observations of neural dynamics.

Neste trabalho estudamos o modelo Olami-Feder-Christensen (OFC). Fortes correlações espaciais e temporais dificultam a obtenção de resultados analíticos para este modelo. Assim, nossas investigações foram realizadas através de simulações computacionais. A fim de identificar o regime estacionário de forma eficiente e econômica desenvolvemos algumas estatrégias. Também percebemos que a escolha adequada da configuração inicial pode antecipar ou retardar o início do regime estacionário. Por fim, a criticalidade do modelo foi estudada através de uma abordagem totalmente nova. Em vez de tentarmos identificar o comportamento crítico do sistema por meio da distribuição de avalanches, definimos uma grandeza , que em um processo ramificado simples seria a taxa de ramificação do sistema. Analisando o comportamento dessa variável em um espaço de fases verificamos que o modelo OFC e sua versão aleatória (que de antemão sabemos que só apresenta criticalidade no regime conservativo) tem um comportamento bastante similar. Obtivemos, ao contrário do que se acreditava, fortes evidências de que o modelo OFC apenas exibe criticalidade no regime conservativo. / We have investigated the Olami-Feder-Christensen model. The model presents strong temporal and spatial correlations what makes it very difficult to perform analytical calculations. So our treatment was numerical. We developed strategies to identify the regime with high level of accuracy. We noticed that depending on the initial configurations, the statistical stationary state can be reached faster. Finally we have investigated the criticality of the model through new strategy. Instead of looking for powers laws, we defined a quantity , very similar to the branching ratio in a simple branching process. We were able to show the behavior of the Olami-Feder-Christensen and the random version of this model are similar. We got strong numerical evidences that, in opposition to previous results, the Olami-Fedel-Christensen model is critical only in the conservative regime.

Neste trabalho estudamos o modelo Olami-Feder-Christensen (OFC). Fortes correlações espaciais e temporais dificultam a obtenção de resultados analíticos para este modelo. Assim, nossas investigações foram realizadas através de simulações computacionais. A fim de identificar o regime estacionário de forma eficiente e econômica desenvolvemos algumas estatrégias. Também percebemos que a escolha adequada da configuração inicial pode antecipar ou retardar o início do regime estacionário. Por fim, a criticalidade do modelo foi estudada através de uma abordagem totalmente nova. Em vez de tentarmos identificar o comportamento crítico do sistema por meio da distribuição de avalanches, definimos uma grandeza , que em um processo ramificado simples seria a taxa de ramificação do sistema. Analisando o comportamento dessa variável em um espaço de fases verificamos que o modelo OFC e sua versão aleatória (que de antemão sabemos que só apresenta criticalidade no regime conservativo) tem um comportamento bastante similar. Obtivemos, ao contrário do que se acreditava, fortes evidências de que o modelo OFC apenas exibe criticalidade no regime conservativo. / We have investigated the Olami-Feder-Christensen model. The model presents strong temporal and spatial correlations what makes it very difficult to perform analytical calculations. So our treatment was numerical. We developed strategies to identify the regime with high level of accuracy. We noticed that depending on the initial configurations, the statistical stationary state can be reached faster. Finally we have investigated the criticality of the model through new strategy. Instead of looking for powers laws, we defined a quantity , very similar to the branching ratio in a simple branching process. We were able to show the behavior of the Olami-Feder-Christensen and the random version of this model are similar. We got strong numerical evidences that, in opposition to previous results, the Olami-Fedel-Christensen model is critical only in the conservative regime.

A capacidade de um sistema sensorial detectar estímulos eficientemente é tradicionalmente dimensionada pela faixa dinâmica, que é simplesmente uma medida da extensão do intervalo de intensidades de estímulo para as quais a rede é suficientemente sensível. Muitas vezes, sistemas biológicos exibem largas faixas dinâmicas, que abrangem diversas ordens de magnitude. A compreensão desse fenômeno não é trivial, haja vista que todos os neurônios apresentam janelas de sensibilidade muito estreitas. Tentativas de explicação baseadas em argumentos de recrutamento sequencial dos neurônios sensoriais, com efeitos essencialmente aditivos, simplesmente não são realísticas, haja vista que seria preciso que os limiares de ativação das unidades também apresentassem um escalonamento por várias ordens de magnitude, para cobrir a faixa dinâmica empiricamente observada em nível macroscópico. Notavelmente, o modelo Kinouchi-Copelli (KC), que carrega o nome de seus idealizadores, mostrou que aquele comportamento pode ser um efeito coletivo (não aditivo) do conjunto de neurônios sensoriais. O modelo KC é uma rede de unidades excitáveis com dinâmicas estocásticas e acoplados segundo uma topologia de grafo aleatório. Kinouchi e Copelli mostraram que a taxa espontânea de disparo dos neurônios (ou atividade média) sinaliza uma transição de fase fora do equilíbrio do tipo ordem-desordem, e que exatamente no ponto crítico desta transição (em termos de um parâmetro ligado às características estruturais da rede) a sensibilidade a estímulos externos é máxima, ou seja, a faixa dinâmica exibe uma otimização crítica. Neste trabalho, investigamos como o ponto crítico depende da topologia, utilizando os modelos mais comuns das chamadas redes complexas. Além disso, estudamos computacionalmente os padrões de atividade (avalanches neuronais) exibidos pelo modelo, com especial atenção às mudanças qualitativas de comportamento devido às mudanças de topologia. Comentaremos também a relação desses resultados com experimentos recentes de monitoramento de dinâmicas neurais. / The capacity of a sensory system in efficiently detecting stimuli is usually given by the dynamic range, a simple measure of the range of stimulus intensity over which the network is sensible enough. Many times biological systems exhibit large dynamic ranges, covering many orders of magnitude. There is no easy explanation for that, since individual neurons present very short dynamic ranges isolatedly. Arguments based on sequential recruitment are doomed to failure since the corresponding arrangement of the limiar thresholds of the units over many orders of magnitude is unrealistic. Notably the so-called Kinouchi-Copelli (KC) model strongly suggested that large dynamic range should be a collective effect of the sensory neurons. The KC model is a network of stochastic excitable elements coupled as a random graph. KC showed the spontaneous activity of the network signals an order-disorder nonequilibrium phase transition and that the dynamic range exhibits an optimum precisely at the critical point (in terms of a control parameter related to structural properties of the network). In this work, we investigate how the critical point depends on the topology, considering the alternatives among the standard complex networks. We also study the burts of activity (neuronal avalanches) exhibited by the model, focusing on the qualitative changes due to alternative topologies. Finally we comment on possible connections among our results and recent observations of neural dynamics.

Ce travail de thèse a pour objectif la reproduction et l'analyse du comportement invariant d'échelle, tel que l'on peut le trouver dans la nature et en particulier au niveau d'une faille tectonique. Pour cela, une expérience reproduisant la structure locale d'une faille par le cisaillement d'un milieu granulaire 2D a été montée et exploitée. L'utilisation de conditions aux limites périodiques dans cette expérience permet l'acquisition d'une statistique très riche, principal défaut des expériences présentes dans la littérature. Le suivi du système est effectué à la fois par des moyens optiques, mécaniques et acoustiques, ces derniers fournissant les informations les plus nombreuses et les plus précises. La dynamique a montré une invariance d'échelle compatible avec les lois statistiques existantes pour les tremblements de terre réels. Les corrélations entre les avalanches et entre les différents moyens de mesure sont analysées. Des expériences complémentaires ont été réalisées pour mieux comprendre les mécanismes à l'œuvre lors du déclenchement d'une avalanche et pendant son déroulement. La principale concerne la propagation d'une onde acoustique au sein d'un milieu granulaire, et a montré qu'une simple mesure de la vitesse de l'onde permet d'obtenir des informations sur la structure interne de l'empilement. L'utilisation d'un matériau photoélastique a permis, suite à une calibration adaptée, de sonder les forces locales au niveau des contacts entre grains et de voir l'évolution des réseaux de force dans le milieu au cours du cisaillement / The aim of this thesis is to reproduce and analyze the scale invariant behavior, as one can find in nature and in particular for a tectonic fault. Thus we set an experiment which reproduces the local structure of a fault thanks to a sheared 2D granular medium. The use of periodic boundaries in the experiment allows us to gain a very rich statistics, which was lacking in previous experiments presented in the literature. The system is monitored by three different methods: optical, mechanical and acoustics, which gives the most numerous and most precise informations. We observed scale invariant dynamics, consistent with statistical laws derived for real earthquakes. We also analyzed the correlations between avalanches and between the three monitoring methods. Additional experiments have been performed to better understand the mecanisms that take place at the triggering and during an avalanche. The main one focuses on sound wave propagation inside a granular pile, and we have shown that a mere velocity measure can give information about the internal structure of the pile. The use of a photoelastic material allows us (with appropriate calibration) to probe local forces at the edges and contacts of the grains and to see the evolution of force chains during the shear

Ce travail de thèse vise à reconstituer les conditions paléo-environnementales ayant eu lieu dans la vallée de l’Oisans à partir des archives naturelles représentées par les sédiments lacustres. Par une approche multi-marqueurs, nous visons à reconstituer tout à la fois : i) les fluctuations glaciaires en Oisans; ii) les modifications des activités humaines en montagne et leurs influence sur les flux de matière lors d’événements extrêmes ; iii) comprendre la relation entre changements climatiques, changements d’usage et évolution des aléas en haute montagne. Cette étude a permis de mettre en évidence les changements globaux, regroupant la variabilité climatique et les pratiques humaines, ayant influencés l’érosion autour des lacs de La Muzelle et du lac du Lauvitel durant les derniers millénaires.Les résultats principaux ont montrés que l’érosion du lac de la Muzelle a été largement dominé par l’activité glaciaire notamment avec la présence à certaines époques de fines particules détritiques liées à l’abrasion sous glaciaire. Ces dernières ont révélées une relation aux pluies torrentielles encore jamais observées auparavant, leur présence en période d’extension glaciaire étant synonyme d’une augmentation du nombre de dépôts lacustres de crues. Les activités humaines, en majeure partie représentées par l’utilisation agro-pastorale de l’espace avoisinant les lacs, ne sont clairement identifiables que lors des 300 dernières années au lac de la Muzelle. Le lac de Lauvitel est situé à une altitude moindre, dont la majorité du bassin versant est maintenant une réserve intégrale. Les études palynologiques ont mis en évidence certaines périodes d’activités humaines plus marquées. Les événements extrêmes enregistrés dans les sédiments dulac regroupent d’une part les crues, dont l’enregistrement permet une comparaison régionale de l’occurrence de ces événements et de mettre en évidence les changements de circulations atmosphériques à l’échelle des Alpes. D’autre part, les avalanches de neige lourde, dont peu d’enregistrements sont disponibles dans la bibliographie. Nous les avons identifiés grâce à l’utilisation du CT scan l’élaboration d’une nouvelle méthodologie basée sur la différence de densité relative des sédiments. Le comptage et la quantification des apports de matériel détritique grossier aux seins d’une matrice de sédiment lacustre fin, a permis d’identifier les apports liés à cet aléa au cours du temps. Nous avons ensuite reconstitué les événements d’avalanche de neige lourde déposés dans le lac de Lauvitel sur les derniers 3500 ans, dont l’occurrence intervient préférentiellement lors des périodes de retraits glaciaires. / This doctoral thesis aims to a paleo-environmental reconstitution of the Oisans valley based on the natural archive of lake sediments. From a multi-proxy approach, we aim to reconstruct: i) Glacial fluctuation reconstruction in Oisans valley; ii) human activities evolution in mountain area and their influences on sediment fluxes especially during extreme events; iii)understand the relationship between climate change, use of mountain lands and natural hazard.Through this study, we identified processes of global change, comprising natural climate variability and human practices, which affected erosion patterns around Lakes Muzelle and Lauvitel during the last millennia. Main results have shown that erosion in the lake Muzelle watershed was dominated in the past by glacial activity, in particular with the presence of fine detrital particles related to subglacial abrasion. They were identified to have a relationship never observed before; during glacial extension their presence is triggering a higher number of flood deposits. At this location, human activities were identified through agro-pastoral activities and more precisely by the coprophilous fungi spore counting, revealing cattle presence since 300 years. Lake Lauvitel is located at lower altitude, which most of the watershed is now situated in an Integral Reserve. Palynological investigations lead to identification of periods of higher human activities in the valley. Extreme events recorded in Lake Lauvitel sediment are on the one hand torrential floods, which allowed a comparison with other reconstructions highlighting changes in the climatic settings in the Alps. On the other hand, wet avalanches deposits were identified with the use of a CT scan and the development of a new methodology based on relative density differences in the sediment. The counting and the quantification of coarse detrital matter within the lacustrine fine sediment matrix allowed income identification of this hazard though time. We then reconstructed wet avalanches events deposited in Lake Lauvitel during the last 3500 years, which occurrence is preferentially during glacial retreats.