In the present thesis, we have explored multiple aspects concerning the stochasticity of magnetic domain wall motion during magnetization reversal, all of which originated from our initial study of magnetic field sensing using planar Hall effect. Magnetic field sensors occupy a very important and indispensable position in modern technology. They can be found everywhere, from cellphones to automobiles, electric motors to computer hard disks. At present there are several emerging areas of technology, including biotechnology, which require magnetic field sensors which are at the same time simple to use, highly sensitive, robust under environmental conditions and sufficiently low cost to be deployed on a large scale. Magnetic field sensing using planar Hall effect is one such feasible technology, which we have explored in the course of the thesis. The work was subsequently expanded to cover some fundamental aspects of the stochasticity of domain wall motion, studied with planar Hall effect, which forms the main body of work in the present study. In Chapter 1, we give an introduction to the phenomenology of planar Hall effect, which is the most important measurement technique used for all the subsequent studies. Some early calculations, which had first led to the understanding of anisotropic magnetoresistance and planar Hall effect as being caused by spin-orbit interaction are discussed. In Chapter 2, we discuss briefly the experimental techniques used in the present study for sample growth and fabrication, structural and magnetic characterization, and measurement. We discuss pulsed laser ablation, which is the main technique used for our sample growth. Particular emphasis is given to the instrumentation that was carried out in-house for MOKE and low field magnetotransport (AMR and PHE) measurement. This includes an attempt at domain wall imaging through MOKE microscopy. Some of the standard equipments used for this work, such as the SQUID magnetometer and the acsusceptometer are also discussed in detail. In Chapter 3 we discuss our work on planar Hall sensors that led to the fabrication of a device with a very simple architecture, having transfer characteristics of 650V/A.T in a range of _2Oe. The sensing material was permalloy (Ni81Fe19), and the value had been obtained without using an exchange biased pinning layer. Field trials showed that the devices were capable of geomagnetic field sensing, as well as vehicle detection by sensing the anomaly in Earth's magnetic field caused by their motion. Its estimated detection threshold of 2.5nT made it well suited for several other applications needing high sensitivity in a small area, the most prominent of them being the detection of macromolecules of bio-medical significance. Chapter 4: The work on Barkhausen noise was prompted by reproducibility problems faced during the sensor construction, both between devices as well as within the same device. Study of the stochastic properties led us to the conclusion that the devices could be grouped into two classes: one where the magnetization reversal occurred in a single step, and the other where it took a 0staircase0 like path with multiple steps. This led us to simulations of Barkhausen noise using nucleation models like the RFIM whence it became apparent that the two different groups of samples could be mapped into two regimes of the RFIM distinguished by their magnetization reversal mode. In the RFIM, the nature of the hysteresis loop depends on the degree of disorder, with a crossover happening from single-step switching to multi-step switching at a critical disorder level. Appropriate changes also appear in the Barkhausen noise statistics due to this disorder-induced crossover. By studying the Barkhausen noise statistics for our permalloy samples and comparing them with simulations of the RFIM, we found nearly exact correspondence between the two experimental groups with the two classes resulting from crossing the critical disorder. What remained was to quantify the 0disorder0 level of our samples, which was done through XRD, residual resistivity and a study of electron-electron interaction effects in the resistivity. All these studies led to the conclusion that the samples reversing in multiple steps were more 0defective0 than the other group, at par with the model predictions. This completed the picture with respect to the modeling of the noise. In experiments, it was found that a high rate of film deposition yielded less 0defective0 samples, which severed as an important input for the sensor construction. These results can be viewed from a somewhat broader perspective if we consider the present scenario in the experimental study of Barkhausen noise, or crackling noise in general. Two classes of models exist for such phenomena: front propagation models and nucleation models. Both appear to be very successful when it comes to experiments with bulk materials, while the comparison with experiments on thin films is rather disappointing. It is still not clear whether the models are at fault or the experiments themselves. Through our study, we could demonstrate that there can be considerable variation in the Barkhausen noise character of the same material deposited in the same way, and what was important was the degree of order at the microscopic level. This may be a relevant factor when experimental papers report non-universality of Barkhausen noise in thin films, which can now be interpreted as either insufficient defects or a sample area too small for the study. Chapter 5: Defects in a sample are not the only cause for stochastic behavior during magnetization. In most cases, random thermal 0events0 are also an important factor determining the path to magnetization reversal, which was also true for our permalloy samples. We studied the distribution of the external fields at which magnetization reversal took place in our samples, and tried to explain it in terms of the popular Neel-Brown model of thermal excitation over the anisotropy barrier. The analysis showed that even though the coercivity behaved 0correctly0 in terms of the model predictions, the behavior of the distribution width was anomalous. Such anomalies were common in the literature on switching field distributions, but there seemed to be no unified explanation, with different authors coming up with their own 0exotic0 explanations. We decided to investigate the simplest situations that could result in such a behavior, and through some model-based calculations, came to the conclusion that one of the causes of the anomalies could be the different magnitudes of barrier heights/anisotropy fields experienced by the magnetic domain wall when the reversal occurs along different paths. Though an exact match for the behavior of the distribution width could not be obtained, the extended Neel-Brown model was able to produce qualitative agreement. Chapter 6 contains a study of some interesting 0geometrical0 effects on Barkhausen noise of iron thin films. By rotating the applied magnetic field out-of plane, we could observe the same single-step to multi-step crossover in hysteresis loop nature that was brought about by varying disorder in Chapter 4. We could explain this through simulations of a random anisotropy Ising model, which, apart from exhibiting the usual disorder induced crossover, showed a transition from sub-critical to critical hysteresis loops when the external field direction was rotated away form the average anisotropy direction. Once again, simulation and experiment showed very good agreement in terms of the qualitative behavior. In the second part of this chapter, a study of exchange biased Fe-FeMn system was carried out, where it was observed that the reversal mode has been changed from domain wall motion to coherent rotation. Barkhausen noise was also suppressed. Though many single-domain models existed for this type of reversal, our system was not found to be strictly compatible with them. The disagreement was with regard to the nature of the hysteresis, which, if present, had to be a single step process for a single domain model. The disagreement was naturally attributed to interaction with the nearby magnetic moments, to verify which, simulations were done with a simplified micromagnetic code, which produced excellent agreement with experiment. In Chapter 7, we have studied the temporal properties of Barkhausen avalanches, to compare the duration distributions with simulation. We had used a permalloy sample that was sub-critical according to avalanche size distributions, and our measurement was based on magneto-optic Kerr effect. We measured duration distributions which showed a similar manifestation of finite-size effects as were shown by the size distributions. The power law exponent was calculated, which was deemed 0reasonable0 upon comparison simulations of the sub-critical RFIM. Appendix A contains a study of high-field magnetoresistance of permalloy, which shows that the dominant contribution to magnetoresistance is the suppression of electron-magnon scattering. An interesting correlation is observed between the magnetization of samples and an exchange stiffness parameter d1, that was extracted from magnetoresistance measurements. Here we also re-visit our earlier observation of permalloy thin films possessing a resistance minimum at low temperature. The origin of this minimum is attributed to electron-electron interaction. Appendix B contains the source codes for most of the important programs used for simulation and data analysis. The programs are written in MATLAB and FORTRAN 95. LabView programs used for data acquisition and analysis are not included due to space requirements to display their graphical source codes. Appendix C discusses the studies on a disordered rare-earth oxide LaMnO3. The re-entrant glassy phase is characterized with ac susceptibility and magnetization measurements to extract information about the nature of interactions between the magnetic 0macrospins0 in the system. Appendix D deals with electron scattering experiments performed with spinpolarized electrons (SPLEED) from clean metal surfaces in UHV. A study of the scattering cross sections as a function of energy and scattering angle provides information about spin-orbit and exchange interactions of the electrons with the surface atoms, and can answer important questions pertaining to the electronic and magnetic structure of surfaces. In the course of this study, planar Hall effect is seen to emerge as a powerful tool to study the magnetic state of a thin film, so that it is interesting to apply it to thin films of other materials such as oxides, where magnetization noise studies are next to nonexistent. What also emerged is that there is still a lot of richness present in the details of supposedly well-understood magnetization phenomena, some of which we have explored in this thesis in the context of stochastic magnetization processes.

Submitted by Biblioteca do Instituto de Física (bif@ndc.uff.br) on 2017-02-15T18:53:58Z No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_Cinthya Valeska Chianca da Silva.pdf: 2780664 bytes, checksum: 6254459484571e92eca2c9372931a17e (MD5) / Made available in DSpace on 2017-02-15T18:53:58Z (GMT). No. of bitstreams: 2 license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Tese_Cinthya Valeska Chianca da Silva.pdf: 2780664 bytes, checksum: 6254459484571e92eca2c9372931a17e (MD5) / Conselho Nacional de Desenvolvimento Científico e Tecnológico / Neste trabalho, estudamos sistemas com características diferentes mas que apresentam comportamento crítico. Na primeira parte, a partir de modelos que estudam avalanches usando blocos e molas e utilizando cadeias de bits simulamos estes sistemas de maneira simples e conseguimos obter resultados que são compatíveis com os já mostrados na literatura. A seguir, descrevemos um método para analisar séries temporais com tendências que serve para retirar tendências de baixas frequências da mesma. Finalmente, descrevemos a realização de uma experiência de fragmentação em gotas de mercúrio que repetimos para tentar obter melhores resultados. / In this work we study critical behaviour in systems with different characteristics. In the first part, beginning with block and spring models for avalanches, we use bit chains for simple simulations of these models and obtain results compatible with those published in the literature. Next, we describe a new method to analyse temporal series which is useful to extract the low frequency tendencies. Finally, we describe the realisation of an experiment using mercury drops, which we repeated to obtain bettes results.

Quand un fluide mouillant visqueux (comme une huile) pénètre un milieu hétérogène tel qu’une fracture, l’interface (entre l’air déplacé et l’huile) développe des corrélations à longue portée menant à une dynamique spatio-temporelle complexe. Dans cette Thèse, nous avons étudié expérimentalement et théoriquement ce processus de transport d’un fluide, appelé imbibition, dans un modèle de fracture ouverte, pertinent dans diverses situations. Notre travail a combiné une étude expérimentale détaillée, avec une analyse précise des données, basées sur des méthodes de physique statistique et non-linéaire. D’abord, la position moyenne de l'interface h(t) est étudiée lors d’expériences de montée capillaire donnant lieu à une nouvelle solution analytique pour h(t). Nous avons ensuite étudié les propriétés d’invariance d’échelle de l’interface et en particulier leur évolution pour des processus d’imbibition forcées, caractérisée par un scénario dit «super-rugueux». Enfin, nous avons étudié et quantifié la dynamique intermittente par avalanches des fronts d’imbibition à partir de l’analyse multi-échelle (spatiales et temporelles) de leurs vitesses. L'ensemble des résultats présentés dans cette Thèse propose une image très générale de la dynamique hors équilibre des fronts d’imbibition se propageant lentement dans des fractures ouvertes. La propagation latérale des fluctuations interfaciales est contrôlée par conservation de la masse locale. L'avancement de l'interface dans la direction de propagation est contrôlé par l’échelle caractéristique du désordre et la vitesse moyenne du front. / The heterogeneous structure of fractured media can lead to complex spatiotemporal fluid invasion dynamics. It thus brings forward challenging fundamental questions in the context of out-of-equilibrium dynamical systems, but also relevant to many processes of interest. The goal of the Thesis is to study the spatio-temporal dynamics of the oil-air interface between displaced air and invading oil, in imbibition through a model open fracture. The research combines exhaustive experimental work with accurate data analysis based on methods of nonlinear statistical physics. The mean postion of the interface h(t) is studied in capillary rise experiments, giving rise to a new analytical solution for h(t). The fluctuations of the interface in forced-flow experiments are analysed in the context of kinetic roughening, characterizing a super-rough scaling scenario. Finally, the burst-like dynamics is studied by analysing the local and global velocities of the front, which are widely distributed and display complex spatio-temporal correlations. We define local and global avalanches whose sizes and durations are also widely distributed, with cutoffs that diverge with the capillary number. Intermittentcy of the global signal is quantified. The ensemble of results presented in this Thesis supports a very general picture of the nonequilibrium dynamics of slowly-driven fronts in open fractures: the lateral propagation of interfacial fluctuations is controlled by local mass conservation, through the lateral correlation length; and the advancement of the interface in the direction of propagation is controlled by the characteristic extent of the disorder d and by the mean front velocity.

This study is an examination of the rate at which organic debris and clastic sediment accumulate in a gully after it is scoured by a debris torrent. Of particular interest is the effect that a change in land use from old-growth to clear-cut conditions may have on these rates. This change should result in a reduction in the delivery of large organic debris (LOD), which is a major factor in sediment storage in gullies. It is hypothesized that this change in land use, and the subsequent reduction in the LOD supply, should result in a significant difference in debris recharge rates between old-growth and clear-cut gullies. Twenty-nine gullies in both land-treatment groups were sampled on the west coast of the Queen Charlotte Islands. Sampling procedures involved the estimation of the volume of LOD and sediment in storage (normalized by the gully surface area) and the determination of the time elapsed since the last debris torrent. These data were then used to estimate recharge rates(3h1)am’year of LOD, sediment, and total debris. Recharge rates of each material were compared between land-treatment groups using the nonparametric Mann-Whitney test. This test revealed that LOD has been delivered to old-growth gullies at a significantly higher rate relative to clear-cut gullies. There was no significant difference in sediment and total debris recharge rates between gullies in the two groups, but this outcome was partially a result of the small samples and the different debris recharge times in each data set. Graphical representations of the data permitted the identification of possible temporal trends in sediment and debris accumulation, which may be strengthened with larger data sets. Debris recharge rates have several applications. The estimate of sediment volume stored in a gully can be used in the construction of local sediment budgets, as one component of a watershed sediment cascade is quantified. The calculation of debris recharge rates will provide insight into the transfer rate of sediment from hillslopes to low order channels and to the storage capacity of the channels. Finally, debris recharge rates can be used to improve knowledge of the frequency-magnitude characteristics of debris torrents in an area. / Arts, Faculty of / Geography, Department of / Graduate

Located in the Columbia River Gorge, The Red Bluff Landslide (18.8 km2) is one of four large landslides that make up the Cascade Landslide Complex. In its current form, the Red Bluff Landslide is a post-Missoula Flood feature made up of two components: an active upper lobe (8.6 km2) that is translational, creeping to the south at 25 cm/yr and spreading laterally to the east at 6 cm/yr over a semi-fixed portion (10.2 km2) of the Red Bluff Landslide area that has been "smoothed" by Missoula Floods. The upper active lobe is the landslide debris accumulated since Missoula Flood time (~15,000 yr. BP). Five separate collapse events have been identified and rock failures along the main scarp headwalls continue. Two rock avalanches on the Red Bluff Landslide were mapped. The Old Greenleaf Basin Rock Avalanche is estimated to have occurred 100 to 150 years ago, represents the fifth collapse event on the Red Bluff Landslide, and covers an area of 200,000 m2. It has a volume of 4.2 million m3; its length is 748 m and has a width of 215 m. On January 3, 2008, the Greenleaf Basin Rock Avalanche occurred, flowing over the Old Greenleaf Basin Rock Avalanche, covering an area of 100,000 m2 and deposited a volume of about 375,000 m3. Its length is 730 m with an average depth of 1.22 m. It contributed approximately 0.058% of the total volume and 0.01% of the surface area to the active upper lobe portion of the Red Bluff Landslide. The Greenleaf Basin Rock Avalanche was determined to be insignificant in the movement of the active part of the Red Bluff Landslide during the winter of 2007-2008. The original Cascade Landslide Complex map (Wise, 1961) included the Mosley Lakes Landslide which has now been removed because it lacked the characteristics of a landslide like a scarp. The original complex (35.5 km2) has been renamed the "Greater Cascade Landslide Complex" (43.0 km2), with the addition of the adjacent Stevenson Slide and the elimination of the Mosley Lakes Landslide.

Geophysical research has long been interdisciplinary, with many phenomena on the Earth's surface involving multiple, linked processes that are best understood using a combination of techniques. This is particularly true in the case of grain flows on sand dunes, in which the sedimentary stratification with which geologists are concerned arises from the granular processes investigated by physicists and engineers, and the water permeation that interests hydrologists and soil scientists determines the seismic velocities of concern to exploration geophysicists. In this dissertation, I describe four projects conducted for the degree of Doctor of Philosophy, using a combination of laboratory experimentation, fieldwork, numerical simulation, and mathematical modelling to link avalanching on dunes to its effects on stratification, on the permeation of water, and on seismic surveys. Firstly, I describe experiments on erodible, unbounded, grain piles in a channel, slowly supplied with additional grains, and I demonstrate that the behaviour of the consequent, discrete avalanches alternates between two regimes, typified by their size statistics. Reconciling the `self-organised criticality' that several authors have predicted for such a system with the hysteretic behaviour that others have observed, the system exhibits quasi-periodic, system-spanning avalanches in one regime, while in the other avalanches pass at irregular intervals and have a power-law size distribution. Secondly, I link this power-law size distribution to the strata emplaced by avalanches on bounded grain piles. A low inflow rate of grains into an experimental channel develops a pile, composed of strata in which blue-dyed, coarser grains overlie finer grains. Associating stopped avalanche fronts with the `trapped kinks' described by previous authors, I show that, in sufficiently large grain piles, mean stratum width increases linearly with distance downslope. This implies the possibility of interpreting paleodune height from the strata of aeolian sandstones, and makes predictions for the structure of avalanche-associated strata within active dunes. Thirdly, I discuss investigations of these strata within active, Qatari barchan dunes, using dye-infiltration to image strata in the field and extracting samples across individual strata with sub-centimetre resolution. Downslope increases in mean stratum width are evident, while measurements of particle size distributions demonstrate preferential permeation of water along substrata composed of finer particles, explaining the strata-associated, localised regions of high water content discovered by other work on the same dunes. Finally, I consider the effect of these within-dune variations in water content on seismic surveys for oil and gas. Having used high performance computing to simulate elastic wave propagation in the vicinity of an isolated, barchan sand dune, I demonstrate that such a dune acts as a resonator, absorbing energy from Rayleigh waves and reemitting it over an extensive period of time. I derive and validate a mathematical framework that uses bulk properties of the dune to predict quantitative properties of the emitted waves, and I demonstrate the importance of internal variations in seismic velocity, resulting from variations in water content.

The heavy precipitation event of November 3-8, 2006 dropped over 60 cm of rain onto the bare southern slopes of Mount St. Helens and generated debris flows in eight of the sixteen drainages outside the 1980 debris avalanche zone. Debris flows occurred on the upper catchments of the Muddy River, Shoestring Glacier, Pine Creek, June Lake, Butte Camp Dome, Blue Lake, Sheep Creek, and South Fork Toutle River. Debris flows were clustered on the west and south-east sides of the mountain. Of the eight debris flows, three were initiated by landslides, while five were initiated by headward or channel erosion. Six debris flows were initiated in deposits mapped as Holocene volcaniclastic deposits, while two were in 1980 pyroclastics on andesite flows. The largest (~975,000 m2) and longest (~8,900 m) debris flow was initiated by landslides in the upper South Fork Toutle River Drainage. The average debris flow initiation zone elevation was 1,750 m, with clusters around 1,700 m and 2,000 m elevation. The lower cluster is associated with basins that host modern or historic glaciers, while the upper is possibly associated with recent pyroclastic deposits. Upper drainages with debris flows averaged 41% slopes steeper than 33 degrees, while those without debris flows averaged 34%. The upper basins with debris flows averaged 6% snow and ice cover, 21% consolidated bedrock, and 74% unconsolidated deposits. Basins without debris flows averaged 3% snow and ice cover, 27% bedrock, and 67% unconsolidated deposits. Drainages with debris flows averaged an 89% loss of glacier area between 1998 and 2009, while those without debris flows lost 68%. Further comparing glacier coverage during that period found that only five of ten glaciers still existed in 2009. On average, the glaciers had reduced in area by 67%, decreased in length by 36%, and retreated by an average of 471 m during that period. Basin attributes were measured or calculated in order to construct a predictive debris flow model based on that of Pirot (2010) using multiple logistic regression. The most significant factors were the percentage of slopes steeper than 33 degrees, unconsolidated deposits in the upper basin, and average annual rainfall. These factors predicted the 2006 debris flows with an accuracy of 94% in a debris flow susceptibility map for Mount St. Helens.

La rhéologie et les lois de comportement des écoulements granulaires font l'objet de nombreuses recherches, aussi bien expérimentales que théoriques. Des études d'écoulements continus de billes en tambour tournant ont mis en évidence que le gradient de vitesse dans la zone à écoulement libre est constant. Nous avons testé expérimentalement la validité de cette expression du gradient de vitesse pour des gammes de paramètres de fonctionnement beaucoup plus larges que celles utilisées jusqu'à présent. Pour de faibles vitesses de rotation du tambour, nous avons montré que les angles d'avalanche et de repos peuvent être homothétiques dans la mesure où les particules sont comparables en terme distribution granulométrique et de morphologie. A plus forte vitesse de rotation, des mesures systématiques d'épaisseur de zone d'écoulement et d'angle d'écoulement ont permis de mettre en évidence différentes transitions de régime. Quand la taille des particules d est grande par rapport à la taille du tambour D (rapport D/d faible), on constate que l'épaisseur de la zone d'écoulement sature à environ 1/3 du rayon du tambour. Dans ce cas, une augmentation de la vitesse de rotation du tambour est accommodée par une augmentation du gradient de vitesse. Dans des systèmes à rapport D/d élevé, on constate que c'est la vitesse maximale des particules à la surface libre qui sature avant l'épaisseur en écoulement. Dans ces conditions, une augmentation de la vitesse de rotation du tambour est accommodée par une augmentation de l'épaisseur, ce qui entraîne une réduction du gradient de vitesse. Ainsi les limites du domaine de validité d'un gradient de vitesse constant ont été définies : la relation est valable tant que ni l'épaisseur en écoulement ni la vitesse maximale des particules ne saturent. Enfin, dans une dernière partie, des expériences de ségrégation ont été réalisées dans le but de mettre en rapport les épaisseurs en écoulement et les vitesses de ségrégation de grosses billes de verre placées dans un milieu de petites billes de même densité.

Considérons une boite remplie de grains et inclinons la progressivement. Au-dessus d'un angle critique, un écoulement de surface se déclenche. Cette avalanche, d'amplitude et de durée finies, fait relaxer l'angle du tas de quelques degrés. Ce processus intervient fréquemment dans la nature, notamment sous la forme d'écoulements de débris qui se produisent aussi bien à la surface de la Terre que dans les fonds marins. Cependant, les écoulements denses de granulaires immergés dans un liquide ont été peu étudiés. Ainsi, le travail expérimental rapporté dans ce manuscrit s'attache à déterminer l'influence d'un fluide environnant, gaz ou liquide, sur l'amplitude et la dynamique des avalanches. Nous mettons en évidence trois régimes d'avalanches contrôlés par deux paramètres sans dimension : le rapport r entre la densité des grains et celle du fluide, et le nombre de Stokes St qui compare l'inertie d'un grain aux effets visqueux du fluide. Dans un gaz (grandes valeurs de r et de St), l'effet du fluide est négligeable. Dans les liquides (petites valeurs de r), l'amplitude des avalanches diminue tandis que leur durée augmente lorsque St diminue. Dans une deuxième partie, nous étudions l'effet d'un confinement du tas entre deux parois latérales sur sa stabilité. Maximale quand l'écart entre parois est minimum, la valeur des angles diminue sur une longueur caractéristique B lorsque l'écart entre parois augmente. Cet effet peut s'expliquer par la redirection d'une partie des contraintes internes au tas vers les parois, ce qui y induit des forces de frottement prévenant ou bloquant l'écoulement. Deux lois d'échelles dépendantes de la taille des grains sont mises en évidence pour la longueur B : l'effet des parois est géométrique pour les gros grains alors qu'un régime cohésif est observé pour les petits grains. Enfin, nous rapportons les résultats de premières expériences dans l'air où la vitesse des grains est mesurée et apparaÎt exponentiellement décroissante avec la profondeur.

Un milieu granulaire peut rester au repos même avec une surface libre inclinée, et il ne se met spontanément en mouvement qu'au dessus d'un angle critique. Il apparaît alors un écoulement de surface dont l'amplitude et la dynamique sont régies par la mobilisation et le dépôt de grains à l'interface avec la phase statique. Malgré son importance pratique, la dynamique de cette transition demeure largement inconnue. Le manuscrit présente une étude expérimentale de ce problème dans deux géométries d'écoulements granulaires différentes. Dans la première expérience, l'équilibre d'une couche de sable déposée dynamiquement sur un plan rugueux est rendu métastable: après avoir augmenté l'inclinaison du plan, une perturbation locale déclenche une avalanche. La mesure du seuil de déclenchement montre que la transition entre équilibre statique et écoulement est sous-critique. La forme et la dynamique des avalanches résultent d'un échange permanent de matière avec la couche statique, mobilisant à l'avant et déposant des grains à l'arrière. De manière surprenante, une saturation de l'amplitude est observée, qui peut être attribuée à la présence du fond solide. Nous avons aussi mis en évidence l'existence de deux types d'avalanches, l'un où la couche n'est mobilisée qu'en aval du point de déclenchement et laissant une trace triangulaire, et l'autre où l'écoulement envahit tout le plan par un front de remontée. L'étude montre que des mécanismes de propagation différents interviennent dans ces deux cas. La deuxième expérience consiste à étudier l'écoulement transitoire à la formation d'un talus. Nous avons observé l'influence prépondérante de la préparation sur l'écoulement. Deux effets sont mis en évidence, l'un associé à la densité et l'autre à la «texture» anisotrope acquise pendant le remplissage. Dans la dernière partie, nous discutons de la validité des modèles existants pour décrire les avalanches sur le plan incliné. En guise de conclusion, nous développons un modèle continu décrivant la dynamique des écoulements granulaires de surface et prenant appui sur les résultats expérimentaux présentés.