O meandramento da Corrente do Brasil (CB) ao sul da Bifurcação de Santos é investigada por meio de imagens satelitárias, dados quase-sinóticos, análise de funções ortogonais empíricas (EOF) de correntômetros de fundeios e um modelo analítico semi-teórico. A análise das imagens satelitárias revelam que em média 1,2 meandros ciclônicos de grande amplitude são formados anualmente nas vizinhanças do Cabo de Santa Marta (∼28,5°S). Os meandros parecem ser geostroficamente instáveis e a taxa de crescimento típica estimada é de 0,05 m s-1 . Eles ainda se propagam para sul com velocidade de fase de 0,07 m s-1 . A seção de velocidade, como a inferida por perfis de L-ADCP obtidos durante cruzeiros hidrográficos, revelam que os meandros do Cabo de Santa Marta possuem estrutura de velocidade distinta daquelas observadas em Cabo Frio (23°S) e Cabo de São Tomé (22°S). Os meandros alcançam profundidades maiores que 1400 m e recirculam Água Tropical, Água Central do Atlântico Sul, Água Intermediária Antártica e Água Circumpolar Superior. Ocasionalmente, a estrutura do vórtice se funde com a camada subjacente da Corrente de Contorno Oeste Profunda. O padrão geostrófico horizontal dos meandros foram mapeados usando dados de temperatura e salinidade de cruzeiros históricos e foi obtido que a estrutura ciclônica do meandro possui número de Rossby (∼0,07) e número de Burger (∼0,06) pequenos. Portanto, vorticidade de estiramento parece ter papel importante na dinâmica de meandramento e, consequentemente, instabilidade baroclínica é o fenômeno primariamente responsável pelo crescimento do ciclone. O número de Burger pequeno também sugere que a dinâmica do meandro é influênciada pela topografia. A análise de EOFs bidimensionais conduzida no transecto WOCE 28°S de fundeios históricos dos anos 90 mostram que o primeiro modo seccional explica cerca de 54% da variância das séries e está relacionado ao meandramento da CB. A amplitude do meandro ciclônico é aproximadamente 200 km uma vez que cruza o transecto e a onda de vorticidade baroclínica associada tem tipicamente 26 dias. Finalmente, um modelo de Dinâmica de Contornos idealizado de 2 camadas é construído para isolar o mecanismo de instabilidade baroclínica e para investigar as razões do crescimento e velocidade de fase para sul. A estrutura do fluxo básico do modelo é construído baseado no ajuste por mínimos quadrados das funções teóricas à média das observações nas espessuras das camadas. A simulação mostrou que o meandro evolui e se desenvolve devido ao fechamento de fase da camada inferior mais lenta relativo à camada superior mais rápida. Além disso, a propagação de fase para sul ocorre como uma consequência direta da componente barotrópica robusta, adquirida pela CB devido o ramo sul da Bifurcação de Santos. / The Brazil Current (BC) meandering south of the so-called Antarctic Intemediate Water\'s Santos Bifurcation is investigated by means of satellite imagery, quasi-synoptic data, empirical orthogonal function (EOF) analysis of currentmeter moorings and a semi-theoretical dynamical model. The analysis of the infrared imagery revealed that on average 1.2 large amplitude cyclonic meanders are formed annualy in the vicinities of Cape Santa Marta (∼28.5°S). The meanders seem to be geophysically unstable and the estimated typical growth rate is of 0.05 days-1 . They also propagate southward with phase speed of 0.07 m s-1 . The sectional velocity distributions, as inferred from L-ADCP profiles obtained during hydrographic cruises, revealed that the Cape Santa Marta meanders have a very distinct vertical structure from those observed off Cape Frio (23°S) and Cape São Tomé (22°S). The meanders reach much depths of 1400 m and recirculated Tropical Water, South Atlantic Central Water, Antarctic Intemediate Water and Upper Circumpolar Waters. Occasionally, the eddy structure melds with the underlying Deep Western Boundary Current. Geostrophic horizontal patterns of the meanders were mapped using T-S information from historical cruises and it is obtained that the meander is a low-Rossby number (∼0.07) and low-Burger(∼0.06) number cyclone feature. Therefore, stretching vorticity seems to play a major role on the meandering dynamics and, consequently, baroclinic instability is the phenomenon primairily responsible for the cyclone growth. The low-Burger number also suggests that the meander dynamics is influenced by the topography. The two-dimensional EOF analysis conducted on the historical 28°S WOCE mooring transect from the 90s shows that the first sectional mode explains about 54% of the series variance and is related to the BC meandering. The amplitude of the cyclonic meander is roughly 200 km as it crosses the transect and the associated baroclinic vorticity wave period is typically 26 days. Finally, an idealized 2-layer Contour Dynamics model is constructed to isolate the baroclinic instability mechanism and to investigate the reasons for the growth and the southward phase speeds. The model\'s basic flow structure is built based on least-square fits of the observations averaged within the two layer\'s vertical extensions. The simulation showed that the meander evolve and grow due to the phase-locking of the slower lower layer relative to the faster upper layer. Also, the southward phase speed occurs as a direct consequence of the robust barotropic component acquired by the BC due to the southern branch of the Santos Bifurcation of the Antarctic Intemediate Water.

This thesis describes a combined model and laboratory investigation of the generation and mutual interactions of fluid waves whose characteristic scales differ by an order of magnitude or more. The principal aims are to study how waves on one scale can generate waves on another, much shorter scale, and to examine the subsequent nonlinear feedback of the short waves on the long waves. The underlying motive is to better understand such interactions in rotating, stratified, planetary fluids such as atmospheres and oceans. The first part of the thesis describes a laboratory investigation using a rotating, two-layer annulus, forced by imposing a shear across the interface between the layers. A method is developed for making measurements of the two-dimensional interface height field which are very highly-resolved both in space and time. The system's linear normal modes fall into two distinct classes: 'slow' waves which are relatively long in wavelength and intrinsic period, and 'fast' waves which are much shorter and more quickly-evolving. Experiments are performed to categorize the flow at a wide range of points in the system's parameter space. At very small background rotation rates, the interface is completely devoid of waves of both types. At higher rates, fast modes only are generated, and are shown to be consistent with the Kelvin-Helmholtz instability mechanism based on a critical Richardson number. At rotation rates which are higher still, baroclinic instability gives rise to the onset of slow modes, with subsequent localized generation of fast modes superimposed in the troughs of the slow waves. In order to examine the generation mechanism of these coexisting fast modes, and to assess the extent of their impact upon the evolution of the slow modes, a quasi-geostrophic numerical model of the laboratory annulus is developed in the second part of the thesis. Fast modes are filtered out of the model by construction, as the phase space trajectory is confined to the slow manifold, but the slow wave dynamics is accurately captured. Model velocity fields are used to diagnose a number of fast wave radiation indicators. In contrast to the case of isolated fast waves, the Richardson number is a poor indicator of the generation of the coexisting fast waves that are observed in the laboratory, and so it is inferred that these are not Kelvin-Helmholtz waves. The best indicator is one associated with the spontaneous emission of inertia-gravity waves, a generalization of geostrophic adjustment radiation. A comparison is carried out between the equilibrated wavenumbers, phase speeds and amplitudes of slow waves in the laboratory (which coexist with fast modes), and slow waves in the model (which exist alone). There are significant differences between these wave properties, but it is shown that these discrepancies can be attributed to uncertainties in fluid properties, and to model approximations apart from the neglect of fast modes. The impact of the fast modes on the slow modes is therefore sufficiently small to evade illumination by this method of inquiry. As a stronger test of the interaction, a stochastic parameterization of the inertia-gravity waves is included in the model. Consistent with the laboratory/model intercomparison, the parameterized fast waves generally have only a small impact upon the slow waves. However, sufficiently close to a transition curve between two different slow modes in the system's parameter space, it is shown that the fast modes can exert a dominant influence. In particular, the fast modes can force spontaneous transitions from one slow mode to another, due to the phenomenon of stochastic resonance. This finding should be of interest to the meteorological and climate modelling communities, because of its potential to affect model reliability.

The thermally-driven rotating annulus is a laboratory experiment used to study the dynamics of planetary atmospheres under controlled and reproducible conditions. The predictability of this experiment is studied by applying the same principles used to predict the atmosphere. A forecasting system for the annulus is built using the analysis correction method for data assimilation and the breeding method for ensemble generation. The results show that a range of flow regimes with varying complexity can be accurately assimilated, predicted, and studied in this experiment. This framework is also intended to demonstrate a proof-of-concept: that the annulus could be used as a testbed for meteorological techniques under laboratory conditions. First, a regime diagram is created using numerical simulations in order to select points in parameter space to forecast, and a new chaotic flow regime is discovered within it. The two components of the framework are then used as standalone algorithms to measure predictability in the perfect model scenario and to demonstrate data assimilation. With a perfect model, regular flow regimes are found to be predictable until the end of the forecasts, and chaotic regimes are predictable over hundreds of seconds. There is a difference in the way predictability is lost between low-order chaotic regimes and high-order chaos. Analysis correction is shown to be accurate in both regular and chaotic regimes, with residual velocity errors about 3-8 times the observational error. Specific assimilation scenarios studied include information propagation from data-rich to data-poor areas, assimilation of vortex shedding observations, and assimilation over regime and rotation rate transitions. The full framework is used to predict regular and chaotic flow, verifying the forecasts against laboratory data. The steady wave forecasts perform well, and are predictable until the end of the available data. The amplitude and structural vacillation forecasts lose quality and skill by a combination of wave drift and wavenumber transition. Amplitude vacillation is predictable up to several hundred seconds ahead, and structural vacillation is predictable for a few hundred seconds.