Les surcotes de tempête sont souvent sous-estimées dans les modèles hydrodynamiques, ainsi que les grandes vagues dans les modèles de vagues. Les causes possibles sont une sous-estimation des vents dans les modèles atmosphériques et/ou une formulation incorrecte de la tension de vent. Les objectifs de cette thèse sont (1) d’estimer les biais par vents forts dans les modèles atmosphériques (2) de développer une nouvelle paramétrisation du coefficient de traı̂née permettant de réduire ce biais (3) d’étudier l’impact des vagues sur la tension de vent. La méthode consiste à étudier la réponse de l’atmosphère et de l’océan à la tension de vent. Dans une première partie, nous utilisons le modèle couplé vagues-atmosphère d’ECMWF. Nous montrons que les vents forts sont sous-estimés, avec un biais de l’ordre de -7 m/s à 30 m/s. Des écarts significatifs existent aussi entre les observations, les bouées et les vents issus de ASCAT-KNMI étant généralement inférieurs à ceux des plateformes et des autres données satellites utilisées dans cette étude (AMSR2, ASCAT-RSS, WindSat, SMOS et JASON-2). La nouvelle paramétrisation développée permet d’obtenir des vents plus forts qu’avec celle d’ECMWF par défaut. Dans une deuxième partie (réponse de l’océan), nous utilisons le modèle global océanique TUGO du LEGOS forcé par le modèle couplé vagues-atmosphère d’ECMWF. Nous montrons qu’une paramétrisation de la tension de vent dépendant des vagues plutôt que du vent est plus appropriée quand l’état de mer est jeune. Elle conduit à des surcotes plus proches des observations (marégraphes et traces altimétriques de JASON-2). L’impact des vagues sur la surcote est significatif, et peut atteindre 20 cm. / Storm surges may be underestimated in hydrodynamic models, as well as large wave heights in wave models. This could come from an underestimation of strong winds in atmospheric models and/or an inappropriate wind stress formulation. The objectives of the present work are (1) to estimate how strong are the biases for high winds in atmospheric models (2) to develop a new drag parameterization that could reduce this bias (3) to investigate the impact of the waves on the wind stress. The method consists of studying the response of the atmosphere and the ocean to the wind stress.In a first part, we use the coupled wave-atmosphere model from ECMWF. We show that strong winds may be underestimated, as much as -7 m/s at 30 m/s.Significant differences also exist between observations, with buoys and ASCAT-KNMI generally showing lower wind speeds than the platforms and other remote-sensing data used in this study(AMSR2, ASCAT-RSS, WindSat, SMOS and JASON-2).The newly empirically adjusted Charnock parameterization leads to higher winds compared to the default ECMWF parameterization. In a second part, we use the global ocean model TUGO fromLEGOS forced with ECMWF coupled wave-atmopshere model. We show that a wave-dependent rather than wind-dependent stress formulation is more appropriate, when the sea state is young and the sea rougher. It yields to simulated surges closer to observations (i.e. tide gauges and JASON-2 altimeter tracks). The wave impact on the surges is significant, and may reach 20 cm.

Resultados da reanálise SODA v2.2.6 (Carton, Giese, 2008) e da Renálise do Século 20 v2 (Compo, et al., 2011) foram analisados para verificar alterações dos modos de variabilidade da TSM (o modo do Gradiente Meridional de Temperatura (GMT) e o Modo Zonal) no Atlântico Tropical (de 1929 a 2008) através de funções empíricas ortogonais (EOF) e funções empíricas ortogonais associadas (jEOF). A evolução do padrão espacial do modo do GMT se inicia com a configuração de dipolo de temperatura, com eixo central em ≈ 5ºN evoluindo para o GMT com variabilidade concentrada no Atlântico Tropical Norte. O Modo Zonal apresenta inicialmente variabilidade associada à região equatorial (entre 5ºS e 5ºN) e à costa sudoeste africana, que evolui para um gradiente meridional de TSM, centrado em ≈ 5ºN. Sua variabilidade concentra-se exclusivamente no Atlântico Tropical Sul. A variabilidade equatorial se degenera ao longo do período, devido ao aumento, gerado pelo vento, da profundidade das isopicnais na termoclina. No equador o acoplamento entre o oceano e a atmosfera ocorre nos períodos de T = 30 meses e T ≈ 34 meses, com o vento antecedendo a temperatura em 1 e 2 meses, respectivamente. O Modo Zonal apresenta acoplamento com o vento durante a segunda metade das análises. O período associado é de T ≈ 34 meses, com o vento antecedendo a temperatura em cerca de 1 mês. O modo do GMT está associado aos ventos no Atlântico Tropical Norte e Atlântico Tropical Sul. Os períodos de acoplamento são de T = 96 e T = 60 meses, com o vento antecedendo a TSM em 3 e ≈ 2 meses respectivamente. / The results of SODA v2.2.6 reanalysis (Carton, Giese, 2008) and 20th Century Reanalysis v2 Project (Compo, et al., 2011) were analyzed in order to verify changes of the SST modes (the Meridional Temperature Gradient mode (GMT) and the Zonal Mode) in the Tropical Atlantic (1929 to 2008) using Empirical Orthogonal Functions (EOF) and joint Empirical Orthogonal Functions (jEOF). The spatial distribution of GMT starts initially as a temperature dipole centred at ≈ 5ºN. It evolves into a meridional gradient with variability concentrated at the Tropical North Atlantic. The zonal mode variability is initially associated with the equatorial region (between 5ºS and 5ºN) and with the northwestern african coast. It evolves into a merdional gradient with central axis located at 5ºN. Its variability is concentrated exclusively in the Tropical South Atlantic. The equatorial variability degenerates throughout the period, due to the inhibition of the isopicnal uplift by the wind. At the equator, the coupling occurs in periods of T = 30 months and T ≈ 34 months, with the wind preceding the TSM by 1 and 2 months, respectively. The zonal mode presents coupling with the wind only during the second half of the analysis. The periods are of T = 34 months, with wind preciding TSM by about 1 month. GMT mode is associated to the winds of both Tropical North Atlantic and Tropical South Atlantic. Coupling periods are of T = 96 and T = 60 months, with the wind preceding TSM in 3 and ≈ 2 months respectively.

Resultados da reanálise SODA v2.2.6 (Carton, Giese, 2008) e da Renálise do Século 20 v2 (Compo, et al., 2011) foram analisados para verificar alterações dos modos de variabilidade da TSM (o modo do Gradiente Meridional de Temperatura (GMT) e o Modo Zonal) no Atlântico Tropical (de 1929 a 2008) através de funções empíricas ortogonais (EOF) e funções empíricas ortogonais associadas (jEOF). A evolução do padrão espacial do modo do GMT se inicia com a configuração de dipolo de temperatura, com eixo central em ≈ 5ºN evoluindo para o GMT com variabilidade concentrada no Atlântico Tropical Norte. O Modo Zonal apresenta inicialmente variabilidade associada à região equatorial (entre 5ºS e 5ºN) e à costa sudoeste africana, que evolui para um gradiente meridional de TSM, centrado em ≈ 5ºN. Sua variabilidade concentra-se exclusivamente no Atlântico Tropical Sul. A variabilidade equatorial se degenera ao longo do período, devido ao aumento, gerado pelo vento, da profundidade das isopicnais na termoclina. No equador o acoplamento entre o oceano e a atmosfera ocorre nos períodos de T = 30 meses e T ≈ 34 meses, com o vento antecedendo a temperatura em 1 e 2 meses, respectivamente. O Modo Zonal apresenta acoplamento com o vento durante a segunda metade das análises. O período associado é de T ≈ 34 meses, com o vento antecedendo a temperatura em cerca de 1 mês. O modo do GMT está associado aos ventos no Atlântico Tropical Norte e Atlântico Tropical Sul. Os períodos de acoplamento são de T = 96 e T = 60 meses, com o vento antecedendo a TSM em 3 e ≈ 2 meses respectivamente. / The results of SODA v2.2.6 reanalysis (Carton, Giese, 2008) and 20th Century Reanalysis v2 Project (Compo, et al., 2011) were analyzed in order to verify changes of the SST modes (the Meridional Temperature Gradient mode (GMT) and the Zonal Mode) in the Tropical Atlantic (1929 to 2008) using Empirical Orthogonal Functions (EOF) and joint Empirical Orthogonal Functions (jEOF). The spatial distribution of GMT starts initially as a temperature dipole centred at ≈ 5ºN. It evolves into a meridional gradient with variability concentrated at the Tropical North Atlantic. The zonal mode variability is initially associated with the equatorial region (between 5ºS and 5ºN) and with the northwestern african coast. It evolves into a merdional gradient with central axis located at 5ºN. Its variability is concentrated exclusively in the Tropical South Atlantic. The equatorial variability degenerates throughout the period, due to the inhibition of the isopicnal uplift by the wind. At the equator, the coupling occurs in periods of T = 30 months and T ≈ 34 months, with the wind preceding the TSM by 1 and 2 months, respectively. The zonal mode presents coupling with the wind only during the second half of the analysis. The periods are of T = 34 months, with wind preciding TSM by about 1 month. GMT mode is associated to the winds of both Tropical North Atlantic and Tropical South Atlantic. Coupling periods are of T = 96 and T = 60 months, with the wind preceding TSM in 3 and ≈ 2 months respectively.

O presente trabalho teve como finalidade estudar os efeitos das variações do vento resultantes de ocorrências do fenômeno El Niño-Southern Oscillation (ENSO) nos padrões da circulação superficial ao longo da costa peruana, através da análise de dados observados e de modelagem numérica. É enfocado o período 1991-2000, quando ocorreram fortes eventos La Niña (1996-97, 1998-2000) e El Niño (1997- 98). Esses eventos tiveram fortes impactos em escala global mas muito pouco se sabe sobre os impactos locais na estrutura da termoclina e no ciclo da dinâmica de Ekman ao longo da costa peruana. Os dados analisados no presente estudo foram dados coletados em duas radiais ao longo das latitudes 5 S e 15 S. O modelo oceânico utilizado foi o Modelo da Universidade de Princeton (Princeton Ocean Model-POM). As simulações numéricas foram forçadas com produtos de vento relativos ao período 1991-2000. Essas simulações reproduziram satisfatoriamente os padrões médios da circulação na região de estudo, confirmando que nas áreas próximas da costa o vento é o principal mecanismo gerador de ressurgência ou subsidência. Os resultados mostraram que durante a ocorrência do forte evento El Niño 1997-1998, nas radiais de 5 S e 15 S, houve uma drástica alteração do sistema de ressurgência costeira em resposta às variações do vento. / The objective of the present work was to study the effects of variability in the wind due to the El Niño-Southern Oscillation events in the surface circulation patterns of the Peruvian coast, by means of data analysis and numerical modeling. It is focused the period 1991-2000, when it was observed strong La Niña (1996-1997, 1998-2000) and El Niño events (1997-98). These events had strong effects worlwide but very little is known on the local impacts on the circulation, thermocline structure and the Ekman dynamics of the Peruvian coast. The data analysed in the present study were collected on transects along 5 S and 15 S. The model used was an implementation of the Princeton Ocean Model (POM). The numerical simulations were forced with wind products relative to the period of interest (1991-2000). The simulations reproduced satisfactorily the mean circulation patterns in the study area, confirming that the in the nearshore region the wind is the main driving mechanism for coastal upwelling/downwelling. The results show that during the strong 1997-1998 El Niño, a drastic alteration of the coastal upwelling system ocurred in response to the changes in the wind.

O presente trabalho teve como finalidade estudar os efeitos das variações do vento resultantes de ocorrências do fenômeno El Niño-Southern Oscillation (ENSO) nos padrões da circulação superficial ao longo da costa peruana, através da análise de dados observados e de modelagem numérica. É enfocado o período 1991-2000, quando ocorreram fortes eventos La Niña (1996-97, 1998-2000) e El Niño (1997- 98). Esses eventos tiveram fortes impactos em escala global mas muito pouco se sabe sobre os impactos locais na estrutura da termoclina e no ciclo da dinâmica de Ekman ao longo da costa peruana. Os dados analisados no presente estudo foram dados coletados em duas radiais ao longo das latitudes 5 S e 15 S. O modelo oceânico utilizado foi o Modelo da Universidade de Princeton (Princeton Ocean Model-POM). As simulações numéricas foram forçadas com produtos de vento relativos ao período 1991-2000. Essas simulações reproduziram satisfatoriamente os padrões médios da circulação na região de estudo, confirmando que nas áreas próximas da costa o vento é o principal mecanismo gerador de ressurgência ou subsidência. Os resultados mostraram que durante a ocorrência do forte evento El Niño 1997-1998, nas radiais de 5 S e 15 S, houve uma drástica alteração do sistema de ressurgência costeira em resposta às variações do vento. / The objective of the present work was to study the effects of variability in the wind due to the El Niño-Southern Oscillation events in the surface circulation patterns of the Peruvian coast, by means of data analysis and numerical modeling. It is focused the period 1991-2000, when it was observed strong La Niña (1996-1997, 1998-2000) and El Niño events (1997-98). These events had strong effects worlwide but very little is known on the local impacts on the circulation, thermocline structure and the Ekman dynamics of the Peruvian coast. The data analysed in the present study were collected on transects along 5 S and 15 S. The model used was an implementation of the Princeton Ocean Model (POM). The numerical simulations were forced with wind products relative to the period of interest (1991-2000). The simulations reproduced satisfactorily the mean circulation patterns in the study area, confirming that the in the nearshore region the wind is the main driving mechanism for coastal upwelling/downwelling. The results show that during the strong 1997-1998 El Niño, a drastic alteration of the coastal upwelling system ocurred in response to the changes in the wind.

To understand and to predict the weather and climate, numerical models are important tools and it is crucial that the controlling processes are described correctly. Since 70% of the global surface is covered with water the description how the ocean and atmosphere communicates has a considerable impact. The ocean–atmosphere exchange occurs through transport of momentum (friction) and heat, governed by turbulent eddies. The sea surface is also an important source of turbulence in both directions. The scales of the turbulent eddies cannot be resolved in ocean and climate models. Therefore, the turbulent exchanges have to be related to mean variables, such as wind speed and temperature differences. By using measurements, new methods to describe the air–sea exchange during two specific processes were developed. These processes are the so-called UVCN-regime (Unstable Very Close to Neutral stratification) and swell, i.e. waves which are not produced by the local wind. These processes were included in an ocean model and in a regional atmospheric climate model and the impact was investigated. The UVCN-regime enhances the heat transport significantly during the autumn and winter months in the ocean model. This results in a shallower well-mixed surface layer in the ocean. Wind-following swell reduces the surface friction, which is very important for the atmosphere. Some secondary effects in the climate model are reduced low-level cloud cover and reduced precipitation by more than 10% over sea areas. Locally and for short periods the impact is large. It is important to include the UVCN-regime and the swell impact in models, to make simulations more reliable.

Atmospheric models are strongly dependent on the turbulent exchange of momentum, sensible heat and moisture (latent heat) at the surface. Oceans cover about 70% of the Earth’s surface and understanding the processes that control air-sea exchange is of great importance in order to predict weather and climate. In the atmosphere, for instance, hurricane development, cyclone intensity and track depend on these processes. Ocean waves constitute an obvious example of air-sea interaction and can cause the air-flow over sea to depend on surface conditions in uniquely different ways compared to boundary layers over land. When waves are generated by wind they are called wind sea or growing sea, and when they leave their generation area or propagate faster than the generating wind they are called swell. The air-sea exchange is mediated by turbulent eddies occurring on many different scales. Field measurements and high-resolution turbulence resolving numerical simulations have here been used to study these processes. The standard method to measure turbulent fluxes is the eddy covariance method. A spatial separation is often used between instruments when measuring scalar flux; this causes an error which was investigated for the first time over sea. The error is typically smaller over ocean than over land, possibly indicating changes in turbulence structure over sea. Established and extended analysis methods to determine the dominant scales of momentum transfer was used to interpret how reduced drag and sometimes net upward momentum flux can persist in the boundary layer indirectly affected by swell. A changed turbulence structure with increased turbulence length scales and more effective mixing was found for swell. A study, using a coupled wave-atmosphere regional climate model, gave a first indication on what impact wave mixing have on atmosphere and wave parameters. Near surface wind speed and wind gradients was affected especially for shallow boundary layers, which typically increased in height from the introduced wave-mixing. A large impact may be expected in regions of the world with predominant swell. The impact of swell waves on air-sea exchange and mixing should be taken into account to develop more reliable coupled Earth system models.