Ces travaux de thèse s'intéressent à la variabilité océanique intrinsèque générée spontanément sous forçage atmosphérique saisonnier dans des simulations océaniques à haute-résolution (1/4 de degré et 1/12 de degré ), en présence de tourbillons de méso-échelle (environ 100 km). L'empreinte de la variabilité intrinsèque sur le niveau de la mer (SLA) et sur la température de surface (SST) est caractérisée dans plusieurs gammes d'échelles spatio-temporelles par des méthodes de filtrage. En particulier, il est montré que la variabilité petite-échelle inter-annuelle à décennale de SLA observée par les satellites altimétriques, a un caractère stochastique et est majoritairement intrinsèque. Aux échelles de temps multi-décennales, la variabilité intrinsèque de SLA dans les régions océaniques turbulentes est d'amplitude comparable à la variabilité interne simulée dans les modèles climatiques couplés (dénués de turbulence océanique), et pourrait constituer une source additionnelle d'incertitudes dans la détection de l'élévation régionale du niveau de la mer d'origine anthropique. Une analyse spectrale montre enfin que l'advection non-linéaire de vorticité relative transfère spontanément de l'énergie cinétique d'ondes frontales hautes-fréquences vers des tourbillons de méso-échelle plus lents, générant in fine de la variabilité intrinsèque basse-fréquence. Les diagnostiques ont été réalisés à l'aide d'outils développés pour traiter de façon optimale les données océaniques haute-résolution. Ces travaux témoignent plus généralement du caractère chaotique de l'océan turbulent, dont l'interaction avec l'atmosphère est encore mal comprise. / This work focuses on the intrinsic oceanic variability spontaneously generated under seasonal atmospheric forcing in high-resolution oceanic simulations (1/4 ? and 1/12 ? ) including mesoscale eddies (~ 100 km). The imprint of intrinsic variability on sea-level (SLA) and sea surface temperature (SST) is characterized in various spatio-temporal ranges using filtering methods. In particular, the small-scale interannual-to-decadal SLA variability observed by satellite altimetry, is stochastic and mostly intrinsic. At multi-decadal timescales, SLA intrinsic variability is comparable to the internal climate variability simulated in climate coupled models (devoid of oceanic turbulence), and may yield additional uncertainties in the detection of human-induced regional sea-level change. A spectral analysis eventually shows that non-linear advection of relative vorticity spontaneously transfers kinetic energy from high-frequency frontal waves to slower mesoscale eddies, ultimately generating low-frequency intrinsic variability. The analyses have been performed using tools developed for optimal processing of high-resolution oceanic dataset. This work generally provides evidence of the chaotic behavior of the turbulent ocean, whose interaction with the atmosphere is still poorly-known.

Two codes have been developed and implemented for use on massively parallelsuper computers to simulate two-dimensional and quasi-geostrophic turbulence.The codes have been found to scale well with increasing resolution and width ofthe simulations. This has allowed for the highest resolution simulations of twodimensionaland quasi-geostrophic turbulence so far reported in the literature.The direct numerical simulations have focused on the statistical characteristicsof turbulent cascades of energy and enstrophy, the role of coherent vorticesand departures from universal scaling laws, theoretized more than 40 yearsago. In particular, the investigations have concerned the enstrophy and energycascades in forced and decaying two-dimensional turbulence. Furthermore, theapplicability of Charney’s hypotheses on quasi-geostrophic turbulence has beentested. The results have shed light on the flow evolution at very large Reynoldsnumbers. The most important results are the robustness of the enstrophycascade in forced and decaying two-dimensional turbulence, the sensitivity toan infrared Reynolds number in the spectral scaling of the energy spectrumin the inverse energy cascade range, and the validation of Charney’s predictionson the dynamics of quasi-geostrophic turbulence. It has also been shownthat the scaling of the energy spectrum in the enstrophy cascade is insensitiveto intermittency in higher order statistics, but that corrections apply to the”universal” Batchelor-Kraichnan constant, as a consequence of large-scale dissipationanomalies following a classical remark by Landau (Landau & Lifshitz1987). Another finding is that the inverse energy cascade is maintained bynonlocal triad interactions, which is in contradiction with the classical localityassumption. / QC 20101029

The dynamics captured in the ocean surface current data provided by satellite altimetry has been a subject of debate since the past decade. In particular, the contribution of surface and interior dynamics to altimetry remains unclear. One avenue to settling this issue is to compare the turbulence (for example, the nature of spectra and interscale fluxes) captured by altimetry to theories of two-dimensional, surface and interior quasigeostrophic turbulence. In this thesis, we focus on mesoscales (i.e., scales of the order of few hundred kms) that are well resolved by altimetry data. Aspects of two dimensional, three dimensional, geotropic and surface quasigeostrophic turbulence are revisited and compared with the observations. Specifically, we compute kinetic energy (KE) spectra and fluxes in five geographical regions (all over the globe) using 21 years of 0.25◦resolution daily data as provided by the AVISO project. We report a strong forward cascade of KE at small scales (accompanied by a spectral scaling of the form k−3) and a robust inverse cascade at larger scales. Further, we show that the small diver-gent part in horizontal velocity data drives the strong forward flux of KE. Indeed, on considering only the non-divergent part of the flow, in accord with incompressible two-dimensional turbulence, the inverse cascade is unaffected, but the forward transfer becomes very weak and the spectral slopes over this range of scales tend to a relatively steeper k−3.5scaling. We note that our results do not agree with interior first bar clinic mode quasigeostrophic (incorrect strength of forward flux) or surface-quasigeostrophic (incorrect spectral slopes) turbulence. Rather, the results are compatible with rotating shallow water and rotating stratified Boussinesq models in which condition of geostrophic balance is dominant but the divergence of horizontal velocity field is not exactly zero. Having seen the “mean” picture of fluxes and spectra from altimetry, in the second part of the thesis we investigate the variability of these entities. In particular, we employ Empirical Or-thogonal Function (EOF) analysis and focus on the variability in the spectral flux. Remarkably, over the entire globe, irrespective of the region under consideration, we see that the first two EOFs explain a large part of the variability in flux anomalies. The geometry of these modes is distinct, the first represents a single signed transfer across scales (i.e. large to small or small to large depending on the sign of the associated principal component), while the second is a mixed mode in that it exhibits a forward/inverse transfer at large/small scales.

Le golfe de Gascogne (GdG), un milieu riche en processus physiques a été étudié à partir de simulations numériques. L’étude est construite autour d'échelles allant du GdG à la sous méso-échelle. Dans la première partie, nous avons examiné la variabilité interannuelle de la température et de la salinité de surface sur une période de 53 ans : nous avons pu décrire deux tendances en lien avec I'Atlantique Nord-Est. Le refroidissement et la dessalure jusqu'en 1976 seraient liés à la grande anomalie de salinité, le réchauffement et la salinification actuels liés à I'augmentation de CO2 atmosphérique. Le GdG se caractérise par un courant de pente, lberian Poleward Current (lPC) : sa variabilité serait liée au vent du Sud-Ouest qui renforce l'lPC par un courant géostrophique dans le Bassin lbérique. L’un des résultats intéressant trouvé dans les simulations et confìrmé par les observations est l'apparition des anomalies froides liées à des upwellings en alternance avec des anomalies chaudes "La Navidad". Ces upwellings seraient liés au vent de Nord dans le Bassin lbérique mais au courant d'Ouest le long des côtes Nord espagnoles. Dans une deuxième partie, nous nous sommes attachés à la méso et sous méso-échelle à travers la détection des tourbillons et la variabilité des spectres d'énergie. Un indice basé sur le rapport entre I'anomalie de densité de surface et I'anomalie de niveau de la mer permet de détecter les tourbillons de subsurface et de les distinguer des tourbillons de surface. Une application de cet indice à partir des données satellites confirme le potentiel de détection des Slope Water Oceanic eDDIES (tourbillons de subsurface caractéristiques du GdG). La description de l'énergie cinétique turbulente (EKE) dans le GdG montre une variabilité spatiale avec un maximum le long de la côte Nord espagnole liée à I'lPC. Les pentes des spectres (k-4.2 pour la SSH, en k-2.4 pour la SST et en k-2.4 pour l'énergie cinétique) sont différents des observations satellites, mais comparables avec les précédentes études. Ces pentes de spectres ont également une variabilité saisonnière avec un maximum en hiver et un minimum en été, liée au cycle saisonnier de I'EKE. / The Bay of Biscay (BoB), an environment rich in physical processes has been studied from numerical simulations. Thestudy is built around scales from the size of BoB until sub mesoscale.ln the first part, we examined the interannual variability of the sea surface temperature and saliniÇ over a period of 53years: we were able to describe two trends related to the North-East Atlantic. Cooling and freshening until 1976 thatcould be related to the Great Salinity Anomaly and current salinification related to the atmospheric increase of CO2.The Bay of Biscay is characterized by a slope current, the lberian Poleward Current (lPC): its variability is linked to theSouth West wind strengthens the IPC by a geostrophic current in the lberian Basin. One of the interesting results foundin simulations and confirmed by observations is the appearance of cold anomalies related to upwellings and alternatingwith warm anomalies 'La Navidad'. These upwellings could be linked to the north wind in the lberian Basin but to the West current along the northern Spanish coast.ln the second part, we are committed to the meso and sub mesoscale eddies through a method of detection and throughthe variability of energy spectra. An index based on the ratio of surface density anomaly and sea level anomaly allowsdetecting subsurface vortices and distinguishing them from the surface ones. The application of this index from thesatellite data confirms the detection potential of Slope Water Oceanic Eddies (subsurface vortices of BoB).The description of the Eddy Kinetic Energy (EKE) in the BoB. shows a spatial variability with maximum along the Spanishnorth coast linked to the lPC. The slopes of the spectra (k-4.2 for SSH, k-2.4 for SST and k-2.4 for the kinetic energy) are different from satellite observations, but comparable with previous studies. These spectral slopes have a seasonalvariability with a maximum in winter and minimum in summe¡ related to the seasonal cycle of EKE.