The sinking of dense shelf waters down the continental slope (or “cascading”) contributes to oceanic water mass formation and carbon cycling. Cascading is therefore of significant importance for the global overturning circulation and thus climate. The occurrence of cascades is highly intermittent in space and time and observations of the process itself (rather than its outcomes) are scarce. Global climate models do not typically resolve cascading owing to numerical challenges concerning turbulence, mixing and faithful representation of bottom boundary layer dynamics. This work was motivated by the need to improve the representation of cascading in numerical ocean circulation models. Typical 3-D hydrostatic ocean circulation models are employed in a series of numerical experiments to investigate the process of dense water cascading in both idealised and realistic model setups. Cascading on steep bottom topography is modelled using POLCOMS, a 3-D ocean circulation model using a terrain-following s-coordinate system. The model setup is based on a laboratory experiment of a continuous dense water flow from a central source on a conical slope in a rotating tank. The descent of the dense flow as characterised by the length of the plume as a function of time is studied for a range of parameters, such as density difference, speed of rotation, flow rate and (in the model) diffusivity and viscosity. Very good agreement between the model and the laboratory results is shown in dimensional and non-dimensional variables. It is confirmed that a hydrostatic model is capable of reproducing the essential physics of cascading on a very steep slope if the model correctly resolves velocity veering in the bottom boundary layer. Experiments changing the height of the bottom Ekman layer (by changing viscosity) and modifying the plume from a 2-layer system to a stratified regime (by enhancing diapycnal diffusion) confirm previous theories, demonstrate their limitations and offer new insights into the dynamics of cascading outside of the controlled laboratory conditions. In further numerical experiments, the idealised geometry of the conical slope is retained but up-scaled to oceanic dimensions. The NEMO-SHELF model is used to study the fate of a dense water plume of similar properties to the overflow of brine-enriched shelf waters from the Storfjorden in Svalbard. The overflow plume, resulting from sea ice formation in the Storfjorden polynya, cascades into the ambient stratification resembling the predominant water masses of Fram Strait. At intermediate depths between 200-500m the plume encounters a layer of warm, saline AtlanticWater. In some years the plume ‘pierces’ the Atlantic Layer and sinks into the deep Fram Strait while in other years it remains ‘arrested’ at Atlantic Layer depths. It has been unclear what parameters control whether the plume pierces the Atlantic Layer or not. In a series of experiments we vary the salinity ‘S’ and the flow rate ‘Q’ of the simulated Storfjorden overflow to investigate both strong and weak cascading conditions. Results show that the cascading regime (piercing, arrested or ‘shaving’ - an intermediate case) can be predicted from the initial values of S and Q. In those model experiments where the initial density of the overflow water is considerably greater than of the deepest ambient water mass we find that a cascade with high initial S does not necessarily reach the bottom if Q is low. Conversely, cascades with an initial density just slightly higher than the deepest ambient layer may flow to the bottom if the flow rate Q is high. A functional relationship between S/Q and the final depth level of plume waters is explained by the flux of potential energy (arising from the introduction of dense water at shallow depth) which, in our idealised setting, represents the only energy source for downslope descent and mixing. Lastly, the influence of tides on the propagation of a dense water plume is investigated using a regional NEMO-SHELF model with realistic bathymetry, atmospheric forcing, open boundary conditions and tides. The model has 3 km horizontal resolution and 50 vertical levels in the sh-coordinate system which is specially designed to resolve bottom boundary layer processes. Tidal effects are isolated by comparing results from model runs with and without tides. A hotspot of tidally-induced horizontal diffusion leading to the lateral dispersion of the plume is identified at the southernmost headland of Spitsbergen which is in close proximity to the plume path. As a result the lighter fractions in the diluted upper layer of the plume are drawn into the shallow coastal current that carries Storfjorden water onto the Western Svalbard Shelf, while the dense bottom layer continues to sink down the slope. This bifurcation of the plume into a diluted shelf branch and a dense downslope branch is enhanced by tidally-induced shear dispersion at the headland. Tidal effects at the headland are shown to cause a net reduction in the downslope flux of Storfjorden water into deep Fram Strait. This finding contrasts previous results from observations of a dense plume on a different shelf without abrupt topography. The dispersive mechanism which is induced by the tides is identified as a mechanism by which tides may cause a relative reduction in downslope transport, thus adding to existing understanding of tidal effects on dense water overflows.

Plane-layer geometry convection models remain useful for modelling planetary mantle dynamics however they yield significantly warmer mean temperatures than spherical shell models. For example, in a uniform property spherical shell with the same radius ratio, f, as the Earth's mantle; a bottom heating Rayleigh number, Ra, of 10^7 and a nondimensional internal heating rate, H, of 23 (arguably Earth-like values) are insufficient to heat the mean temperature, θ, above the mean of the non-dimensional boundary value temperatures (0.5), the temperature in a plane-layer model with no internal heating. This study investigates the impact of this geometrical effect in convection models featuring uniform and stratified viscosity. To address the effect of geometry, heat sinks are implemented to lower the mean temperature in 3D plane-layer isoviscous convection models. Over 100 models are analyzed, and their mean temperatures are used to derive a single equation for predicting θ, as a function of Ra, H and f in spherical and plane-layer systems featuring free-slip surfaces. The inclusion of first-order terrestrial characteristics is introduced to quantitatively assess the influence of system geometry on planetary scale simulations. Again, over 100 models are analyzed featuring a uniform upper mantle viscosity and a lower mantle viscosity that increases by a factor of 30 or 100. An effective Rayleigh number, Raη, is defined based on the average viscosity of the mantle. Equations for the relationship between θ, Raη, and H are derived for convection in a spherical shell with f = 0.547 and plane-layer geometries. These equations can be used to determine the appropriate heating rate for a plane-layer convection model to emulate spherical shell convection mean temperatures for effective Rayleigh numbers comparable to the Earth’s value and greater. Comparing cases with the same H and Raη, the increased lower mantle viscosity amplifies the mismatch in mean temperatures between spherical shell and plane-layer models. These findings emphasize the importance of adjusting heating rates in plane-layer geometry models and have important implications for studying convection with temperature-dependent parameters in plane-layer systems. The findings are particularly relevant to the study of convection in super-Earths where full spherical shell calculations remain intractable.

Plane-layer geometry convection models remain useful for modelling planetary mantle dynamics however they yield significantly warmer mean temperatures than spherical shell models. For example, in a uniform property spherical shell with the same radius ratio, f, as the Earth's mantle; a bottom heating Rayleigh number, Ra, of 10^7 and a nondimensional internal heating rate, H, of 23 (arguably Earth-like values) are insufficient to heat the mean temperature, θ, above the mean of the non-dimensional boundary value temperatures (0.5), the temperature in a plane-layer model with no internal heating. This study investigates the impact of this geometrical effect in convection models featuring uniform and stratified viscosity. To address the effect of geometry, heat sinks are implemented to lower the mean temperature in 3D plane-layer isoviscous convection models. Over 100 models are analyzed, and their mean temperatures are used to derive a single equation for predicting θ, as a function of Ra, H and f in spherical and plane-layer systems featuring free-slip surfaces. The inclusion of first-order terrestrial characteristics is introduced to quantitatively assess the influence of system geometry on planetary scale simulations. Again, over 100 models are analyzed featuring a uniform upper mantle viscosity and a lower mantle viscosity that increases by a factor of 30 or 100. An effective Rayleigh number, Raη, is defined based on the average viscosity of the mantle. Equations for the relationship between θ, Raη, and H are derived for convection in a spherical shell with f = 0.547 and plane-layer geometries. These equations can be used to determine the appropriate heating rate for a plane-layer convection model to emulate spherical shell convection mean temperatures for effective Rayleigh numbers comparable to the Earth’s value and greater. Comparing cases with the same H and Raη, the increased lower mantle viscosity amplifies the mismatch in mean temperatures between spherical shell and plane-layer models. These findings emphasize the importance of adjusting heating rates in plane-layer geometry models and have important implications for studying convection with temperature-dependent parameters in plane-layer systems. The findings are particularly relevant to the study of convection in super-Earths where full spherical shell calculations remain intractable.

Observations of Jupiter show a super-rotating (prograde) equatorial jet that has persisted for decades. Shallow water simulations run in the Jovian parameter regime reproduce the mixture of robust vortices and alternating zonal jets observed on Jupiter, but the equatorial jet is invariably sub-rotating (retrograde). Recent work has obtained super-rotating equatorial jets by extending the standard shallow water equations to relax the height field towards its mean value. This Newtonian cooling-like term is intended to model radiative cooling to space, but its addition breaks key conservation properties for mass and momentum. In this thesis the radiatively damped thermal shallow water equations are proposed as an alternative model for Jovian atmospheres. They extend standard shallow water theory by permitting horizontal variations of the thermodynamic properties of the fluid. The additional temperature equation allows a Newtonian cooling term to be included while conserving mass and momentum. Simulations reproduce equatorial jets in the correct directions for both Jupiter and Neptune (which sub-rotates). Quasi-geostrophic theory filters out rapidly moving inertia-gravity waves. A local quasi-geostrophic theory of the radiatively damped thermal shallow water equations is derived, and then extended to cover whole planets. Simulations of this global thermal quasi-geostrophic theory show the same transition, from sub- to super-rotating equatorial jets, seen in simulations of the original thermal shallow water model as the radiative time scale is decreased. Thus the mechanism responsible for setting the direction of the equatorial jet must exist within quasi-geostrophic theory. Such a mechanism is developed by calculating the competing effects of Newtonian cooling and Rayleigh friction upon the zonal mean zonal acceleration induced by equatorially trapped Rossby waves. These waves transport no momentum in the absence of dissipation. Dissipation by Newtonian cooling creates an eastward zonal mean zonal acceleration, consistent with the formation of super-rotating equatorial jets in simulations, while the corresponding acceleration is westward for dissipation by Rayleigh friction.

Ocean submesoscales are the subject of increasing focus in the oceanographic literature; with instrumentation now more capable of observing them in situ and numerical models now able to reach the resolution required to more fully capture them. Submesoscales are typified by horizontal spatial scales of O(1 − 10) km, vertical scales O(100) m and time-scales of O(1) day and are known to be associated with regions of high vertical velocity and vorticity. Occurring most commonly at density fronts at the ocean surface they can control mixed layer restratification and provide an important control on fluxes between the atmosphere and the deep ocean. This thesis sets out to better understand the fundamental physical processes underpinning submesoscale instabilities using a number of idealised process models. Linear stability analysis complemented by non-linear, high-resolution simulations will be used initially to explore the ways in which submesoscale instabilities in the mixed layer may compete and interact with one another. In particular, we will investigate the way in which symmetric and ageostrophic baroclinic instabilities interact when simultaneously present in a flow, with focus on the growth rates and energetic pathways of previously unexplored dynamic instabilities that arise in this paradigm; three-dimensional, mixed symmetric-baroclinic instabilities. Further, these non-linear simulations will allow us to investigate the transition to dissipative scales that can occur in the classical Eady model via a multitude of small-scale secondary instabilities that result from primary submesoscale instabilities. Finally, observational data, taken aboard the SMILES project cruise to the Southern Ocean, helps to motivate the consideration of a new dynamical paradigm; the Eady model with superimposed high amplitude barotropic jet. Non-linear simulations investigate the extent to which the addition of such a jet is capable of damping submesoscale growth. The causes of this damping are then investigated using linear analysis. With this approach eventually demonstrated as being unable to fully explain growth rate reductions, we introduce a new framework combining potential vorticity mixing by submesoscale instabilities with geostrophic adjustment, which relaxes the flow back to a geostrophic balanced state. This framework will help to explain, conceptually, how non-linear eddies control the linear stability of the flow.

To understand the mechanisms of energy exchange between the tropics and the midlatitudes, it is necessary to develop simplified climate models. Motivated by linear wave theory, one such model is derived below. It captures the nonlinear interaction between barotropic and first baroclinic modes. In particular, it allows for the study of the barotropic response to a baroclinic forcing. Numerical methods for handling this nonlinear system are carefully developed and validated. The response generated by a physically realistic Kelvin wave forcing is studied and is found to consist mainly of one eastward propagating wave (phase-locked to the forcing) and two westward propagating (Rossby) waves. The Rossby waves are shown to be highly constrained by the initial parameters of the forcing and an explanation of this result is proposed.

In studies of the ocean it has become conventional to retain only the component of the Coriolis force associated with the radial component of the Earth’s rotation vector, the so-called “traditional approximation”. We investigate the role of the “non-traditional” component of the Coriolis force, corresponding to the non-radial component of the rotation vector, in transporting abyssal waters across the equator. We first derive a non-traditional generalisation of the multi-layer shallow water equations, which describe the flow of multiple superposed layers of inviscid, incompressible fluid with constant densities over prescribed topography in a rotating frame. We derive these equations both by averaging the three-dimensional governing equations over each layer, and via Hamilton’s principle. The latter derivation guarantees that conservation laws for mass, momentum, energy and potential vorticity are preserved. Within geophysically realistic parameters, including the complete Coriolis force modifies the domain of hyperbolicity of the multi-layer equations by no more than 5%. By contrast, long linear plane waves exhibit dramatic structural changes due to reconnection of the surface and internal wave modes in the long-wave limit. We use our non-traditional shallow water equations as an idealised model of an abyssal current flowing beneath a less dense upper ocean. We focus on the Antarctic Bottom Water, which crosses the equator in the western Atlantic ocean, where the bathymetry forms an almost-westward channel. Cross-equatorial flow is strongly constrained by potential vorticity conservation, which requires fluid to acquire a large relative vorticity in order to move between hemispheres. Including the complete Coriolis force accounts for the fact that fluid crossing the equator in an eastward/westward channel experiences a smaller change in angular momentum, and therefore acquires less relative vorticity. Our analytical and numerical solutions for shallow water flow over idealised channel topography show that the non-traditional component of the Coriolis force facilitates cross-equatorial flow through an almost-westward channel.

Cette thèse considère quatre systèmes physiques complexes pour lesquels il est exceptionnellement possible d’identifier des variables lentes qui contrôlent l'évolution à temps long du système complet. La séparation d'échelle de temps entre ces variables lentes et les autres variables permet d'utiliser la technique de moyennisation stochastique pour obtenir une dynamique effective pour les variables lentes. Cette thèse considère la possibilité de prédire les événements rares dans le système solaire. Nous avons étudié deux types d’événements rares. Le premier est un renversement possible de l'axe de rotation de la Terre en l'absence des effets de marée de la Lune. Le second est la désintégration de l'ensemble du système solaire interne suite à une instabilité dans l'orbite de Mercure. Pour chacun des deux problèmes, il existe des variables lentes non triviales, qui ne sont pas données par des variables physiques naturelles. La moyennisation stochastique a permis de découvrir le mécanisme physique qui conduit à ces événements rares et de donner, par une approche purement théorique, l'ordre de grandeur de la probabilité de ces phénomènes. Nous avons également montré que la déstabilisation de Mercure sur un temps inférieur à l'âge du système solaire obéit à un mécanisme d'instanton bien décrit par la théorie des grandes déviations. Le travail effectué dans cette thèse ouvre donc un nouveau champ d'action pour l'utilisation d'algorithmes de calcul d'événements rares. Nous avons utilisé pour la première fois les théorèmes de moyennisation stochastique dans le cadre de la mécanique céleste pour quantifier l'effet stochastique des astéroïdes sur la trajectoire des planètes. Enfin, une partie du travail porte sur un problème de turbulence géophysique: dans l'atmosphère de Jupiter, on peut observer des structures zonales (jets) à grande échelles évoluant beaucoup plus lentement que les tourbillons environnants. Nous montrons qu'il est pour la première fois possible d'obtenir explicitement le profil de ces jets par moyennisation des degrés de liberté turbulents rapides. / The present thesis describes four complex dynamical systems. In each system, the long-term behavior is controlled by a few number of slow variables that can be clearly identified. We show that in the limit of a large timescale separation between the slow variables and the other variables, stochastic averaging can be performed and leads to an effective dynamics for the set of slow variables. This thesis also deals with rare events predictions in the solar system. We consider two possible rare events. The first one is a very large variation of the spin axis orientation of a Moonless Earth. The second one is the disintegration of the inner solar system because of an instability in Mercury’s orbit. Both systems are controlled by non-trivial slow variables that are not given by simple physical quantities. Stochastic averaging has led to the discovery of the mechanism leading to those rare events and gives theoretical bases to compute the rare events probabilities. We also show that Mercury’s short-term destabilizations (compared to the age of the solar system) follow an instanton mechanism, and can be predicted using large deviation theory. The special algorithms devoted to the computation of rare event probabilities can thus find surprising applications in the field of celestial mechanics. We have used for the first time stochastic averaging in the field of celestial mechanics to give a relevant orders of magnitude for the long-term perturbation of planetary orbits by asteroids. A part of the work is about geophysical fluid mechanics. In Jupiter atmosphere, large scale structures (jets) can be observed, the typical time of evolution of which is much larger than that of the surrounding turbulence. We show for the first time that the mean wind velocity can be obtained explicitly by averaging the fast turbulent degrees of freedom.

Approved for public release, distribution is unlimited / occurring most frequently. For the 217 large-error cases due to midlatitude influences, the most frequent error mechanisms were E-DCI (midlatitude), excessive response to vertical wind shear, excessive midlatitude cyclogenesis (E-MCG), insufficient midlatitude cyclogenesis (I-MCG), excessive midlatitude cyclolysis (E-MCL) and excessive midlatitude anticyclogenesis (E-MAG), which accounted for 68% of all large errors occurring in both NOGAPS and GFDN. Characteristics and symptoms of the erroneous forecast tracks and model fields are documented and illustrative case studies are presented. Proper identification and removal of the track forecast displaying an error mechanism could form a selective consensus that will be more accurate than a non-selective consensus. / Captain, United States Air Force

This thesis presents a numerical exploration of the dynamics governing rotating flow driven by a surface stress in the " sliced cylinder " model of Pedlosky & Greenspan (1967) and Beardsley (1969), and its close relative, the " sliced cone " model introduced by Griffiths & Veronis (1997). The sliced cylinder model simulates the barotropic wind-driven circulation in a circular basin with vertical sidewalls, using a depth gradient to mimic the effects of a gradient in Coriolis parameter. In the sliced cone the vertical sidewalls are replaced by an azimuthally uniform slope around the perimeter of the basin to simulate a continental slope. Since these models can be implemented in the laboratory, their dynamics can be explored by a complementary interplay of analysis and numerical and laboratory experiments. ¶ In this thesis a derivation is presented of a generalised quasigeostrophic formulation which is valid for linear and moderately nonlinear barotropic flows over large-amplitude topography on an f-plane, yet retains the simplicity and conservation properties of the standard quasigeostrophic vorticity equation (which is valid only for small depth variations). This formulation is implemented in a numerical model based on a code developed by Page (1982) and Becker & Page (1990). ¶ The accuracy of the formulation and its implementation are confirmed by detailed comparisons with the laboratory sliced cylinder and sliced cone results of Griffiths (Griffiths & Kiss, 1999) and Griffiths & Veronis (1997), respectively. The numerical model is then used to provide insight into the dynamics responsible for the observed laboratory flows. In the linear limit the numerical model reveals shortcomings in the sliced cone analysis by Griffiths & Veronis (1998) in the region where the slope and interior join, and shows that the potential vorticity is dissipated in an extended region at the bottom of the slope rather than a localised region at the east as suggested by Griffiths & Veronis (1997, 1998). Welander's thermal analogy (Welander, 1968) is used to explain the linear circulation pattern, and demonstrates that the broadly distributed potential vorticity dissipation is due to the closure of geostrophic contours in this geometry. ¶ The numerical results also provide insight into features of the flow at finite Rossby number. It is demonstrated that separation of the western boundary current in the sliced cylinder is closely associated with a " crisis " due to excessive potential vorticity dissipation in the viscous sublayer, rather than insufficient dissipation in the outer western boundary current as suggested by Holland & Lin (1975) and Pedlosky (1987). The stability boundaries in both models are refined using the numerical results, clarifying in particular the way in which the western boundary current instability in the sliced cone disappears at large Rossby and/or Ekman number. A flow regime is also revealed in the sliced cylinder in which the boundary current separates without reversed flow, consistent with the potential vorticity " crisis " mechanism. In addition the location of the stability boundary is determined as a function of the aspect ratio of the sliced cylinder, which demonstrates that the flow is stabilised in narrow basins such as those used by Beardsley (1969, 1972, 1973) and Becker & Page (1990) relative to the much wider basin used by Griffiths & Kiss (1999). ¶ Laboratory studies of the sliced cone by Griffiths & Veronis (1997) showed that the flow became unstable only under anticyclonic forcing. It is shown in this thesis that the contrast between flow under cyclonic and anticyclonic forcing is due to the combined effects of the relative vorticity and topography in determining the shape of the potential vorticity contours. The vorticity at the bottom of the sidewall smooths out the potential vorticity contours under cyclonic forcing, but distorts them into highly contorted shapes under anticyclonic forcing. In addition, the flow is dominated by inertial boundary layers under cyclonic forcing and by standing Rossby waves under anticyclonic forcing due to the differing flow direction relative to the direction of Rossby wave phase propagation. The changes to the potential vorticity structure under strong cyclonic forcing reduce the potential vorticity changes experienced by fluid columns, and the flow approaches a steady free inertial circulation. In contrast, the complexity of the flow structure under anticyclonic forcing results in strong potential vorticity changes and also leads to barotropic instability under strong forcing. ¶ The numerical results indicate that the instabilities in both models arise through supercritical Hopf bifurcations. The two types of instability observed by Griffiths & Veronis (1997) in the sliced cone are shown to be related to the western boundary current instability and " interior instability " identified by Meacham & Berloff (1997). The western boundary current instability is trapped at the western side of the interior because its northward phase speed exceeds that of the fastest interior Rossby wave with the same meridional wavenumber, as discussed by Ierley & Young (1991). ¶ Numerical experiments with different lateral boundary conditions are also undertaken. These show that the flow in the sliced cylinder is dramatically altered when the free-slip boundary condition is used instead of the no-slip condition, as expected from the work of Blandford (1971). There is no separated jet, because the flow cannot experience a potential vorticity " crisis " with this boundary condition, so the western boundary current overshoots and enters the interior from the east. In contrast, the flow in the sliced cone is identical whether no-slip, free-slip or super-slip boundary conditions are applied to the horizontal flow at the top of the sloping sidewall, except in the immediate vicinity of this region. This insensitivity results from the extremely strong topographic steering near the edge of the basin due to the vanishing depth, which demands a balance between wind forcing and Ekman pumping on the upper slope, regardless of the lateral boundary condition. The sensitivity to the lateral boundary condition is related to the importance of lateral friction in the global vorticity balance. The integrated vorticity must vanish under the no-slip condition, so in the sliced cylinder the overall vorticity budget is dominated by lateral viscosity and Ekman friction is negligible. Under the free-slip condition the Ekman friction assumes a dominant role in the dissipation, leading to a dramatic change in the flow structure. In contrast, the much larger depth variation in the sliced cone leads to a global vorticity balance in which Ekman friction is always dominant, regardless of the boundary condition.