Turbulence and Convection in Southern Ocean Circulation

Circulation in the Southern Ocean is a key component of the global climate system as it controls the exchange of water masses between the Atlantic, Pacific and Indian oceans, thereby modulating heat and carbon uptake and biological activity. In the Southern Ocean, surface winds and buoyancy produce a circulation characterised by processes spanning a wide range of length and time scales, from large-scale overturning and zonal flow to small-scale turbulence and convection. A major source of uncertainty in numerical studies of the Southern Ocean lies in their characterisation of convection. Convection, an advective process triggered by a destabilising buoyancy gradient, can manifest over a small area (for instance, as open-ocean convection) or broad region (such as along the Antarctic coast). Convection in the ocean is a turbulent process, and is not resolved by large-scale ocean models. In addition, it can be difficult to capture field observations of convection at a sufficiently high resolution. As a consequence, the impact of convection on Southern Ocean circulation remains uncertain. In this thesis, we use a turbulence-resolving Direct Numerical Simulation (DNS) to investigate the influence of turbulent convection on Southern Ocean circulation. Our DNS model resolves scales of motion down to the turbulent grid-scale. This allows us to explore the effect of wind stress and convection on energetics and volume transport in the region. The DNS also facilitates understanding problems with the characterisation of small-scale flows in large-scale ocean models. We find that energy fluxes in the model are modified by turbulence, convection and surface winds. Diapycnal mixing is highest where turbulent convection is active, and dissipation is highest in the viscous sub-layer underlying surface wind stress. Both mixing and dissipation increase with increasing winds, and dissipation exhibits a higher sensitivity to winds than mixing. Dissipation is smaller than mixing in the wind stress regime relevant to the Southern Ocean. Using a scaling theory derived from the DNS results, we predict that mixing represents a sink of 75% of the total mechanical energy supply in the Southern Ocean, higher than the canonical assumption of 20%. Zonal transport in the model is eddy saturated, so it does not vary with wind stress. In addition, the abyssal overturning cell is stronger and larger than the upper overturning cell, enabled primarily by turbulence and convection. Both the abyssal and upper overturning cells strengthen with increasing wind, with the upper cell exhibiting more sensitivity to wind than the abyssal cell. In the wind stress regime relevant to the Southern Ocean, the abyssal overturning is stronger than the upper overturning. The high sensitivity of mixing and volume transport to small-scale flows underscores the need to develop more accurate turbulence and convection parameterisations. With this goal in mind, we compare idealised simulations of stratified open-ocean convection in the DNS and in a large-scale ocean model. The lack of turbulent vertical convection in the large-scale ocean model delays the formation of baroclinic eddies, resulting in anomalously large mixed layer depths prior to quasi-equilibrium. To address this behaviour, we propose the development of a stochastic convection parameterisation which would more accurately capture the internal variability of turbulent convection. The work presented in this thesis advances our understanding of the ways convection and turbulence can impact global ocean circulation. There are ongoing research questions about the link between small-scale flows (including turbulence and convection) and large-scale circulation (such as gyres, jets and overflows). Ocean models are far from being able to resolve small-scale features, so it remains critical to continue to use novel research tools like DNS to bridge the gap between small- and large-scale processes in the future