This thesis concerns theory, numerical simulations and observations of double-diffusion in polar settings. Double diffusion refers to processes occurring due to the difference in molecular diffusivities between two components that both contribute to the density. Specifically, these processes occur in the ocean due to the much slower diffusion of salinity compared to temperature. Within polar regions, thermohaline staircases have been frequently observed. These are layered structures in both temperature and salinity that can form due to double-diffusive processes, that give a characteristic `staircase' shape to profiles of temperature and salinity. Thermohaline staircases provide observational evidence of the importance of double diffusion to small scale ocean mixing, and so motivate our discussion of double-diffusive convection in polar environments.

After an introduction to the topic, the first results chapter discusses the energetics of double diffusion, developing a new model for the flow of energy within double-diffusive fluids. The second results chapter is motivated by observations of thermohaline staircases beneath George VI Ice Shelf, Antarctica. We conducted Large-Eddy-Simulations to explore the interaction of double diffusive convection with turbulence forced at a prescribed rate. Utilising the theory developed in chapter 1, the transition between double diffusive convection and stratified turbulence is identified and a criterion is developed for that transition in terms of profiles in temperature, salinity, and turbulence rate.

The third results chapter considers observational turbulence data collected in the Chukchi Sea in the marginal seas of the Arctic Ocean. This data shows an oceanographic section of a warm core intrahalocline eddy, where thermohaline layering was observed. We develop a criterion to predict the observed turbulent dissipation rates using fine-scale temperature and salinity data, assuming double-diffusive convection is active. This criterion is based on the energetic model from the first results chapter and assumes a lateral stirring of `spice’ variance (compensated thermohaline variance) along isopycnals is the driver of turbulence. The final results chapter consists of an analysis of mooring data from beneath George VI Ice Shelf, at the same location as thermohaline staircases were observed. We find that shear-driven turbulence cannot explain the observed dissipation rates. Utilising the method from the third results chapter, we show that lateral variations in spice can explain the observed turbulent mixing, suggesting it exerts control over the ice shelf basal melt rate.