Submesoscale flows, current systems 1-100 km in horizontal extent, are increasingly coming into focus as an important component of upper-ocean dynamics. A range of processes have been proposed to energize submesoscale flows, but which process dominates in reality must be determined observationally. We diagnose from observed flow statistics that in the thermocline the dynamics in the submesoscale range transition from geostrophic turbulence at large scales to inertia-gravity waves at small scales, with the transition scale depending dramatically on geographic location. A similar transition is shown to occur in the atmosphere, suggesting intriguing similarities between atmospheric and oceanic dynamics. We furthermore diagnose from upper-ocean observations a seasonal cycle in submesoscale turbulence: fronts and currents are more energetic in the deep wintertime mixed layer than in the summertime seasonal thermocline. This seasonal cycle hints at the importance of baroclinic mixed layer instabilities in energizing submesoscale turbulence in winter. To better understand this energization, three aspects of the dynamics of baroclinic mixed layer instabilities are investigated. First, we formulate a quasigeostrophic model that describes the linear and nonlinear evolution of these instabilities. The simple model reproduces the observed wintertime distribution of energy across scales and depth, suggesting it captures the essence of how the submesoscale range is energized in winter. Second, we investigate how baroclinic instabilities are affected by convection, which is generated by atmospheric forcing and dominates the mixed layer dynamics at small scales. It is found that baroclinic instabilities are remarkably resilient to the presence of convection and develop even when rapid overturns keep the mixed layer unstratified. Third, we discuss the restratification induced by baroclinic mixed layer instabilities. We show that the rate of restratification depends on characteristics of the baroclinic eddies themselves, a dependence not captured by a previously proposed parameterization. These insights sharpen our understanding of submesoscale dynamics and can help focus future inquiry into whether and how submesoscale flows influence the ocean's role in climate.Thesis supervisor: Raffaele Ferrari Title: Breene M. Kerr Professor of Oceanography