Untangling the effects of vertical mixing schemes and convective adjustment in the Mediterranean Sea: insights from a sensitivity study

Abstract. The Mediterranean Sea provides a natural laboratory for investigating ocean circulation processes of global relevance due to its complex dynamics, active deep and intermediate water formation, and sensitivity to climate variability. Regional ocean circulation models' skill strongly depends on the parameterization of subgrid-scale processes, among which turbulent vertical mixing and convection play a major role. To evaluate their impact in the Mediterranean Sea two-year-long simulations were conducted using three vertical closure schemes: the Richardson-number dependent parameterisation, the Turbulent Kinetic Energy (TKE) scheme, and the Generalised Length Scale (GLS) scheme. Each scheme was tested both with and without the convective adjustment approach, resulting in a set of comparative experiments designed to isolate the contribution of the adjustment process and its combined effect with each parameterization. Model results are evaluated against all available Argo floats data, both at the basin scale and in key deep and intermediate water formation regions. The simulations show that adding a convective adjustment is crucial to accurately reproduce observations with the Richardson-number dependent parameterization, where it improves all key variables, while for the TKE scheme it is particularly important for representing the mixed layer depth across the basin and in deep water formation areas. For more physics-based vertical schemes, like the GLS closure, the convective adjustment is mostly redundant and can occasionally degrade results. Overall, the GLS scheme without any convective adjustment provides the most accurate representation of the mixed layer depth as well as the vertical structure and variability both at basin scale and in key regions of deep and intermediate water formation.