Scalar mixing in oceanic boundary layers: a method to guide model development

Abstract:

The K-profile parameterization (KPP) is widely used in coarse-resolution models to represent the effects of turbulent mixing. This model aims to represent turbulent fluxes by separating them into a diffusive (local) and a nondiffusive (nonlocal) component. However it is not clear how to properly separate the turbulent flux into two components without making a priori assumptions. Furthermore this separation should change based on the oceanic regime (nondiffusive fluxes, for example, are likely to be more important in flow regimes populated with large eddies), making the problem even more challenging.

In order to bridge this gap, we introduce a method that estimates this separation of fluxes for KPP based on data alone (i.e. without needing any a priori assumptions of shape or scaling). The method is based on an optimization procedure and it is applied to large-eddy simulation data. We use this method to investigate how the turbulent fluxes of passive scalars should be partitioned between diffusive and nondiffusive components across a wide range of oceanic regimes, which can be used to modify KPP in an informed manner. Based on those results, we also make efforts to create a version of KPP for passive scalars that is more adapted to wavy regimes than previous versions. In particular this new KPP version includes a change in the shape of both the eddy diffusivity and the nondiffusive terms in relation to previous versions.