#O04 Vertical Mixing in MPAS-O

Thursday, November 10, 2016 - 16:00
Add to Calendar

An accurate representation of ocean boundary layer (OBL) turbulent mixing is critical to a climate model.  Here we present results of an evaluation of the vertical turbulent mixing parameterization in MPAS-O, K-profile parameterization (KPP).  KPP assumes that the vertical turbulent fluxes are given by a strictly specified via a cubic polynomial within the OBL.  The maximum of this profile is determined by the surface forcing.  We have compared KPP to numerous large eddy simulation (LES) configurations.  In LES the dominant eddies, which are responsible for boundary layer mixing, are explicitly resolved. Further, in LES there are very few tuning parameters and hence minimal sources of bias.  The LES configurations have widely different surface forcing and initial configuration.  The strength of wind, temperature, and salinity forcing are varied as well as the initial temperature and salinity profile.  Results will be presented that have been used to improve the simulation of the OBL in MPAS-O.  Despite many improvements to KPP and MPAS-O, critical issues remain.  For example, KPP can cause warming when the only surface forcing is cooling and can lead to negative concentrations of BGC in the presence of strong surface forcing and small OBL concentrations.  Further, KPP can create negative diffusivities in the presence of large interior diffusivities near the OBL.

Given these results, we suggest a possible new representation for vertical mixing (a variant of the CLUBBE scheme) that assumes an underlying joint probability density function for fluctuations of momentum and tracers.  With this assumption only seven turbulence equations must be predicted in total.  While for a standard higher order turbulence closure needs 15 equations for second order and a closure for higher orders.  This formulation can also be used to derive an explicit equation for boundary layer entrainment and is amenable to any coordinate system.  Initial results where LES data is used in the proposed scheme to reconstruct turbulent fluxes are encouraging.

Presentation file(s):