Towards super-parameterization of ocean surface boundary layer turbulent mixing in MPAS-Ocean

Qing Li, Los Alamos National Laboratory, Los Alamos, NM, United States and Luke Van Roekel, Los Alamos National Laboratory, Fluid Dynamics and Solid Mechanics Group, Los Alamos, United States
Turbulent motions in the ocean surface boundary layer (OSBL) control the exchange of heat, momentum, and trace gases such as CO2 between the atmosphere and ocean, and thereby affect the weather and climate. Effects of these turbulent motions are commonly parameterized by simple one-dimensional vertical mixing schemes in regional and global ocean general circulation models (GCM) due to their small scales and non-hydrostatic nature. However, significant deficiencies are found among many vertical mixing schemes, highlighting the uncertainties in our understanding of turbulent mixing in the OSBL. Improvements in the OSBL turbulent mixing parameterizations require better constraints under different conditions from either observations or realistic high-resolution simulations where the turbulent motions are resolved, such as large eddy simulations (LES).

In this study, we take the super-parameterization approach by embedding a three-dimensional LES inside each horizontal grid cell of an ocean GCM, MPAS-Ocean. Small scale turbulent motions in LES and large scale currents in MPAS-Ocean are coupled by exchanging information at each MPAS-Ocean time step. In particular, variations in the large scale stratification and currents are applied in LES as external forcings, and simulated tendencies due to turbulent fluxes are then returned to MPAS-Ocean to represent the turbulent mixing. This loose coupling is different from the coupling across scales in, e.g., an LES over a large domain, but allows for sensitivity tests on the coupling time scale. The goal of this development is twofold: (1) a global super-parameterization in MPAS-Ocean; (2) a framework to study the interactions between small scale turbulent mixing and large scale variations under various conditions. To reduce the computational cost, the LES is running on the general-purpose graphics processing unit (GPGPU). Progresses in the development and initial assessment of the performance on GPGPU will be presented. Sensitivity tests on the coupling between the small scale turbulent mixing in LES and large scale variations in the stratification and currents in MPAS-Ocean in some idealized cases will also be discussed.