Abstract
Deficiencies in upper ocean vertical mixing parameterizations contribute
to tropical upper ocean biases in global coupled general circulation
models, affecting their simulated ocean heat uptake and ENSO
variability. To better understand these deficiencies, we develop a suite
of ocean model experiments including both idealized single column models
and realistic global simulations. The vertical mixing parameterizations
are first evaluated using large eddy simulations as a baseline to assess
uncertainties and evaluate their implied turbulent mixing. Global models
are then developed following NOAA/GFDLâ\euro™s
0.25$\degree$ nominal ocean horizontal grid spacing OM4
(uncoupled ocean) configuration of the MOM6 ocean model, with various
modifications that target improvements to biases in the original model.
We identify a variety of enhancements to the existing mixing schemes
that are evaluated using observational constraints from TAO moorings and
Argo floats. In particular, we find that we can improve the diurnal
variability of mixing in OM4 via modifications to its mixing scheme, and
that we can improve the net mixing in the upper thermocline by reducing
the background vertical viscosity, allowing for more realistic, less
diffuse currents. The improved OM4 model better represents the mixing
and its diurnal deep-cycle variability, leading to more realistic
time-mean tropical thermocline structure, mixed layer depths, SSTs, and
a better Pacific Equatorial Undercurrent.