Quantifying the Uncertainty of Upper Ocean Turbulence Parameterizations

Andre Souza1, Gregory LeClaire Wagner2 and Raffaele M Ferrari2, (1)Massachusetts Institute of Technology, Cambridge, United States, (2)MIT, Cambridge, United States
A significant uncertainty in climate prediction is associated with the parameterization of sub-grid-scale processes. The parameterization of upper ocean processes is particularly important because it affects the transient climate sensitivity on timescales of years to decades. We introduce a Bayesian framework for assessing parameterizations of upper ocean turbulence. First, we run large eddy simulations of the upper ocean that resolve the small-scale turbulent processes that we wish to parameterize in climate models. Then we estimate the optimal parameter values, as well as their uncertainty, that need to be used in the popular K-Profile Parameterization to best reproduce the results of the large eddy simulations. We conclude by discussing how this methodology can be used to both improve sub-grid-scale parameterizations and to quantify the climate simulation uncertainty associated with the parameterizations.