Atmosphere Ocean Science Colloquium
Data-driven parameterizations of ocean mesoscale eddies
Speaker: Pavel Perezhogin, Courant
Location: Warren Weaver Hall 1302
Date: Wednesday, December 6, 2023, 3:30 p.m.
Mesoscale eddies are critical in setting ocean transport of momentum and tracers. Eddies are poorly represented in climate ocean models, and therefore their effects on the large-scale circulation must be parameterized. Classical parameterizations, which represent the bulk effect of the unresolved eddies, can be improved with new subgrid models learned directly from data. We will present two different and complementary approaches for developing data-driven parametrizations and their impact on the flow. Zanna and Bolton (2020) (ZB20) applied an equation-discovery algorithm to reveal an interpretable expression parameterizing the subgrid mesoscale fluxes. In the first part of the talk, we will discuss the physical properties and the sensitivity of the ZB20 parameterization in the primitive-equation GFDL MOM6 ocean model. The ZB20 improved the climatological mean flow and energy distributions, compared to the current state-of-the-art energy backscatter parameterizations. In the second part, we will discuss further development of data-driven mesoscale parameterizations, in particular the importance of inclusion of the physical constraints.