Applications of Deep Learning to Ocean Data Inference and Subgrid Parameterization

Abstract

Geostrophic eddies contribute to the mixing of heat, carbon, and other climatically important tracers. A passive tracer driven by satellite-derived surface velocity fields is used to study the regional and temporal variability of lateral eddy mixing in the North Atlantic. Using a quasi-Lagrangian diffusivity diagnostic, we show that the upstream region (80°–50°W) of the Gulf Stream jet exhibits a significant mixing barrier (with diffusivity of ≈1 × 103 m2 s−1), compared to the downstream region (50°–10°W), which displays no mixing suppression (≈10 × 103 m2 s−1). The interannual variability is 10%–20% of the time mean in both regions. By analyzing linear perturbations of mixing-length diffusivity expression, we show that the across-jet mixing in the upstream region is driven by variations in the mean flow, rather than eddy velocity. In the downstream region, both the mean flow and eddy velocity contribute to the temporal variability. Our results suggest that an eddy parameterization must take into account the along-jet variation of mixing, and within jets such diffusivities may be a simple function of jet strength.

Type
Publication
J. of Phys. Oceanogr., 49(10), 2601-2614
Laure Zanna
Laure Zanna
Professor of Mathematics & Atmosphere/Ocean Science [She/Her]

My research interests include Climate Dynamics, Physical Oceanography and Data Science.