A posteriori learning of quasi-geostrophic turbulence parametrization: an experiment on integration steps

Hugo Frezat (1) , Julien Le Sommer (2) , Ronan Fablet (3, 4) , Guillaume Balarac (1) , Redouane Lguensat (5)

1 LEGI - Laboratoire des Écoulements Géophysiques et Industriels [Grenoble]
2 IGE - Institut des Géosciences de l’Environnement
3 IMT Atlantique - MEE - Département Mathematical and Electrical Engineering
4 Lab-STICC_OSE - Equipe Observations Signal & Environnement
5 NEMO R&D - Nucleus for European Modeling of the Ocean

Modeling the subgrid-scale dynamics of reduced models is a long standing open problem that finds application in ocean, atmosphere and climate predictions where direct numerical simulation (DNS) is impossible. While neural networks (NNs) have already been applied to a range of three-dimensional problems with success, the backward energy transfer of two-dimensional flows still remains a stability issue for trained models. We show that learning a model jointly with the dynamical solver and a meaningful $\textit{a posteriori}$-based loss function lead to stable and realistic simulations when applied to quasi-geostrophic turbulence.

Domains: Ocean, Atmosphere

Hugo Frezat, Julien Le Sommer, Ronan Fablet, Guillaume Balarac, Redouane Lguensat. A posteriori learning of quasi-geostrophic turbulence parametrization: an experiment on integration steps. 2021. ⟨hal-03456259v1⟩