Abstract
Abstract
Recent years have seen a surge in interest for leveraging neural networks to parameterize small-scale or fast processes in climate and turbulence models. In this short paper, we point out two fundamental issues in this endeavor. The first concerns the difficulties neural networks may experience in capturing rare events due to limitations in how data is sampled. The second arises from the inherent multiscale nature of these systems. They combine high-frequency components (like inertia-gravity waves) with slower, evolving processes (geostrophic motion). This multiscale nature creates a significant hurdle for neural network closures. To illustrate these challenges, we focus on the atmospheric 1980 Lorenz model, a simplified version of the Primitive Equations that drive climate models. This model serves as a compelling example because it captures the essence of these difficulties.
Funder
Office of Naval Research
National Science Foundation
H2020 European Research Council
Reference70 articles.
1. Numerical forecasting with the barotropic model;Bolin;Tellus,1955
2. On complete filtering of gravity modes through nonlinear initialization;Baer;Mon. Weather Rev.,1977
3. On the dynamics of gravity oscillations in a shallow water model, with applications to normal mode initialization;Machenhauer;Beitr. Phys. Atmos,1977
4. Normal mode initialization;Daley;Rev. Geophys.,1981
5. Nonlinear normal mode initialization and quasi-geostrophic theory;Leith;J. Atmos. Sci.,1980