|Abstract: The climate system comprises motions with scales ranging from meters (e.g., turbulence) to thousands of kilometers (weather systems), governed by a set of nonlinear equations. These motions interact across timescales, challenging our ability to understand and simulate climate. In particular, global climate simulations often cannot resolve critical processes such as clouds or turbulence and rely on ad-hoc relationships (known as parameterizations). In this talk, I demonstrate how we can discover equations from data for turbulence parameterizations, representing scale interactions in the ocean. I will discuss how the newly discovered equations transfer momentum and energy across scales and then use these new equations in coarse-resolution climate models. We demonstrate that multiscale climate simulations can be improved with these new representations of ocean turbulence, increasing the fidelity of climate projections. Some of the ideas can be extended to different problems in science, for example, astrophysics, in which scale interactions are also significant.
Link to the Event Video