Prediction of Extreme Events in Multiscale Simulations of Geophysical Turbulence using Reinforcement Learning

Accurate subgrid-scale closures are essential for weather/climate models, where predicting extreme events is critical. Traditional closures have structural errors, e.g., producing excessive diffusion that dampens extremes. Artificial intelligence has gained attention for closure modeling, but the pr