Tests of Normal versus Anomalous Diffusion of Tropical Cyclones using Huge Ensembles of Machine-Learning-based Climate Emulators

Studying low-likelihood high-impact climate events in a warming world requires massive ensembles of hindcasts and forecasts to capture their statistics. It is extremely challenging to generate these ensembles using traditional weather or climate models, especially at sufficiently high spatial resolution. We describe how machine learning (ML) can generate ensembles at orders-of-magnitude lower cost and higher speed than conventional numerical methods.

We illustrate the power of this approach by generating a huge ensemble (HENS) of 7,424 members initialized for each day of June through August 2023, the second-hottest summer in 2000 years. We show how HENS can quantify the diffusion of tropical cyclones in the general circulation. To predict where tropical cyclones make landfall, it is critical to know whether ensembles of predicted cyclone paths obey subdiffusion, Brownian motion, or superdiffusion. Existing observational analyses suggest that cyclones are superdiffusive. We show how HENS can confirm that cyclone paths obey superdiffusion and, in fact, approach ballistic trajectories.