Studying low-likelihood high-impact climate events in a warming world requires massive ensembles of hindcasts and forecasts to capture their statistics. At present, it is extremely challenging to generate these ensembles using traditional weather or climate models, especially at sufficiently high spatial resolution.

We describe how to bring the power of machine learning (ML) to generate climate hindcasts at four to five orders-of-magnitude lower computational cost than conventional numerical methods. We show how to evaluate ML climate emulators using the same rigorous metrics developed for operational numerical weather prediction.

Furthermore, 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 at least the last 2000 years. We show how HENS can be used to quantify the intensity of atmospheric rivers in the Southern Hemisphere, the diffusion of tropical cyclones in the general circulation, and the severity of unprecedented heatwaves characteristic of last summer.

We conclude with the prospects of extending machine-learning emulators to make skillful predictions of future climate change.