probabilistic forecasting is crucial to decision-making under uncertainty
about future weather. The dominant approach is to use an ensemble of forecasts
to represent and quantify uncertainty in operational numerical weather
prediction. However, generating ensembles is computationally c