Ensembles of numerical weather prediction (NWP) models are increasingly being used to make weather forecasts, but they also can provide valuable insights into the atmospheric processes that contribute to or inhibit the development of high-impact weather events. This presentation will discuss the use of ensemble data from multiple scales to investigate midlatitude convective systems that produce heavy rainfall.

First, global operational ensemble forecasts are used to assess forecast skill and uncertainty in several cases of widespread, long-lived heavy rainfall. In comparison with rainfall associated with extratropical cyclones and tropical cyclones, the forecasts of slow-moving cutoff circulations and their associated precipitation have low skill and large uncertainty. One event, associated with a convectively generated vortex, had particularly poor forecasts. The ensemble forecasts are also used to examine the processes responsible for the development and maintenance of this vortex. The maintenance of the vortex, and its slow movement, is found to be closely related to the strength of a closed midlevel anticyclone in the southwestern US and the strength of a midlevel ridge in the northern Plains. Relatively small differences in the wind and mass fields early in the ensemble forecast, in conjunction with modifications of the synoptic and mesoscale flow by deep convection, lead to very large spread in the resulting precipitation forecasts.

Then, the discussion will consider smaller scales, with data from a high-resolution ensemble of models that explicitly predict convection. Previous research identified that when a strong low-level jet interacts with a mesoscale vortex in a moist environment, extreme local rainfall often results. The ensemble is used to investigate this relationship further by contrasting members with jets and vortices of varying intensity and examining the resulting precipitation amounts. General implications of these results for probabilistic weather prediction will be discussed, as will issues relating to presenting and communicating forecasts with large uncertainty and spatial variability.