Evaluation of different ensemble configurations for the analysis and prediction of flash-flood-producing mesoscale convective systems

Probabilistic prediction of high-impact convective weather has advanced greatly in recent years with the development of ensembles of numerical weather prediction models at convection-allowing resolutions. Likewise, these high-resolution ensembles have enabled in-depth analysis of the mesoscale and storm-scale processes that contribute to the development of severe weather and heavy precipitation. However, the optimal design of such ensembles remains in question, including the application of perturbations to the initial conditions and the model physics.

In this presentation, different ensemble configurations will be evaluated for their utility in analyzing and predicting mesoscale convective systems (MCSs) that produce extreme precipitation and lead to flash flooding. In particular, the Center for the Analysis and Prediction of Storms (CAPS) Storm-Scale Ensemble Forecast (SSEF) system, which was run for real-time experimental forecasts and includes initial-condition perturbations drawn from the coarser Short-Range Ensemble Forecast (SREF) system, will be compared with ensembles using initial conditions drawn from an ensemble-based data assimilation method, and using different combinations of physical parameterizations. Several heavy-rain-producing MCS cases will be analyzed here. In addition to an evaluation of the different ensemble configurations for prediction, their utility for ensemble-based diagnostic analysis will also be discussed.