Exploring a Practical Path to More Skilful Ensemble Forecasts

A diagnostic framework has been proposed recently that could aid forecast system development. The approach is based on the identification of flow-types that contribute most to the overall un-reliability in short forecasts. Since reliability and refinement are complementary components of forecast skill, these flow-types are not necessarily the ones associated with the largest uncertainty growth-rates; they could be associated with more benign yet more frequent flow-situations, which are poorly represented in the forecast model. This study will discuss initial results, based on the ECMWF ensemble, in the context of synoptic flow-types that occur in the North Atlantic region.