Using ECMWF’s Forecasts (UEF2024)

Whilst the users and the uses of extended range forecasts are inevitably wide-ranging, a common, basic user requirement for map products is visualising these two factors: (i) where there is a shift in the distribution of the ensemble member forecasts, and what the sign and magnitude of that shift is, and (ii) how ensemble member forecast spread compares with climatological (reforecast) spread.

Currently available ensemble mean products, which use a statistical significance test, have their place, but do not fit these basic needs very well. This presentation will provide, with examples, an idea for an alternative approach, in which we show, on one map, both the ensemble mean anomaly, where it differs from some near-normal range, and a spread-related metric based on comparing the interdecile ranges of the forecasts and the reforecasts. Significance tests may not be needed. For most variables this overall approach should work well, although some difficulties are envisaged - for example for rainfall when rain is forecast in an arid climate. These difficulties will also be discussed. A key aim of this talk is to glean some early user feedback on these ideas.