Using ECMWF’s Forecasts (UEF2024)

An update is given on ECMWF's forecast performance across lead times and for a range of upper-air and surface parameters. We take a look at how ECMWF is doing relative to other global forecasting centres, identifying specific strengths and weaknesses. The model intercomparison will also include data-driven forecasts such as from the newly developed AIFS, and how these compare to those of other machine-learning models. With the help of ERA5 forecasts as a reference (being made by a 'frozen' forecasting system) one can separate the positive effects of the latest upgrade (48r1 in June 2023) from inter- and intra-annual variability. Among other changes, in 48r1 there has been an increase in the ENS horizontal resolution and in the number of ensemble members of the extended-range forecast, both of which have led to increased skill. On the seasonal timescale, the good performance with regard to the 2023 El Niño will be discussed, and to what extent this translated into enhanced predictive skill of near-surface temperature across the globe.