@misc{misc,
  author = {Thomas Haiden and Martin Janousek and Frederic Vivart and Michael Maier-Gerber and Cathy Wing Yi Li and Fernando Prates and Matthieu Chevallier},
  title = {Evaluation of ECMWF Forecasts},
  abstract = {<p class="MsoNoSpacing">This report provides a summary of ECMWF{\textquoteright}s forecast performance, covering medium, sub-seasonal, and seasonal forecast ranges. Headline scores adopted by ECMWF in collaboration with its member states monitor the evolution of various aspects of forecast skill. The report gives updates on these scores, as well as supplementary scores to provide a more complete assessment of forecast skill. The implementation of IFS Cycle 50r1 in May 2026 has led to further improvements in forecast skill. The primary focus of this summary is the medium range, specifically the forecast performance for upper-air variables. It is shown that in this respect, ECMWF continues to have an overall lead among centres. For surface variables, other centres have partially taken the lead, especially in the short range. ECMWF is actively working on this aspect, but the gap has not yet been fully closed. The data-driven forecasting system AIFS is included in many of the plots, showing superior skill in most upper-air and surface parameters. In the sub-seasonal forecast range, increasing skill of the IFS above persistence is being observed. On the seasonal timescale, the change in 2025-26 of Pacific SSTs from La Ni{\~n}a to a strong El Ni{\~n}o was well predicted by SEAS5, but cold temperature anomalies over continental areas, especially Eurasia, remain difficult to predict. This report also includes scores from higher-resolution forecasts (DestinE Continuous Extremes Digital Twin), and atmospheric composition forecasts from the\&nbsp;Copernicus Atmosphere Monitoring Service (CAMS).</p>
},
  year = {2026},
  journal = {Technical Memoranda},
  number = {938},
  month = {09/2026},
  publisher = {ECMWF},
  url = {\&nbsp;},
  doi = {10.21957/f66a9ec531},
}