The impact of Aeolus observations on wind and rainfall predictions

Abstract. Previous studies showed substantial improvements in upper-level wind and mass field forecasts from assimilating Aeolus wind observations. This study extends those analyses using the improved reprocessed Aeolus dataset (version B16) in experiments with the global ECMWF forecasting system spanning more than three years. Results show that zonal wind forecasts improve across the entire troposphere during the first forecast week, propagating gradually into the stratosphere, with average reductions in the Root Mean Square Error of about 0.5 %, reaching over 1.5 % in the tropical upper troposphere. These improvements lead to more accurate rainfall forecasts in some regions and seasons, as measured by the Fraction Skill Score (FSS) and the Stable Equitable Error in Probability Space (SEEPS). They are largest during the winter half-year in the extratropics, particularly in the Southern Hemisphere, and appear primarily at the grid scale. The largest FSS improvements, reaching several percent, occur for 5–10 day leads and heavy rainfall categories, while SEEPS indicates modest but consistent gains in categorical precipitation skill. These results suggest that assimilating Aeolus winds improves large-scale dynamics, such as jet streams and Rossby waves, leading to more accurate long-term predictions of cyclones and fronts, and ultimately better local wind and heavy rainfall forecasts.