Global wind stilling and the role of sub-monthly variability in explaining deficiencies in atmospheric reanalyses

Terrestrial near-surface (10-m) wind speed decreased over the Northern Hemisphere between 1980 and 2010. We revisit this ‘global stilling’ by comparing station-based observations and multiple atmospheric reanalysis products and assessing their ability to capture the past changes in near-surface wind. We show, using a station-based wind speed dataset from the Met Office (HadISD3), that global stilling is robust and not an artefact of analysis method or observational network characteristics. In contrast to a previous study, we find that atmospheric reanalyses fail to reproduce the observed changes in wind speed over the Northern Hemisphere land, Europe and China. By decomposing wind speed variability into contributions from sub-monthly and monthly timescales, we show that the error in capturing the global stilling trend is primarily due to trends in the sub-monthly component. The exception is the products from the Japanese Meteorological Agency (JRA55 and JRA55C), which reproduce global stilling over the Northern Hemisphere and Europe with high skill, though this appears to be attributable to correction when producing the dataset rather than being a fundamental property of the forecast systems used to produce the reanalysis. Our results highlight the importance of resolving sub-monthly wind variability in reanalyses and raise critical questions about the drivers of global stilling and the reliability of reanalysis-based prediction systems for understanding past and future variations of the near-surface wind speed.