Inconsistent wind speed trends in current twentieth century reanalyses

Reanalysis data underpin much research in atmospheric and related sciences. While most reanalysis only cover the last couple of decades, National Oceanic and Atmospheric Administration (20CR) and European Centre for Medium‐Range Weather Forecasts (ERA20C and CERA20C) also developed reanalyses for the entire twentieth century that theoretically allow investigation of multidecadal variability. However, the approaches adopted to handle the massively evolving number of observations can cause spurious signals. Here we focus on wind speeds, as its assimilation is a key difference among these two products. We show that ERA20C and CERA20C feature significant trends in the North Atlantic and North Pacific wind speeds of up to 3 m/s per century. We show that there is a good relation between the trends in the reanalysis and assimilated wind speeds. In contrast, 20CR and the European Centre for Medium‐Range Weather Forecasts free model run ERA20CM do not show positive trends in the same regions. As a consequence, conclusions drawn from any single twentieth century reanalysis should be treated cautiously in particular in sectors with a strong wind dependency (e.g., wind energy)