The ERA5 Reanalysis

by Hans Hersbach (ECMWF)



The C3S ERA5 climate reanalysis provides detailed hourly snapshots of the Earth’s atmosphere, land surface and ocean waves from 1950 with timely updates with a latency of 5 days. It is a highly popular product on the Climate Data Store (CDS) and is used for a wide range of applications. To date there are over (ERA5-Land included) 46,000 users worldwide and around 400 Terabyte is downloaded every single week.

ERA5 uses a version of the ECMWF Numerical Weather Prediction forecast model and data assimilation system from 2016 (Cy41r2) to assimilate both in-situ and satellite observations, many of which stem from reprocessed data records. This presentation provides a concise overview of the ERA5 system and a basic evaluation of characteristics and performance.

Special focus will be on the ERA5 back extension, the segment from 1950 to 1978 that was made available in autumn 2020 and supplements the previously published segment covering 1979-present. It features the assimilation of many conventional observations, as well as improved use of early satellite data. The fidelity of the extension is illustrated by the accurate depiction of the North Sea Storm of 1953, and the events leading to the first discovery of Sudden Stratospheric Warmings in 1952. Time series of ERA5 global surface temperature anomalies show temperatures to be relatively stable from 1950 until the late 1970s, in agreement with the other contemporary full-input reanalyses covering this period and with independent datasets, although there are significant differences in the accuracy of representing specific regions. The evolution of upper air temperatures, humidities and winds show smoothly varying behaviour, including tropospheric warming and stratospheric cooling, modulated by volcanic eruptions. The Quasi Biennial Oscillation is well represented throughout.

Despite these good characteristics, the ERA5 back extension did use historical observations for tropical cyclones in a sub-optimal way, which, given their sparsity is a challenging subject in itself. As a result, a number of tropical cyclones are far too intense. Although this has an occasional local impact, it does affect the consistency of extreme events and ocean waves on larger scales. For these reasons it was decided to publish the back extension as a separate, preliminary data set in the CDS, while preparations for an improved version of the back extension have been taking place.