CERRA, the Copernicus European Regional Reanalysis system

Abstract

AbstractA regional reanalysis has been produced for a domain covering entire Europe from 1984 to 2021. The reanalysis is produced as part of the Copernicus Climate Change Service. The Service provides the high‐resolution deterministic Copernicus European Regional Reanalysis (CERRA), run at a horizontal resolution of 5.5 km, a 10‐member ensemble run at 11‐km resolution as well as an offline surface analysis, CERRA‐Land. CERRA‐EDA uses an ensemble data assimilation (EDA) technique to perturb the initial condition of the different members. Apart from the horizontal resolution the CERRA and CERRA‐EDA setups are the same; for example, the same data assimilation scheme, same physics parameterization as well as the same vertical levels. These new systems are built from HARMONIE cy40 version, including some back‐phased physics from a newer model version (cy42). Conventional observations, satellite‐based radiances, atmospheric motion vector winds and bending angle from radio occultation observations are used. In addition, ground‐based zenith total delay (ZTD) from global navigational satellite systems (GNSS) and local surface observations, rescued from historical archives at the local National Meteorological Services, are used. Another new feature is the construction of the background error statistics for the data assimilation. Information from the ensemble run, CERRA‐EDA, is used in the derivation of the background error statistics for the high‐resolution CERRA runs. These background error statistics are updated every second day. By doing so, daily environment variation is taken into account as well as all variations over the 37 years of production. The reanalyses and reforecasts from CERRA show an added value compared to the global ERA5 for almost all variables at the surface level. This becomes particularly clear when selecting smaller areas with complex terrain where the high resolution is beneficial. In the free atmosphere it is primarily the analyses and short forecasts, 3–6 hours, that give an added value.