The ECMWF's first atmospheric reanalysis of the 20th Century, ERA-20C, provides a long, gridded dataset of many climate variables, spanning 1900-2010. Observations assimilated include surface pressures from the International Surface Pressure Databank v3.2.6 and ICOADS v 2.5.1, and surface winds over the oceans from ICOADSv2.5.1. Upper-air and satellite data are omitted. Due to the limited observations used, the ERA-20C does not provide the best estimate of the atmospheric state for the better observed period since 1979; users investigating this period have several, more comprehensive reanalyses from which to choose.
Compared to the NOAA 20th-Century Reanalysis (20CR), ERA-20C is shorter, but provides finer horizontal and vertical resolution. Due to subtle differences in the data assimilated and the data assimilation schemes, ERA-20C may perform better than 20CR in relatively well observed regions such as western Europe, while 20CR may take better advantage of sparse observations in regions such as the Southern Hemisphere middle to high latitudes. See the "Expert Guidance" tab for further comparison of the 20CR and ERA-20C. ECMWF has also released an "AMIP style" version of the 20th Century reanalysis, known as ERA-20CM, which includes no data assimilation. ERA-20C and ERA-20CM are forced with the same time-varying SSTs, sea ice concentrations, and radiative forcings.
The following was contributed by Dr. Paul Poli at Météo-France, January, 2016:
This post explores the specificities of the two 20th century reanalyses from NOAA and ECMWF, respectively 20CR and ERA-20C.
Both reanalyses attempt to fairly represent the evolution of meteorology and climate throughout the 20th century. Like all reanalyses, they provide gridded datasets that cover many variables, and which can be sliced along any dimension of interest, thereby serving potentially many applications.
Neither of these two reanalyses continues in near-real-time. Also, by voluntarily omitting upper-air and satellite observations, neither reanalysis attempts to provide "best estimates" for the years when these observations are available in large numbers, which is approximately since 1957 for in situ upper-air observations, and since 1979 for satellite observations. Consequently users searching "best estimates" for those time periods should turn to more complete reanalyses, such as JRA-55, MERRA, CFSR, or ERA-Interim.
The global changes in observation coverage mean that regional trends suggested by these two reanalyses in the Southern hemisphere are at best to be treated with caution. In the Northern hemisphere, the best-observed areas throughout the record, and the areas downstream of them, are the most likely to feature reliable changes in circulation and precipitation.
The non-assimilation of any humidity observations leaves the analysis free of any increment that would affect the water cycle. Consequently, the trends in precipitation are mostly driven by the changes in forcings and atmospheric circulation, and not by sudden jumps in humidity observation coverage that would correct for biases in the atmospheric general circulation model. This means that the absolute level of precipitations may not be optimal in either of the two reanalysis, but the qualitative changes may be.
Although both reanalyses use very similar observation input -- only surface -- the subtle differences in the choices of variables assimilated, the thinning imposed by 20CR, and the quality controls of either data assimilation system means that the coverage in assimilated observations can differ significantly. For example, tropical cyclone bogus observations are largely rejected in ERA-20C because these observations do not fit with the large-scale expected background errors.
The 6-hour EnKF allows for fast-evolving and locally targeted background errors, and helps optimally analyze scarce observations in otherwise largely unobserved areas. This clearly advantages 20CR for the synoptic analysis for most of the record in the Southern hemisphere. In contrast, the 24-hour 4D-Var enables synergetic exploitation of the atmospheric model diurnal cycle and observations reported several times a day by stations or ships. This advantages ERA-20C in well-observed areas such as Western Europe, where the assimilation of winds also assists to better represent synoptic systems.
The availability of multiple analyses (56 ensemble members) from 20CR enables users to "flow" this uncertainty into their application. However, the absolute level of uncertainty may need tuning (e.g., inflating the spread), to match the estimate of uncertainty for the application in question. Such adjustment is not trivial unless the distribution of the ensemble member solutions is reduced to a parametric distribution, which negates the advantages of considering ensemble members separately.
The availability of a model integration dataset, ERA-20CM, to accompany the ERA-20C dataset, is key to assess whether changes in ERA-20C for a variable of interest results from the assimilation of observations.
The assimilation methodology of ERA-20C, compatible with present-day Numerical Weather Prediction ECMWF operations, makes it easier for future repeats of 20th century reanalyses at ECMWF to gradually incorporate additional observing systems such as in situ upper-air and satellite ones. ##
For data via NCAR RDA: European Centre for Medium-Range Weather Forecasts. 2014, updated daily. ERA-20C Project (ECMWF Atmospheric Reanalysis of the 20th Century). Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. http://dx.doi.org/10.5065/D6VQ30QG. Accessed† dd mmm yyyy.
Click the thumbnails to view larger sizes
Poli, Paul & National Center for Atmospheric Research Staff (Eds). Last modified 29 Mar 2017. "The Climate Data Guide: ERA-20C: ECMWF's atmospheric reanalysis of the 20th century (and comparisons with NOAA's 20CR)." Retrieved from https://climatedataguide.ucar.edu/climate-data/era-20c-ecmwfs-atmospheric-reanalysis-20th-century-and-comparisons-noaas-20cr.
Funding: NSF | National Science Foundation
Based at: NCAR | National Center for Atmospheric Research
A Project of: Climate Analysis Section in Climate and Global Dynamics Laboratory
Created by: Climate Data Guide PIs and Staff