This data set, which is prepared for the Stratosphere–troposphere Processes And their Role in Climate (SPARC) Reanalysis
Intercomparison Project (S-RIP), provides several zonal-mean diagnostics
computed from reanalysis data on pressure levels. Diagnostics are currently
provided for a variety of reanalyses, including ERA-40, ERA-Interim, ERA-20C,
NCEP–NCAR, NCEP–DOE, CFSR, 20CR v2 and v2c, JRA-25, JRA-55, JRA-55C,
JRA-55AMIP, MERRA, and MERRA-2. The data set will be expanded to include
additional reanalyses as they become available. Basic dynamical variables
(such as temperature, geopotential height, and three-dimensional winds) are
provided in addition to a complete set of terms from the Eulerian-mean and
transformed-Eulerian-mean momentum equations. Total diabatic heating and its
long-wave and shortwave components are included as availability permits,
along with heating rates diagnosed from the basic dynamical variables using
the zonal-mean thermodynamic equation. Two versions of the data set are
provided, one that uses horizontal and vertical grids provided by the various
reanalysis centers and another that uses a common grid (CG) to facilitate
comparison among data sets. For the common grid, all diagnostics are
interpolated horizontally onto a regular

List of reanalyses represented in the S-RIP zonal-mean data set.

Reanalysis products are commonly used to study weather and climate variability and to validate climate models. By combining numerical forecast models and various observations through data assimilation procedures, reanalyses aim to produce a best estimate of the state of the atmosphere. However, differences among the model and assimilation components of reanalysis systems, as well as differences in the assimilated observations, result in different representations of the historical state and behavior of the atmosphere. These discrepancies contribute to uncertainties in our understanding of the atmosphere and its variability.

The Stratosphere–troposphere Processes And their Role in Climate (SPARC)
Reanalysis Intercomparison Project

The data set comprises two major components. The first component provides
variables on an original latitude–pressure grid defined by the
corresponding reanalysis center (original grid, OG). Note that this grid is
typically not defined by the model resolution, nor is it necessarily unique,
as some reanalysis products are distributed on a range of grids
(Table

The characteristics of this zonal-mean data set on pressure levels are
described in this paper. The reanalysis data sets included in the comparison
are listed and briefly described in Sect.

Vertical levels of the CG and OG data sets. Pressure levels provided in the OG data set are indicated with x and pressure levels provided in the CG data set are highlighted in bold-italic type.

The zonal-mean data set on pressure levels includes most major reanalysis
products (Table

Horizontal grid sizes in degrees for the reanalysis products used to produce
the OG zonal-mean data set are listed in Table

The highest pressure level provided is also listed in Table

With the exception of the model-generated diabatic heating rates (see
Sect.

Potential temperature is calculated on pressure levels as follows:

Differences in the preparation of the OG and CG data sets are illustrated in
Fig.

Flowchart illustrating differences in the calculation of diagnostics in the original grid (OG) and common grid (CG) data sets.

A three-point stencil is used to evaluate all derivatives. In the case of
meridional derivatives the three-point stencil is expressed as

The core of the data set consists of simple zonal-mean diagnostics.
Zonal-mean variables such as zonal wind, meridional wind, temperature, and
geopotential height are provided (Table

Core zonal-mean variables.

Several covariance terms are provided (Table

Covariance terms.

For the covariance terms of individual zonal wavenumbers,

The zonal-mean tendency of zonal wind

Eulerian-mean momentum diagnostics.

For the contribution of individual zonal wavenumbers,

TEM momentum diagnostics.

For the contribution of individual zonal wavenumbers,

Transformed-Eulerian-mean (TEM) momentum diagnostics (Table

The quasi-geostrophic (QG) version of the TEM equation

Zonal-mean diabatic heating rates generated by a subset of the reanalysis
forecast models are provided at 6 h resolution in units of K day

Model-generated diabatic heating diagnostics.

Diabatic heating is a fundamental component of the temperature budget, as
expressed by the thermodynamic equation in pressure coordinates:

A complementary set of heating rates is diagnosed using a subset of the
zonal-mean dynamical quantities introduced above. These heating rates have
been calculated based on analysis of the zonal-mean thermodynamic equation:

Diabatic and dynamical heating diagnostics.

Table

All diagnostics are provided on two distinct grids (see Fig.

Selected geopotential height contours on 1 January 1980 at
00:00 UTC for three isobaric surfaces: 1 hPa (top;

The impact of the grid transformation on variables provided in these data
sets is tested for some selected diagnostics. Figure

Vertical profiles of zonal wind averaged from 40 to 90

Vertical profiles of EP flux divergence averaged over 30 to
85

The differences between the CG and OG profiles shown in
Fig.

EP flux divergence as a function of latitude for the
1 hPa

Although zonal-mean quantities are largely insensitive to grid spacing and
interpolation, flux terms may be more sensitive. Figure

Figure

The sensitivity of momentum diagnostics to numerical resolution has been
evaluated separately in both horizontal and vertical dimensions by

Time series of total diabatic heating on the 50 hPa isobaric
surface based on

Overall, differences between the OG and CG diabatic heating diagnostics are
similar to those for other variables: very small in zonal-mean fields,
slightly larger for area averages, and typically much smaller than
inter-reanalysis differences. The latter two features are illustrated in
Fig.

Zonal-mean distributions of potential temperature tendencies
(K day

Figure

Figure

The S-RIP zonal-mean data set of reanalyses on pressure levels provides preprocessed zonal-mean diagnostics using unified NetCDF-4 classic file format. The main purpose of making this data set publicly available is to reduce the workload of researchers contributing to the S-RIP project by providing diagnostics that are commonly needed for reanalysis intercomparison, particularly in the middle atmosphere. The provision of preprocessed data will also save users the need to download and store dozens of terabytes of data. Producing the data set locally using a standardized set of computer codes ensures that the diagnostics are consistent among the reanalyses.

The dynamical

The S-RIP zonal-mean data set of reanalyses on pressure levels aims to facilitate the comparison of reanalysis data sets for the S-RIP community and the general atmospheric science community at large. In its current iteration, the data set includes 14 reanalyses and ancillary products from multiple research institutes. It covers the satellite era (1979–present) and extends backward in time to 1958 when data are available. Diagnostics provided include zonal-mean variables, diabatic heating, covariance and variance terms, and complete diagnostics from the Eulerian-mean and transformed-Eulerian-mean momentum equations. The diagnostics are provided on two grids, the original grid (OG) where diagnostics are performed on the original files acquired from each reanalysis center and the common grid (CG) where data are interpolated to a unified grid before advanced diagnostics are performed. The data set will grow in time to include more reanalyses and variables, as dictated by the evolving needs of the S-RIP community.

Websites and dates of access for core reanalysis variables.

Websites and dates of access for model-generated reanalysis diabatic heating products.

PM, JSW and NZ wrote the scripts to produce the data sets and prepared the figures shown in this paper. MF and PM worked towards archiving the data sets at CEDA. All authors contributed to the writing of this paper.

The authors declare that they have no conflict of interest.

This article is part of the special issue “The SPARC Reanalysis Intercomparison Project (S-RIP) (ACP/ESSD inter-journal SI)”. It is not associated with a conference.

We appreciate the support of the British Atmospheric Data Centre (BADC) of the UK Centre for Environmental Data Analysis (CEDA) for hosting the data set on their servers. We thank the reanalysis centers for providing support and access to data products. We also thank Amy Butler, Andrew Orr, Peter Hitchcock, and Edwin Gerber for bringing coding errors and missing data to our attention; James Anstey and the BADC for helping to secure storage space for archiving and processing the raw data; and Chi-Fan Shih for assistance with resolving data access problems. Masatomo Fujiwara's contribution was financially supported in part by the Japan Society for the Promotion of Science (JSPS) through Grants-in-Aid for Scientific Research (26287117 and 16K05548). Patrick Martineau was partially supported by the US CAREER grant AGS-1742178 that was awarded to Gang Chen through Cornell University and UCLA and by a Japan Society for the Promotion of Science Kakenhi grant. Jonathon Wright and Nuanliang Zhu were supported by the Ministry of Science and Technology of the People's Republic of China (2017YFA0603900) and the National Natural Science Foundation of China (41761134097). Edited by: Gabriele Stiller Reviewed by: two anonymous referees