Direkt zum Inhalt | Direkt zur Navigation

Personal tools

You are here: Home » Data » ENES Model Data and Metadata » CORDEX » CORDEX Data Structure

CORDEX Data Structure

Domains, experiments, variable names and other centralized properties

Climate simulation data are frequently needed for a limited spatial region but in a finer resolution than Global Climate Models (GCM) can provide. CORDEX tries to fill the gap in a systematic manner. CORDEX data were calculated using Regional Climate Models (RCM) together with a technique called dynamical downscaling. A GCM, the driving model, is used to embed the domain in a simulated global climate, i.e. data from a global cliomate simulation is used as input for a RCM. For this forcing, usually 6hr or 3hr GCM data at the boundary of the domain is used, the so-called forcing data. In the CORDEX project, most of the forcing data are taken from CMIP5. Up to now, this dynamical downscaling has been used in the CORDEX project; utilization of statistical methods (statistical downscaling) shall follow soon.

Domains and grid resolution

Simulation output for 13 domains is in preparation or already available, e.g. for the EUR domain (Europe). The number behind the domain abbreviation is the grid resolution in the native CORDEX simulation. Native EUR CORDEX simulations have a grid resolution of 0.44 or 0.11 degrees (EUR-44, EUR-11) and rotated poles.

To facilitate the use of CORDEX data, also data on normal grids are available or will be calculated. The poles have been rotated back and the grid resolution has slightly been modified. Since these data have been calculated by interpolation, an i has been added to the domain name, e.g. EUR-44i. EUR-44i data have a grid resolution of 0.5 degrees, EUR-11i data a resolution of 0.125 degrees, latitude and longitude. Interpolation will usually be done for monthly and semi-annual data only.


Corresponding to the forcing data taken from CMIP5, CORDEX experiments have the same names as in the CMIP5 project. Exception is the evaluation experiment, which is forced by reanalysis data.

evaluation forced by ERAINT data from ECMWF, for testing of regional models with observational data
historical 1950-2005 hindcast, forced by CMIP5 historical
rcp26 Representative Concentration Pathways for 2006-2100, forced by the corresponding CMIP5 RCPs

Originally, AMIP and three decadal experiments (initialized in 1980, 1990 and 2005) were also in the list of projected CORDEX experiments. It is unclear if these additional experiments will be realized.


CMIP5 calculations have been done for an ensemble of starting states. The names of the ensemble members in CORDEX are that of the CMIP5 data used for forcing.

Centralized variable names

CORDEX variable names have been overtaken from CMIP5 unless a variable has been calculated for several altitudes. In the latter case the pressure level in hPa is appended to the CMIP5 short variable name, e.g. ta500 for the air temperature at the 500 hPa level. CORDEX variables are only a subset of CMIP5 variables since CORDEX RCM have less model components, e.g. usually no ocean model.

Centralized file format