Filenames of upload data MUST follow the
CORDEX archive specifications. This is crucial since an automatic program identifies the content of each file in the incoming path from its filename. These info are then used to fill the relational database, to create the right outgoing path and to move there the file. Threfore files with a wrong filename structure will not enter tha data base and will be left in the incoming path where its producer uploaded it.
Variable names of oceanic fields are not part of the CORDEX specifications, thus for them we use the IPCC variable names. (see also
Climate and Forecast Standard Names of the NERC Vocabulary Server of the
Each file contains data of a single variable, produced for a given realm (ocean or atmosphere) over a given geographic area (domain), by a given GCM, within a given CMPIP5 experiment, a given CMPIP5 Ensamble, by a given version of a given RCM, having output with a certain frequency and, if applicable, covering a given period of time.
Name of the data files are made of the following tokens:
VariableName_Domain_GCMModelName_CMIP5ExperimentName_CMIP5EnsembleMember_
RCMModelName_RCMVersionID_Frequency[_StartTime-EndTime].nc
where the square brackets indicates a token which can also be missing, depending on the values of the other tokens .
Files are placed into a directory structure named MEDCORDEXPATH having the following scheme:
/domain/institution/GCMModelName/
CMIP5ExperimentName/CMIP5EnsambleMember/RCMModelName/
RCMVersionID/Frequency/VariableName
Obviously each file is placed in the MEDCORDEXPATH made of the same tokens as the file. For example the file
tas_MED-44_ECMWF-ERAINT_evaluation_r1i1p1_CNRM-ALADIN52_v1_mon_199101-200012.nc
uploaded by CNRM will be moved into the following MEDCORDEXPATH:
/MED-44/CNRM/ECMWF-ERAINT/evaluation/r1i1p1/CNRM-ALADIN52/v1/mon/tas/
As an extension of the CORDEX archive specification, a special directory named /ALL has been created. It is a flat directory with links to all the netcdf files present in the whole MEDCORDEXPATH. Using this flat directory is much easier to locate files, being enough to name them and not also the MEDCORDEXPATH. /ALL is not by means listable and its files are available only to who knows their name.
At the moment the HyMeX database file structure is as follow:
- /HYMEX/area/experiment/institution/CORE
where area is any of atmosphere, ocean, land
where experiment is any of TTM3a-lop, TTM3b-HINDCAST, TTM3c-SCENARIO, TTM3de-TEST
where institution is any of the above institution shortnames (those in red)