Relations consist of three parts. Let’s look at the following example:
coastDat-3_COSMO-CLM_ERAi has been forced with Era Interim.
The three parts that describe this relation are:
The left object (here coastDat-3_COSMO-CLM-ERAi)
the right object (here Era Interim)
The description of the relation (here has been forced with)
Each relation has an inverse, in this case the inverse is
Era Interim forces coastDat-3_COSMO-CLM-ERAi
Sometimes, a relation between two items can also be interpreted as a permission. If CoastDat is created by Hereon, the dataset owner of Coastdat implicitly grants the Hereon group the permission to list this dataset as one of theirs.
Roles within the model data explorer are purely considered from the metadata perspective (see Authors and Contact Persons) and describes how an author or datagroup has been involved in the creation of a dataset. When interpreting metadata of a dataset, e.g. by reading the ISO INSPIRE metadata (see Metadata of datasets), the model data might suggest to grant permissions to users based on the role that the user had in the dataset.
Link to interpretation of roles and permissions
Add Django graphs for user model
A user is a person with a login for the Model Data Explorer. A user can be uniquely identified from the email address. Every user is related to one specific author.
Add Django graphs for dataset model
A dataset (formerly known as model run) is a collection of variables with a temporal, spatial and optional vertical dimension. It can be the output of a single run of your model, or other raster data, e.g. derived from satellite measurements. A dataset might comprise multiple parameters. All data that is represented by one single dataset has been generated with one specific methodology and one specific set of input parameters.
In the language of CERA and CMIP6, one dataset in the Model Data Explorer corresponds to one experiment, e.g. coastDat-3_COSMO-CLM_ERAi in CERA or rcp45 in CMIP6.
Each dataset in the model data explorer will have a unique persistent identifier (PID) and a unique page where it’s possibilities (visualization on the map, statistical analysis, download, metadata are accessible).
One dataset in the model data explorer might correspond to - the output that you generated with one runscript of your model - an ensemble mean of multiple models and/or experiments - multiple realizations (ensemble members) produced with variations of the same scenario (in CMIP6: all data for a given model (e.g. MPI-ESM-LR) in a given experiment (e.g. rcp85)) - a single ensemble member with multiple variables for one CMIP6 scenario of one model with a specific forcing - data from a specific satellite derived with a specific method
Examples that would not be a dataset are: - a subset data, e.g. 1 out of 99 years of a model run (unless you want to throw away the other 98 years) - Multiple experiments (e.g. a combination of rcp26, rcp45, rcp60 and rcp85) of one model participating in CMIP6 (unless you only refer to ensemble statistics) - Multiple runs of the same model (unless they are follow-ups of each other, e.g. via restart files) - Data from multiple models (unless you only refer to the combined ensemble statistics)
Add Django graphs for datagroup model
A data group is a group of datasets (see above) that are managed by a certain set of users.
Each data group will have an own webpage and a unique handle where all the datasets are accessible and the metadata of the group is shown.
A data group is owned by a certain set of users (see below) and can have regular members.
Datasets can be associated with groups and provide edit and access rights.
a research center, e.g. Hereon
a department or institute, e.g. the Institute of Coastal Systems - Analysis and Modeling
a unit, e.g. Regional Land and Atmosphere Modeling
a project, e.g. CMIP6, MOSES, or MuSSEL