Within the AtMoDat project (Atmospheric Model Data, www.atmodat.de), a
standard has been developed which is meant for improving the FAIRness of
atmospheric model data published in repositories. Atmospheric model data
form the basis to understand and predict natural events, including
atmospheric circulation, local air quality patterns, and the planetary
energy budget. Such data should be made available for evaluation and
reuse by scientists, the public sector, and relevant stakeholders.
Atmospheric modeling is ahead of other fields in many regards towards
FAIR (Findable, Accessible, Interoperable, Reusable, see e.g. Wilkinson
et al. (2016, doi:10.1101/418376)) data: many models write their output
directly into netCDF or file formats that can be converted into netCDF.
NetCDF is a non-proprietary, binary, and self-describing format,
ensuring interoperability and facilitating reusability. Nevertheless,
consistent human- and machine-readable standards for discipline-specific
metadata are also necessary. While standardisation of file structure and
metadata (e.g. the Climate and Forecast Conventions) is well established
for some subdomains of the earth system modeling community (e.g. the
Coupled Model Intercomparison Project, Juckes et al. (2020,
https:doi.org/10.5194/gmd-13-201-2020)), other subdomains are still
lacking such standardisation. For example, standardisation is not well
advanced for obstacle-resolving atmospheric models (e.g. for urban-scale
modeling). The ATMODAT standard, which will be presented here, includes
concrete recommendations related to the maturity, publication, and
enhanced FAIRness of atmospheric model data. The suggestions include
requirements for rich metadata with controlled vocabularies, structured
landing pages, file formats (netCDF), and the structure within files.
Human- and machine-readable landing pages are a core element of this
standard and should hold and present discipline-specific metadata on
simulation and variable level.