Recipe: Format and file organization

Back to G5NR Data Access Guide.

GEOS-5 files are generated with the Network Common Data Form (NetCDF-4) library, which uses Hierarchical Data Format Version 5 (HDF-5) as the underlying format. NetCDF-4 is an open-source product of UCAR/Unidata (https://www.unidata.ucar.edu/software/netcdf/) and HDF-5 is developed by the HDF Group (http://www.hdfgroup.org/). One convenient method of reading GEOS-5 files is to use the netCDF library, but the HDF-5 library can also be used directly. These files can be easily read by applications such as IDL, Matlab, GrADS, FERRET, NCL, Panoply as well in Python using the netCDF4, h5py or PyTables packages.

Each GEOS-5 file contains a collection of geophysical quantities that we will refer to as “fields” or “variables” as well as a set of coordinate variables that contain information about the grid coordinates. While the coordinate variables are COARDS and CF-1.0 compliant, the metadata associated with the data variables may not strictly meet all recent CF requirements.

All products are chunked and internally compressed with a GZIP-based method that is transparent to the user. This method degrades the precision of the data, but every effort has been made to ensure that differences between the product and the original, uncompressed data are not scientifically meaningful. Once the precision has been degraded, the files are written using the standard GZIP deflation available in NetCDF-4. When reading these files the NetCDF-4 Library will automatically decompress these files without any specific input from the user.