Guidelines_netcdf

My guidelines for netCDF files: what kind of attributes / parameters are needed, usual dimension names, etc. The motivation for this repository is that I always end up struggling to find the necessary instructions / list of points to follow to package my netCDF files following CF conventions. Strongly fitted to / following what is recommended by adc.met.no and inspired from the content there.

The main points to follow to be compliant / easy to reuse is to i) use well formed netCDF files, ii) follow metadata conventions. There are different kinds of metadata:

ACDD (Attribute Conventions Dataset Discovery) is discovery metadata, used to search for datasets / nc files,
CF (Climate and Forecast metadata conventions) is use metadata, used to understand the content of the files.

Note that netCDF files can be edited (to change dimensions / variables names, attributes names and contents, add / remove any of these) separately from the data they contain using the tooling from netCDF Operator (NCO):

https://nco.sourceforge.net/ (documentation), https://github.com/nco/nco (code development), among others (see there for more information):
- ncatted netCDF ATTribute EDitor
- ncrename netCDF RENAMEer

This makes it easier to edit a file to make it netCDF-CF+ACDD compliant.

Resources about metadata

ACDD metadata

homepage: https://wiki.esipfed.org/Category:Attribute_Conventions_Dataset_Discovery
- current release: http://wiki.esipfed.org/index.php/Attribute_Convention_for_Data_Discovery
- for example, 1-3 attributes: https://wiki.esipfed.org/Attribute_Convention_for_Data_Discovery_1-3

CF metadata

homepage: http://cfconventions.org/index.html
- latest working draft of the conventions (link on the homepage): http://cfconventions.org/cf-conventions/cf-conventions.html
list of standard variable names:
- http://cfconventions.org/Data/cf-standard-names/current/build/cf-standard-name-table.html
- not sure what the difference is, but another possible source: https://vocab.nerc.ac.uk/standard_name/
misc
- files should end in .nc
- when a variable has a dependence in date or time (T), height or depth (Z), latitude (Y), longitude (X), recommend to order the dimensions as (T, Z, Y, X)
- if variables provide information about another variable, use ancilliary variable description; see "Ancillary data" in http://cfconventions.org/cf-conventions/cf-conventions.html

Metadata checkers

Note: on my machine, I have installed all compliance checkers in a conda env:

~> eval "$(/home/jrmet/miniconda3/bin/conda shell.bash hook)"
(base) ~> conda activate netcdf_utils
(netcdf_utils) ~> cfchecks -h
 cfchecks [OPTIONS] file1 [file2...]
...
(netcdf_utils) ~> cchecker.py -V
IOOS compliance checker version 5.0.2
(netcdf_utils) ~> ncrename --help
...
(netcdf_utils) ~> ncatted --help
...