CFThe CF Conventions are the dominant metadata standard for scientific data stored in NetCDF files, providing a shared vocabulary that makes climate and geoscience datasets self-describing and interoperable. Recommended by Unidata and adopted by major data centers worldwide, CF has become the de facto standard for atmospheric, oceanic, and increasingly broader Earth science data exchange.
Background
The Climate and Forecast (CF) conventions emerged in the early 2000s from the need to standardize metadata in NetCDF files produced by climate and numerical weather prediction models. Prior to CF, different modeling centers used inconsistent variable naming and metadata practices, making cross-comparison of datasets difficult. The conventions built on earlier work such as the COARDS conventions (1995) and the GDT conventions, unifying and extending them into a comprehensive, community-driven standard. Development is entirely open, conducted through GitHub discussions and governed by a transparent process documented on the CF website.
Purpose & Scope
CF defines metadata that provides a definitive description of what the data in each variable represents, along with the spatial and temporal properties of the data. This includes coordinate systems, cell methods (describing how data values represent variation within cells), data packing, compression, and flags. The conventions enable users from different sources to determine which quantities are comparable and facilitate building applications with powerful extraction, regridding, and display capabilities.
While initially focused on gridded model output for atmosphere and ocean, CF has expanded to cover observations, satellite data, and other geoscience domains. The conventions apply to any data stored in NetCDF format and are increasingly used with other formats that can carry CF-style metadata.
Key Components
Standard Name Table -- The centerpiece of CF is its controlled vocabulary of standard names, each mapping a string identifier to a precise physical quantity with a canonical unit. The standard name table is maintained separately from the conventions document and is updated more frequently.
Area Type Table -- Defines controlled values for area types used in cell methods and other contexts.
Standardized Region List -- Provides canonical names for geographic regions.
Convention Rules -- Cover coordinate variables, dimensionality, cell boundaries, ancillary data, and discrete sampling geometries (for point, profile, trajectory, and timeseries data).
Governance & Maintenance
CF is developed and maintained by an open community with no single institutional owner. Governance is formalized through published rules covering how proposals are made, discussed, and decided. Changes are proposed as GitHub issues, discussed publicly, and require community consensus. A conventions committee and information management team oversee the process. The CF community meets regularly at workshops, with the most recent held in 2024.
The standard is released under CC0, placing it in the public domain and encouraging unrestricted adoption.
Notable Implementations
CF is used by virtually all major climate and weather data providers, including CMIP (Coupled Model Intercomparison Project), ESGF (Earth System Grid Federation), Copernicus Climate Data Store, NOAA, NASA, and the UK Met Office. Libraries such as cf-python, Iris, xarray, and nco natively understand CF metadata. The convention is the recommended metadata standard for data archived at CEDA, NCEI, and many other national data centers.
Related Standards
CF is designed specifically for the NetCDF format, which is maintained by Unidata at UCAR. The CF standard name table overlaps with and is sometimes aligned with GCMD keywords and WMO vocabulary efforts. For Earth observation validation data, GEOMS provides a complementary standard using HDF.