cfconventions.org

One-sentence overview

cfconventions.org is an open-source metadata standards project maintained by the international climate research community. Its full name is the “Climate and Forecast Metadata Conventions.” It is not a commercial product, but a set of rules for naming and describing climate, ocean, and atmospheric data in NetCDF files. Developers use it because it helps keep variable names, units, and time coordinates consistent across different models, observation stations, and reanalysis datasets, allowing data exchange and visualization tools such as Panoply and Matplotlib to read them directly without manual translation.

Business details

The core role of cfconventions.org is to maintain and publish the CF standards documentation — currently the latest version is CF-1.11 — and to provide online lookup and downloads for the Standard Name Table. The standard was initiated by the international atmospheric and ocean science community in the early 2000s. It is currently hosted by a research team at the UK Met Office and accepts change requests from the global community via its GitHub repository. In terms of industry status, it is the de facto international standard in the climate data field: almost all major climate model outputs, such as CMIP6 and CORDEX, as well as satellite remote sensing products such as NASA’s MODIS and NOAA’s AVHRR, are required to follow CF conventions. Its users include climate scientists, data engineers at meteorological agencies, data analysts at environmental consulting firms, and graduate students in university Earth science departments.

Who it is for

• Climate/ocean data developers: If you need to write scripts to read NetCDF files, the CF standard saves you the trouble of manually interpreting what variables mean.
• Data publishing organizations: Meteorological agencies, environmental agencies, and similar institutions need to ensure that their data can be used directly by international peers and avoid creating “data silos.”
• Open-source tool maintainers: Libraries such as xarray, CDO, and NCL rely heavily on CF conventions, so understanding the standard is helpful for contributing code.
• Not suitable for: Pure business users, such as climate policy makers who only read final reports; developers who do not work with the NetCDF format; or commercial users who need a real-time weather API.

Key features and highlights

• Standard Name Table: Provides 5000+ registered names for physical quantities, such as “air_temperature” and “sea_surface_salinity,” along with units, descriptions, and GRIB code mappings.
• Metadata constraint rules: Specifies dimension names such as “time,” “lat,” and “lon,” coordinate variable attributes such as units: “days since 1900-01-01”, and conventions for marking missing values.
• Grid mapping conventions: Supports metadata descriptions for irregular grids, such as rotated latitude-longitude grids and triangular grids, which are required by many climate model outputs.
• Version control and community voting: All standard name updates must go through GitHub Issue discussions and CF meeting votes, ensuring authority and consistency.
• Open source with no licensing restrictions: The documentation and Standard Name Table use CC0 or similar licenses and can be freely embedded in commercial software or academic papers.
• Deep integration with NetCDF: Almost all tools that support NetCDF include CF validation features, with error messages pointing directly to the relevant CF documentation.

Pricing analysis

In terms of pricing, cfconventions.org is completely free. All standards documents, name tables, and tools are provided as open source, with no subscription or licensing fees required. Among comparable standards, it sits in the “zero-cost” tier. Compared with commercial weather data services such as IBM Weather Data or AccuWeather APIs, which charge based on usage, the CF standard itself generates no direct cost. However, there may be hidden costs in practice: for example, you need to spend time learning NetCDF syntax and CF rules; if your data does not comply with the CF specification, manually correcting metadata can also require human effort. In short, there are no hidden fees, but the learning curve does require time investment.

How Chinese users can use it

• Network accessibility: cfconventions.org is generally accessible directly from mainland China, with relatively fast page loading and no need for a VPN. The GitHub repository used for submitting change requests may occasionally be unstable, but downloading standards documents is not affected.
• Payment methods: No payment is required, so Alipay or WeChat Pay are not relevant.
• Need for a VPN: Only GitHub’s Issue system may occasionally require circumvention, depending on the network environment. The main website and Standard Name Table lookup are fully usable.
• Domestic alternatives in China: The China Meteorological Administration has introduced meteorological data metadata standards, such as the QXT series, but their international compatibility is weaker. Common domestic tools such as the MICAPS system also use similar internal conventions, but these are not public standards. For teams that need to collaborate with the international community, the CF standard remains the preferred choice.
• Invoices: Since there is no transaction, invoices cannot be issued. If a company needs reimbursement internally, it may consider paid support services for CF-based open-source tools such as xarray, but cfconventions.org itself does not provide them.

Pros and cons

Pros:

• ✅ Completely free, with no licensing restrictions; suitable for academic teams with limited budgets.
• ✅ Extremely high international recognition and compatible with major projects such as CMIP and CORDEX.
• ✅ Community-driven, with timely updates; the Standard Name Table is updated 2-3 times per year.
• ✅ Detailed documentation, with example NetCDF files for reference and learning.
• ✅ Naturally integrated with mainstream toolchains such as xarray, NCL, and CDO.

Cons:

• ❌ Steep learning curve: beginners need to understand the NetCDF data model and attribute syntax.
• ❌ No technical support: issues can only be handled through mailing lists or GitHub Issues, and response speed depends on community activity.
• ❌ The Standard Name Table is not fully comprehensive: variables in some emerging fields, such as urban climate and biogeochemistry, may not yet be registered.
• ❌ Not friendly to non-NetCDF formats: users working with HDF5 or Zarr need to convert metadata separately.
• ❌ Documentation is in English only, with no official Chinese translation, creating a barrier for non-English users in China.

Comparison with similar products

• GRIB coding standard (WMO): Used for operational weather forecasting and more focused on binary formats rather than metadata description. CF emphasizes self-describing data, while GRIB focuses more on compression and transmission efficiency.
• ISO 19115 geographic information metadata: Broader in scope, covering remote sensing and maps, but less detailed than CF for climate-specific data. CF is “small but focused,” while ISO is “broad and comprehensive.”
• NCAR’s NCL built-in conventions: NCL has its own metadata rules, but they are highly compatible with CF. CF’s advantage is cross-tool interoperability, while NCL rules are more oriented toward its own syntax.

Summary and recommendation

Best suited for: If you are developing a climate dataset that needs long-term archiving or international exchange, such as regional climate model outputs or historical observation compilations, it is strongly recommended to follow the CF standard from the beginning. This can save significant costs in later data cleaning. For supplementary materials in academic papers, using the CF standard allows reviewers to validate the data directly. Not suited for: If you are only processing a few NetCDF files temporarily and do not care about reproducibility, or if your data is only used inside a closed domestic system in China, there is no need to invest time learning it. Recommendation: First read the “quick start” section of the CF documentation, which is about 20 pages, then try using xarray’s open_dataset to read a standard CF file, which can be downloaded from an ESGF node, and check whether you can parse it correctly. No payment is required; you can use it directly.