Aidan
24 October 2025 03:30
2
Thanks @ClaireT !
I’m interested in standards compliance for global metadata and I’m working with @joshuatorrance to improve and standardise the global metadata of ACCESS model outputs.
Some related GitHub issues:
main ← esm16-output-file-schema
opened 11:00PM - 06 Oct 25 UTC
As part of the under-development ESM1.6 model output specification I have a new … draft of the new schema in a new branch and I'm contemplating adding some more CI checks for the schema.
I'm attempting to draft the new version of the schema with multiple files to hopefully simplify the specification and allow for easy changes in the future.
Still To do:
- [x] Check relative references/ids work as intended in a github repo particularly with tests etc. (they work locally...)
- I think this is working, no problems with a couple of validators although may need to specify $id in top level file in some circumstances (e.g. online validator)
- [ ] What should the new version of the schema be? Likely `2-0-0` or `1-1-0`
- [ ] What extra CI can/should be added?
- [x] json schema metavalidater, linter, formatter, tests
- I wasn't able to get pre-commit autoformatting working in my initial tests so it's just a CI test at the moment
- [ ] Is there a way to support schema versions using git commits/tags/releases/something rather than duplicate files in the repo?
- [ ] Should the version of the schema be mentioned explicitly in each schema/sub-schema file?
- Given the number of files this would need to be automated
- [ ] Remove title from attributes? Attribute key/name is the same string.
- [ ] details of global metadata schema need reviewing
- [x] `realm` should be simplified to a single `realm` per data file
- [x] `creation_date` needs a look (for now follow cmip/cordex)
- [ ] How should `id` work? `experiment_id` (CMIP) or just `id` (ACDD)?
- [x] Remove `_Format` & `_IsNetCDF4`?
- [ ] Do we want to dictate the order of the attributes? There's no way to do this with jsonschema, would need to do it elsewhere.
- [ ] What else?
- [x] variable metadata
- [x] Can this be defined with a unknown variable name?
- [x] Make nothing required since we don't know if a given .nc variable is a coordinate or data?
- [x] Build auto-generating user docs from schema (likely in [this](https://github.com/ACCESS-NRI/ACCESS_Output_data_specifications) repo)
- drafted [here](https://github.com/ACCESS-NRI/ACCESS_Output_data_specifications/pull/2)
opened 05:55AM - 16 Apr 25 UTC
In some cases there are modifications to metadata for model output files that is… required.
Best case would be for the model to include this metadata, but it can be difficult/impossible to enable this.
It would be good if there was a standard way to modify metadata, with a standard way to configure it.
One option is to resuscitate the [addmeta](https://github.com/coecms/addmeta) project. Either call this in a `userscript`, or incorporate this functionality directly in `payu`.
I need a tool to use in CI pipelines to check standards compliance.
The last release of cf-checker was four and half years ago. Maybe it’s feature complete, but it seems there isn’t much resourcing for support .
Maybe compliance checker is a better option, but it is quite verbose and I get the impression from their docs they think it is a tool that should be run by a person.