Zarr Convention Specification

[mwengren]

Adding a reference to the nascent Zarr Conventions specification:

https://github.com/zarr-conventions/zarr-conventions-spec

Updates to ACDD might want to follow progress here and ensure alignment (i.e. that a new ACDD specification version be compatible or usable as a Zarr Convention).

Not clear what requirements this would add, but we should research it as a new ACDD is developed.

[mwengren]

There's a more comprehensive overview of how the Zarr conventions are designed here: https://github.com/zarr-conventions, including a table titled 'Share your Convention'.

Right now, conventions listed include:

• geo-proj
• spatial
• multiscales
• license
• uom

So, there's a model for how an ACDD convention might be added, if desired in the future. And, potentially some crossover/duplication already between attribution that's within scope of existing ACDD 1.3 (license).

I think reaching out to this community should yield some fruitful discussion on our path forward, at least in respect to Zarr, hopefully.

I don't think a future ACDD version should be specific to Zarr, and compatibility with other data formats should be a fundamental requirement, but a Zarr-convention flavor of ACDD might make sense.

[ethanrd]

Hi Micah @mwengren - GeoZarr is working within the Zarr conventions framework and is working (with CF involvement) to integrate with (or map between?) CF. Given the mutual netCDF history of CF and ACDD, there might be some good lessons learned in that effort (or vice versa).

[mwengren]

Thanks @ethanrd, it seems like this rather active thread from the past five days is highly-relevant:

https://github.com/orgs/cf-convention/discussions/461

The Zarr conventions I mentioned were referenced in some comments.

I have not read all that's been said here, but wanted to make the connection based on your comment. Hopefully someone with some spare time can read through that from an ACDD perspective to capture possible implications for our goals here.