Skip to content

Export virtual datasetsΒ #1196

@TomNicholas

Description

@TomNicholas

Earthaccess allows users to search CMR, then combine the results efficiently into a single datacube using VirtualiZarr behind the scenes.

But after you do this work, there's no way to persist it! IIUC earthaccess.open_virtual_mfdataset returns a non-virtual dataset, i.e. actual loadable lazy numpy arrays, not virtualizarr.ManifestArrays. So even though internally it assembled a virtual datacube, there is no way to serialize those virtual refs to a persistent format such as kerchunk/icechunk.

I think that would be very useful, because then users could create Icechunk stores/Kerchunked Zarr stores directly from CMR, for later use even without needing to call earthaccess again on every access!

(But for it to work with Icechunk would also require Earthdata Login support in Icechunk.)

cc @jbusecke

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    Status

    πŸ†• New

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions