ACCESS-NRI Intake catalog: a way to find, load and share data on Gadi

Catalog Update

The default ACCESS-NRI Intake Catalog now uses a file id system based on the grids on which variables are defined, rather than filenames. This system is substantially simpler than the previous system, with the following benefits:

  • User created catalogs (via eg. use-esm-datastore) should now automatically group all assets into xarray datasets correctly, regardless of how the files are named
  • There are now fewer distinct datasets per experiment, with these datasets grouped by grid. This should massively reduce the likelihood of needing to experiment with search keys until a single dataset can be obtained.
  • It enables us to begin virtualising experiment output, which should greatly improve performance.

Warning

As a result of this update, file_id’s in most esm-datastores in the ACCESS-NRI Intake Catalog will have changed. If you need to use a catalog with the old style file id’s, you can do so via:

import intake
intake.cat.access_nri(version="v2025-07-30")

which will load the latest catalog built with the old file id system.

User created catalogs will not have been affected by this change.