Frequently Asked Questions (FAQ)#

How should I cite EOForestSTAC?#

Please use the following citation when referencing EOForestSTAC in your work:

Besnard, S. (2025). EOForestSTAC: A toolbox for accessing the EO forest data catalog. GitHub repository: simonbesnard1/eoforeststac

How do I find out which collections and versions are available?#

Use DiscoveryProvider:

from eoforeststac.providers.discovery import DiscoveryProvider

disc = DiscoveryProvider(
    catalog_url="https://s3.gfz-potsdam.de/dog.atlaseo-glm.eo-gridded-data/collections/public/catalog.json",
    endpoint_url="https://s3.gfz-potsdam.de",
    anon=True,
)

disc.list_themes()
disc.list_collections(theme="biomass-carbon")
disc.list_versions("CCI_BIOMASS")

Or browse the interactive STAC Browser at https://simonbesnard1.github.io/eoforeststac/.

I get a ValueError: Asset 'zarr' not found. What is wrong?#

Some products (e.g. GAMI_AGECLASS) are stored at multiple spatial resolutions. Their assets are named zarr_1deg, zarr_0.25deg, etc. rather than just zarr. Pass the resolution= argument:

ds = provider.open_dataset("GAMI_AGECLASS", "3.0", resolution="0.25deg")

The error message also lists the available resolution keys.

Why is loading data slow?#

Data is streamed from GFZ Ceph S3 over HTTPS. The first call to .compute() fetches only the chunks that fall within your requested spatial and temporal subset. To minimise data transfer:

  1. Subset spatially and temporally before calling .compute().

  2. Load only the variables you need: ds[["agb"]].compute().

  3. For repeated access to the same region, consider caching locally with xr.Dataset.to_zarr().

What coordinate names do datasets use?#

All products follow CF conventions: spatial dimensions are named latitude and longitude; the time dimension is time. Coordinates carry units, long_name, and standard_name attributes.

Can I load data without an internet connection?#

No. All data is hosted on GFZ Ceph S3. The package does not support local file caching natively. Use xr.Dataset.to_zarr() or xr.Dataset.to_netcdf() to save a subset locally after loading.

Can I use EOForestSTAC from a Jupyter notebook?#

Yes. The package works in any Python environment. For interactive notebooks, install jupyter and Dask’s dashboard:

pip install jupyter dask[distributed]

Then enable the progress bar in a notebook:

from dask.distributed import Client
client = Client()   # starts a local cluster with a dashboard

How do I contribute to EOForestSTAC?#

Contributions are welcome! See Contributing to EOForestSTAC for guidelines on submitting bug reports, feature requests, and pull requests.

How do I add a new product to the catalog?#

Adding a new product requires three steps:

  1. Product config (eoforeststac/products/<product>.py): define spatial extent, temporal extent, keywords, license, and asset template.

  2. Catalog module (eoforeststac/catalog/<product>.py): create the STAC collection and items using the factory functions.

  3. Writer (eoforeststac/writers/<product>.py): implement BaseZarrWriter to read raw input data and write an analysis-ready Zarr store.

See the CCI_BIOMASS or GAMI_AGECLASS implementations for a complete example.