Overview: Why EOForestSTAC?#

EOForestSTAC is a Python package built to simplify access to global forest Earth Observation datasets hosted on cloud-native object storage. It exposes a curated STAC catalog of analysis-ready Zarr archives and provides a clean Python interface for discovery, loading, subsetting, and combining multiple products.

The motivation behind EOForestSTAC#

Working with large-scale forest EO datasets presents several challenges:

  • Fragmented products: biomass, disturbance, canopy height, forest age, and land-use datasets are published by different groups in different formats, often as multi-gigabyte GeoTIFFs.

  • Version proliferation: most products have multiple versions and spatial resolutions; tracking what is available and what is current requires manual bookkeeping.

  • Download overhead: traditional workflows require downloading complete datasets before any analysis, even when only a small region and time range are needed.

  • Grid inconsistency: combining products from different sources requires reprojection and resampling to a common grid.

EOForestSTAC was designed to address all of these by combining a STAC metadata catalog with Zarr cloud-native storage, enabling lazy streaming access with no local downloads.

What EOForestSTAC enables#

  • Zero-download access: datasets are streamed from Ceph S3 directly into an xarray.Dataset; only the requested spatial and temporal subset is transferred.

  • Unified catalog: all products, versions, and spatial resolutions are registered in a single STAC catalog browsable via Python or the interactive STAC Browser.

  • Multi-resolution support: products like GAMI age-class fractions are available at five spatial resolutions; the resolution= parameter selects the right asset automatically.

  • Spatial subsetting: subset() clips any dataset to a GeoDataFrame geometry with automatic CRS reprojection.

  • Multi-product alignment: DatasetAligner reprojects and resamples multiple datasets onto a common grid in one call.

  • Lazy evaluation: all operations are backed by Dask and xarray; computation is deferred until .compute() is called.

Catalog structure#

The catalog is organized as a three-level hierarchy:

catalog.json  (root)
├── biomass-carbon/          ← theme catalog
│   ├── CCI_BIOMASS/         ← collection (one per product)
│   │   ├── CCI_BIOMASS_v6.0 ← item (one per version)
│   │   └── CCI_BIOMASS_v7.0
│   ├── SAATCHI_BIOMASS/
│   └── LIU_BIOMASS/
├── disturbance-change/
│   ├── EFDA/
│   ├── HANSEN_GFC/
│   └── ...
├── structure-demography/
│   ├── POTAPOV_HEIGHT/
│   └── GAMI_AGECLASS/       ← multi-resolution: one item, 5 resolution assets
└── ...

Each item holds one or more Zarr assets (one per spatial resolution for multi-resolution products). The ZarrProvider and DiscoveryProvider classes navigate this hierarchy transparently.

Core components of EOForestSTAC#

  1. eoforeststac.providers.discovery.DiscoveryProvider: navigates the STAC catalog hierarchy to list themes, collections, versions, and asset keys.

  2. eoforeststac.providers.zarr.ZarrProvider: opens a specific dataset version as an xarray.Dataset streamed from Ceph S3.

  3. eoforeststac.providers.subset.subset: clips a dataset to a spatial geometry and optional time range.

  4. eoforeststac.providers.align.DatasetAligner: reprojects and resamples multiple datasets onto a shared reference grid.

Available product themes#

Theme

Products (examples)

biomass-carbon

CCI Biomass, Saatchi Biomass, Liu Biomass, GAMI

disturbance-change

EFDA (forest disturbances), Hansen GFC, RADD Europe, JRC TMF, JRC GFC, Robinson Carbon Recovery

structure-demography

Potapov Canopy Height, GAMI Age-Class Fractions (multi-resolution)

land-use

Potapov LCLUC, RESTOR Land Use, ForestPaths Genus

Goals and aspirations#

EOForestSTAC’s primary objective is to provide a reproducible, interoperable platform for forest EO research, including:

  • Multi-source synthesis: load biomass, disturbance, canopy height, and forest age data onto a common grid for integrated analyses.

  • Temporal change detection: access all available versions of a product and compare them with .sel(time=...).

  • Carbon cycle studies: combine multiple products to quantify forest carbon stocks, changes, and uncertainty.

—

Acknowledgments#

The development of EOForestSTAC was supported by the European Union through the FORWARDS project.

The author thanks Konstantin Ntokas (Brockmann Consult) for technical assistance, and the EO-LINCS project for valuable feedback on data structure and improvements to the STAC catalog design.

—

A project from GFZ Potsdam#

EOForestSTAC was developed by Simon Besnard from the Global Land Monitoring Group at the Helmholtz Centre Potsdam GFZ German Research Centre for Geosciences. The project applies cloud-native data principles established in the gedidb and icesat2db packages to global gridded forest EO products. It is open-source (EUPL-1.2) and welcomes contributions from the research community.

—