Olmoearth Pretrain: Satellite Imagery for Remote Sensing Foundation Models

A pre-training dataset for the OlmoEarth remote sensing foundation models, created by AllenAI. The dataset includes Sentinel-2 L2A and Sentinel-1 GRD IW imagery from the European Space Agency. It was last updated on November 1, 2025.

Use Cases

• Pre-training remote sensing vision models based on Sentinel-2 L2A imagery.
• Developing models for synthetic aperture radar (SAR) analysis based on Sentinel-1 GRD IW vv+vh imagery.
• Training multimodal geospatial models that combine optical and radar data sources.
• Benchmarking foundation model performance on Earth observation tasks.

Strengths

• Includes data from authoritative sources like the European Space Agency's Sentinel program.
• Released under the CC BY 4.0 license, which permits sharing and adaptation.
• Documentation is available on a dedicated GitHub repository.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and exact size are unknown, which may limit suitability assessment.
• Data may reflect geographic or temporal bias inherent to the Sentinel satellite coverage.

Provenance

Source

European Space Agency (Sentinel-1 and Sentinel-2 data), compiled by AllenAI.

Collection Method

Likely aggregated and processed from open satellite data sources for model training.

Freshness

Last updated 2025-11-01 04:25:39; freshness should be verified.