FM4CS — Foundation Models for Climate & Society

The FM4CS project developed a multimodal foundation model (THOR) for Earth Observation (EO) using large-scale self-supervised learning to reduce reliance on labeled data. It natively supports Sentinel-1 (SAR), Sentinel-2 (MSI), and Sentinel-3 OLCI & SLSTR, with resolutions ranging from 10 m to 1000 m. Benchmarks across diverse EO tasks show strong robustness and adaptability, establishing FM4CS as a reliable base for climate and society applications.

Partners: The Danish Meteorological Institute, the National Meteorological Administration of Romania, the Norwegian Water Resources and Energy Directorate, Polar View ApS, UiT Arctic University of Norway

Funding: The European Space Agency (ESA) – Φ-lab

Coming soon: model weights, code, pretraining dataset, and article.