The Allen Institute for Artificial Intelligence (Ai2) has launched OlmoEarth v1.3, a satellite map embedding model covering the entire planet. The tool converts a year of satellite observations into numerical fingerprints for land patches, enabling the identification of forests, crops, and mining sites.
The model processes radar and optical data from Sentinel-1, Sentinel-2, and Landsat missions. Compared to previous versions, v1.3 is more efficient, requiring 53% of the computing time of v1.2 and 18% of the time used for the first version, while reducing carbon emissions by 50% and 22% respectively.
Ai2 also introduced the OlmoEarth Embeddings Explorer, a browser-based tool that allows users to generate maps without requiring GPU resources or model fine-tuning. The platform enables analysis of environmental changes, such as tracking illegal mining activities or agricultural expansion, in seconds.
The model provides data at a resolution of 10-by-10 meters, with additional options for 64-square-meter and 128-meter resolutions. While fine-tuning remains an option for specific research tasks, the new explorer simplifies initial data analysis for various environmental monitoring projects.
