AlphaEarth Foundations processes petabytes of satellite and environmental data into a unified, AI-powered mapping tool with 24% higher accuracy than previous models.
The model enables consistent, on-demand mapping of Earth’s land and coastal ecosystems at 10×10 meter precision, with up to 16x less storage requirement.
Used by over 50 partners—including the UN FAO and MapBiomas—AlphaEarth Foundations supports faster, more accurate mapping for conservation, agriculture, and climate initiatives.
Google has introduced AlphaEarth Foundations, a powerful AI model designed to decode our planet’s complexities through advanced geospatial mapping. Developed as part of the company’s new EarthAI initiative, the model leverages massive volumes of Earth observation data to create consistent, detailed representations of terrestrial and coastal regions.
“Today we’re launching a new AI model, AlphaEarth Foundations, to help understand our planet in magnificent detail,” Google announced.
AlphaEarth Foundations acts as a virtual satellite, capable of analyzing petabytes of multimodal data—from optical satellite imagery to 3D laser mapping and climate simulations. By organizing the planet into 10×10 meter blocks, the system generates compact, high-fidelity summaries of each area, enabling users to detect changes with unmatched precision.
The model’s technical advantage lies in its efficient embeddings, which require 16 times less storage than comparable AI systems. This dramatically lowers the cost and time required for global-scale environmental analysis.
“This work is a significant step forward in understanding our changing planet and we’re excited to see how it can accelerate work in vital mapping initiatives,” Google stated.
To expand accessibility, Google has released annual embeddings from AlphaEarth Foundations as a Satellite Embedding dataset within Google Earth Engine. Over the past year, more than 50 organizations have tested the model across diverse applications—ranging from food security and deforestation to urban growth and water management.
RELATED ARTICLE: Google Maps Expands Sustainable Travel Features Across Europe
Nick Murray, Director of the James Cook University Global Ecology Lab and Global Science Lead of the Global Ecosystems Atlas, praised the tool:
“The Satellite Embedding dataset is revolutionizing our work by helping countries map uncharted ecosystems – this is crucial for pinpointing where to focus their conservation efforts.”
In Brazil, MapBiomas is using the model to track environmental changes and guide development in ecologically sensitive regions like the Amazon.
“The Satellite Embedding dataset can transform the way our team works—we now have new options to make maps that are more accurate, precise and fast to produce—something we would have never been able to do before,” said Tasso Azevedo, founder of MapBiomas.
Benchmark testing showed AlphaEarth Foundations consistently outperformed both traditional methods and competing AI models, achieving a 24% lower error rate on average. The model was especially effective when training data was limited—a frequent challenge in remote sensing.
Google envisions future applications where AlphaEarth Foundations could be paired with general reasoning AI models like Gem
Follow ESG News on LinkedIn
