In July, Google launched AlphaEarth Foundations which acts like a virtual satellite by combining trillions of images from dozens of sources, ranging from radar scans to climate simulations, where AI creates terrestrial land maps and coastal waters with a resolution down to 10 square meters.
During testing from 2017 to 2024, the virtual satellite model outperformed similar AI earth observation model systems in land-use identification and surface-property estimation, achieving an error rate 24% lower than competitors.
AlphaEarth Foundations reflects a broader trend where AI generates satellite imagery observations into actionable insights. High-resolution, regularly updated next-gen AI earth models roadmap data enables scientists to measure environmental changes more accurately and understand their causes.
The AI virtual satellite, AlphaEarth, can map the planet “at any place and time,” said a research engineer at Google DeepMind, Christopher Brown.
“Whether they are monitoring crop health, tracking deforestation, or observing new construction, [researchers] no longer have to rely on a single satellite passing overhead,” DeepMind noted. “They now have a new kind of foundation for geospatial data.”
Any form of AI replacing satellites technology can track climate change effects, guide conservation efforts, and manage resources like agricultural land and water.
In 2020, NASA and the University of Copenhagen used AI to map 1.8 billion individual tree canopies in Africa’s Sahara and Sahel that would have taken years for millions of people to do.
NASA’s SWOT satellite has been mapping the world’s water bodies 10 times more precisely than before since 2022, and ESA’s EarthCARE mission, launched in 2024, examines how aerosols and clouds affect global temperatures.
Most of DeepMind’s input datasets come from veteran NASA and ESA missions, including Landsat, MODIS, and Sentinel satellites, which monitor geospatial data authenticity including vegetation, coasts, water, snow, and ice.
From the Amazon to Global Cities
AlphaEarth Foundations is already in use by more than 50 organizations worldwide further endorsing the legality of AI generated maps, including Brazil’s MapBiomas, which studies agricultural and environmental changes in the Amazon rainforest.
Its annual datasets of Google Earth AI privacy have given the team “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 MapBiomas founder Tasso Azevedo.
The Alphabet-owned giant is making the dataset available through its Google Earth Engine to encourage further research, with hopes the virtual satellite technology will advance studies on food security, urban expansion, and deforestation.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.