This model, constructed using NASA’s satellite data, will be accessible through Hugging Face, marking the largest geospatial foundation model on the platform and the first-ever open-source AI model developed in partnership with NASA.
Climate science faces the challenge of rapidly changing environmental conditions that require up-to-date data. Despite efforts to increase their volume of data, with NASA projecting 250,000 terabytes of new mission data by 2024, researchers still struggle to analyze these immense datasets effectively.
To address this issue, IBM has collaborated with NASA on a Space Act Agreement, to embark on the creation of an AI foundation model dedicated to geospatial data. The model’s release on Hugging Face, an open-source repository for transformer models, aims to democratize AI access to generate innovations for climate and Earth science.
IBM and NASA, in collaboration with Clark University, are adapting the geospatial model for various applications, including time-series segmentation and similarity research. The model uses Harmonized Landsat Sentinel-2 satellite data and, according to its creators, outperforms existing techniques, demonstrating a 15 percent improvement, using just half the labeled data. It also extends to tasks like deforestation tracking, crop yield prediction and greenhouse gas detection.
This move follows IBM’s commitment to collaborate with NASA to accelerate satellite image analysis and contribute to NASA’s Open-Source Science Initiative. NASA’s declaration of 2023 as the Year of Open Science aligns with its decade-long effort to foster an inclusive and collaborative scientific community.
IBM’s AI and data platform, watsonx, launched in July, facilitates scalable and impactful AI deployment. The commercial version of the geospatial model, included in IBM’s Environmental Intelligence Suite (EIS), will be available later this year.
Sriram Raghavan, VP of IBM Research AI, said: “The essential role of open-source technologies to accelerate critical areas of discovery such as climate change has never been clearer.
“By combining IBM’s foundation model efforts aimed at creating flexible, reusable AI systems with NASA’s repository of Earth-satellite data, and making it available on the leading open-source AI platform, Hugging Face, we can leverage the power of collaboration to implement faster and more impactful solutions that will improve our planet.”
Jeff Boudier, head of product and growth, Hugging Face, said: “AI remains a science-driven field, and science can only progress through information sharing and collaboration.
“This is why open-source AI and the open release of models and datasets are so fundamental to the continued progress of AI, and making sure the technology will benefit as many people as possible.”
Kevin Murphy, chief science data officer, NASA, said: “We believe that foundation models have the potential to change the way observational data is analyzed and help us to better understand our planet.
“And by open sourcing such models and making them available to the world, we hope to multiply their impact.”