IBM and Hugging Face have joined forces with NASA to release an open-source AI model for climate science. The collaboration aims to democratize access to AI technology in the field of climate science, allowing researchers and scientists to effectively analyze large datasets and address the challenges posed by rapidly changing environmental conditions.
The geospatial foundation model, developed by IBM and NASA, is based on satellite data from NASA and is set to be the largest of its kind on the Hugging Face platform. It marks the first open-source AI foundation model created in collaboration with NASA.
Access to the latest data is crucial in climate science, but scientists and researchers often struggle with analyzing vast amounts of data efficiently. NASA estimates that by 2024, there will be 250,000 terabytes of data from new missions. To alleviate this issue, IBM and NASA entered into a Space Act Agreement to build an AI foundation model for geospatial data.
By making the geospatial model openly available on Hugging Face, IBM and Hugging Face aim to foster collaboration and accelerate progress in climate and Earth science. Emphasizing the significance of information sharing and collaboration, Jeff Boudier, Head of Product and Growth at Hugging Face, believes that open-source AI and the release of models and datasets are essential for ensuring AI benefits as many people as possible.
The geospatial model, jointly trained by IBM and NASA on Harmonized Landsat Sentinel-2 satellite data (HLS) over a year, has demonstrated promising results. Even with just half the labeled data, the model showed a 15% improvement over state-of-the-art techniques.
With further refinement, the geospatial model can be adapted for various tasks such as deforestation tracking, crop yield prediction, and greenhouse gas detection. This versatility makes it a valuable tool for addressing climate challenges effectively.
IBM’s collaboration with NASA aligns with NASA’s Open-Source Science Initiative, which aims to promote accessibility and inclusivity within the scientific community. NASA, alongside other federal agencies, has designated 2023 as the Year of Open Science, celebrating the benefits of openly sharing data, information, and knowledge.
Kevin Murphy, Chief Science Data Officer at NASA, believes that open-sourcing foundation models can transform how observational data is analyzed, leading to a better understanding of our planet.
IBM’s geospatial model is part of its broader initiative to create and train AI models with transferable capabilities across different tasks. The watsonx.ai geospatial model will also be integrated into IBM’s Environmental Intelligence Suite (EIS), which will be available commercially later this year.
By leveraging the power of open-source technologies, the collaboration between IBM, Hugging Face, and NASA seeks to effectively address climate challenges and contribute to a more sustainable future. With the release of the geospatial foundation model, researchers and scientists worldwide will have access to advanced AI tools to accelerate their work in climate and Earth science.
