IBM and Hugging Face have teamed up to introduce a groundbreaking AI model for climate science. This new model, called the watsonx.ai geospatial foundation model, aims to revolutionize climate research and accelerate innovations in this critical field. By leveraging NASA’s satellite data and open-source AI platform, the partnership intends to democratize access to AI and foster collaboration within the scientific community.
The realm of climate science faces a significant challenge in accessing the most up-to-date data due to constantly changing environmental conditions. Despite the growing influx of data, analyzing these vast datasets remains a daunting task for scientists and researchers. To address this concern, IBM developed the geospatial foundation model earlier this year through a collaboration with NASA under a Space Act Agreement.
By making the geospatial foundation model available on the open-source AI platform Hugging Face, the collaboration aims to facilitate greater cooperation and information sharing in the AI community. This move is expected to expedite the development of impactful solutions that benefit the planet.
The geospatial foundation model was trained jointly on Harmonized Landsat Sentinel-2 satellite data across the continental United States over a year. The model has showcased an impressive 15 percent improvement over existing techniques while requiring only half the labeled data. Researchers can further fine-tune the model for various tasks such as deforestation tracking, crop yield prediction, and greenhouse gas detection and monitoring. Collaboration between IBM, NASA, and Clark University is also exploring applications such as time-series segmentation and similarity research.
IBM’s geospatial model harnesses its foundation model technology, which is part of the company’s broader efforts to create and train AI models for diverse tasks by leveraging transferable knowledge across scenarios. In July, IBM launched Watsonx, an AI and data platform enabling enterprises to scale and accelerate the impact of advanced AI using reliable data. A commercial version of the geospatial model will be integrated into the IBM Environmental Intelligence Suite (EIS) and is expected to be released later this year.
This partnership between IBM, Hugging Face, and NASA’s satellite data represents a promising opportunity to drive scientific progress and deepen our understanding of the Earth’s climate. The open-source nature of the model is poised to empower researchers and scientists worldwide in tackling pressing environmental challenges. By fostering collaboration and democratizing AI access, this initiative aims to pave the way for accelerated innovations and solutions that benefit our planet.