O access to the most recent data remains a significant challenge in climate science, where environmental conditions change almost daily. And despite the growing amount of data, estimates from NASA suggest that, by 2024, scientists will have 250.000 terabytes of data from new missions – scientists and researchers still face obstacles in analyzing these large datasets.
As part of a Space Law Agreement with NASA, IBM decided earlier this year to build an Artificial Intelligence (AI) base model for geospatial data. And now, by making available a geospatial foundation model via Hugging Face, a recognized leader in open source platforms and a well-known repository for all transformer models, efforts can move forward to democratize the access and application of AI to drive innovations in climate and earth sciences.
IBM, Hugging Face and NASA have teamed up to create a model of open source geospatial foundation that will serve as the basis for a new class of climate and Earth science. They will be AI systems capable of monitoring deforestation, predicting the yield of plantations and respective harvests, and assessing greenhouse gas emissions.
For this project, IBM leveraged its newly released Watsonx.ai to serve as a fundamental model using information from one year of satellite data Harmonized Landsat Sentinel-2 (HLS) from NASA. This data is collected by the pair of satellites Sentinel-2 by ESA, being built to acquire high-resolution optical images over terrestrial and coastal regions in 13 spectral bands.
