AutoTerm: Automated outlet-glacier Terminus extraction for Greenland

The Greenland Ice Sheet is losing ice at an accelerating rate and is one of the largest contributors to global sea-level rise. Understanding how rapidly its outlet glaciers are changing is essential for improving projections of future coastal impacts. However, measuring glacier change remains a labor-intensive process that cannot keep pace with the rapidly growing volume of satellite observations. This project will create a sustainable, automated system that continuously maps the fronts of Greenland's marine-terminating glaciers from satellite imagery and delivers these measurements as freely available public data products. By combining advances in artificial intelligence with modern data infrastructure, the project will provide researchers, educators, and decision makers with timely, standardized information needed to better understand ice sheet change and improve projections of future sea-level rise. The project will also strengthen the nation's scientific cyberinfrastructure through open-source software, cloud-based workflows, publicly accessible data products, and training opportunities for graduate and undergraduate students. These resources will lower barriers to applying artificial intelligence in the geosciences while supporting a growing workforce skilled in data science, remote sensing, and Earth science. This project will modernize and expand an existing deep-learning framework for automated glacier terminus detection into a sustainable, community-supported cyberinfrastructure platform. The research will integrate satellite observations from multiple optical and radar missions spanning more than five decades to generate regularly updated records of terminus position for all marine-terminating glaciers in Greenland. The investigators will improve the underlying machine-learning algorithms through updated software architecture, expanded training data, automated quality-control procedures, uncertainty estimation, independent validation against manually interpreted datasets, and community-based review of generated products. The resulting system will produce four primary data products: glacier terminus traces, annually updated glacier identifiers, land-ice-ocean masks, and records of glacier retreat through time. The software will be containerized, documented, and deployed in cloud computing environments to ensure reproducibility, portability, and long-term sustainability while enabling researchers to adapt the framework to other regions and satellite datasets. All software and data products will be openly distributed through established national data repositories with comprehensive documentation and version control. By transforming glacier terminus mapping from a largely manual process into an automated, continuously updated community resource, this project will provide a critical observational foundation for studies of glacier dynamics, ice-ocean interactions, and future sea-level rise while advancing reproducible cyberinfrastructure for the polar research community. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.