AI Helps Track the World’s Shrinking Glaciers

Researchers at the Friedrich-Alexander University of Erlangen–Nuremberg in Germany developed an AI-based method that improves the monitoring of glacier retreat using satellite imagery. By combining a deep learning model with just one manually labeled image per glacier, seasonal reference images, and maps of underlying rock formations, the team reduced boundary-detection errors from more than 1 kilometer to less than 70 meters—roughly comparable to the accuracy of human analysts.