On Greenland's Store Glacier, a team of researchers led by T. R. Chudley has cracked one of the ice sheet's most crucial timing problems: understanding exactly when meltwater drains from the surface down into the depths below—knowledge that could sharpen our predictions of sea-level rise in the decades ahead.
Every spring, meltwater fills crevasses that spider across the glacier's surface like a network of cracks in ice. Some of these crevasses drain as summer progresses; others hold their water stubbornly through the season. For years, scientists have known that timing matters enormously. When water drains through a glacier, it changes how the ice flows, can trigger ice to break apart, and sets off feedback loops that lead to faster ice loss. But they couldn't reliably predict when that drainage would happen, making it harder to model how quickly glaciers shed ice into the ocean.
Chudley's team approached the problem by linking ice mechanics directly to water movement. They used satellite imagery from the Sentinel-2 mission to track water in crevasses between 2016 and 2022, focusing especially on 2019 when satellite coverage was most complete. Rather than make educated guesses about drainage timing, they fed those observations into a convolutional neural network—an artificial intelligence tool—to map water coverage through the season and search for patterns between the mechanical forces squeezing and stretching the ice and the actual drainage of crevasse ponds.
The discovery was elegant and powerful: seasonal ice dynamics are the dominant factor controlling when crevasse meltwater drains. Specifically, when seasonal changes cause ice to stretch, crevasses can suddenly release the water they've been holding. This stretching and squeezing is driven by gravity pulling the glacier downhill, by the ice sliding over subglacial water, and by how the glacier interacts with the ocean. The findings appear in AGU Advances, a peer-reviewed journal published by the American Geophysical Union.
The practical importance cannot be overstated. The Greenland Ice Sheet sheds trillions of gallons of water each year. Understanding when that drainage occurs matters not only for predicting sea-level rise, but also for modeling how the glacier slides across its bedrock and when meltwater reaches the ocean. These processes are not isolated to Store Glacier—the researchers believe their findings likely illuminate similar dynamics in other fast-flowing outlet glaciers and ice sheets worldwide.
What makes this work particularly promising is that it offers a path forward for improving how ice behavior is represented in the climate models scientists use to forecast future change. As climate models become more sophisticated, they increasingly need to account for these kinds of mechanical details. The research shows that linking satellite observations to physics-based modeling can reveal hidden patterns that simpler assumptions might miss.
Chudley and his colleagues have shown that sometimes the most powerful predictor of what ice will do is not a chemical process or a hidden variable, but something visible to the naked eye: the physical forces that squeeze and stretch the glacier with the turning of the seasons.