February 5, 2026 Press Release
Researchers at the British Antarctic Survey (BAS) have developed a new artificial intelligence tool that can, for the first time, automatically track the lifecycle of icebergs as they drift, melt and fragment in the world’s oceans, reconstructing detailed ‘family trees’ of giant ice layers.
The system uses satellite imagery to identify individual icebergs, assigning each a unique ID and tracking it over time. Importantly, when a large iceberg breaks, the AI can connect the scattered “child” pieces to the original “parent” iceberg. This has not been possible at scale until now.
This is important because when icebergs melt, large amounts of fresh water are released into the ocean. When that fresh water enters the ocean, it changes ocean currents, affects marine ecosystems, and can influence global climate patterns. Until now, scientists have struggled to track icebergs after they break into smaller pieces, leaving major blind spots in climate models.
Until recently, scientists have tracked only the largest named icebergs by visually tracking them one by one via satellite. As a result, the fate and impact of the thousands of tiny pieces flung across the Southern Ocean remain largely unknown.
Ben Evans, lead author and machine learning expert at the British Antarctic Survey, said:
“What’s interesting is that this finally gives us observations that we’ve missed so far. We’ve gone from tracking some famous icebergs to building complete family trees. For the first time, we know where each piece comes from, where it goes, and why it matters to the climate. ”
The system works by analyzing the characteristic geometry of icebergs captured in satellite images. When an iceberg breaks, AI effectively performs a digital “jigsaw” puzzle, piecing together the pieces that once belonged together. The system was tested with real satellite observations of Arctic icebergs taken from the Peterman Glacier and other areas in northwestern Greenland.
By revealing where iceberg meltwater is released into the ocean, this method provides important new information that can be input into global climate and ocean models, improving predictions of when ice loss from polar ice sheets is expected to increase as the climate warms. Data feeds into the NEMO ocean model, which is part of the UK Earth System Model.
This approach is useful for a variety of applications, including navigation, allowing ships operating in polar waters to better understand where dangerous ice comes from and how it may move.
This research was funded by EPSRC Grant EP/Y028880/1, the Alan Turing Institute, and the British Antarctic Survey’s Polar Science Program for a Sustainable Planet.
