Research project Using AI to predict the retreat of glaciers
This project explores some of the most inaccessible fjords in the Northern Hemisphere to map the seafloor and trace evidence of past and present glacier movement.
The aim is to improve understanding of how marine-terminating glaciers respond to climate change, particularly where they meet the ocean in fjord environments. Using advanced sonar technology, satellite data, and artificial intelligence, the research contributes both to global efforts in seabed mapping and to the development of more reliable models of glacier retreat in a warming climate, with implications for sea-level rise.
Photo: Magnus Bergström
The project explores some of the most inaccessible fjords in the Northern Hemisphere with the aim of mapping the seafloor, tracing imprints of past and present glacier dynamics, and improving our understanding of how marine-terminating glaciers respond to climate change. By combining advanced echo sounding technology, satellite data, and artificial intelligence, the research contributes both to global efforts in seabed mapping and to the development of more robust and reliable ice-sheet models.
In August 2024, the Swedish icebreaker Oden carried out its third scientific expedition to northern Greenland, led by Professor Martin Jakobsson together with Professor Nina Kirchner. The expedition focused on mapping the seafloor of the remote Victoria Fjord, where the C.H. Ostenfeld Glacier meets the ocean—an area that had never previously been accessed by a ship.
The expedition, known as the GEOEO – North of Greenland Expedition, brought together approximately 40 researchers from ten international universities and research institutions. It contributes to the global Seabed 2030 initiative, which aims to map the entire ocean floor by the year 2030. High-resolution maps of the seafloor were produced using multibeam echosounder technology, revealing landforms shaped by past and ongoing glacier activity.
By analysing these seabed features in combination with satellite observations and sediment data, researchers can reconstruct how marine-terminating glaciers have advanced and retreated under different climatic conditions. This knowledge is essential for improving models of glacier and ice-sheet behaviour and for reducing uncertainties in projections of future sea-level rise.
A central component of the project is the development of new methods using artificial intelligence and machine learning to automate the interpretation of the collected echosounder data. These approaches enable faster and more detailed analyses of large datasets, enhancing both the efficiency and the scientific value of seabed mapping efforts. Beyond its importance for climate and cryosphere research, the results support a wide range of applications, from planning underwater infrastructure to the development of offshore and marine-based energy systems.