A summer ice-free Arctic Ocean could become a reality as early as the 2030s, with global consequences including intensified heatwaves, stronger storms, and disrupted ecosystems. In the new IcyAlert project, researchers from the Danish Meteorological Institute (DMI), the Royal Meteorological Institute of Belgium (RMI) and the Technical University of Denmark (DTU) will use a combination of climate models, causal methods and artificial intelligence (AI) to develop a new early warning system for ice-free summers in the Arctic and its impacts.
IcyAlert – The project
IcyAlert stands for “Intelligent Climate Early Warning Alert for Arctic Ice-Free Summers” and will be funded by the Danish Novo Nordisk Foundation from late 2025 until late 2031. The main goal of the project is to develop an advanced forecasting model to better predict ice-free Arctic summers months in advance.
The project will be led by DMI (PI: Tian Tian), in collaboration with RMI (PI: Stéphane Vannitsem) and DTU (PI: Tommy S. Alstrøm). It will combine the expertise from the three groups to develop the early warning system: climate modeling (DMI), causal analysis (RMI), and large-scale AI (DTU). Through this project, the consortium will receive privileged access to the AI supercomputer Gefion.
Why is it important?
With ongoing climate change, the Arctic has been warming nearly four times faster than the global average and summer Arctic sea-ice extent has been declining fast. Sea ice is frozen seawater that floats on the ocean surface. It covers about 15 million km² of the Arctic Ocean in late winter and 5 million km² in late summer. With the current rate of ice melting, the Arctic could face its first ice-free summer (total sea-ice extent below 1 million km²) as early as the 2030s.
Loss of Arctic sea ice has already made shipping routes more accessible. However, its potential adverse impacts are numerous, including large changes in atmospheric circulation at mid-latitudes, increased risk of extreme events (e.g. severe cold winters, marine heatwaves, storms, etc.) not only over the Arctic but also over Europe, North America and Asia, and threats to Arctic communities and biodiversity. Thus, it becomes increasingly important to better forecast the timing of an ice-free Arctic.
What will be the role of the Royal Meteorological Institute?
In IcyAlert, RMI will provide its expertise in causal analysis to identify the predictors of ice-free Arctic conditions, as well as its climate impacts. The Institute has already used such causal methods in a variety of climate studies (e.g. for studying North Pacific and Atlantic regions, as well as for Arctic sea ice). Causal methods allow to go beyond traditional correlation analyses and identify cause-effect relationships between climate variables.
RMI will work in close collaboration with DMI and DTU to combine causal methods with global climate models and explainable AI (Figure 1), which will allow to develop a powerful forecasting tool. Stéphane Vannitsem will lead the supervision of the project, with David Docquier working as a postdoctoral researcher (Dynamical Meteorology and Climatology Unit). “With this project, we not only want to advance the field of climate science, but also to contribute to a better preparedness of our communities in the current context of climate change”, says David Docquier.
Figure 1: Diagram showing the main fields of expertise of the three IcyAlert institutes (AI: artificial intelligence; ML: machine learning), the three key objectives O1-O3 (CMIP: Coupled Model Intercomparison Project; IFA: ice-free Arctic), and some examples of impacts (symbols in the middle) that will be analyzed in IcyAlert (e.g. heatwaves, coldwaves, droughts, shipping, flooding). Credit: Tian Tian (DMI).
