Focus Period Linköping 2024

Machine Learning for Climate Science

September 23 – October 25, 2024

Climate change is a complex and urgent problem that requires accurate and reliable models of the Earth’s climate systems. These models can help us understand the causes and consequences of climate change, and to evaluate the effectiveness of different policies and actions. Traditionally, mechanistic models have been used for this purpose. However, the Earth’s climate systems are extremely complex and these models are inevitably oversimplifying. For many highly critical use cases, in particular involving local events such as accurately predicting flooding, heatwaves, and storm surges, existing models need to be improved. At the same time, today we have access to a plethora of data from multiple dense networks of ground-based monitoring stations as well as both passively and actively remote-sensed data from drones, aircraft, and satellites. We are therefore seeing and increasing interest in using machine learning for improving and complementing the traditional modeling approaches. Machine learning can help us discover new patterns and relationships in the data, improve the accuracy and speed of the simulations, and result in better uncertainty quantification.

The purpose of this focus period is to bring together researchers who share an interest in machine learning for climate science in general, and machine-learning-augmented modeling and analysis of climate and weather systems in particular, to lay the foundation for novel research on how machine learning can be used to address critical problems in climate science currently lacking satisfactory solutions.

The focus period will run in Linköping between September 23 and October 25, 2024. A number of international scholars interested in these research challenges will be invited to participate, and to interact with leading scientists as well as ELLIIT researchers. In the third week (October 8-10) we will have a 3-day symposium, with several invited top-level researchers.