As renewable energy sources are dependent on the local weather, they do not only contribute to mitigating climate change but are also affected by it. Quantifying how climate change affects current and future energy resources therefore sits at the core of our work. Atmospheric processes are highly non-linear and chaotic, yet recurring patterns give rise to predictable tendencies, from short-term weather to long-term climate shifts, and we develop machine learning algorithms, including generative approaches and masked token models, which are suited to uncover these structures. An additional focus is here  proper uncertainty quantification and uncertainty which helps downstream decision-making.

Main Research Interests:

• Probabilistic climate model downscaling of energy relevant variables
• Decision-calibrated uncertainty estimates
• Energy system optimisation under climate uncertainty

Integrating renewable energy sources into the electricity grid comes with new challenges and opportunities. The aim in this area is enhancing the efficiency and functionality of energy systems, ranging from small-scale household infrastructures to large-scale transmission networks. As energy systems represent a web of interdependent components on these multiple scales that are rich in prior knowledge, they present an opportunity for machine learning models.

Main Research Interests:

• Optimisation of Solar Thermal Systems (e.g. Fault Detection)
• Graph Coarsening for Energy System Models
• Probabilistic Forecasting of Energy Demand and Supply