Hi.

I’m Philipp, a postdoc at the Potsdam Institute for Climate Impact Research and the Technical University of Munich.

I am particularly interested in the following research topics:

• Generative machine learning for improving weather and climate simulations. I am interested in how generative machine learning methods for image and video data can be extended and applied, e.g., for correcting systematic errors in simulations, increasing their resolution, improving forecasst skills, or learning fast emulators of computationally expensive processes.

• Predicting extreme events, such as rainfall extremes, accurately in the context of climate modelling and weather forecasting, e.g., by developing suitable loss functions, data transformations, and sampling strategies.

• Generalization under distribution shifts of data-driven machine learning models. When modelling climate phenomena and many other real-world problems, the data distribution seen during training might undergo changes over time that lead to out-of-sample predictions during the model inference.

• Hybrid modelling that combine process knowledge, e.g. physical conservation laws, with data-driven machine learning algorithms.