I’m a doctoral researcher aiming to use machine learning (ML) methods to create large and physically-consistent artificial weather datasets.

Studying rare weather and climate event and their associated risks requires large amounts of data and even more data is needed for research into multivariate compound events (such as simultaneous hot and dry conditions). The observational period is too short to reliably quantify the probability of such rare events. Climate model simulations offer long, continuous time series from multiple ensemble members, but are spatially coarse and contain model-specific biases. Weather forecasts have finer spatial resolution, but are run for short time horizons only.

I plan to draw from data sources such as these and use ML methods to create long, continuous weather datasets while focusing on maintaining physical consistency and capturing the relevant processes of the climate system for given applications. These datasets can then be used to drive impact models, enabling researchers to explore the effects of climatic extremes on impacts such as floods, wildfires, or forest mortality.

I am part of the compound weather and climate events group, led by Prof. Jakob Zscheischler, and also affiliated with ScaDS.AI.