Atmospheric dynamics on weather and climate timescales, including a broad range of phenomena such as gravity waves, Rossby waves, the jet stream, storm tracks, extratropical cyclones and extreme weather, as well as modelling across complexity levels, including very high-resolution models, idealized models, the development of machine learning parameterisations to data-driven forecasting.
Research Area
Sebastian Schemm is a member of the Atmosphere-Ocean Dynamics group at DAMTP. His research interests are in the field of atmospheric sciences. In particular, the physics and dynamics of weather and climate in the extratropics across spatio-temporal scales from turbulence to synoptic- and to planetary scales. A second interest is in high resolution atmospheric modelling and some machine learning. This involves modeling at different complexity, the use of observations, basic theory and the development of diagnostic tools. In recent years, research has focused on the life cycle of extratropical cyclones, the dynamics of the jet stream and the storm tracks, Rossby waves and teleconnection patterns. Beyond these topics, ongoing research is on parameter estimations using data assimilation in LES simulations, diabatic modification of the atmospheric circulation, data-driven parameterisations and km-scale global models, but also large-scale machine learning models such as ECMWF's WeatherGenerator.
Project Interests
Physics-informed machine learning for sub-grid-scale parameterisations with the goal to develop next generation weather and climate models that learn from data and generalise.
Develop parameter estimation using Bayesian methods to improve models using observations by assimilating observations.
The global circulation response to warming, which includes changes in the jet stream and storm tracks and their downstream impact on extreme weather.
Diabatic-adiabatic process interactions to improve the mechanistic understanding of how clouds affect the jet stream. Dynamical systems theory applied to the atmosphere to understand changes in predictability.
