From the perspective of Earth System predictions, the use of machine learning, and in particular deep learning, is still in its infancy. There are many possible ways how machine learning could improve model quality, generate significant speed-ups for simulations or help to extract information from numerous Earth System data, in particular satellite observations. However, it has yet to be shown that machine learning can hold what it is promising for the specific needs of the application of Earth System predictions. This session aims to provide an overview how machine learning can/will be used in the future and tries to summarise the state-of-the-art in an area of research that is developing at a breathtaking pace.
Martin Schultz, Felix Kleinert, Lukas Leufen, Jessica Ahring, Susanne Theis, Jan Keller, Gordon Pipa, Johannes Leugering, Pascal Nieters, Peter Baumann, Vlad Merticariu, Andreas Hense, and Rita Glowienka-Hense