Learning from Big Data in the Atmospheric Sciences

In modern atmospheric physics, it is often found that the models used do not correctly represent complex situations in weather and climate. Modern machine learning methods offer a new way to develop models. In the project, these methods are to be applied to important questions in atmospheric physics for large amounts of data. In particular, clouds and their representation in models as well as the predictability of difficult weather situations will be considered. To this end, the methods must be further developed and expanded for application in interdisciplinary work; technical challenges also play a major role here. The infrastructure developed is to be made available to other sciences at the University of Mainz beyond the scope of the project.