<\/p> \n \n
The Earth’s climate is a highly complex system and as such many processes and phenomena cannot be resolved in terms of fundamental physical equations, either because they are unknown or because they would be to costly to explicitly include in numerical simulations. While there is a long history of complementary, data-driven modelling approaches, recent developments in deep learning promise substantial advances.<\/p> \n \n
Here Prof. Boers presents some recent results on machine learning approaches to address several challenges in nonlinear Earth system dynamics, including the prediction of climate phenomena such as monsoon onsets or ENSO, hybrid modelling combining process-based with deep learning model components, as well as bias correction and downscaling of Earth system models with generative machine learning models.<\/p> \n \n
<\/p> \n \n
Niklas Boers is professor of Earth system modelling at the Technical University of Munich, leads the Future Lab on Artificial Intelligence at the Potsdam Institute for Climate Impact Research, and is Honorary Professor of Mathematics at the University of Exeter. He obtained his PhD from Humboldt University Berlin and spent several years at the ENS Paris and Imperial College London. As a physicist and mathematician by training, he studies theoretical climate dynamics with a combination of physical and data-driven approaches. In his research he develops methods rooted in dynamical system theory, complexity, science, and machine learning and applies them to different problems of nonlinear dynamics, with a focus on modelling and predicting extreme events and abrupt transitions in the Earth system.<\/p> \n \n
<\/p> \n \n