Universität zu Köln

Peter Dueben (ECMWF) will present recent progress at the ECMWF.

To predict the weather is a difficult task as it requires to model the Earth System -- a huge system that consists of many individual components and shows chaotic dynamics. Based on a huge amount of data that is available from both observations and modelling, there are a large number of machine learning applications which have the potential to improve the different components across the workflow of numerical weather predictions. However, whether these approaches will succeed is still unclear as there are also a number of challenges for the application of machine learning tools in weather predictions. This talk will present recent progress on the use of deep learning tools at the European Centre for Medium-Range Weather Forecasts and discuss the challenges. It will also put progress on the use of deep learning into the context of modern supercomputing and scalability challenges that the weather and climate modelling community is currently facing.