Seminar: Using machine learning for regional forecasting and climate downscaling of the atmosphere and ocean - can we throw out our numerical models now?

Abstract:

The recent advances in ML or AI modelling of the atmosphere and the ocean have upended decades of conventional wisdom - namely that the way forward is higher resolution and better parameterisations of what remains unresolved. Here I will present a handful of examples of how forecasting and modelling the atmosphere and the ocean can be done using graph neural networks and more traditional convolutional neural networks. The big question is then whether we are headed toward a future where models in the traditional sense become obsolete? I will argue that on the contrary, we need the models to guide (supervise) machine learning and artificial intelligence. However, the current use of numerical models is not fit for purpose and we need to rethink what type of numerical models we use for the training. We also need to be aware of the common pitfalls in machine learning - perhaps most importantly how ML models handle previously "unseen" cases, whether these come in the form of extreme weather events or in modelling a future climate very different from what the models have been trained on.

Biography:

Professor Øyvind Breivik is Head of Division for Oceanography and Marine Meteorology at MET Norway. He has more than 25 years of experience in wind and wave climate research and was involved in the development of the ocean-surface wave coupling at ECMWF. He oversees the development of the Norwegian wave forecast system and the OpenDrift oil drift trajectory forecast models (OpenDrift) for oil drift and search and rescue. He was involved in the development of the NORA10 hindcast and is responsible for the development of a new high-resolution hindcast archive for the Norwegian Sea.