Climate prediction is a rapidly developing field that aims to provide society with tools for planning activities in the near-term. The skill of these predictions remains challenged by accurate use of observations for forecast initialisation and systematic model error. Data assimilation is a powerful approach for using scarce observations to constrain a dynamical system.

I will introduce some recent advances in its use to improve the initialisation of climate predictions. In addition, I will introduce the supermodelling approach in which multiple models are interactively combined to compensate errors and thereby create a superior model.

Finally, I will present an application of supermodelling to multi-annual predictions of marine ecosystems. In particular, I will show that skilful decadal-scale prediction of fish habitat and distribution shifts are already possible in the North Atlantic.

Pour en savoir plus

Noel Keenlyside is a professor at the Geophysical Institute, University of Bergen and Bjerknes Centre for climate research. He leads the Bjerknes Climate Prediction Unit and also holds adjunct positions at the Nansen Environmental and Remote Sensing Center in Bergen. Keenlyside has extensive expertise in climate dynamics and predictability, with much of his research focusing on the Atlantic and Pacific sectors.

He has led the development of the Norwegian Climate Prediction model and pioneered novel climate modelling approaches, such as supermodelling. His research interests extend to developing climate-based predictions of marine ecosystems, and he leads major international initiatives on this topic (e.g., the EU H2020 TRIATLAS project). As part of his sabbatical leave from Bergen, Keenlyside is a Visiting Professor in the ENS’s Geosciences Department until July 1st and planning to interact in particular with Michael Ghil and Fabio D’Andrea.

ENS – salle Froidevaux – E314 • 24, rue Lhomond 75005 PARIS