UMR CNRS 5805 EPOC
Environnements et Paléoenvironnements
Océaniques et Continentaux

        

Consignes de Sécurité et Numéros d'Urgence
(à l'usage des personnels du laboratoire)

le 18-03-2022 à 11:00

Lieu Salle Atmosphère, Bâtiment B18N, OASU, Université de Bordeaux Intervenant(s) Leo BORCHERT, LMD, ENS, Paris

Near-term climate prediction for the next few years, so-called decadal climate prediction, is a potentially powerful tool for decision makers and society. However, such predictions are notoriously uncertain, because climate models struggle to simulate decadal-scale changes in many climatic indices. Here, decadal climate predictions are analysed for their agreement with observations, their skill, in the North Atlantic and European region.
This presentation presents two recent advances in the field of decadal prediction. First, an approach for analysing decadal predictions using physical storylines is presented. This approach can separate the contributions of different forcing agents or modes of climate variability to any diagnosed prediction skill, enabling process understanding in the context of climate predictions. Such an activity can help users of climate information estimate the expected skill of a decadal climate forecast more accurately, and can even yield skill where there used to be none. Second, the skill of decadal climate predictions from the recently released CMIP6 archive is systematically analysed and compared to CMIP5, illustrating advances in the field of climate modelling as a whole.