Séminaire
Decadal prediction (or near-term climate prediction) was made a WCRP Grand Challenge in 2016. Since then decadal prediction has moved towards international collaboration through the WMO. This has made new research and insights possible. In this talk I will show that decadal climate predictions are more skilful than previously thought, including skill for rainfall and sea level pressure (as well as temperature) in many regions. Partly, this is because the signal-to-noise ratio is too small in models, so a large ensemble is needed to extract the predictable signal (hence the need for a multi-model ensemble). The benefit of initialisation with observations is also larger than previously thought, I will present a new more powerful approach developed at the Met Office to show this. Finally, internal variability does not appear to play a large role, and the
improvements from initialisation may arise from correcting the modelled response to external forcings, especially solar and aerosols.
matthew.menary@locean-ipsl.upmc.fr