Séminaire
The new field of decadal climate prediction is recognized as a major part of the Coupled Model Intercomparison Project phase 5 (CMIP5), which is the suite of model experiments intended to inform the IPCC assessment process. Decadal predictions are certainly of increasing scientific interest because they potentially represent a benefit to society through improvements in the development of climate services and of adaptation strategies. Decadal prediction aims to provide climate information on time scales of 10 to 30 years into the future while exploring the benefits of initializing climate models. The initialization tries to add contemporaneous information about the upper-ocean heat content, to achieve forecast quality beyond that provided by the externally-forced signal; its objectives include capturing low-frequency internal variability and correcting the model response to climate change forcings and commitment. An overview will be provided showing results from recent works that assess multi-year forecast quality for global-mean surface temperature, Atlantic multi-decadal variability (AMO), and Pacific decadal variability (IPO/PDO).
Two methodological issues for near-term climate prediction will also be discussed: the impact of the choice of the observational reference period, and of the number of years included in the forecast average. Results derive from the assessment of CMIP5 decadal hindcasts and historical runs as well as the ENSEMBLES multi-model and perturbed-parameter (DePreSys) decadal re-forecasts. Multi-year prediction skill of the dominant West African monsoon rainfall regimes, i.e. the Guinean and Sahelian modes, will also be illustrated.
martin.vancoppenolle@locean-ipsl.upmc.fr