EC-EARTH and CMIP5

The 5th Coupled Model Intercomparison Project (CMIP5) of the World Climate Research Program (WCRP) is about to start. All global climate models are invited to participate in CMIP5 and submit their results to the database that will be a major resource for the coming IPCC Assessment Report 5.

The EC-EARTH consortium will contribute to several CMIP5 experiments with the current model version 2.2. The horizontal resolution will be T159 in the atmosphere (1.125 deg) and 1 deg in the ocean. The Rossby Centre will use the EC-EARTH model for centennial simulations as well as climate predictions.

Centennial simulations with reference concentration pathways

The centennial simulations are the classical type of climate model experiments, where models start from a pre-industrial spun-up state in the mid 1800’s, followed by a hindcast simulation forced with observed concentrations of greenhouse gases, aerosols and land-use changes up to present day. The experiment will then extend into the future with assumptions about the forcing by greenhouse gases and aerosols following reference concentration pathways (RCP) that end at well defined levels for the radiative forcing at the end of the 21st century. The Rossby Centre in collaboration with MISU plans to contribute to RCP8.5 and RCP4.5 experiments with several ensemble members, where RCP8.5 corresponds to a positive perturbation of the radiative budget at the top of the atmosphere of 8.5Wm-2 at 2100 and RCP4.5 to a perturbation of 4.5Wm-2.

Centennial run with RCP

Decadal prediction for CMIP5

Decadal prediction is a novel method for assessing the climate evolution of the coming ~10 to 30 years. The model used for climate predictions is the same as for the long simulations, but the initial state is carefully controlled to match current conditions as much as possible. Starting from an observed state, it is believed, will improve model skill in predicting the climate of the nearest future, at least in regions where there exists potential predictability on these time-scales. Decadal predictions are still highly experimental and the results for CMIP5 will require careful analysis. To test methods for model initialisation and to assess predictability levels, the CMIP5 decadal predictions will include a large number of start dates from the past, allowing for a direct comparison between the predictions (termed hindcasts when made over the past period) and observations. This effort will be crucial to guide understanding and use of the decadal predictions made into the near future.

CMIP5, decadal, RCN 2010

In addition to centennial simulations and decadal predictions, there will also be atmosphere-only simulations with pre-scribed se surface temperature and sea-ice. These simulations compare the atmosphere components of climate models and test their ability to reproduce the mean state and variability of the atmosphere given “perfect” surface boundary conditions.