Influence of internal variability on sea ice predictability

Abstract:

Uncertainty in climate model predictions of the Arctic sea ice cover steams from many sources, including model biases, scenario uncertainty, and internal variability. Ensembles of climate simulations give a range of projections, but it is unclear how large these ensembles need to be to provide an accurate assessment of the influence of internal variability on Arctic sea ice projections. As large ensembles of fully coupled globak climate models are expensive to run, the maximum ensemble size in the CMIP5 archive consisted of ten simulations, with many models represented by only one to six ensemble members. An analysis of the Arctic sea ice simulation from the Community Earth System Model (CESM) large ensemble with 30 members for RCP8.5 and the CESM medium ensemble with 15 members for RCP4.5 demonstrates how many ensemble members are needed to constrain the range of uncertainty in Arctic sea ice predictions for the 21^th century in one of the CMIP5 models. The goal of this analysis is to aid in the interpretation of smaller ensembles and encourage larger ensembles for CMIP6.