Internal Variability Limits the Predictability of a Summer Ice-Free Arctic

Thursday, 18 December 2014: 11:05 AM
Alexandra Jahn1, Marika M Holland1 and Jennifer E Kay2, (1)National Center for Atmospheric Research, Boulder, CO, United States, (2)University of Colorado Boulder, Atmospheric and Oceanographic Sciences and CIRES, Boulder, CO, United States
The transition to a summer ice-free Arctic Ocean has attracted much attention, but climate model simulations in CMIP5 vary widely in their prediction of when we will reach a summer ice-free state. To quantify the contribution of internal variability to the spread of projections and to assess the limit of predictability of the year we will reach a summer ice-free Arctic we will present results from the new Large Ensemble with the Community Earth System Model (CESM). The CESM is a state of the art climate system model that has shown skill in capturing the observed decline of Arctic sea ice. The CESM Large Ensemble has 30 members for 1920-2100 that only differ through round-of level perturbations in the initial state. It is an unprecedented opportunity to analyze the contribution of natural variability to the simulated climate evolution within one model, as it provides enough ensemble members to characterize the probability distribution of many climate phenomena. Focusing on Arctic sea ice predictability, we find that the individual ensemble members first reach an ice-free Arctic in September (defined as a sea ice extent of 1 million km2) over a period of 21 years (2032-2053), with the majority of the ensemble members first reaching September ice-free conditions in the 5-year period from 2040-2044. This large spread shows the large impact of internal variability on the timing of reaching an ice-free Arctic, which severely limits the predictability of a summer ice-free Arctic we can expect from CMIP-type climate models. In addition to results on the predictability of an ice-free Arctic, we will also present results on the probability of hiatus periods in sea ice loss and of rapid ice loss events (RILEs), based on the CESM large ensemble.