Arctic Sea Ice in Two Configurations of the Community

17 We provide an assessment of the current and future states of Arctic sea ice simulated by 18 the Community Earth System Model version 2 (CESM2). The CESM2 is the version of 19 the CESM contributed to the sixth phase of the Coupled Model Intercomparison Project 20 (CMIP6). We analyze changes in Arctic sea ice cover in two CESM2 configurations with 21 differing atmospheric components: the CESM2(CAM6) and the CESM2(WACCM6). Over 22 the historical period, the CESM2(CAM6) winter ice thickness distribution is biased thin, 23 which leads to lower summer ice area compared to CESM2(WACCM6) and observations. 24 In both CESM2 configurations, the timing of first ice-free conditions is insensitive to the 25 choice of CMIP6 future emissions scenario. In fact, the probability of an ice-free Arctic 26 summer remains low only if global warming stays below 1.5◦C, which none of the CMIP6 27 scenarios achieve. By the end of the 21st century, the CESM2 simulates less ocean heat 28 loss during the fall months compared to its previous version, delaying sea ice formation and 29 leading to ice-free conditions for up to 8 months under the high emissions scenario. As a 30 result, both CESM2 configurations exhibit an accelerated decline in winter and spring ice 31 area under the high emissions scenario, a behavior that had not been previously seen in 32 CESM simulations. Differences in climate sensitivity and higher levels of atmospheric CO2 33 by 2100 in the CMIP6 high emissions scenario compared to its CMIP5 analog could explain 34 why this winter ice loss was not previously simulated by the CESM. 35 Plain Language Summary 36 We provide a first look at the current and future states of Arctic sea ice as simu37 lated by the Community Earth System Model version 2 (CESM2), which is part of the 38 newest generation of large-scale climate models. The CESM2 model has two configurations 39 that differ in their representation of atmospheric processes: the CESM2(CAM6) and the 40 CESM2(WACCM6). We find several differences in the simulated Arctic sea ice cover be41 tween the two CESM2 configurations, as well as compared to the previous generation of 42 the CESM model. Over the historical period, the CESM2(CAM6) model simulates a win43 ter ice cover that is too thin, which leads to lower summer ice coverage compared to the 44 CESM2(WACCM6) model and observations. In both CESM2 configurations, the proba45 bility of the Arctic becoming nearly ice free at the end of the summer only remains low 46 if global warming stays below 1.5◦C. In addition, the specific year a first ice-free Arctic is 47 reached is not sensitive to the future greenhouse gas emissions trajectories considered here. 48