Abstract: The Coupled Model Intercomparison Project Phase Six (CMIP6) organized by the World Climate Research Project (WCRP) is in progress. Nine earth climate system models from China contribute to CMIP6. The seasonal cycle, long-term linear trend, and intra-annual variability of Arctic sea ice extent (SIE) from the nine models are evaluated by comparing them with observations from 1980 to 2014. The results show that eight models are capable of reproducing the seasonal cycles of Arctic SIE well, except one of nine models in which the maximum value of seasonal cycle is delayed by one month. Most of the models (8/9) overestimate the maximum sea ice extent values of seasonal cycle. In terms of long-term trends, five models overestimate the declining trends of Arctic sea ice in March, and four models underestimate the declining trends of Arctic sea ice in September. Compared with the results of the CMIP6 multi-model ensemble mean, it is found that there is one model for which the seasonal cycle and long-term linear trend of SIE are both within the range of the multi-model ensemble mean’s standard deviation. The difference in long-term September and March SIE trends leads to a significant increasing trend of SIE intra-annual variability as measured by the standard deviation of SIE within a calendar year. Two models can reproduce this feature reasonably well. Finally, it is worth pointing out that four models from the same institution that contributed to both CMIP5 and CMIP6 show improvements in terms of SIE seasonal cycle and its long-term linear trend of annual averaged SIE.

Key words: earth climate system model, CMIP6, Arctic sea ice, sea ice extent