Exploiting SMILEs and the CMIP5 Archive to Understand Arctic Climate Change Seasonality and Uncertainty

Arctic Amplification (AA) exhibits a distinct seasonal dependence; it is weakest in boreal summer and strongest in winter. Here, we analyze simulations from single-model initial-condition large ensembles and Coupled Model Intercomparison Project Phase 5 to decipher the seasonal evolution of Arctic climate change. Models agree that the annual maximum AA shifts from autumn into winter over the 21st century, accompanied by similar shifts in sea-ice loss and surface turbulent heat fluxes, whereas the maximum precipitation shifts only into late autumn. However, the exact seasonal timing and magnitude of these shifts are highly uncertain. Decomposing the uncertainty into model structural differences, emission scenarios, and internal variability reveals that model differences dominate the total uncertainty, which also undergo autumn-to-winter shifts. We also find that the scenario uncertainty is unimportant for projections of AA. These results highlight that understanding model differences is critical to reducing uncertainty in projected Arctic climate change.Plain Language Summary The Arctic climate has experienced considerably greater warming compared with the global mean in recent decades in response to increasing amounts of atmospheric greenhouse gases. This phenomenon is called Arctic amplification (AA), and global climate models predict that it will continue in the future. AA exhibits a distinct seasonal dependence: it is weakest in summer and strongest in winter of the Northern Hemisphere. Here, we find that global climate models predict that the annual maximum AA (i.e., the greatest difference between Arctic and global temperatures) shifts from autumn into winter over the 21st century. However, this seasonal shift can differ considerably between models in terms of its strength and the number of corresponding months, indicating that substantial uncertainty exists. Our findings highlight the importance of identifying model structural uncertainty in projections of Arctic climate change.