The Impacts of Springtime Cloud and Radiation Properties on Arctic Sea Ice Changes in CESM

The cloud-radiation feedback is believed to play an important role in the amplification of Arctic warming. Previous studies have demonstrated through both surface and satellite observations that increasing clouds and their warming effects are partially responsible for accelerated Arctic sea ice decline. Here we investigate the role of springtime cloud-radiation feedback in modulating seasonal-to-interannual scale Arctic sea ice change utilizing the NCAR Community Earth System Model Large Ensemble (CESM-LE). Based on CESM-LE simulations, it appears that the observed positive linear trends of Arctic clouds in the early 21st century (2006-2021) may be due to internal variability. The cloud-radiation relationships in CESM-LE are compared with NASA Clouds and the Earth’s Radiant Energy System (CERES) satellite retrievals and other surface observations over the Arctic. The simulated downward longwave flux at the surface is more sensitive to changes in cloud water path (CWP) than cloud fraction (CF), which is attributed to a relatively large negative model bias in CWP over the Arctic. Therefore, the CWP in the spring has larger impacts on September sea ice changes than CF, although they both exhibit negative correlations with September sea ice concentration over the Arctic. On a seasonal basis in the present climate, the cloud warming (longwave) effect on the sea ice melting increases from March to June in CESM-LE, while it is found to be strongest in April and May from satellite retrievals. With decreased sea ice under the representative concentration pathway 8.5 (RCP8.5) scenario, the impacts of cloud and radiation properties on sea ice melting tend to shift to early spring in the middle and late 21st century. To further identify the causality in the relationship between changes in sea ice with changes in clouds and radiation, we examine the cloud and radiation response under different September sea ice extend linear trends derived from selected CESM-LE ensemble members during 2006-2021 in an Atmospheric Model Intercomparison Project (AMIP) experiment.