A missing physics in climate models for the simulation of Southern Ocean: longwave radiative coupling between surface and atmosphere

Climate models usually neglect the treatments of surface spectral emissivity and cloud scattering in their longwave (LW) radiation scheme. Recent studies have demonstrated that this negligence causes biases for the simulated polar climate. This study aims to quantify the influence of these LW treatments on the simulated climate of the Southern Ocean. Using a modified NCAR CESM ver 1.1.1 slab-ocean model that includes both surface spectral emissivity and ice cloud LW scattering, we carried out three numerical experiments: noScat_black (non-scattering ice clouds with blackbody surface), Scat_black (scattering ice clouds with blackbody surface), and Scat_emis (scattering ice clouds with realistic surface spectral emissivity). Each experiment consists of four members and each member ran for 35 years, with the last 30 years being used in analysis. By contrasting the simulated mean climate of each experiment, we can quantify the individual effects of surface emissivity, cloud LW scattering, and the combined effect together. The results show that the inclusion of either surface emissivity or ice cloud LW scattering can each lead to an increase of zonal-mean surface air temperature by 1 K, and correspondingly, a decrease in sea ice fraction by 5%. These changes have a distinct dependence on season, with the largest difference in winter and the smallest difference in summer, presumably because the LW radiation coupling plays a more important role in the high-latitude winter than in summer. In terms of precipitation changes, each effect increases zonal-mean precipitation rate by 5%, even larger increases being seen in marginal sea ice regions. The effects of surface spectral emissivity and cloud LW scattering are largely additive. Mechanisms for these changes as well as the influences on cloud fields, surface energy budget, and other climate features will be further discussed.