We have investigated how the inclusion of realistic and consistent surface emissivity in both land-surface and atmospheric components of the CESM coupled-climate model affects a wide range of climate variables. We did this by replacing the unit emissivity values in RRTMG_LW for water, fine-grained snow, and desert scenes with spectral emissivity values, and by replacing broadband emissivity values in surface components with the Planck-curve weighted counterparts. We find that this harmonized treatment of surface emissivity within CESM can be important for reducing high-latitude temperature biases. We also find that short-term effects of atmospheric dynamics and spectral information need to be considered to understand radiative effects in higher detail, and are possible with radiative kernels computed for every grid and time point for the entire model integration period.
We find that conventional climatological feedback calculations indicate that sea-ice emissivity feedback is positive in sign, but that the radiative effects of the difference in emissivity between frozen and unfrozen surfaces exhibit seasonal dependence. Furthermore, this seasonality itself exhibits meridional asymmetry due to differences in sea-ice response to climate forcing between the Arctic and the Antarctic. In the Arctic, this seasonal, temporally higher order analysis exhibits increasing outgoing surface emissivity radiative response in a warming climate. While the sea-ice emissivity feedback and seasonal sea-ice emissivity radiative response amplitudes are a few percent of surface albedo feedbacks, the feedback analysis methods outlined in this work demonstrate that spatially and temporally localized feedback analysis can give insight into the mechanisms at work on those scales which differ in amplitude and sign from conventional climatological analyses.
We note that the inclusion of this realistic physics leads to improved agreement between CESM model results and Arctic surface temperatures and sea-ice trends. This reduction of persistent high-latitude model biases suggests that the current unrealistic representation of surface emissivity in model component radiation routines may be an important contributing factor to cold-pole biases.