The influence of clouds on Earth's radiation budget in global climate models

Abstract: Considering the high sensitivity of climate to changes in cloudiness, the way clouds might change in a perturbed climate is important for the total temperature response. In fact, the cloud feedback has been identified as the main uncertainty in future projections, as simulated by global climate models.To increase the confidence of future scenario simulations it is necessary, although not sufficient, that the models manage to represent the present-day climate in a realistic manner. We compare the simulations of cloudiness to available observations in the marine subtropics and the Arctic, two regions in which changes in cloudiness are believed to have large impact on the climate.In terms of the annual cycle of Arctic cloud properties, climate models show large disagreement with each other and with observations. There exists a tentative across-model relationship, such that models with higher amounts of clouds in the winter are also associated with larger surface cloud forcing. However, across-model differences in the wintertime surface cloud forcing cannot explain differences in the simulated surface temperatures. Rather, we identify across-model differences in temperature and moisture properties of the air entering the Arctic region to be of greater importance.We find that climate models in general underestimate the amount of low clouds in the marine subtropics but still overestimate the regional averaged cloud radiative cooling. As a consequence we suggest that models are likely to overestimate the radiative response to changes in the cloudiness. We also demonstrate the potential of satellite derived cloud top heights to be used as model diagnostics in the climatologically important transition from stratus-topped to cumulus-topped marine boundary layers in the subtropics.