Comment on acp-2022-259

Abstract

The Southern Ocean radiative bias continues to impact climate and weather models, including the Australian Community Climate and Earth System Simulator (ACCESS). The radiative bias, characterised by too much shortwave radiation reaching the surface, is attributed to the incorrect simulation of cloud frequency and phase. In this work, we use k-means cloud clustering, combined with nudged simulations of the latest generation ACCESS atmosphere model, to evaluate cloud and radiation biases when cloud types are correctly and incorrectly simulated. We find that even if the ACCESS model correctly simulates the cloud type, biases of equivalent, or in some cases greater, magnitude then when they are incorrectly simulated remain in the cloud and radiation fields examined. Furthermore, we find that even when radiative biases appear small on average, cloud property biases, such as liquid or ice water paths or cloud fractions remain large. Our results suggest that simply getting the right cloud type (or the cloud macrophysics) is not enough to reduce the Southern Ocean radiative bias. Furthermore, in instances where the radiative bias is small, it may be so for the wrong reasons. Considerable effort is still required to improve cloud microphysics, with a particular focus on cloud phase.