Inter-model variability in convection-resolving simulations of subtropical marine low clouds

Abstract

We analyze a multi-model ensemble at a convection-resolving resolution based on the DYAMOND models and a resolution ensemble based on the limited-area model COSMO over 40 days to study how tropical and subtropical marine low clouds are represented at a kilometer-scale resolution. The analyzed simulations produce low cloud fields that look in general realistic in comparison with satellite images. The evaluation of the radiative balance, however, reveals substantial inter-model differences and an underestimated low cloud cover in most models. Models that simulate increased low cloud cover are found to have a deeper marine boundary layer (MBL), stronger entrainment, and an enhanced latent heat flux. These find-ings demonstrate that some of the fundamental relations of the MBL are systematically represented by the model ensemble, which implies that the relevant dynamical processes start to become resolved on the model grid at a kilometer-scale resolution. A sensitivity experiment with the COSMO model suggests that differences in the strength of turbulent vertical mixing may contribute to the inter-model spread in cloud cover.