Evaluation of autoconversion schemes in a single model framework with satellite observations

We examined the performance of autoconversion (mass transfer from cloud water to rainwater by the coalescence of cloud droplets) schemes in warm rain, which are commonly used in general circulation models. To exclude biases in the different treatment of the aerosol-cloud-precipitation-radiation interaction other than that of the autoconversion process, sensitivity experiments were conducted within a single model framework using an aerosol-climate model, MIROC-SPRINTARS. The liquid water path (LWP) and cloud optical thickness have a particularly high sensitivity to the autoconversion schemes, and their sensitivity is of the same magnitude as model biases. In addition, the ratio of accretion to autoconversion (Acc/Aut ratio), a key parameter in the examination of the balance of microphysical conversion processes, also has a high sensitivity globally depending on the scheme used. Although the Acc/Aut ratio monotonically increases with increasing LWP, significantly lower ratio is observed in Kessler-type schemes. Compared to satellite observations, a poor representation of cloud macrophysical structure and optically thicker low cloud are found in simulations with any autoconversion scheme. As a result of the cloud-radiation interaction, the difference in the global mean net cloud radiative forcing (NetCRF) among the schemes reaches 10 Wm^-2. The discrepancy between the observed and simulated NetCRF is especially large with a high LWP. The potential uncertainty in the parameterization of the autoconversion process is nonnegligible, and no formulation significantly improves the bias in the cloud radiative effect yet. This means that more fundamental errors are still left in other processes of the model.