Prediction of cloud droplet number in a general circulation model

Abstract

A predictive treatment of droplet number is applied to both a single-column cloud model and a global circulation model. Droplet number is predicted from the droplet number balance, which accounts for droplet nucleation, mixing, and droplet loss due to autoconversion of droplets to rain and collection by rain, ice, and snow. Droplet nucleation is parameterized in terms of the parameters of a lognormal aerosol size distribution and in terms of a Gaussian probability distribution of vertical velocity within each grid cell. The predicted droplet number is found to be significantly less than observations unless vertical resolution provides at least 10 levels within the planetary boundary layer. When droplet number is simply diagnosed from the number nucleated, droplet concentrations are found to be consistently greater than with the predictive treatment. Predicted droplet number concentrations are found to be nonlinearly related to aerosol number concentration.