A parcel model-based approach for simulating cloud droplet number concentrations and climate

Atmospheric aerosols influence microphysical and optical properties of clouds and therefore cause indirect effects on climate. A major challenge for the climate research community to date has been to determine radiative forcings associated with indirect effects of aerosols on climate. A key parameter for cloud microphysical processes in warm clouds and aerosol indirect effects is the cloud droplet number concentration. Very different parameterizations are currently employed in climate models to represent effects of aerosols on cloud droplet number concentration, ranging from empirical relationships based on observations to complex physically-based parameterizations of droplet nucleation from activation of aerosols. On the other hand, detailed but numerically expensive cloud parcel models are often successfully used in process studies to analyze aircraft observations of cloud droplet number concentrations and for validation of parameterizations. Parcel models generally do not employ any of the approximations which are commonly used in climate models. Here we show, for the first time, that cloud droplets can be accurately simulated by using a global climate model with an embedded parcel model. A numerically highly efficient parcel model has been developed and implemented in a developmental version of the Canadian Atmospheric Global Climate Model with aerosol microphysics (CanAM4-PAM). In CanAM4-PAM, simulated aerosol size distributions and aerosol chemical composition are used for prognostic simulations of the activation and growth of aerosol particles to cloud droplets in a single updraft above cloud base. The approach accounts for subgrid-scale turbulence and entrainment of air from the clear-sky environment of the clouds. Simulation results for cloud droplet number concentrations in low clouds from CanAM4-PAM compare well with retrieved cloud droplet number concentrations based on satellite data from MODIS. An analysis of simulated cloud properties and radiative forcings provides evidence for substantial model improvements compared to an empirical parameterization of cloud droplets that is currently used in the operational version of CanAM4.