Linking Cloud-Radiation Parameterization Performance to Large-Scale Dynamics

Parameterization of cloud-radiation interactions in an atmospheric general circulation model using a stochastic approach allows for a statistical representation of cloud field geometry. Previous work indicates that the stochastic approach to cloud-radiation parameterization performs better than typical plane-parallel algorithms in situations where the horizontal cloud fraction is between 0.2 and 0.8. However, this does not completely describe the physical situations when a stochastic approach to modeling cloud-radiation interactions is appropriate. Cluster analysis has been applied to data of the atmospheric state, cloud physical, and dynamical characteristics, and radiative transfer model results to determine the situations in which an atmospheric general circulation model would make best use of a stochastic cloud-radiation parameterization. Preliminary results from the ARM Climate Research Facility (ACRF) Tropical Western Pacific (TWP) locale indicate that situations involving deep convection may benefit the most from a stochastic approach.