Traditionally, the goal of parameterization has been seen as the calculation of unclosed, subgrid-scale quantities such as fluxes and cloud fraction. An alternative viewpoint is that the goal of parameterization is the prediction of the subgrid-scale probability density function (PDF). The PDF viewpoint is more general, since the appropriate joint PDF contains all necessary information to diagnose fluxes and cloud fraction and so forth.

PDF methods were first developed many years ago. They offer the possibility of ameliorating many of the difficulties facing current parameterizations, such as closing upgradient fluxes, avoiding trigger functions, maintaining consistency among the various aspects of cloud parameterization, and accounting for within-cloud variability. However, one impediment for PDF methods has been choosing a simple yet sufficiently general family of PDFs.

In this presentation, we will assess the merits and drawbacks of the PDF method. Then we will present data from boundary layer clouds that shed light on which families of PDFs are most suitable for use in mesoscale models.