A Novel Framework for Introducing Full Subgrid Variability into Physics Parameterizations in Climate Models

Monday, May 12, 2014 - 07:00
Add to Calendar

Subgrid-scale variability is one of the main reasons why parameterizations are needed in large-scale models. Although some parameterizations started to address the issue of subgrid variability by introducing a subgrid probability distribution function (PDF) for each relevant quantity, spatial structure has been typically ignored and thus the subgrid-scale interactions cannot be accounted for in a physical manner. Coherent structures are found at scales ranging from droplet clusters to organized cloud systems. Here we first present a new statistical-physics-like approach whereby the spatial autocorrelation function and PDF are used to physically capture the effects of subgrid cloud interaction with radiation. The new approach is able to faithfully reproduce the Monte Carlo 3D simulations with several orders less computational cost, allowing for more realistic representation of cloud radiation interaction in large-scale models. We then present empirical analysis of long-term measurements at the ARM sites, with a focus on lower-order moments (mean, standard deviation and skewness), PDF characterization, and scale-dependence.