PDF cloud scheme
Our research into cloud schemes focuses on how to represent clouds and cloud processes in numerical models which do not have sufficient resolution to resolve clouds. As both weather forecast or climate models fall into this category this research has high societal and academic relevance.
There are many possible way to parametrize clouds, but the approach we focus on is called a PDF scheme. PDF stands for probability density function, and the scheme receives its name from the analytical PDF it uses to describe the distribution of total water in each model cell. The scheme we use is derived from the scheme Adrian Tompkins presented in his 2002 paper “A Prognostic Parameterization for the Subgrid-Scale Variability of Water Vapor and Clouds in Large-Scale Models and Its Use to Diagnose Cloud Cover” which Vera Schemann revised and extended in her 2014 PhD Thesis “Towards a scale aware cloud process parametrization for global climate models”.
As part of the HD(CP)2 Project we evaluate and extend the PDF scheme using information from recently developed super-domain LES Simulations, an example of which is plotted above. These Simulations have sufficiently high resolution to resolve clouds explicitly, and provide new and unique chances to test the assumptions on which our PDF scheme is based and quantify the PDF scheme performance.
Here is an example of the total water histogram of a 55×55 km subdomain of the LES Simulation containing over 125000 grid cells. Overlaid are two PDFs from our scheme with slightly differing assumptions, as well as the resulting cloud fractions and how much they overlap with the histogram.
With these tools we can now quantify many aspects of cloud parametrizations which were previously not measurable. Such aspects include how parametrization performance changes with model resolution, and breaking down the various error sources.
Related publication: Griewank et al, 2018: https://doi.org/10.1029/2018MS001421