Simulating Deep Convection Using Translating Large-Eddy Simulations for MC3E

 
Poster PDF

Authors

William I. Gustafson — Pacific Northwest National Laboratory
Raj K Rai — Pacific Northwest National Laboratory
Jiwen Fan — Pacific Northwest National Laboratory
Zhe Feng — Pacific Northwest National Laboratory
Scott Giangrande — Brookhaven National Laboratory
Joseph Clinton Hardin — Pacific Northwest National Laboratory
Die Wang — Brookhaven National Laboratory

Category

Convective clouds, including aerosol interactions

Description

Deep convection is unresolvable in climate models for the foreseeable future. Even so-called “high-resolution” climate models envisioned for a decade from now with grid spacings on the order of a few kilometers will require parameterization for cloud-scale convective motions. The weather forecast community has learned this already, and they continue to parameterize deep convection when using sub-10-km grid spacings. For example, the High-Resolution Rapid Refresh model uses the Grell-Freitas deep convective parameterization while using 3-km grid spacing. Redesigning parameterizations for kilometer scales requires benchmark simulations at much higher resolution, which in this case is in the large-eddy simulation (LES) realm. Domains wide enough to contain large portions of organized convection’s life cycle, i.e., many hundreds of kilometers, are impractical with LES for anything beyond “hero” simulations that are too expensive to be run for the ensembles required for parameterization development and sensitivity studies. Alternatively, we have begun using translating LES domains to simulate mesoscale convective systems, where the domain can be much smaller. Instead of having the entire storm track contained within the domain, the domain translates with the storm and is wide enough to contain a slice of the convectively active region perpendicular to the frontal band. This presentation shows the impact of using a 300-km-wide translating domain by comparing it to a static 800-km-wide domain for the 20 May 2011 MC3E case. While some details differ between the simulations, the translating domain generates similar storm statistics while enabling one to do multiple sensitivity tests with a magnitude less computer resources, depending on the specifics of the domain size. Translating LES domains are potentially a very useful methodology for cloud studies within the DOE Atmospheric System Research community for applications in both process studies of deep convection and parameterization development. This methodology is also an option for future LASSO modeling scenarios.