Early Career: Reducing Scale Dependence of Physics Parameterizations for Global Cloud Resolving Climate Models
EMSL Project ID
47809
Abstract
New climate models with variable resolution grids combined with the anticipated two-order reduction in climate model grid spacing, from roughly 100's to 1's km grid spacing, in the next decade offers great potential for providing more detailed information for regional planning purposes as well as increased accuracy for processes that cannot be estimated well at coarser scales. However, multi-resolution grids introduce significant barriers for current atmospheric physics parameterizations in climate models, i.e. the handling of sub-grid phenomena such as clouds, radiation, aerosols, and turbulence. Ultimately, a set of parameterizations is needed that senses the underlying grid spacing and adapts accordingly to use assumptions appropriate for that resolution. This will allow users to pick the grid spacing appropriate for their problem, and will enable the use of variable grid resolutions for global model domains. Unfortunately, knowledge about the resolution dependence of the current generation of parameterizations is limited due to interactions between different parts of the parameterization suite, which mask the behavior of specific components. To get around this issue, this project has developed a framework called the Separate Physics and Dynamics Experiment (SPADE) whereby individual parameterizations can be run at arbitrary resolutions, enabling one to determine which parameterizations are most susceptible to resolution induced differences in behavior. This methodology can be configured to either completely isolate particular parameterizations to understand how they behave at different resolutions given input from a constant resolution, or it can be configured to allow interactions between model components at different scales such that the model scale is constant for most processes but varies for the processes of interest. This provides a unique tool for improving model performance.
In addition to developing and perfecting the SPADE framework, the objectives of this project include quantifying the resolution dependence of current physics parameterizations, identifying best practices for developing resolution aware physics parameterizations for climate models, and ultimately to develop methodologies that enable parameterizations to work from the cloud-scale resolving to the mesoscale model scales. Through these objectives, this project will quicken the adoption of multi-resolution grids in climate models, which will both simplify using these models for targeted downscaling applications, as well as improve model accuracy and efficiency. Until a suitable resolution aware parameterization suite is developed, the new climate model cores with multi-resolution grids will be severely hampered and unable to reach their full potential.
EMSL resources are required to achieve these objectives through the use of the EMSL supercomputer and the ability to achieve better turnaround time on the EMSL managed supercomputer for mid-level commodity computing than on other DOE facilities.
Project Details
Project type
Large-Scale EMSL Research
Start Date
2013-10-01
End Date
2015-09-30
Status
Closed
Released Data Link
Team
Principal Investigator
Team Members
Related Publications
Xiao H, WI Gustafson, Jr, and H Wang. 2014. "Impact of subgrid-scale radiative heating variability on the stratocumulus-to-trade cumulus transition in climate models." Journal of Geophysical Research. D. (Atmospheres) 119(7):4192–4203. doi:10.1002/2013JD020999