Philip Duffy, Bala Govindasamy, Jose Milovich, and Starley Thompson, Lawrence Livermore National Laboratory

Wintertime precipitation in the U.S. as observed and as simulated by our climate model at three different resolutions: 300 km, 75 km, and 50 km. As the model resolution becomes finer, the results converge towards observations.

Research Objectives
Our goal is to perform global climate simulations using a very high-resolution model of the atmosphere, thus improving the realism of the model and the quality of predictions on both global and regional scales.

Computational Approach
We run the CCM3 global atmospheric model at spectral truncations of T170 and T239. The corresponding grid cell sizes are ~75 km and ~50 km, respectively. This is in contrast to a grid size of ~300 km in typical global climate simulations. We use a version of this model known as CCM 3.10.11 (366physics), which achieves distributed memory parallelism using a combination of MPI-based message passing and multiple threading using OpenMP. The model physics were tuned to give good results at high resolution in collaboration with the model developers at the National Center for Atmospheric Research (J. Hack et al.).

Accomplishments
Using NERSC's IBM SP and supercomputers at Lawrence Livermore National Laboratory, we ran a global climate simulation at 50 km resolution, the highest spatial resolution ever used for a global climate simulation. Compared to a typical global climate simulation, this 50 km simulation has 32 times more grid cells and takes roughly 200 times longer to run. Our goal for the 50 km simulation is to evaluate how well the model simulates the present climate at this resolution. Thus far we have run about five simulated years; preliminary analysis of the results seems to indicate that the model is very robust to a large increase in spatial resolution.

Significance
Global climate simulations are typically performed on a latitude-longitude grid, with grid cell sizes of about 300 km. Although simulations of this type can provide useful information on continental and larger scales, they cannot provide meaningful information on regional scales. Thus, coarse-resolution global climate simulations cannot provide information on many of the most important societal impacts of climate change, such as impacts on water resource management, agriculture, human health, etc. By using much finer spatial resolution, we hope to improve the realism of the models and produce better predictions of future climate, specifically of anthropogenic climate change, on both global and regional scales.

Publications
P. B. Duffy, B. Govindasamy, J. Milovich, S. Thompson, and M. Wehner, "Simulation of global climate using a high resolution atmospheric general circulation model" (in preparation).

http://en-env.llnl.gov/cccm/