Advanced Computing and Data Visualization

Christopher L. Kerr and Isidoro Orlanski
Geophysical Fluid Dynamics Laboratory, OAR

The GFDL is pleased to announce the completion of Project TERRA. Project TERRA was conceived for running a one day simulation with the solo explicit convection ZETAC model for the globe with an averaged grid resolution of 10Km to 12Km. This is the first GMCM (General Mesoscale Circulation Model) run at GFDL and since it is non-hydrostatic with explicit convection, probably the first anywhere.

Global models of similar horizontal resolution, but hydrostatic, have been run successfully on the Earth Simulator (Japan's Supercomputer). Most if not all the meteorological centers around the world run hydrostatic GCM's with coarser resolutions. Although they are simpler and faster to run they have a major drawback that can not simulate cloud evolution explicitly. Those models depend on physical parameterization to derive the cloud structure. The cloud distributions are of great importance in weather and climate. The correct distribution of latent heat produced by the moist convection feeds directly into the dynamics that generate weather systems.  The vertical distribution of cloud systems is also of paramount importance in the radiative characteristics of the atmosphere that affects climate and climate change. The present GCM's for weather and climate prediction uses parametrized convection that tries to simulate how cloud distribution develops in each weather system.

The model used in project TERRA is of a different breed, being non- hydrostatic with explicit moist convection (cloud resolving). The model can simulate from a single cloud to the entire global circulation. It is presently used at GFDL in a limited area mode (1/6 of the globe) for studying the impact of sea-surface temperatures in storm track dynamics. TERRA was an ambitious project to push to the limit our ability to run such models on current-generation computer systems. The computer configuration was an Silicon Graphics system with 512 processors. Although the present speed of the computer systems makes the use of such model impractical for weather prediction or climate research, it is envisioned that in a few years' time we will be able to have a major breakthrough in weather forecasting and climate prediction by using cloud resolving models like ZETAC. We have proven that the model is robust and ready; we only need the appropriate hardware for running faster.