An important task for ORNL and NCAR will be to perform climate change simulations in support of the Intergovernmental Panel on Climate Change Fourth Assessment. The global climate change assessment community is eagerly awaiting the results, which will reflect the input from a number of institutions worldwide. The results will be announced in 2007. The collaboration between ORNL and NCAR pools the vast simulation resources and scientific talent of the two institutions, which have considerable staff and computational hardware that can be used to explore novel experimental and computational designs. John Drake of the lab's Computer Science and Mathematics Division envisions the partnership leading to innovative solutions to problems and to more sophisticated models that focus on global carbon cycle, dynamic vegetation in the land model, more realistic hydrology and river routing, and progress in a number of other areas. As part of this effort, the Center for Computational Sciences is providing dedicated computing resources to perform climate change simulations. The simulations could require several months to complete, even with 12 nodes (384 processors) of the IBM Cheetah's 27 nodes assigned to the task. And, equally important, the collaboration focuses on making the most of NCAR and ORNL's vast computing resources by benchmarking new architectures and evaluating programming models required for climate simulations. "The re-emergence of Cray for scientific computation has changed the landscape of parallel computing," Drake said. "We expect to make very good use of the Cray X1 for atmosphere, ocean, land and sea ice climate simulations." In recent years, ORNL and NCAR collaborated in the development of the Community Climate System Model and continue to study a broad range of scientific and technical challenges in modeling the Earth's climate. The CCSM simulates Earth's climate by performing modeling in the atmosphere, ocean, land and sea ice. The model is fully integrated and provides state-of-the-art computer simulations of the Earth's past, present and future climates. Global climate modeling is considered vital to understanding potential effects of increased greenhouse gases and to evaluating likely effects of policies. As computing power continues to increase, models can become more sophisticated and take into account many more variables, thereby increasing the accuracy. "The sustained partnership between NCAR and ORNL has contributed to our increasingly robust understanding of the causes and characteristics of past changes in Earth's climate," said NCAR Director Timothy Killeen. "The next generation of model, made possible though our continued collaborative work, will enable exciting new tools for scientists and decision makers interested in the nature and extent of future changes in the Earth system." The Center for Computational Sciences, established in 1992 as a DOE high-performance computing research center, is a designated user facility focused on grand challenges in science and engineering. The center is housed in a new 170,000-square-foot building with a 40,000-square-foot computer center. It is the nation's largest facility for unclassified scientific research.