Search Magazine 
Article IndexNext ArticlePrevious ArticleFeedback to EditorORNL Review Home Page
Some of the world’s largest global climate models are being run on ORNL’s supercomputers, providing insights for national and international assessments of the effects of global warming caused by human activities.

World-Class Climate Modeling

When Warren M. Washington, a member of the National Science Board, which advises the Executive Branch and Congress on science-related matters, conducts his research, he ponders what will happen 10, 50, or 100 years from now. As head of the Climate Change Research Section at the National Center for Atmospheric Research (NCAR) in Boulder, Colorado, Washington, along with his colleagues, tries to predict how the climate will change in the next century under various conditions. And for Washington, the most interesting hardware right now is the new supercomputer at ORNL named after the cheetah, the fastest land mammal.

Cheetah is the new IBM Power4 supercomputer, a machine rated at 4 teraflops (4 trillion calculations per second) that has 24 “regatta” nodes, each having 32 processors. This supercomputer has 1 terabyte of memory and 24 terabytes of disk space. It is located at DOE’s Center for Computational Sciences (CCS) at ORNL, which is also home for the IBM RS/6000 SP and Compaq AlphaServer SC machines, which together provide 1.5 teraflops of computing power.

Washington and his colleagues have been using the ORNL supercomputers for century-long climate model runs to simulate changes in the world’s climate from 1870 to 2170 under several different greenhouse-gas scenarios. In the business-as-usual case, atmospheric carbon dioxide (CO2) levels steadily increase, trapping heat and causing global warming. Under the “stabilization” scenario, atmospheric CO2 concentrations rise and then level off in response to various nations’ carbon management strategies. The stabilization scenarios assume that CO2 emissions from fossil fuel power plants are reduced and that enhanced absorption and sequestration of CO2 in land plant life and ocean waters result in a significant slowing of the accumulation of CO2 in the atmosphere.

David Erickson and José Hernandez compare raw and statistically-analyzed data
David Erickson (left) and José Hernandez compare raw and statistically-analyzed data from a global climate simulation in ORNL’s CAVE. (Both photos by Curtis Boles)

The DOE-sponsored Parallel Climate Model being run on ORNL supercomputers results from a joint effort involving NCAR, ORNL, DOE’s Los Alamos National Laboratory (LANL), the Naval Postgraduate School, and the U.S. Army Corps of Engineers’ Cold Regions Research and Engineering Laboratory. John Drake in ORNL’s Computer Science and Mathematics Division (CSMD) led the effort that enabled the climate model to be modified so that it could be run on massively parallel supercomputers at ORNL. The Parallel Climate Model brings together the NCAR Community Climate Model version 3, the LANL Parallel Ocean Program, and a sea-ice model from the Naval Postgraduate School in a massively parallel computer environment.

Drake and David Erickson (formerly with NCAR, now with CSMD) are using the ORNL supercomputers for climate predictions in which the interactions of the atmosphere with land and with the oceans are simulated. They plan to couple carbon and climate models together with help from researchers Mac Post and Tony King in ORNL’s Environmental Sciences Division. The project will be multi-institutional, involving colleagues from other DOE labs, as well as a number of universities. Because CO2 is emitted by energy production and because atmospheric increases in the greenhouse gas contribute to global warming, combining carbon and climate modeling fits well into the DOE mission of finding ways to produce energy in an environmentally acceptable fashion. The ORNL researchers plan to perform simulations using a coupled climate-carbon model to determine how the carbon budget would change in a greenhouse-warmed world.

In addition to the prescribed increases in CO2, the model will be allowed to “run free” with the climate and carbon cycle evolving in unison to reach different future greenhouse-gas, or climate, states. The combined models will address these questions: How will increased atmospheric carbon dioxide affect desert size? How will it affect the frequency of droughts, precipitation, and severe storms—such as tornadoes and hurricanes—in different regions?

“Eventually, we hope to do climate modeling that is useful to public health service planners,” Erickson says. “Our modeling could indicate when and where temperature and moisture conditions are likely to be conducive to various insect-borne diseases.”

These images show how wind, salinity, and temperature vary between January and July in the northern portion of South America.
These images, produced by computer simulations, show how wind, salinity, and temperature vary between January and July in the northern portion of South America.

When climate modeling experts look into the details of the climate system, what they see is often cloudy. Emissions of particles from fossil fuel plants can have confounding effects on the warming of the earth’s surface, making predictions less certain. While CO2 and other greenhouse gases in the atmosphere absorb infrared radiation from the earth’s surface and prevent the escape of heat, sulfate aerosols from coal-fired power plant emissions can have a cooling effect, moderating the temperature signal and changing weather patterns. Other challenges faced by the climate modelers are the uncertain effects on the radiative signal of clouds, the motion of sea ice, airborne dust from African deserts, and variability in Pacific Ocean surface temperatures. Research areas include the response of the oceans to rising temperatures, the effect of aerosols on cloud formation, and the feedback between the climate and carbon and water cycles.

The availability of the ORNL supercomputers allows U.S. climate researchers—for the first time—to make an ensemble of predictions for each future climate scenario. This capability enables a more detailed assessment of the variability and error estimates in the simulations, thus reducing uncertainties of model predictions. These climate predictions by Oak Ridge researchers and by NCAR’s Washington and others using ORNL supercomputers will provide timely information for national climate change assessments and reports compiled by the Intergovernmental Panel on Climate Change (IPCC).

Based partly on results from global climate models, the IPCC concluded in 2001 that “there is new and stronger evidence that most of the warming observed over the last 50 years is attributable to human activities” and projected that, by the end of 2100, the global average temperature of the earth could rise by 1.5 to 5.8°C. Policymakers are now paying attention to what the IPCC sees as the future of the earth’s climate.

Beginning of Article

Related Web sites

National Center for Atmospheric Research
ORNL Computer Science and Mathematics Division
Center for Computational Sciences

Search Magazine  
Article Index Next Article Previous Article Feedback to Editor ORNL Review Home Page

Web site provided by Oak Ridge National Laboratory's Communications and External Relations
ORNL is a multi-program research and development facility managed by UT-Battelle for the US Department of Energy
[ORNL Home] [Communications] [Privacy and Security Disclaimer]