July 9, 1999

 Global Climate Model Succeeds at Forecasting Local Events
Berkeley Lab Science Beat

Lab website index

Lawrence Berkeley National Lab home page

Search Lab science articles archive
 Advanced Search  
Search Tips
In the first week of May, tornadoes hit Kansas and Oklahoma in one of the most destructive outbreaks in recorded weather history. In the same month, a new world record for annual snow accumulation was reported on Oregon's Mount Hood. Yet the 1990s have been reported as the warmest consecutive years in 150 years of climate record keeping.

Recently climatologists have seen an increase in extreme weather worldwide, including heavy precipitation, flooding, and extended drought -- some of the most unusual weather patterns since weather and climate data collection began. Is this natural climate variability -- or climate change? Or some combination of these processes? The answer is not yet clear.


Working to understand these phenomena and provide communities with weather and climate information they can use, scientists process a flood of data from numerous observation sites on the ground, from satellites in orbit -- and from computer climate simulations as well.

"People have been developing GCMs -- global climate models -- since the late sixties," says Norman Miller, leader of the Regional Climate Center in Berkeley Lab's Earth Sciences Division, "but GCMs, which link global atmospheric and oceanic circulation models, run at fairly coarse resolution -- for example, a global climate model may detect the Rocky Mountains as a single upslope from the Pacific and a downslope to the Midwest, missing the Coastal Ranges and the Sierra Mountains altogether."

While models with resolution at too fine a scale can overwhelm even state-of-the-art supercomputers, Miller says, "our Regional Climate System Model downscales GCMs and other global-scale information to simulate events on the local scale more accurately, such as the danger of flooding in a particular river valley."

The Regional Climate Center’s ultimate goal is accurate prediction at spatial scales from less than a hundred meters to as big as the western United States, and over time scales that vary from hours to seasons to a century or more. While there is a long way to go in this ambitious scientific goal, computational tools such as the Regional Climate System Model (RCSM) pioneered by Miller and Jinwon Kim have already achieved notable successes.

During the winter of 1994-1995, a series of intense storms moved over California and the Southwest. At that time, Miller and Kim were running an early version of the RCSM over the California-Nevada region and on two Northern California watersheds, the Russian River and the North Fork of the American River. Their system linked an atmospheric model, a soil-plant-snow model, and a hydrology-riverflow model to produce simulated predictions of local precipitation and river flow. Miller and Kim compared their simulations against the real storms then taking place.

In the case of the Russian River, 20 years of historical precipitation and river-flow data served to calibrate and verify the hydrology model. It was able to predict the river's disastrous flooding in early January and March of 1995, two days in advance of the actual events.

On the North Fork of the American River, water levels depend upon the accumulation and melting rate of snow, which in turn are primarily controlled by surface temperatures. In 1995 no historical data was available to fully calibrate and verify the hydrologic model; still, Miller and Kim were able to simulate the relative response of the river's flow during the spring melt.

The RCSM's predictive capability was demonstrated again, with improved accuracy, during January 1997 flooding in California. In addition to 48-hour forecasts, promising new seasonal predictions were obtained during this El Niņo winter, firmly establishing the Regional Climate Center's approach to predicting local water resources and weather impacts by linking global climate models to the RCSM.

January of 1999 saw the inauguration of the California Water Resources Research and Applications Center, sponsored by NASA, in which co-investigators from the University of California at Berkeley and UC Santa Barbara work with the Lab's Regional Climate Center in partnership with state and federal agencies. From satellites and other data sources, the new center will produce snow runoff and river forecasting, sediment transport modeling, tracking of landslide and wildfire risks which depend on changing soil moisture, identification of water quality and contaminant problems, and other tangible information. Among the users will be institutions and individuals with a stake in agriculture, forestry, the environment, insurance, and emergency response.

The Regional Climate Center links numerous universities, public and private institutions, and agencies of state and local governments, who provide data to the center and use its output as well. The center works with over a dozen collaborating groups in the United States and with agencies in Australia, Korea, and China, where the RCSM is being used for regional studies in Northeast Australia and East Asia.

Among the center's other wide-ranging activities: a mandate to contribute to the U.S. National Assessment of the Potential Consequences of Climate Variability and Change for the Nation, and the International Panel on Climate Change; a program to advance the RCSM for evaluating the vulnerability of California's Central Valley to flood and drought and its impact on water quality, agriculture, and rural economy, in collaboration with UCLA and UC Davis and sponsored by the Environmental Protection Agency's STAR program (Science to Achieve Results); and ongoing assessments of the Southwestern United States in a close collaboration with the University of Arizona's NASA-sponsored RESAC and other groups.

"We work as a synergistic group of centers," says Miller, "sort of a big web with lots of complementary nodes."

The power of massively parallel computers at the National Energy Research Scientific Computing Center (NERSC) is crucial to refining the spatial resolution of the RCSM and increasing its usefulness. Working on the Cray T3E under a NERSC grant, the Regional Climate Center has completed an initial study of the long-term effects of increased atmospheric carbon dioxide on the climate of the western United States. NERSC's new IBM RS/6000 will provide an even more powerful means of bringing climate simulations down to earth, making it possible to model the entire continental U.S. at a resolution below 36 kilometers.

There are signs that the climate is indeed changing, says Miller. "Increased heating of the lower atmosphere is generally accepted by the scientific community; the increased atmospheric energetics and the ability to hold more water in the lower atmosphere may lead to more extreme weather events occuring more often."

Specific weather events such as the recent, historically destructive tornadoes in the Great Plains can not be ascribed to climate change directly, Miller says, but he would not be surprised to see an increase in unusually powerful storms in the future.

The Department of Energy's Office of Science is emphasizing the vital importance of climate models, accurate prediction, and useful information accessible to local communities through its Accelerated Climate Prediction Initiative. Berkeley Lab's Regional Climate Center is already demonstrating that the job can be done.

Additional Information: