The U.S. government is launching a $50-million effort to enable supercomputer-powered climate models to deliver regional impacts
David Biello Scientific American 23 Mar 10;
Nobody lives in the global average climate. Nor are the massive grid cells favored by climate models run on today's supercomputers as useful as they could be for planning purposes, given that they can encompass 10,000 square kilometers. Now the National Science Foundation (NSF), along with the U.S. Energy and Agriculture departments are teaming up to financially support the development of new computer models aimed at revealing the anticipated effects of climate change at the regional level.
"The impacts of climate change are becoming more immediate and profound than anticipated," NSF Director Arden Bement said Monday during a Webcast for journalists. "We must be able to predict how climate change will impact... regions in the next 10 to 20 years."
The goal will be to deliver a scientific basis for regional planning purposes, whether that involves adaptation to a disappearing coastline or to the expected severity of droughts. "Some will get down to parts of the U.S., part will get down to statewide levels. Some might even deal with megacities," Bement said. "That's going to take time as we continue to peel the onion down to smaller scales."
The agencies will invest nearly $50 million annually in the new program, dubbed "Decadal and Regional Climate Prediction Using Earth System Models," or EaSM for short, and expect to deliver improved versions within the next three years. Those models will look at impacts such as regional average temperature change, sea-level rise, ocean acidification, and the sustainability of soils and water as well as the impacts of invasive species on food production and human health. Nor will it be confined to ecological concerns. "One really can't have energy sustainability without environmental sustainability. And you can't have either one without economic sustainability," Bement noted.
A big part of the effort will rely on advances in computer power; the Department of Energy (DoE) now hosts the world's most powerful supercomputer at its Oak Ridge National Laboratory. Such petaFLOP-scale (quadrillion-operation-per-second) computers will help scientists to improve both the time and spatial scales of their models. "We love to be able to go exoscale—another 1,000 times faster and bigger," William Brinkman, head of the DoE Office of Science, said during the Webcast. "Climate modeling is probably the driving force to continue up that direction, more than any other modeling."
Of course, such computer models are only as good as the entered data, so additional information from ongoing observation campaigns—whether that is the DoE's Atmospheric Radiation Measurement program or the NSF's National Ecological Observatory Network—will be needed to provide a fuller understanding of how soot and other aerosols impact the global climate as well as the role of clouds.
Ultimately, the biggest impacts to be felt regionally may be on agriculture. "Producers of food will need to know what to expect in the future to be ready for the kinds of changes that are anticipated," said Department of Agriculture chief scientist, Roger Beachy. "We are concerned about the impact on our ability to grow food."
For its part, Agriculture hopes to be able to determine what the overriding impacts and concerns might be for a given multistate region as well as offer advice on farming practices that might curtail agricultural contributions to greenhouse gas emissions (via such processes as fertilizer production and use, plowing practices, and rice paddy methane release). "We need to have the data before we embark on big policy changes that will have impacts on all of us," Beachy said. "This will have long-term impacts for generations and generations. It's important to get this right."
