Taming the ‘butterfly effect’
Supercomputer is expected to reduce the chaos factor in weather science.
CHEYENNE, Wyo. — Here in the shortgrass prairie, where being stuck in the ways of the Old West is a point of civic pride, scientists are building a machine that will, in effect, look into the future.
This month, on a barren Wyoming landscape dotted with gopher holes and hay bales, the federal government is assembling a supercomputer 10 years in the making, one of the fastest computers ever built and the largest ever devoted to the study of atmospheric science.
The National Center for Atmospheric Research’s supercomputer has been dubbed Yellowstone, after the nearby national park, but it could have been named Nerdvana. The machine will have 100 racks of servers and 72,000 core processors, so many parts that they must be delivered in the back of a 747. Yellowstone will be capable of performing 1.5 quadrillion calculations — a quadrillion is a 1followed
by 15 zeros — every second.
That’s nearly a quarter of a million calculations, each second, for every person on Earth. In a little more than an hour, Yellowstone can do as many calculations as there are grains of sand on every beach in the world.
The study of climate and weather patterns has always been hamstrung by volatility — by elements of chaos in the seas and the air. That challenge is most famously summed up by the “butterfly effect,” the idea that the flapping of a butterfly’s wings on the coast of Africa can determine whether a hurricane will strike New Orleans.
The sheer speed of Yellowstone is designed to burst through the limits of chaos theory — the difference, allegorically, between predicting the odds of blackjack after playing five hands versus playing a million. The machine is expected to give scientists a clearer image of the state of the planet, and its future, revolutionizing the study of climate change, extreme weather events, wildfires, air pollution and more.
“These are chaotic systems, but it’s just math,” said Richard Loft, director of technology development at NCAR’s Computational and Information Systems Laboratory. “We play statistics in the climate game. We feed in the basic laws of science, and out comes something that looks like the Earth’s climate. It’s an instrument. This is a mathematical telescope.”
NCAR is in the business of research, not forecasting, but the tools and advances produced from its research could have a profound effect on forecasting. Armed with a high-fidelity portrait of Earth systems, scientists around the United States can begin to pinpoint the regional impact of changes in the weather and atmosphere.
Rather than warning of a tornado risk in the central U.S. between noon and 9 p.m., scientists might one day warn of a tornado risk in Woodson County, Kan., between 1 and 3 p.m. Rather than warning of a hurricane striking the coast of Texas, they hope to be able to warn of a hurricane striking the town of Freeport, with a top wind speed of 90 mph and a tidal surge of 4 1/2 feet.
That regional accuracy is particularly critical in the study of climate change. “The disaster of climate change happens on a regional scale,” Loft said. “Everything is connected.”
For example, once scientists use Yellowstone to help predict the melting of ice at the North Pole, which means significant change in nearby waters, they can better predict the patterns of storms that form in the Gulf of Alaska. Then Yellowstone can help predict how those storms will deposit snow atop the Sierra Nevada, down to precise changes in elevation on individual faces of mountains.
That snow will melt, and the water will run downhill — which means Yellowstone can help predict how much water California will have to drink, even the most efficient locations to build the state’s reservoirs.
“It’s taking the macro information and applying it to the things that matter,” said Richard Neale, an NCAR project scientist.
The computer will be housed in a futuristic, $70million compound west of Cheyenne. The National Science Foundation, which funds NCAR, is paying $50 million of the tab. The state of Wyoming will pay for the rest. In exchange, the state will occupy a dedicated chunk of the computer’s power and memory. University of Wyoming scientists hope to use Yellowstone to advance “carbon sequestration,” a promising method of storing harmful gases underground to combat climate change and open new avenues in industry.
Yellowstone will replace NCAR’s Bluefire system, a supercomputer in its own right, though this one will have roughly 30 times the throughput of the old system.
Yellowstone will hold 600 sets of atmospheric data in its vast memory bank — temperatures, humidity, wind motion, rainfall. Information gleaned from the world’s data-collection systems — buoys in the ocean, wind monitors fastened to the top of telephone poles — will be added to the archive.
The Wyoming compound is one of the most energy-efficient sites of its kind. Heat generated by the computer will be recycled to warm workers’ offices, and pipes carrying the coolant water will have few 90-degree angles; pumping water through pipes that bend at gentle angles requires less energy.
The bulk of the machine will arrive at the 24-acre compound this month. Once the racks are lashed together, scientists will spend weeks “basically trying to break the system,” said Anke Kamrath, director of operations and services at NCAR’s systems lab. If they can’t, they’ll open it to researchers from across the nation, probably in August. Scientists will make proposals to book an “allocation” on the computer, similar to using minutes on a cellphone plan. Most will access the computer remotely.
Many are champing at the bit. Some hope to predict migration patterns of animals, others the success and failure of certain farm crops, others specific hillsides that would be the most efficient spots for wind turbines.
NCAR scientist Michael Wiltberger studies solar flares, superheated gas that emanates from the sun, with the potential to be enormously disruptive on Earth.
“Right now, we don’t know why a particular configuration of the magnetic field of the sun is going to erupt,” Wiltberger said. “We need to know — and now we can run millions times more models to provide meaningful predictions.”
Armed with better predictions of what will happen when solar flares reach Earth — and where, precisely, they will occur — scientists could warn energy companies to protect against power surges. Global positioning systems could be disrupted, so farmers that use GPS to map crops could be warned to suspend planting operations.
NCAR senior scientist Morris Weisman specializes in a tricky corner of science: severe, high-impact weather events, which are by definition so rare that they are difficult to predict. “Scientifically non-satisfying” is how Weisman puts it — but with such a leap in computer modeling, he said, scientists could theoretically predict an extreme weather event “within an hour, within a few kilometers.”
“We can conceive of that now,” he said. “It’s really exciting.”
Loft marveled that such a dizzying array of experiments will be done using time-tested and sometimes rudimentary math — 19th century laws of thermal dynamics, rules of mechanics devised by Isaac Newton after an apple supposedly bonked him on the head and got him thinking about gravity. Yellowstone will use the same, just a whole lot of it at once.
“Newton was thinking about apples, and the moon. He never would have imagined that the same principles would have dictated the behavior of hurricanes,” Loft said. “This science stuff works.”
The scientists behind Yellowstone shrug at a bitter reality: cutting edge doesn’t last long in their world. The Wyoming facility was built with enough space to accommodate the next generation of computer, which is already being contemplated, before this one is put together.
“We won’t be cool for long,” Loft said. “This business is ephemeral. There’s not much room for nostalgia.”