In the last century, the Earth has gotten warmer … fast. Glaciers are melting in the Arctic— swallowing islands, drowning polar bears, and changing regional patterns of ocean circulation. With techniques weathermen have been using for 30 years, today’s climatologists have just begun to forecast these changing ocean currents. And, as of now, their models’ predictive value—like the morning weather report— leaves much to be desired. But now, an applied mathematician from the Whiting School and a physical oceanographer from Hopkins’ Krieger School of Arts and Sciences are joining forces to tackle this problem.
“Our ultimate goal, really, is to figure out what’s going to happen to the climate in the critically sensitive region of the North Atlantic,” says the Whiting School’s Greg Eyink, professor of applied mathematics and statistics.
The cooling and sinking of water in the Earth’s polar regions is critical to large-scale ocean circulation across the globe. In this “global conveyor belt,” heated water from the equator moves north toward the Arctic, where it releases heat into the atmosphere. After the release the water is cooler, and denser, so that it sinks downward, returns south, upwells back to the surface, and repeats the cycle.
“This overturning circulation has a substantial effect on northwest European climate,” explains the Krieger School’s Tom Haine, professor of physical oceanography in the Department of Earth and Planetary Sciences. Norway, for instance, lies within the warm northward current of the North Atlantic, making the Norwegian climate several degrees warmer than say, Greenland, which is at the same latitude.
But recent global warming might be slowing down the Atlantic part of the conveyor belt. When freshwater glaciers melt in the Arctic, the water becomes much less salty, and thus much less dense, and much slower to sink to the bottom. The effects on climate are difficult to predict with confidence. But they could be substantial, perhaps involving rapid cooling of northern Europe by up to several degrees Celsius.
Policy-makers would like to predict these local climate changes. For the last few years, Haine has made studies of the North Atlantic using data assimilation technology, which combines many kinds of data, sampled at different times and locations, coupled with state-of-theart ocean models. “Data assimilation has been used in atmospheric science for 30 years,” Haine says, “particularly when people make forecasts of the weather.”
Which begs the question, jokes Eyink, “of just how well is this going to work?” He says most of the forecasting strategies so far, including Haine’s, come up with simple predictions of the most likely behavior, without any idea of how uncertain that prediction is.
But the pair has just begun a new project— which applies some of Eyink’s sophisticated statistical techniques to Haine’s data assimilation system—to quantify these uncertainties. Eyink compares it to making weather predictions: “Instead of saying that tomorrow the weather will be 80 degrees,” he explains, “we’d like to say there’s a 90 percent probability that it will be within five degrees of 80.”
Though the big picture may seem simple, “these are challenging computational problems,” Haine says. “You move slowly, incrementally. But I’m very hopeful that these will be revolutionary ideas.”
The team doesn’t expect to have initial results for at least a year. They’ll need to develop codes and prepare data sets for high-powered supercomputers. And just one experiment might take a supercomputer a month to compute. “Data assimilation is painstaking, and very labor intensive,” Haine says.
Still, Haine and Eyink are both happy to be part of the unique collaboration. “In other places I’ve worked it’s not been possible,” Haine says. But at Hopkins, he says, “there’s a lot of cross-department mixing.” Eyink agrees, saying it wouldn’t be possible to demonstrate the usefulness of his abstractions, “without someone like Tom to give them flesh.”