Climate Machine: Steady as She Goes

Forecasters see the Atlantic's warm "conveyor belt" running steadily years into the future

While weather forecasters still can't peer much more than a week into the future, climate forecasters are now claiming they can predict the speed of the North Atlantic's great climate-changing current as much as 4 years ahead. That bodes well for forecasting decade-scale climate change around the North Atlantic.

Forecasters quickly lose track of coming weather in the atmosphere's maddeningly chaotic churnings. But modeler Daniela Matei of the Max Planck Institute for Meteorology in Hamburg, Germany, and her colleagues report this week in Science that they have gotten a handle on the mighty Atlantic Meridional Overturning Circulation (AMOC). Being able to model a key ocean property driving the AMOC lets them forecast the current's behavior not months but years into the future.

The AMOC is more widely known as the "conveyor belt" current of the Atlantic. The shallow current flowing out of the Florida Strait becomes the Gulf Stream that delivers massive amounts of heat northward. That heat tends to warm the far North Atlantic and thus keeps the land downwind of it—Western Europe—warmer than it would otherwise be. But the pace of the AMOC can vary naturally over years, influencing climate. And scientists have worried that a gush of meltwater from a warming Arctic might dramatically slow the AMOC for centuries or even kill it off.

So Matei and colleagues set a climate model running that simulates the behavior of the global atmosphere and ocean, including the AMOC. In separate simulations, they got the model running in the same way the real-world atmosphere and ocean were operating at the start of each year from 2004 through 2007. Then they ran each simulation through 2009.

Comparing the model AMOC with the real one in those years, the researchers found considerable similarity. Given a realistic start, the model could reasonably accurately simulate the behavior of a key ocean property driving the AMOC. That allowed the model to forecast the actual behavior of the current 4 years into the future, according to the authors. The model did so well, they write, that, after starting the model in 2010 and running it into the future, "we confidently predict a stable AMOC at least until the end of 2014."

The new modeling "is a definite advance," says climate forecaster Doug Smith of the Hadley Centre for Climate Prediction and Research in Exeter, U.K. "It's consistent with what we're finding from a range of models. We do think there's predictability" of the AMOC in the models.

"This evidence is encouraging for the development of long-term forecasting systems," says meteorologist Rowan Sutton of the University of Reading in the United Kingdom. He adds, however, that "there are quite a few caveats and reasons for caution." As a result, the confident prediction for 2014 "is stronger than I would have chosen, on the evidence presented," he says. A few years will tell.