Deciphering the big thaw

The last great ice age was at its peak from 26,500 years ago to about 19,000 years ago, a time called the Last Glacial Maximum (LGM). Just before that, Earth’s axis was tilted only about 22 degrees, as close to vertical as it gets. Glaciers covered northern Europe, blanketed most of Canada and extended into the United States, reaching the Ohio and Missouri rivers.

Beginning around 25,000 to 22,000 years ago, Earth began to nod, presenting the Northern Hemisphere to the summer sun. But at that time, the season also was migrating along Earth’s orbit, and thousands of summers came and went when the planet was far from its star. About 22,000 years ago summer began to creep closer to the sun, warming Earth.

That much was accepted. But the reason for the Southern Hemisphere’s early warm-up was the focus of a contentious debate. For nearly four decades, the problem was seen as the fly in the ointment of the Milankovitch theory.

Only computational simulations could reveal the complex interactions among major climate-related systems as the planet heated up, but it took years to develop the capable software and hardware. Early simulations modeled oceans as slabs – uniform and mainly static bodies of water. Changes in key variables, such as atmospheric carbon dioxide concentrations, had to be introduced all at once. With the advent of more powerful computers, simulated oceans have become more dynamic, with complex and realistic currents, and variables can be introduced gradually.

In 2008, the research team began a new series of simulations. “The throughput of the Jaguar was really the game changer,” He says. It continued to run simulations day and night for years, with reasonable downtime, producing 300 terabytes of results for analysis.

“We used to run 45 model years per day on Phoenix, the predecessor of Jaguar. On Jaguar, we got 120 model years per day. So for a 22,000-year simulation, it takes about 1½ years to run on Phoenix but only about half a year to run on Jaguar.”

The team ran five simulations, beginning each in the middle of the LGM. Four modeled changes in key climate forcings separately: Earth’s movements, atmospheric carbon dioxide, meltwater and ice cover. Of these, only the changing ocean currents associated with meltwater gave a recorded increase in Southern Hemisphere temperature at the end of the LGM.

The fifth simulation incorporated all four sources of change and used the Community Climate System Model version 3 (CCSM3) to ensure that heat released by any one of the key climatic players – land, sea, ice and air – would be accounted for in the others. It neatly matched ancient temperatures calculated from ocean-sediment and ice cores.

So what caused the Southern Hemisphere to warm up so fast? A slowdown in the Atlantic Meridional Overturning Current. This stream is partly generated when cold, salty (and therefore dense) water sinks in the North Atlantic. That deep water flows south to the Antarctic Ocean and into the Indian and Pacific Oceans, forcing warm, upper seawater layers to flow back across the southern oceans and ultimately into the North Atlantic. This thermohaline circulation cycle, or the conveyor belt, warms coastal nations in the North Atlantic.

During the LGM, when increased sunlight began to melt the ice sheets, huge amounts of low-density freshwater flowed into the North Atlantic, reducing the conveyor belt’s flow and leaving warm sea water to gather in the Southern Hemisphere instead of moving north.

The team’s research, reported earlier this year in Nature, has implications for studies of today’s climate. At the end of the LGM, the warming waters and melting sea ice released carbon dioxide into the atmosphere from the Southern Ocean around Antarctica, accelerating a powerful feedback loop. “For modern climate change, our results show that atmospheric CO₂ is capable of producing global warming, (just) as it provided the critical feedback on global deglaciation in the past,” says He, the paper’s lead author.

The study also helps validate computational modeling of climate change, a field that’s often under attack. CCSM3 is one of the models providing input to the Intergovernmental Panel on Climate Change (IPCC), a United Nations-backed scientific group that periodically reports on the state of climate science.

He adds, “We showed that at least one of the IPCC models used to predict the future can reproduce the past.”