In Deep Water

Quickly shifting currents can radically alter Earth's climate.

By Robert Kunzig|Sunday, December 01, 1996
We can almost see it whole, the round-the-world journey that seawater takes. We can imagine taking the trip ourselves.

It begins north of Iceland, a hundred miles off the coast of Greenland, say, and on a black winter’s night. The west wind has been screaming off the ice cap for days now, driving us to ferocious foaming breakers, sucking every last ounce of heat from us, stealing it for Scandinavia. We are freezing now and spent, and burdened by the only memory we still have of our northward passage through the tropics: a heavy load of salt. It weighs on us now, tempts us to give up, as the harsh cold itself does. Finally comes that night when, so dense and cold we are almost ready to flash into ice, we can no longer resist: we start to sink. Slowly at first, but with gathering speed as more of us join in, and as it becomes clear that there is nothing to catch us--no water below that is denser than we are. We fall freely through the tranquil dark until we hit bottom, more than a mile and a half down.

There we join a pool of other cold, salty water parcels that fills the Greenland and Norwegian basins. From time to time the pool overflows the sill of the basins, an undersea ridge that stretches between Greenland and Iceland and Scotland. Then the falling starts again. Now it is not a parachute drop but a headlong rush, downslope and tumbling like a mountain stream, but more powerful even than Niagara: a giant underwater waterfall, cascading into the Atlantic abyss. Falling, we pull shallower water in behind us. From our right flank, as we reach the latitude of Newfoundland, we are joined by a cohort from the Labrador Sea; not quite as dense as we are, this water settles in above us, headed south along the slope of North America. Near Bermuda our ranks are swelled on the left by spinning blobs of warm Mediterranean water, even saltier than we are; they sail like Frisbees out of the Strait of Gibraltar and cross the ocean to join us. Greenland water, Labrador water, Med water--we all fall in together, and gradually we mingle: we are North Atlantic Deep Water now. Mediterranean salt seeps through us like a dye. Though at every step on the road some of us lose heart and turn back north, still our mighty host advances, 80 Amazon Rivers marching along the ocean floor, toward the equator and across it.

All through the South Atlantic our army remains intact, hugging the western slope of the ocean basin. But that reassuring guide ends where South America does, and in the stormy Southern Ocean we are scattered by the great centrifuge, the Mixmaster, the buzz saw--what metaphor can do justice to the Antarctic Circumpolar Current? Sweeping around the frozen continent from west to east, with no land to stop it, it carries now some 800 Amazons of water. It blends the waters of the world, obscuring their regional roots. The fierce winds drag us--ever so briefly--to the surface off Antarctica, where we absorb a blast of cold and quickly sink again. We spread north now into all the oceans, mostly at a depth of half a mile or so, some back into the Atlantic, some into the Indian Ocean, many of us into the Pacific. In that vast and empty basin we drift northward until we reach the equator; there the trade winds part the waters, and tropical heat mixes down into us, buoying us to the surface. It is time to head for home.

Blasting and wending our way through the confusion of Indonesia, with its near-impenetrable wall of islands, we cross the Indian Ocean, collecting salt from the hot shallows of the Arabian Sea. Southward then down the coast of Mozambique, and we are picking up speed, in preparation for our triumphant return--but rounding the Cape of Good Hope is not easy. Again and again we are beaten back. Only by detaching ourselves in spinning eddies from the main current do some of us manage to sneak into the South Atlantic. There we are joined by water that never bothered with Indonesia and Africa but instead took the colder shortcut around South America, through the Drake Passage.

One last obstacle remains for us all--the equator, where this time we must cross the 12-lane highway of east-west surface currents set up by the trade winds. We do it again in eddies, giant ones that spin us north along the Brazilian and Venezuelan coasts before they finally shatter in the Caribbean and in the process dump us into the Gulf Stream at its source off Florida. This is the homestretch, at last; Iceland looms ahead. A millennium has passed since we left.

Oceanographers call this global journey the thermohaline circulation, because it is driven primarily by heat (in Greek, therme) and salt (in Greek, hals, which also meant sea). The thermohaline circulation is more than a natural curiosity. It spreads solar heat from the tropics to the high latitudes; it is what keeps Europe, for instance, warm and habitable. Given its tremendous force and its antiquity--it has been going on for tens of millions of years--one might imagine that nothing short of continental drift could change it. And one might dismiss as preposterous the notion that human beings, of all feeble agencies, could affect it at all. But the evidence suggests otherwise. We might already be on our way to shutting it down, with consequences for our climate we can only dimly foresee.

Wallace Broecker, or Wally to just about everybody--as in The Glacial World According to Wally, the title of one of his self-published books--dates from an era when oceanography was young and a boy could ask big questions about the ocean without huge tomes of technical literature tumbling off the shelves to crush him--questions like: What does the seafloor look like? Why is there a Gulf Stream? What causes ice ages? Back in the late 1950s, when Broecker was pursuing his Ph.D. at Columbia’s Lamont-Doherty Earth Observatory in Palisades, New York, his adviser urged him to answer that last question in the conclusion to his thesis. You might say I’m still writing the last chapter, Broecker says.

Broecker is still at Lamont today. He has been studying the thermohaline circulation for decades now, except that he has a different name for it: he calls it the conveyor belt. For an article once, Broecker had an artist draw a picture of the conveyor. It showed a broad band of deep water sweeping down the center of the Atlantic to the Antarctic, spreading into the Indian and Pacific, welling up to the surface there, and returning as an equally broad and unwavering band to the North Atlantic. This image drives some oceanographers crazy because they have spent the past few decades realizing just how complicated the flow of water in the ocean really is. Of course Broecker knows that the conveyor belt image is a crude simplification. But he also knows that, notwithstanding its complexity, the thermohaline circulation does something very simple and important: it transports heat into the North Atlantic and salt out of it. In that sense it is like a conveyor.

Broecker remembers exactly when it was that he first made the connection between the conveyor belt and climate change. It was in 1984, in Switzerland, while he was listening to physicist Hans Oeschger of the University of Bern. Oeschger was lecturing on the climate record contained in a mile-and-a-quarter-deep ice core extracted from the Greenland ice sheet, which is a relic of the last ice age. By that time there was a well- developed theory of ice ages; it attributed them to cyclical changes in Earth’s orbit that change the seasonal distribution of sunlight falling on the Northern Hemisphere. Those cycles--the so-called Milankovitch cycles-- seemed to explain why over the past 700,000 years or so, northern ice sheets had repeatedly advanced and retreated, with fits and starts lasting tens to hundreds of thousands of years.

But the Milankovitch theory could not account for what Oeschger was seeing in the core from Dye 3 in southern Greenland: evidence for far more rapid climate fluctuations during the last ice age. One strand of evidence was the ratio of oxygen isotopes in the ice. The heavier isotope, oxygen 18, is less prone to evaporate from the sea surface than light oxygen 16, and more likely to rain or snow out of the atmosphere sooner when it does evaporate. During an ice age, when a lot of water is removed from the ocean and locked up in continental ice sheets, the heavy isotope tends to remain behind in the ocean, and thus marine sediments become enriched with it. Meanwhile the ice in places like Greenland becomes depleted of oxygen 18: the colder the air is, the less likely it is that water vapor containing the heavy isotope will make it to Greenland before precipitating out of the atmosphere. Thus the oxygen isotope ratio in the Greenland ice is a thermometer. It measures how cold the air was over Greenland when the ice was laid down.

Oeschger’s second strand of evidence was actual samples of that ancient air--tiny bubbles that became trapped inside the ice when it formed. He and his colleagues had discovered they could analyze the chemical composition of those bubbles by putting a half-inch ice cube in a vacuum chamber and crushing it between beds of needles. In 1982 they had reported that the atmosphere during the last glaciation was different in a very important way from the preindustrial atmosphere, the one that existed right before we started aggressively burning fossil fuels: it contained only about two-thirds as much carbon dioxide. That made sense, since carbon dioxide tends to warm Earth by trapping heat. But it was not easy to see how small fluctuations in Earth’s orbit could change the CO2 level.

And the findings Oeschger reported in 1984 seemed even more distant from the Milankovitch theory. By then he and his colleagues had analyzed one section of the Dye 3 core in great detail, measuring changes over small time intervals. The ice in that section had been deposited 40,000 to 30,000 years ago, during the height of the last ice age. Yet, remarkably, its oxygen isotopes showed that during that period the climate had not been unwaveringly cold. Abrupt fluctuations in the isotope ratio revealed that the mean annual temperature over Greenland had risen as much as 13 degrees Fahrenheit within just a decade or two, then stayed high for a millennium before falling just as rapidly. And when the Swiss researchers popped the air bubbles in the ice, they found something more remarkable still. The carbon dioxide concentration of the ancient atmosphere seemed to have fluctuated in lockstep with air temperatures. In just a thousand years or so it had risen and fallen by as much as a quarter.

The temperature fluctuations had been seen before. Willi Dansgaard, the Danish researcher who had first suggested that ice cores would make good climate records, had found similar oxygen-isotope swings along the whole length of the Dye 3 core. Dansgaard had suggested that these swings might be caused by shifts between two different quasi- stationary modes of atmospheric circulation. But Oeschger’s carbon dioxide measurements seemed to eliminate that possibility. The atmosphere could certainly not change its own carbon dioxide concentration by 25 percent. In his talk that day in Bern, Oeschger hinted that the answer might lie in the ocean, which is a giant reservoir of dissolved carbon dioxide. At that point Broecker’s mind leapt into a quasi-stationary mode from which it has yet to emerge. Maybe it was ocean circulation that was changing, he thought: I said, oh my God, if you turned on and off the conveyor, it would do exactly what you want.

Even today no one, including Broecker, can say exactly how changes in the thermohaline circulation might have produced dramatic changes in atmospheric CO2. And for the moment the question is moot, anyway--because no one, including Oeschger, has been able to detect the rapid CO2 fluctuations in ice cores from other regions of the world. Although no one doubts that ice age CO2 levels were far lower than today’s, there is considerable doubt that they fluctuated dramatically. The sharp peaks and valleys in the oxygen-isotope record, on the other hand, are definitely real; they have been seen in cores from all over the world. During the last ice age the climate really did lurch back and forth between cold and relatively warm conditions. Broecker calls these lurches Dansgaard-Oeschger events. And his explanation for them, though it was inspired by Oeschger’s CO2 results, has fared better than those results themselves. (Science itself sometimes lurches forward in mysterious ways.) The conveyor belt really does seem to have switched states in the past--and in so doing to have changed the amount of heat it transports to the North Atlantic.

The best-documented case, naturally, is the most recent one. Long ago paleobotanists had discovered that the final retreat of the ice sheets did not go smoothly. It started rapidly and promisingly enough, around 16,000 years ago--but then around 12,500 years ago the temperature plummeted again. For more than a millennium, Europe was plunged back into glacial conditions. The forests that had only lately taken over the landscape gave way again to Arctic shrubs and grasses, including a wildflower, Dryas octopetala, that--thanks to its well-preserved remains-- ended up giving its name to the whole sorry period: the Younger Dryas.

Broecker proposed that this resurgence of cold had been triggered by a collapse of the conveyor belt. During the coldest parts of the ice age, he says, when sea ice spread south past Iceland, deep water formation was shut off. As the ice began its rapid retreat 16,000 years ago--driven ultimately by the Milankovitch variations in sunlight--warm, salty water again reached the region north of Iceland. There it gave up its heat to the cold west winds, which shipped most of it to Europe. The chilled, salty water sank to the seafloor, thus starting up the conveyor. As the conveyor transported more and more heat to the north, it accelerated the retreat of the ice.

Then something curious happened. In North America, in what is now southern Manitoba, a giant lake of glacial meltwater had formed to the west of the lobe of continental ice that protruded south into the central United States. This body of water--called Lake Agassiz, after the nineteenth- century Swiss-born naturalist Louis Agassiz, who had recognized the reality of ice ages--was larger than all the present Great Lakes combined. At first its water drained down the Mississippi into the Gulf of Mexico. But as the ice sheet retreated north, a new and shorter path to the sea was opened: through the Great Lakes Basin and into the St. Lawrence. Thirty thousand tons a second of freshwater began rushing into the North Atlantic from this new source, right into the northward-bound leg of the conveyor belt. All that freshwater substantially diluted the water in the conveyor--in fact, the seawater was no longer salty enough to sink to the ocean floor by the time it reached Greenland. Without that sinking, the conveyor was shut off. So was the heat the conveyor delivers to the North Atlantic region. The ice advanced again, and Dryas flowers began blooming again on the plains of northern Europe.

Just as sediments in the Gulf of Mexico record this diversion of glacial meltwater (their isotope ratio went up during the Younger Dryas), sediments in the Atlantic itself record the throttling of the conveyor. The first evidence of this was uncovered in 1987, not long after Broecker proposed his theory. It came from a broad seafloor elevation called the Bermuda Rise, 400 miles northeast of the island, where mud washes up in thick drifts that make for detailed climate records. Ed Boyle of mit and Lloyd Keigwin of Woods Hole reported that the Younger Dryas was readily discernible in a sediment core from the Bermuda Rise--or rather, in the shells of microscopic creatures known as foraminifera, some species of which float at the surface while others live in the mud. During warm periods like today, they found, the forams absorb into their shells the distinctive chemical imprint of the North Atlantic Deep Water that washes over them. But during the Younger Dryas, the forams were stamped instead by Antarctic Bottom Water, invading from the south and apparently meeting little resistance. The North Atlantic Deep Water must have been weak then-- which is another way of saying the conveyor belt was weak, and possibly had turned off altogether. This result was very gratifying to Wally Broecker.

Oceanographers soon began finding other records of rapid climate fluctuations. And they began to realize that, just as the Younger Dryas was only the last in a long series of climate swings recorded in the Greenland ice sheet, the North Atlantic Deep Water spigot had been turned on and off, or at least down, many times during the last ice age. During its weak intervals, Antarctic water had advanced right up to the base of Iceland. Judging from the sediments, there was never any peace at all in this 100,000-year north-south war of the water masses; the front surged back and forth constantly, rapidly--on the timescale of centuries, anyway--with each shift in fortunes corresponding to a major shift in the operation of the conveyor.

All these shifts, obviously, could not be blamed on the capricious drainage of Lake Agassiz. Nor does there seem to have been an abundant supply of other giant lakes waiting to be diverted at regular intervals into the North Atlantic. On the other hand, there certainly was an abundant supply of ice.

Sediment cores suggest . . . --the phrase scarcely does justice to the suffering of sedimentologists, and to the painstaking labor that goes into extracting even a single clue to Earth’s climate history from a long column of seafloor mud. Extracting the core itself is not the half of it. During the 1950s and 1960s, Lamont scientists were directed to pull up a core every day they were at sea, wherever they might be. Today, as a result, the Lamont archive contains more than 18,000 cores of seafloor mud in various states of desiccation.

Finding the right core for your purposes is one problem, but Gerard Bond has an advantage there; his office adjoins the core archive, and his wife, Rusty Lotti, is the archive’s curator. The bigger problem is teasing climate information out of the core once you have it, with nothing to sustain you through the long hours of tedium but faith--faith that in the end, a scattering of sand grains and microscopic shells may vouchsafe to you the reality of a dramatic change in Earth’s climate tens of thousands of years ago. A rearrangement of ocean currents and winds, a surging of ice sheets--all this is there in a handful of sand or less, if you know what to make of it. To that end Bond and Lotti have spent the better part of the past five years scalpeling through a few select sediment cores. Bond reckons that he personally has counted 700,000 sand grains, one by one under a microscope, sorting them by type. No geologist in his right mind would ever do anything like this, he says--except, perhaps, a geologist who has strayed into the orbit of Wally Broecker.

Bond came late to the study of marine sediments, or at least recent ones. His career had been devoted to the study of sedimentary rocks on land, mostly half-billion-year-old Cambrian formations in the Canadian Rockies. In the late 1980s, though, he conceived the idea that he could see evidence of Milankovitch cycles in the shifting colors of the strata. As a way of testing that idea, he started looking at recent sediment cores, in which the evidence for Milankovitch cycles was well established. The dried- out cores themselves did not show color variations very well, but fortunately for Bond the researchers who extracted the cores had routinely photographed them while they were fresh and wet, and published those photographs in books--page after page of section after section of mud. Bond cut up an article devoted to one core, called dsdp 609, and pasted the photographs end to end on the wall outside his office. He now had 700,000 years of climate history running down a 30-foot hallway. Looking at the photographs from an angle, he could readily see the sequence of ice ages and warmer interglacials marching down the hall in a kind of binary code: dark, light, dark, light, dark, light. And when he digitized the photographs and measured the core’s color more precisely, he could tell that it varied tremendously on a much more rapid timescale than that of ice age and interglacial.

Bond decided this variability was worth studying and wrote up a proposal to secure the necessary grant. He still thought of the project as little more than a brief detour out of the Cambrian Period. And he did not expect much when, as a courtesy, he sent a copy of the proposal to Broecker, whose professional turf he was proposing to tread on. Broecker was far from resenting the intrusion. Wally knew all about ice cores and these problems of abrupt climate change--I knew nothing about that at the time, Bond recalls. He came tearing over to my office. He saw the gray- scale shifts and he said, ‘That’s just like the ice-core record.’ So that was how I got started. Wally really twisted my arm.

By then Lamont scientists had long since figured out what the light and dark stripes in an Atlantic sediment core represented. The light sediment consisted mostly of calcareous foram shells, deposited in a period of relatively equable climate. The dark sediment, on the other hand, came from far away: it consisted of grains of rock scraped off the land by advancing ice sheets, carried out to sea by icebergs, and deposited on the ocean floor when the icebergs melted. Thick stripes of iceberg debris at a latitude of 50 degrees, where Bond’s dsdp 609 came from--the latitude of the south coast of England--obviously must have been deposited in periods that were pretty cold. But until Bond started quantifying the color variations in his core, no one had realized that they indicated much more rapid fluctuations in climate.

With Broecker urging them on, Bond and Lotti and a couple of technicians started dissecting dsdp 609 as no core had been dissected before. They cut samples out of every one of its 800 centimeters--out of every century and a half of climate history. Each thimbleful of mud then went through filtering, to separate out the microscopic shells and grains of rock. Those tiny particles were then spread onto a palm-size tray that had been gridded off into 45 compartments, to facilitate counting, and subjected to several stages of analysis. First one technician would pick over the sample looking for surface-dwelling forams; if they were predominantly of a polar species whose shell coils to the left, it meant that the sea surface over the sediment core had been very cold during that period. Then another technician would go over the same sample to pick out the bottom-dwelling forams, scanning the scattered grains under a binocular microscope and gently lifting out the white, toothlike shells with the moistened tip of a fine paintbrush. It took an hour to do one sample, and after that you might end up with no forams at all; but if you had at least two or three, you could measure their oxygen-isotope and carbon-isotope ratios. Finally Bond himself scanned the sample to sort the rock grains. Those grains could tell him, a sedimentary petrologist with decades of experience, where the icebergs had come from. It would have taken him years to train a technician to do that reliably.

One of the first things Bond noticed was that there was something wrong with equating light sediments with forams and dark sediments with ice-rafted rock. There were places in the core that were light and yet foram-free--because they were crammed with grains of white limestone. It really shocked me, Bond recalls. You would think that with icebergs coming from all these different sources, there would be a mix of things. And the layers above and below this were the normal mix of quartz and feldspar and very minor amounts of limestone. Then all of a sudden, boom, there was this enormous amount of limestone, a huge change in the composition of the grain. There aren’t that many places where that kind of stuff can come from.

In fact there was only one place that was plausible, one place on the North Atlantic rim where an advancing ice sheet was likely to have ground over limestone bedrock: the Hudson Strait, at the mouth of Hudson Bay in Labrador. Bond soon learned that the limestone layers were present in cores from the Labrador Sea, too--and being closer to the source, they were much thicker than the ones in dsdp 609. And from Broecker, Bond learned that a German oceanographer, Hartmut Heinrich, had identified the same layers a few years earlier in a core a couple of hundred miles southeast of dsdp 609.

An astonishing vision took shape in Bond’s mind: a vision of a giant ice sheet surging through the Hudson Strait, its underside melting and refreezing around shattered bits of limestone, and of a vast armada of icebergs setting sail from the thunderously collapsing edge of that ice sheet. Drifting down the Labrador Sea and out across the North Atlantic on the prevailing current, they gradually melted and dropped limestone on their way. A couple of glaciologists later tried to estimate how much sediment might have been deposited in just one of these Heinrich events, and they came up with a figure of around a trillion tons. Bond himself estimated how much freshwater the melting icebergs might have shed into the surface layer of the North Atlantic. He put the concentration at 1 part in 30, which is about what you would get by dropping an ice cube into every quart of ocean. That would be more than enough to freeze the conveyor belt.

Heinrich events happened every 7,000 to 10,000 years or so during the last ice age. But as Bond and Lotti tore deeper into dsdp 609 and another core from the eastern Atlantic, they began to see that Heinrich events were just the tip of the iceberg, as it were. Dense layers of dark rock grains between the Heinrich layers indicated that smaller iceberg armadas had been launched more frequently--but not from the Hudson Strait, because the grains were not limestone. After sorting the dark grains, Bond found that 2 of the 15 separate types he’d defined stood out: black volcanic glass from Iceland, whose active volcanoes at the time poked through a thick ice cap; and redstone--quartz and feldspar coated with iron-rich hematite--that seemed to come from the Gulf of St. Lawrence. Judging from the spacing of the dark layers, iceberg fleets had departed from those ports every 1,500 years, and every fifth or sixth one of them had encountered an even larger Heinrich armada from the Hudson Strait. More important, nearly all the iceberg fleets coincided with Dansgaard-Oeschger events, that is, with periods of sharply cooler air over Greenland.

Every 1,500 years, then, the following events occurred in the North Atlantic region: the air over Greenland, having suddenly warmed nearly to interglacial temperatures, plunged back into deepest cold in the space of a decade. Ice sheets in North America and Iceland, and possibly elsewhere as well, discharged fleets of icebergs that drifted as far south as 45 degrees of latitude. And the formation of deep water in the North Atlantic was stopped or sharply curtailed. Sediment cores suggest that the conveyor belt was weakened during the last ice age but never turned off entirely. Water continued to sink in the North Atlantic, but it was apparently not salty enough to sink all the way to the bottom. It settled instead at an intermediate depth, flowing southward, with Antarctic water sloshing northward underneath it.

All these events happened repeatedly in the last ice age--but unfortunately, researchers cannot be sure in what order. When they look up from their sediment or ice cores, they are haunted by the specter of the chicken and the egg. Perhaps the ice sheets, responding to their own internal rhythm of growth and decay, launched their iceberg armadas whenever they got too fat; the melting ice then clamped down on the conveyor; and the weakened conveyor transported less heat to the North Atlantic, thereby cooling the air over Greenland. But then why would at least two different ice sheets decide to purge themselves simultaneously, as Bond discovered? Perhaps instead the air got colder first, which caused all the ice sheets around the North Atlantic to surge into the sea, which turned down the conveyor, which made things colder still. But then what cooled the atmosphere in the first place?

Add to this dilemma another one: geography. When Broecker first started thinking about Dansgaard-Oeschger cycles, and the Younger Dryas in particular, he was looking to explain how temperatures in the North Atlantic region could ever have taken a sudden millennial nosedive. Computer models of Earth’s climate, chiefly the one developed by Syukuro Manabe at the Geophysical Fluid Dynamics Laboratory in Princeton, confirmed Broecker’s hunch that the conveyor belt could do the job by switching abruptly to a weakened state. They even reproduced the regional extent of the Younger Dryas cooling, which at the time was thought to have been felt primarily in Europe and to a lesser extent in eastern North America. But in the last decade the evidence has changed. The Younger Dryas and the other Dansgaard-Oeschger events are no longer merely North Atlantic curiosities. No way can I get gigantic cooling everywhere, grumbles Manabe. Yet that is what the evidence points to, and it comes from some unusual places.

Huascarán, Peru, is not the first spot most researchers would think to look for the causes or effects of changes in the North Atlantic. It is a glacier-covered mountain in the Andes, 9 degrees south of the equator and 200 miles north of Lima. The highest of its twin peaks reaches 22,205 feet. Lonnie Thompson of Ohio State University did not make his drilling team climb that high; they stopped just shy of 20,000 feet with their six tons of equipment, at a saddle point between the two peaks, where the ice was more than 700 feet thick.

Thompson is used to skepticism from his scientific colleagues. He has been drilling into mountain glaciers for nearly two decades now, ever since he got bored with drilling in Greenland and Antarctica. Not long after he started, Willi Dansgaard, the polar-drilling pioneer, wrote a letter to him and to his funding agency saying that the technology did not exist to do what Thompson wanted to do. This did not help Thompson’s cause. But he knew Dansgaard was right. He had already discovered that on his first expedition, in 1979, to a glacier called Quelccaya in southern Peru.

We were naive, he recalls. We thought we could use a helicopter and bring a drill up from Antarctica, and we’d get it up there and drill the core and that would be it. But the elevations we work in, above 19,000 feet, are really out of the range of most helicopters, and when you have a lot of convective activity in the mountains, it makes flight very difficult and dangerous. We’d be flying along at 19,000 feet and the helicopter would just fall. There was no way we could get near the surface. Because the technology did not exist to land a big ice-drill on an Andes peak, Thompson logically concluded that he would have to build a drill light enough to carry up on his back--and the backs of his graduate students and a few dozen porters and mules. If the technology did not exist, he would invent the technology.

Fourteen years after that first failure, Thompson found himself camped on Huascarán with a carbon-fiber drill and 60 solar panels to power its heated, ring-shaped tip through the ice. As each length of ice core was extracted from the borehole, it went into insulated packing material and then into a walk-in storage cave that Thompson and his crew had dug into the glacier. When the cave was full, the porters were called. Working in the pitch darkness of 3 a.m.--the coldest, and so most desirable, time of day--they hoisted the ice onto their backs and carried it down a 50-foot ladder that sloped across an 80-foot-deep crevasse; then on to the edge of the glacier, where mules waited to take it to the foot of the mountain, where trucks waited to take it to a fish freezer in the town of Huaraz. Some of Thompson’s graduate students did not appreciate the beauty of that crevasse, which widened steadily as the expedition wore on (Sometimes they made career choices when they looked at the ladder, Thompson says), but fortunately porters were plentiful. We happened to drill this core at the height of the Shining Path guerrilla activities in Peru, says Thompson. On one side that was a problem, because there was danger. But on the other side, we had a complete hotel to ourselves, so we could set up a laboratory. And we had all the porters we needed. Not that Thomas himself spent much time in the hotel: he camped out on Huascarán for 45 days, working sunup to sundown in winds that ripped his tents and in air that was half as thick as at sea level.

To what end this amazing effort? When Thompson drilled his cores to bedrock at Huascarán, he got access to a deep past: the ice at the bottom was 20,000 years old. It had survived intact since the last peak of the last glaciation.

The conventional wisdom had been that the ice age had left the tropics largely untouched. The Huascarán cores give that view the lie: the oxygen isotopes in them indicate that at the height of the glaciation the temperature on the mountain was 15 to 22 degrees below what it is today. If you extrapolate that temperature down to sea level, as Thompson did, you find that the surface of the tropical Atlantic, where the snow falling on Huascarán comes from, was at least 9 degrees colder than today. Like the atmosphere at high latitudes, the tropical atmosphere was also much drier in the ice age: the strata from the bottom of the Huascarán cores contain 200 times more dust than falls on the mountain today. That dust was apparently blown in from Venezuela and Colombia, where vast tracts of land that are now savanna were then covered by dune fields.

Most surprising of all, the Younger Dryas shows up clearly in Huascarán ice.

A few years ago, while Broecker was writing the first edition of The Glacial World According to Wally, he developed a severe case of writer’s block as he approached the last section, in which he had hoped to set forth his grand hypothesis of what had driven all the climate change during the last glaciation. It was more or less the same last chapter he had failed to write for his Ph.D. thesis, only now the facts had gotten considerably more complicated. Broecker found he still did not have a coherent hypothesis. By 1996, though, he was groping toward one. It was inspired by the work of Lonnie Thompson on Huascarán.

That ice core offers the strongest of several strands of evidence that the tropical atmosphere was extremely dry during the ice age--Thompson and Broecker estimate it contained only 80 percent as much water vapor as it does today near the surface, and only 40 percent as much at high altitudes. Today the tropics are the planet’s largest source of water vapor; it rises there off the warm sea surface and is carried by winds toward the poles. Along the way it precipitates as rain and snow, and at the same time serves another critical function: it is the most important greenhouse gas, more important even than carbon dioxide. If the water vapor concentration in the last ice age was substantially lower, then that alone would have cooled the planet substantially.

In Broecker’s hypothesis, rapid changes in the water vapor concentration, caused somehow by changes in the conveyor belt, are what produced the millennial global climate swings of the last ice age. The most likely trigger, he says, is still a shot of freshwater to the North Atlantic. Icebergs streaming off the North American ice sheet could weaken the conveyor over the course of centuries; but when the last berg had melted and the atmosphere was in the coldest and driest trough of a Dansgaard-Oeschger cycle, such that not much snow was falling on the northern latitudes, then the North Atlantic would quickly grow salty again, salty enough to sink into the deep off Greenland, and the conveyor would spring back to life. Models such as Manabe’s show that the conveyor can rebound rapidly when it stops getting hosed with freshwater. And a hypothesis such as Broecker’s explains how a sudden warming of the North Atlantic can propagate rapidly through the atmosphere to the Peruvian Andes and other points south--provided that somehow the resurgent conveyor can pump water vapor back into the tropical atmosphere.

The operative word is somehow. The equatorial ocean is a zone of major upwelling currents, which might be expected to influence the amount of water that evaporates from the sea surface, and which might in turn be under the influence of the conveyor. And in the equatorial Pacific off Peru, at least, the upwelling shuts down from time to time, during the phenomenon known as El Niño. That suggests to Broecker that the tropical atmosphere may have discrete states of operation as well, like the conveyor belt, and that it might flip in response to a flip of the conveyor. But he grows a bit exasperated when he is pressed for a more precise link between the two. The only part of the system that we know about that has multiple states is thermohaline circulation, he says. Okay? And we know from evidence in sediment that thermohaline circulation did change. Okay? So the working hypothesis has to be that these changes in thermohaline circulation have far-reaching effects. And what I’m trying to tell you is that we don’t know what the link is. What you’re asking for is the big missing piece of the whole puzzle. I mean, we have every other piece in place, and we’re missing a major piece.

Would that it were really only one.

In 1991, when lonnie Thompson went back to Quelccaya, the Peruvian glacier he had first climbed 12 years earlier, he found that it was melting. There were three lakes downhill from the ice cap that had not been there before. Thompson was disappointed but not surprised. In Venezuela, three glaciers have disappeared altogether since the early 1970s. Three have disappeared from Mount Kenya in Africa as well; since the early 1960s glaciers there have lost two-fifths of their mass. It’s throughout the tropics, says Thompson. Every glacier that we have any data on shows a very rapid retreat taking place. You have to ask why that might be.

Thompson’s hunch is that his vanishing glaciers are an early sign of man-made global warming. Even a slight warming caused by the carbon dioxide we have added to the atmosphere might be enough to evaporate a lot more water off the tropical ocean. The water vapor might then amplify the warming enough to melt the ice. Thompson, unlike Broecker, is inclined to believe that the tropical atmosphere drives the conveyor belt, rather than the other way around.

An experiment that Manabe did a few years ago with his climate model lends some support to that view. Manabe allowed the concentration of carbon dioxide to keep increasing at the rate it is now, about 1 percent per year, until after 140 years its atmospheric concentration had quadrupled. From then on he let it remain constant. As Earth’s temperature rose, so did the amount of water vapor in the atmosphere, and winds carried it to high latitudes, where it fell as rain and snow. In Manabe’s model world, the rivers of the far north--the Mackenzie, the Ob, the Yenisei-- became torrents emptying into the Arctic. From there the water made its way south into the Greenland Sea. By the 200th year of the simulation, the thermohaline circulation had stopped dead.

It is possible that the carbon dioxide concentration will not quadruple over the next century and a half--that Earth’s fractious community of nations, with their burgeoning head counts, will agree on the drastic economic and technological changes needed to limit the growth of fossil fuel emissions. If the CO2 level were only to double, Manabe’s model predicts that the conveyor belt would merely weaken for two or three centuries and then restore itself--much as it may have done in the Younger Dryas. There is also a more plausible reason to believe the conveyor belt may survive. Every time we burn fossil fuels, especially coal, along with the carbon dioxide we emit sulfur dioxide, to the extent that we do not scrub it out of the smokestack plume. If we emit enough sulfur dioxide, thereby worsening the world’s acid rain and smog problems, it could help protect us from the worst of global warming. Sulfur dioxide is a parasol gas--it reflects sunlight back into space--and Manabe’s model did not take this effect into account in predicting thermohaline collapse. He thinks it could prevent that dire outcome--particularly if the Chinese burn through their vast deposits of coal without worrying about acid rain.

On the other hand, Manabe’s model also did not take into account the possibility that the Greenland ice sheet might melt in a CO2-warmed world. It is hard to imagine just how the conveyor belt would handle that kind of freshwater jolt to its soft spot. We cannot eliminate completely the possibility of the ‘drop dead’ scenario, says Manabe.

What would happen if it did drop dead? Some good might come of either a collapse or a weakening of the conveyor belt. Manabe’s model suggests that global warming might be somewhat moderated around the North Atlantic rim, particularly in Europe, by a Younger Dryas-type cooling effect. But the truth is we really do not know how a change in the conveyor belt would affect the world’s climate. The only thing we can safely conclude from Manabe’s model and from the sediment and ice-core evidence is that a rapid change in the thermohaline circulation is possible now, even when the world is not in the midst of an ice age.

The thermohaline circulation has been around for tens of millions of years at least, but some researchers date it in its present form to just 3 million years ago. That is when the Isthmus of Panama emerged from the sea, connecting North and South America and dividing the Atlantic from the Pacific. In a recent book, paleontologist Steven Stanley of Johns Hopkins proposed that this change was crucial to our own evolution. The establishment of the modern conveyor belt, Stanley argues, paved the way for the ice ages; and as Africa grew cooler and drier and forests gave way to savannas and deserts, our australopithecine ancestors were forced to come down from the trees. Stanley’s book is called Children of the Ice Age, but it might as well have been called (had the publisher been indifferent to its sales potential) Children of the Thermohaline Circulation.

After 3 million years, the children have grown up now, sort of: they have acquired the power to slay their parent. Have they grown up enough to stay their own hand? Wally Broecker is not optimistic. Little has changed since Roman times, Broecker wrote in the conclusion to one of his own books, How to Build a Habitable Planet. Man fiddles and hopes that somehow the future will take care of itself. It surely will, but mankind may not like the course it takes.
Comment on this article

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar

Log in to your account

Email address:
Remember me
Forgot your password?
No problem. Click here to have it emailed to you.

Not registered yet?

Register now for FREE. It takes only a few seconds to complete. Register now »