Journey to the Center of the Ice

David Pacchioli
May 01, 1998

When the sun was out and low in the sky, Todd Sowers remembers, standing on the summit was an amazing experience. His goose-down swaddled form, and the similarly bulky shapes of his colleagues, would cast their shadows hundreds of meters down, to the stark landscape of the Bolivian altiplano below. Not that he had much time to contemplate such magnificent vistas. Sunny days were particularly precious on Nevado Sajama, not least because in order to penetrate the mountain's icy cap Sowers and the others were using a solar-powered drill. "When the sun's out," he says, remembering those two vivid weeks last June, "you're working like a madman."


Sowers, an assistant professor of geosciences at Penn State, was part of an expedition organized by Lonnie Thompson of Ohio State University, a six-week adventure in mountaineering and logistics whose object was to drill through an ice cap atop the highest peak in Bolivia to harvest 15,000 years' worth of ice—and haul it back safe to North America for careful analysis. The composition of the ice and the ancient air trapped within these cores is expected to yield a fair sample of what the Earth's climate was like at the end of the last ice age.

Sowers was no rookie to this business. He had spent two prior summers drilling in Antarctica, where the ice runs a lot deeper and the daytime temperatures a lot colder than Sajama's relatively balmy -12 degrees C. But Antarctica featured none of the climbing and hauling prerequisite to doing science at 22,000 feet above sea level. "In Antarctica," Sowers says, "they drop you off in a plane and you're there." After three days at Sajama's summit, the lightheadedness and the headaches were gone, but every exertion still demanded an inordinate fee in fatigue. "You're burning up so much energy in that low oxygen that the weight just falls off you," Sowers says. "Your metabolism speeds up and your body starts eating muscle."

By the 26th of June, when the actual drilling began, Thompson and Sowers and several others had been preparing for almost two weeks, working between storms and bouts of 50-knot winds. ("We ripped up a few tents," Sowers says.) Besides the intermittent conditions, there had been problems below. Dysentery was raging at the high camp halfway down the mountain, striking ill more Bolivian porters every day. All the while, people and equipment continued to stream periodically up the mountain. A separate party of three climatologists from the University of Massachusetts was erecting a satellite-linked weather station—the highest in the world—to provide recent data for calibrating records from the ice cores. A team of hot-air balloonists was assembling at the high camp, getting ready to float the ice samples off the mountain under a balloon dubbed the Soaring Penguin. (The air at 22,000 feet is too thin for helicopters.)

At last, on the 24th, Thompson's team had hoisted the drill mast, and anchored it with guy wires. After a final testing of components, they had started to drill. They worked steadily, two men operating the drill, with Thompson and Sowers logging and processing core as it emerged in one-meter long, 10-centimeter diameter chunks. By the 2nd of July, with the drill sunk 80 meters, halfway to bedrock on the second of three planned holes, Sowers had seen what he had come to see. The drilling would go on for another week, but the time for his real work would be later. "Plus," he says, "I had a four-month-old baby waiting at home. I figured it was time to go.

"Coming down," he remembers, "I got about halfway and I was just totally out of energy. I could hardly move my legs, but I was not going to spend the night at the high camp—it was really nasty there. So I kept on going." He staggered into base camp at 9:30 that night, "completely dead," using his ice axe as a crutch, his way lit by headlamp; he had dropped 15 pounds in a fortnight.

"I didn't have any macho reason for going up there," Sowers says, sitting in his cluttered office in the Deike building back at University Park. He is tall and lanky, with a dark mustache; shod in tooled gray cowboy boots. As a "gas person," Sowers' part in the Sajama expedition is to fix definite dates on the fossil ice cores that Thompson brought back from the Andes. It's exacting and delicate work: to be accomplished out of the wind, in the relatively cozy confines of his laboratory, and he'd just as soon wait for the samples to come to him, he says. "It's kind of an unwritten law, though. If you're going to work on the cores, you have to help get "em."

At the surface of an ice sheet or glacier, in those places on Earth where the temperature never meets freezing, every year brings a new covering of snow. With each fresh layer on top, the ones underneath grow more compacted. Powdery grains become crystals, and crystals grow as they are compounded; density increases until eventually, at a depth from 10 to 40 meters below the surface, the snow turns into ice. This snow-to-ice process is called firnification: firn, the in-between stage, is German for "of last year."

Captured inside the ice as it is buried continually deeper are tiny bubbles of air, pressurized remnants from the spaces between snowflakes. These bubbles are what Sowers is after. "Measuring the trapped gases," he explains, "provides a means of reconstructing the composition of the atmosphere back through time."

Ice cores recently drilled from up to two miles deep in the vast ice sheets in Greenland and Antarctica have already yielded a wealth of ancient climate data. The so-called GISP-II core from Greenland, the most thoroughly studied of these, provides a continuous temperature record that extends far into the Weichselian glacial period, 110,000 years before the present. Even older are the cores retreived from Vostok Station in Antarctica, which date back more than 400,000 years, a passage through four full ice ages.


Comparing the GISP-II and Vostok cores offers tantalizing glimpses of the unwritten history of the Earth's climate. According to Sowers, the samples are first of all remarkable for their essential similarity. The "signal" from each place—the gleanable record of temperature and atmospheric composition plotted over time—follows basically the same pattern. That the two poles seem to echo each other this way reinforces climatologists' ideas of a global system woven together via teleconnections—hidden long-distance linkages that lie beneath more obvious climatic occurrences. ("Tele" is Greek for "distant.ö El Ni±o is the most obvious present-day example of a teleconnection.)

Perhaps more interesting, however, are the differences. As Sowers explains it, the GISP-II signal is an "amplified" version of its southern counterpart, distinguished by large and rapid variations, including temperature swings of up to 12 degrees C over time scales as brief as a single decade. Antarctic cores for the same periods show much gentler trends. Change at Vostok appears subtler, more gradual. Thus the prevailing notion of how climate change occurs. As Sowers puts it, "The prevailing wisdom holds that climate change originates in the North Atlantic, with Antarctica responding in muted fashion." What drives this system, and which forces account for such rapid changes in the North Atlantic, are two of the large theoretical holes that Sowers and his colleagues hope to shrink by looking toward the tropics.

Up to now, the wide middle latitudes have provided little in the way of data on continental climate: History, in warmer climes, is a perishable thing. But venerable deposits of ice do exist atop the highest of tropical mountain peaks; and although these caps are nowhere near as deep as the polar ice sheets, they can, theoretically at least, provide a useful look at least as far back as the last ice age.

In 1993, Thompson, a research scientist at Ohio State's Byrd Polar Research Center, led an expedition to Huascaran, a 20,000-foot peak in the Cordillera range of Peru. On Huascaran, he and his team drilled two holes to bedrock, about 160 meters down, bringing home samples over 10,000 years old, the first tropical cores with glacial ice at the bottom. Precise dating of the Huascaran cores, however, has so far remained elusive. Changes in temperature during their long and difficult transit back to North America may have compromised some of the integrity of the trapped bubbles as atmospheric fossils. To remedy this, and for a second point of reference in the tropics, Thompson chose Sajama, an inactive volcano in southwest Bolivia that is just 18 degrees south of the equator. "We think the volcano has been dormant for a long time," Sowers says. "We're praying it's been at least 15,000 years."

From a scrap bucket next to a table saw in the walk-in freezer behind the Deike building, Sowers plucks a disc of ice, larger in diameter but flatter than a hockey puck, and holds it up to the light. A thousand tiny bubbles glimmer in the glassy matrix. "Some of the bubbles we see in samples like this hold air that is half a million years old," Sowers says.

Without precise, accurate dating, this old ice holds little value. But good dating is a difficult proposition. For one thing, snowfall varies greatly, even in polar regions. Greenland gets an average of 10 inches a year, 10 times what falls on Antarctica. (The lack of moisture in very cold air accounts for the paucity in both places.) Compaction rates vary too, according to complex physical factors. So a given depth of ice does not necessarily represent the same age.

One solution is to read the core much as one would read a tree ring. A trained eye can pick up the difference between adjacent layers of winter and summer snow, the latter being cloudy or gray. An even better annual marker, for Greenland ice anyway, is the dust layer deposited every spring by the northward stirrings of the jet stream. Sowers' Penn State colleague Richard Alley, professor of geosciences, has been at the center of a painstaking effort to document the GISP-II cores. "People have actually counted every dust layer back to 50,000 years ago," Sowers says. "Because of this work, we now have an absolute time scale."

But dating by dust doesn't work for tropical cores. As Sowers explains it, the maximum depth of an ice sheet is constrained by its "footprint," the diameter of its base. The summit of Sajama is about 200 meters across, surface enough to sustain an ice sheet about 130 meters deep. Compared to the continental sheets that bore the GISP-II cores, over 3,000 meters deep, and Vostok, over 3,500, it isn't much. "Time is extremely compressed in tropical cores," Sowers says.

In order to fix dates on such tight-packed samples, researchers have had no choice but to rely solely on the projections of computer models—an approximate solution at best. That's where Sowers comes in. For his Ph.D. thesis, working at the University of Rhode Island during the late 1980s with paleoclimatologist Mike Bender, Sowers developed a method for dating cores by analyzing the isotopic composition of the oxygen present in their bubbles.

The technique is based on the existence of two isotopes of oxygen in the atmosphere. Oxygen-16 and Oxygen-18, Sowers explains, are chemically identical but subatomically slightly different: O-16 has eight neutrons and eight protons, while O-18, or heavy oxygen, has an extra two neutrons. The relative abundance of these two isotopes in atmospheric oxygen has varied throughout earth's history. Fortuitously, this variation can be correlated with changes in climate. "The amount of O-18 in the atmosphere," Sowers says, "is dictated by the amount present in seawater, which changes with the volume of continental ice."

Earth's atmosphere is extremely well-mixed, he notes, so much so that "particles of smoke from a cigarette smoked in State College today will be in Antarctica in two or three years." So a core taken from anywhere on Earth will, for the same date, have the same O-18 level. And because he has done atmospheric O-18 studies on the firmly dated GISP-II core, he has a yardstick: a record of O-18 levels for the last hundred millenia. To date a tropical core he has only to take an O-18 reading and fit it into this sturdy frame of reference.

A trip to Sowers' lab shows the apparatus he has constructed to recover the trapped gas from ice core samples. A 15-gram chunk of ice is placed into a glass vessel that looks like a large light bulb. To separate the fossil air from the ice matrix, he places the contents of the bulb under a vacuum, drawing out all the present-day lab air, then he simply allows the sample to melt. "You can actually see the bubbles burst, because they're under pressure," he says. "Eventually you're left with a layer of glacial water in the bottom of your vessel, and a quantity of paleo-air in the headspace above it."

After refreezing the water ("to squeeze out any remaining air"), Sowers transfers the fossil air into a stainless steel tube which is inserted into the center of a large metal flask filled with liquid helium. At -270 degrees C, the tube is cold enough to freeze both oxygen and nitrogen solid. This concentration step allows efficient transfer of the fossil air sample to a mass spectrometer, where, after it returns to room temperature, the gas sample is analyzed into its elemental and isotopic composition.

"The ultimate goal of this research," Sowers has written, "is to gain insight into the mechanisms which caused the Earth to move out of the last glacial period."

What exactly happened 10,000 years ago, to put an end to the last ice age and render large expanses of Earth's mid-latitudes habitable to shirt-sleeved humans? Already the polar cores have given some clues. About 18,000 years ago, they show, there was a "significant warming event" over Antarctica. This occurrence coincided with a decrease in dust (suggesting a shift in atmospheric circulation), increases in levels of the greenhouse gases carbon dioxide and methane, and a rise in sea level. Some three thousand years later, a similar warming swept the North Atlantic.

What Sowers and his colleagues do not yet know, however, is what caused these changes, or exactly how (or even if) they are related. Why did warming happen first in Antarctica, traditionally viewed as the tail of the dog? And was the observed increase in greenhouse gases actually a cause of that warming, or merely a result of it? "Water vapor in the atmosphere is the most important greenhouse gas," Sowers notes. "A significant change in the amount of atmospheric moisture could produce a change in ocean circulation, which would change the CO2 content in the atmosphere. That's one way climate change could have preceded a change in CO2."

In the North Atlantic, he continues, one theory for the warming at what he and other climatologists refer to as "termination" has to do with the "switching on" of something called the North Atlantic deep water, a kind of global conveyor belt for heat.


The tropical latitudes receive a preponderance of the sun's heat, and are also thickly blanketed with a layer of water vapor (clouds), preventing that heat's re-release back to space. In order for the global heat budget to maintain its equilibrium, therefore, there has to be a means of moving heat from the tropics to the polar regions, where excess heat can leave via longwave radiation into space. The North Atlantic deep water is an important part of this movement. Every winter, when temperatures get cold enough in the Greenland Norwegian seas, the surface layer of ocean becomes more dense than the water beneath it, sinks, and heads south along the bottom of the Atlantic. This bottom current runs opposite to the northbound Gulf Stream.

During the last glacial period, however, current evidence suggests, this treadmill was slowed substantially; at times, in fact, the North Atlantic deepwater was turned "off." So how did the Earth adjust? Where did the excess heat go? And: What function eventually forced the conveyor back on? The polar record alone can't solve this puzzle.

"Forcing functions are a black box right now," Sowers admits. "We're grasping for clues." But evidence from the tropical cores, he says, should sharpen the picture. "It's easier to envision climate change originating in the tropics—these are very dynamic systems, with all kinds of events and feedbacks. Hurricanes, cyclones—the majority of the action is sitting right here." Conjecture aside, however, cause and effect remains unproven. Do things really start out in the tropics—or do they end up there? Is the tropics the first domino, or the last?

The prerequisite to understanding the tropics' true role, Sowers suggests, is putting them onto the timeline: That's why accurate dating assumes such importance. "If we can pin down the sequence of events," he says, "we're on our way to understanding the physics of the system." And if climatologists can fathom the physics that have governed the planet in the past, it follows, they can plug that knowledge into their general circulation models, the global computer simulations that are being used to predict climate change in the future.

"In order to provide a good time scale for Sajama," Sowers says, "it was important for me to see the drilling operation, to understand the history of the cores."

The problem with the Huascaran data from Peru, he suspects, is that the cores were not kept cold enough during their long transport on foot down from the summit. For analysis of the ice itself, he explains, "you don't have to be as careful—just so you keep things below freezing"; but when you're studying those fragile bubbles, temperatures have to be kept much lower, below -10 degrees C. Any warmer and the ice will start to expand, causing microfractures which can allow fossil air to escape. "This time," Sowers says, "we were much more careful with temperature. That's one of the reasons I went along."

Thompson's plan to use a balloon to ferry samples down from Sajama was an attempt to address the transit problem. A balloon, he reckoned, could get the cores from summit to base camp in half an hour, much quicker than the five to six hours it would take for even the most experienced porters to run them down. There would be that much less chance of their getting overheated.

Balloon transport, however, was only the beginning of the logistical challenge that was the assault on Nevado Sajama. Ten thousand pounds of equipment, including the drilling rig itself and the 60 solar panels required to run it, had to be carried to the summit, a 6,000-foot ascent. ("There was some technical stuff," Sowers says, describing the climbing, not the equipment. "Nothing too serious, but once we got onto the ice we were roped up for safety.") In order to acclimate the scientists to low oxygen, the trip was split into at least two days: the first to get from base camp to high camp, and the second (ideally after a few days to get adjusted) to push on to the summit. A team of 20 Bolivian porters did most of the hauling once the load had been carried as far as the base camp by burros, and Peruvian mountaineers served as guides.

Before they ever set eyes on the mountain, however, they had to get all that equipment through customs at La Paz, and then gain all the required permissions from government officials. That process took a week. Then, when they got to the village of 500 people at the base of the mountain, there were further obstacles.

"Sajama is actually the highest point in Bolivia," Sowers says, "and for that reason it gets a lot of tourist climbers—it's really a tourist trap. But the mountain is also the local deity—not only their livelihood. When we showed up, even though we had the necessary permits, the local officials said no way. They weren't going to allow us to rape their god."

There was a town meeting, at which Thompson rose to plead the expedition's case: that they were using only solar-powered drills, they would not destroy anything, they would carry out everything they carried in. Finally, after a closed-door session, the village leaders emerged and announced that they would allow the expedition to go forward, with two provisos. The first, palatable enough, was that copies of any publications based on the expedition's research would be donated to the tiny village library. The second requirement was that the scientists participate in the sacrifice of a white llama before they set out, as a gesture of appeasement to Sajama. "It was an amazing ritual," Sowers remembers. "It was the first time it had been done in a couple of decades."

When it was complete, Sowers was part of the first team of five to go on to the summit. "The first couple days we had nice weather," he remembers, "Then we got socked in for three days by a storm—nothing to do but sit in your tent and read." To make things worse, the dysentery outbreak at the high camp put a temporary kink in the food supply line. "Luckily, Lonnie had been through this before, so we had enough dry food to survive, but it was not fun for a couple of days." By the end of a week, however, all the drill parts and solar panels had arrived, and the drilling operation could commence. "The hardest part is always getting through the firn-ice transition layer," which in this case happened at about 40 meters, Sowers says. "That took five or six days. After you get into the ice, things are pretty routine."

The rest of the expedition played out relatively smoothly. Everything, that is, except for the balloon transport system. "Three of the five pilots the company sent down couldn't handle altitude," Sowers reports. "The other two couldn't get the system operating right, between winds, weather, and mechanical problems." The cores were carried down on foot after all, three meter-lengths to a backpack, at two or three in the morning, when temperatures were coldest. "We had a Land Rover type vehicle at the base camp to carry them the couple of miles to the nearest real road," Sowers says. Waiting there was an ice-cream truck that had been hired in La Paz for the demand-driven fee of some $20,000. The truck took them all the way to the international airport. From there, it was insulated boxes on a commercial carrier to Miami ("You can imagine coming through customs with this shipment from South America, and trying to look like scientists."), and finally onto a freezer truck for the 30-hour drive to Columbus, Ohio.

Amazingly, Sowers reports, the cores arrived intact. Matchbook-sized sensors he slipped in with samples as they were packaged at the summit recorded temperature histories that stayed well below the danger zone. "This time," he says, "I think we've got some very good ice."

Todd A. Sowers, Ph.D., is assistant professor of geosciences in the College of Earth and Mineral Sciences, 447 Deike Building, University Park PA 16802; 814-863-8093; Lonnie Thompson, Ph.D., chief scientist for tthe Sajama expedition, is professor of geological science and research scientist at the Byrd Polar Research Institute at the Ohio State University. Sowers' work related to the Sajama expedition is funded by the National Science Foundation; the expedition itself was funded by the National Oceanographic and Aeronautic Administration. For more information about the work on Sajama, check the following sites on the World Wide Web: and

Last Updated May 01, 1998