“Even before climate change came onto the scene, we were worried about the potential contribution to sea level of West Antarctica,” says Robert Bindschadler, a glaciologist at NASA Goddard Space Flight Center in Greenbelt, Maryland. Unlike other ice sheets, the WAIS rests on ground that sits below sea level (without ice, West Antarctica would be an archipelago). This leaves it uniquely sensitive to warming oceans. It is the only marine ice sheet to have survived beyond the last ice age, and just as scientists predicted 30 years ago, the parts of it that are shrinking are those that are raked by sea currents.

The very shape of the WAIS makes it vulnerable, says geophysicist Donald Blankenship of the University of Texas at Austin. Moving inland from the edge of the WAIS, the glacier bed plummets farther below sea level and the ice gets thicker. As the glaciers retreat inland, more and more of their ice will come in contact with warmer water, while the area on which new snow can gather will decrease—a double hit that many people think will cause the WAIS glaciers to melt more and more quickly. “Once they get short enough they’re not stable,” Blankenship says.

Another wild card lies beneath the WAIS. The sea basin that it straddles is a rift valley, a geologically active zone with volcanoes and geothermal heat accelerating the melting of ice off the sheet’s underside. That water gushes in rivers beneath the ice and collects in lakes, which periodically flood. No one knows how much that water influences slippage of the ice above it, but finding out is critical to predicting how the WAIS will respond to warming temperatures.

Tulaczyk has come to Antarctica to find out. He and Joughin (back in Seattle) will monitor changes in the Whillans Ice Stream as closely as doctors monitor an ICU patient’s vital signs. The goal is to understand how the flow of water controls ice movement.

“That’s the total holy grail right now, to see if we can pin down the response of the ice stream to lakes’ filling and draining,” says Helen Fricker of the Scripps Institution of Oceanography in San Diego. Fricker uses a laser altimetry satellite, called Ice­Sat, to monitor these lakes as they fill and drain by measuring the rise and fall of the ice above them. But Tulaczyk and Joughin hope to gather information that no satellite could by installing GPS sensors on the ice, simultaneously measuring its vertical and lateral movements to determine the effect of water running underneath.

It is 10 p.m. on the WAIS, and sun filters through the fog like fluorescent light. In this nondescript spot 25 miles from camp, Tulaczyk connects boards on the first GPS unit. The instrument must survive conditions notorious for sapping electronics—four months of sunless winter, with temperatures down to –75 degrees Fahrenheit. A solar panel will recharge its four 70-pound batteries during summer, a wind-powered generator during winter. Every 10 seconds for the next two years, this unit will triangulate radio signals with GPS satellites overhead to measure movement of the ice to the nearest half inch.

By now we have developed a field routine: 9-hour workdays alternating with 14-hour workdays. We return to camp as late as 1 a.m., which hardly matters in the 24-hour summer light.

Back at camp we anchor our equipment in the snow with bamboo poles so it doesn’t blow away. We pee at a patch of yellow snow. At night we sleep in one-person tents and use bottles when the need arises (my bottle, issued to me in McMurdo, bears the words Karen’s pee written in black marker). We store our food in snow trenches, and in the morning we use a hacksaw on slabs of frozen egg to fry up. The temperature ranges from 5 to 15 degrees F. “It is the banana belt of Antarctica,” Tulaczyk had assured me months before our trip, but I still sleep with my laptop computer to keep its electronics from going haywire in the cold, and I must warm the icy object on my bare belly so it will start up.

We install a total of 10 GPS units, some of them atop Lakes Whillans, Mercer, and Seven, subglacial lakes that Fricker, the scientist at Scripps, discovered by satellite a few months before. Pettersson’s ice radar identifies water channels under the ice, above which we place other GPS units. The lakes and channels are undetectable by human senses, since they lie beneath half a mile of ice. Only our GPS coordinates tell us when we’ve arrived on top of them.

We ride our snowmobiles as much as 10 hours a day as we install the GPS units. Those long rides present the periods of greatest discomfort as hands, feet, and thighs gradually numb with cold—but also the best moments. The Antarctic ice, sculpted by wind into the layered texture of Monument Valley sandstone, flies past, and the snowmobile’s shocks judder and clack over ice ridges called sastrugi. The glare of sunlight turns the ice silver-gray, and at times we seem to be zooming through a desert.

By the time we make our second trip to Lake Mercer, the icescape feels as familiar to me as the hills of Northern California. We’re retracing our tracks from two days earlier. When we stop for a moment, I hop off my snowmobile and walk a few steps along the tracks.

It’s only after my right foot has sunk six inches that I begin to feel that something is wrong. Horribly wrong. My right foot sinks deeper in the snow and I shift my weight to the left. By the time I stop my leg from sinking further, it’s halfway to the knee. I pull it out and look into the hole that I’ve created. It has no bottom.

I’m peering into a hidden world inches beneath our feet. Two crisp walls of ice drop vertically. Cold blue light seeps in. The crack, a foot or more wide, plunges 20 feet, 30 feet, and disappears into bottomless black. It’s concealed beneath a crust of snow 12 inches thick. On the surface only the faintest impression—as thin as a shoelace—trickles across our snowmobile tracks.

I share a couple of quiet seconds with this thing, staring into its maw, mostly—believe it or not—in wonder. Then I step back gingerly from its lip. “Hey, guys,” I yell. “Crevasse.” Pettersson motions us onto our snowmobiles; they’re safer, since they create a bigger footprint for distributing weight. Drive forward half a kilometer, he says. In 30 seconds we cross the mark of another crevasse, then another and another—four or five total. The marks cut through our tracks from two days ago, meaning that the cracks have widened in just the last 48 hours. Our crevasse-detecting radar missed them.

I spend the rest of that afternoon in solitude, bouncing for hours on the radar-bearing cargo sled as Pettersson drags it across the ice plain. We are measuring the contours of the landscape below. All the while, the crevasse I stepped into plays on an endless loop in my mind, often with a horrible ending: wedged in a crack without any help. That image will quicken my pulse at unexpected moments for weeks to come. But ironically, the memory of what really did happen—looking into this cold blue portal and walking away unharmed—will remain one of my most cherished from the trip.

Given the extreme costs and dangers of putting boots on the ice, much Antarctic research these days is done with satellites. But satellite monitoring is a young science, which presents limitations. When Bindschadler, the NASA Goddard glaciologist, wanted to study how the WAIS had changed over decades, the problem was that decades ago few satellites had orbited over the area. Bindschadler’s research led him to an archive of newly declassified spy satellite photos from the cold war. Among them he found two images taken by one of the CIA’s Corona satellites on October 29 and October 31, 1963. These satellites, launched to monitor Soviet military movements, captured photos on massive spools of film that were parachuted back to earth and snagged in midair by U.S. planes.

A comparison of these photos with images from 1992 revealed surprisingly rapid changes: the Whillans Ice Stream had widened by 2.5 miles and had eroded 10 miles off a wedge of slow-moving ice to the north called Engelhardt Ridge.

Even people using modern satellites to monitor ice suffer from a lack of options. Due to competing scientific priorities, few NASA satellites are optimally designed to monitor the earth’s polar ice, and those that are sometimes hit snags. The IceSat satellite, which Fricker uses to monitor subglacial lakes and which others use to monitor the sagging tops of melting glaciers, can function only 66 days per year due to a technical glitch.

Scientists also face shortfalls in other satellite functions that they need for monitoring the Antarctic ice: high-frequency radar for measuring the speed of glaciers and low-frequency radar for measuring ice thickness. “It’s a major impediment to developing realistic ice sheet models when you don’t even know how thick some of these outlet glaciers are,” says Eric Rignot, a remote-sensing glaciologist at the NASA Jet Propulsion Laboratory in Pasadena, California. Even NASA’s gravity-sensing GRACE satellites, which have provided stunning ice mass data since their launch in 2002, are nearing the end of their planned life. “This is probably the best method to look at mass changes of ice sheets if you want to get a number that you can trust,” Rignot says.

In February 2008 NASA announced its satellite missions for the next decade. They include a new IceSat satellite, but it won’t come online until years after the expected end date of the one it is replacing. NASA also has plans for new gravity-sensing satellites, but not for a decade or more—well after the current GRACE mission has ended. “We’re definitely going to have an instrument gap,” Fricker says. “We’re going to be blind for a while.”

That blind spot comes at a time when ice sheets in West Antarctica and Greenland are changing more quickly than anyone expected. Glaciologists would like to know what’s happening. The shortage of satellite coverage underscores the need for field projects like Tulaczyk’s and Joughin’s to succeed.

My nylon tent roars in the wind, nearly flattening to a trapezoid with each gust. Katabatic winds, caused by cold air accelerating as it slides off the polar plateau, have battered us for 36 hours. I have hardly slept.

I kick at the wall of the tent to break away a shell of ice and slip outside with Karen’s pee bottle in hand. As I stumble through blowing snow, the ghost of a bamboo flagpole looms into view. We planted these flags the previous day to mark the route to the cooking tent, for which I’m thankful. I turn my back to the wind, look up, and despite the whiteout I see a circle of blue sky directly above—the quintessential sign of an Antarctic storm. Very little new snow is actually falling; most of it is being redistributed, wind-blasted off the ice or blown off the Transantarctic Mountains 40 miles away.

The storm gives way to an afternoon of digging cargo sleds and food out of the snow. A flag marks the location of each object; 47 of them flap and clap nonstop. The morning after that day of digging, Tula­czyk, Pettersson, and I depart for what we expect to be our toughest day—following a serpentine route to place a GPS unit near heavily crevassed ice, 30 miles away. We anticipate a 10-hour ride, slow and roped together, but after only 30 minutes I begin to glimpse the telltale marks of covered crevasses. We slow down and Pettersson stands on his snowmobile to scan the snow ahead. A minute later he raises his hand for us to stop. He dismounts, plays out some rope, and slides an avalanche probe into the snow to check for crevasses. Jab after jab, the probe sinks easily to the hilt.

Pettersson announces the verdict: A buried crevasse, 10 feet wide, blocks our path. It is covered by just two feet of fragile snow. The gentle slope of the ice predicts bigger cracks ahead. Despite months of planning, our day is over. We turn around.

Over the next several days, we install the last GPS units, one of them near the crevasse that I stepped into and another near the 10-footer that stopped us. The crevasses are interesting, Tulaczyk says, because they mark spots where, half a mile below, the ice grinds over a high spot in the underlying rock—sort of a thorn in its belly. The ice stretches over these sticking points twice per day as the tide lifts up and sets down the Ross Ice Shelf, a 1,100-foot-thick slab of ice the size of Spain that hangs over the ocean several hundred miles away. “You could actually get daily cycles of opening of these cracks,” Tulaczyk says.

Our time in the field ends with a day spent chipping our sleeping tents out of the ice that clings tenaciously to their edges, and digging the cooking tent out of drifts as deep as six feet.

As planes ferry us back to McMurdo on the first leg of our trip home, Tulaczyk and Pettersson open their laptops at every opportunity to peek at the data they’ve collected so far. The GPS stations have tracked the movement of the Whillans Ice Stream and confirmed a curious fact: The Whillans sits still most of the time. Twice per day it lurches into motion, slipping 18 inches forward across its entire 5,000-square-mile expanse. That slip corresponds to ocean tides lifting up and setting down the Ross Ice Shelf. (The Ross acts as a brake that holds the ice sheet back; its tension is released at these twice-daily moments.)

The seismometers that we planted show the curvy peaks of ice quakes occurring each time the glacier lurches. Several months after we return home, another team will report in the journal Nature that these twice-daily ice quakes measure a staggering magnitude 7—strong enough to topple cities and kill thousands if they were to happen in a populated area. Only because the quakes occur in slow motion were we not tumbled out of bed every morning during our stay on the ice sheet. Tula­czyk’s own calculations put them at a more modest magnitude 4, still large enough to jostle jars of honey off kitchen shelves in Los Angeles or San Francisco.

The bigger questions that brought Tulaczyk to Antarctica will take longer to answer, but the ice sheet already seems to be cooperating. Shortly after we return to California, the GPS unit on Lake Mercer will start to report—via satellite link—that the ice is rising, a sign that the lake is filling with water. At some point the lake will bloat past its holding capacity and issue forth a subglacial flood [pdf]. Tulaczyk’s and Joughin’s sensors will report whether the ice sheet suddenly speeds up when that happens.

Tulaczyk and Joughin should start to see within two years how water controls movement of the ice. That information will feed into models that tell us how the WAIS will fare in the next 100 years and whether its glaciers are capable of a massive speedup, as some people fear.

“We have the technology” to predict the ice sheet’s fate, Tulaczyk says as we chat in a quiet corner of McMurdo’s dining hall a few hours before leaving Antarctica. “The commitment to invest the money is the limiting factor.”

As we sit there, exhausted, my mind wanders. In a few minutes I’ll head to my dormitory and take my first shower in a month. Then I’ll catch an hour of sleep before boarding a 2 a.m. flight on a military cargo jet back to New Zealand—December 21, last flight before Christmas. Twelve hours from now I’ll walk the streets of Christchurch and revel in things that I haven’t experienced in six weeks: sunset, stars, living things, the color green—and hot Indian curry.

For eons our species and its ancestors have lived according to the faraway rhythms of the WAIS as it expanded and collapsed time and again, helping to drive sea levels up and down around the world. Only now is humanity exploring this remote area that affects its fate.

“Many of the places we went have probably never been crossed by people,” Tulaczyk says. His beard is full and his skin darkened by the ultraviolet rays that glaciologists come to expect from working on a reflective ice sheet in 24-hour sunlight. “Almost every time you make measurements on the ice, something really new pops up,” he says. “That’s part of why people put up with this type of work.”

A quarter-inch crack indicates the location of a crevasse. An avalanche probe showed that the underlying crevasse, hidden beneath 2 feet of snow, is ?10 feet wide.