Glaciers are melting. Seas are rising. We already know ocean water will move inland along the Eastern Seaboard, the Gulf of Mexico and coastlines around the world. What scientists are urgently trying to figure out is whether the inundation will be much worse than anticipated—many feet instead of a few. The big question is: Are we entering an era of even faster ice melt? If so, how much and how fast? The answer depends greatly on how the gigantic Thwaites Glacier in West Antarctica responds to human decisions. It will determine whether the stingrays cruising seaside streets are sports cars or stealthy creatures with long, ominous tails.

Global warming is melting glaciers up in mountainous areas and expanding ocean water, while shrinking ice at both poles. Averaged over the planet’s oceans for the past 25 years, sea level has risen just over a tenth of an inch per year, or about a foot per century. Melting the rest of the globe’s mountain glaciers would raise the sea a little more than another foot. But the enormous ice sheets on land in the Arctic and Antarctic hold more than 200 feet of sea-level rise; a small change to them can create big changes to our coasts. Ice cliffs many miles long and thousands of feet high could steadily break off and disappear, raising seas significantly.

Well-reasoned projections for additional sea-level rise this century have remained modest—maybe two feet for moderate warming and less than four feet even with strong warming. Scientists have solid evidence that long-term, sustained heating will add a lot to that over ensuing centuries. But the world might be entering an era of even more rapid ice melt if the front edges of the ice sheets retreat.

To learn whether this could happen, we look for clues from changes underway today, aided by insights gained about Earth’s past and from the physics of ice. Many of the clues have come from dramatic changes that started about two decades ago on Jakobshavn Glacier, an important piece of the Greenland Ice Sheet. Glaciers spread under their own weight toward the sea, where the front edges melt or fall off, to be replaced by ice flowing from behind. When the loss is faster than the flow from behind, the leading edge retreats backward, shrinking the ice sheet on land and raising sea level.

During the 1980s Jakobshavn was among the fastest-moving glaciers known, racing toward Baffin Bay, even though it was being held back by an ice shelf—an extension of the ice floating on top of the sea. In the 1990s ocean warming of about 1.8 degrees Fahrenheit (one degree Celsius) dismantled the ice shelf, and the glacier on land behind it responded by more than doubling its speed toward the shore. Today Jakobshavn is retreating and thinning extensively and is one of the largest single contributors to global sea-level rise. Geologic records in rocks there show that comparable events have occurred in the past. Our current observations reveal similar actions transforming other Greenland glaciers.

If Thwaites, far larger, unzips the way Jakobshavn did, it and adjacent ice could crumble, perhaps in as little as a few decades, raising sea level 11 feet. So are we risking catastrophic sea-level rise in the near future? Or is the risk overhyped? How will we know how Thwaites will behave? Data are coming in right now.

Waffles on the coast

Calculating Thwaites’s threat is complex. To make sense of it, let’s begin with breakfast. If you pour batter on a waffle iron, your mound will spread across the iron’s crosshatched grid. Physically, the weight of the batter pushes the mound outward against the friction on the grid below it. This spreading slows as cooking stiffens the batter—or if you hold the batter back with your spatula.

Credit: Mapping Specialists; Source: “How Much, How Fast?: A Science Review and Outlook for Research on the Instability of Antarctica’s Thwaites Glacier in the 21st Century,” by T. A. Scambos et al., in Global and Planetary Change, Vol. 153; June 2017

Glacial ice sheets are like big waffles, up to two miles thick and a continent wide. Snow falls on top and is squeezed into ice under the weight of subsequent snowfalls. These huge ice mounds are strong—I have landed on them in heavy ski-equipped military transport planes—but they still spread. Their temperature is often within a few degrees of the melting point, making the ice soft enough to slowly flow from the high, central region toward the edges, where it more readily melts and breaks off. Thicker or steeper mounds such as those on Greenland and Antarctica spread faster.

Credit: Ben Gililand; Sources: “Sub-Ice-Shelf Sediments Record History of Twentieth-Century Retreat of Pine Island Glacier,” by J. A. Smith et al., in Nature, Vol. 541; January 5, 2017; Natural Environment Research Council

Left to itself, an ice sheet grows until it is thick and steep enough for the spreading, melting and breaking to balance the ongoing, additional snowfall. The mound can stay at one size for a long time. But that is not the case on our warming planet. The moisture in the snow that falls on Greenland and Antarctica each year, which almost entirely comes from the sea, is equal to a layer of water evaporated from all oceans, just over a quarter of an inch deep. The ice sheets are now returning about 15 percent more than this amount to the oceans, by meltwater runoff or icebergs that “calve” off, raising sea level a little. If melting remains greater than snowfall for long enough, an ice sheet can disappear. But that could take almost 100,000 years at recent rates. If warming rises, however, the melting quickens. That is the case we are facing globally.

Terrible beauty

An ice sheet’s flow depends on how strong the mound is, how well lubricated it is underneath on land, and whether or not it is held back by a spatula—an attached, floating ice shelf. General atmospheric warming can soften ice and thaw the places where the ice bottom is frozen to the rock below, allowing the ice to slide faster toward the sea. But the heat takes a long time to be conducted through two-mile piles. The big ice sheets have not finished warming from the rising air temperatures that ended the most recent ice age more than 10,000 years ago!

A speedier way to warm the ice and its bed is for water melting on top to pour down into crevasses. In some places on the flanks of Greenland’s ice, meltwater in summer collects in large hollows on the surface, forming big, beautiful blue lakes. The water, being denser than ice, tends to wedge open crevasses that can reach the bed at the bottom and drain the lake. An expanding lake can break through half a mile of ice or more, creating a flow of water greater than Niagara Falls. In an hour, that can warm the bed as much as would have occurred over 10,000 years.

This process is important, and we are studying it eagerly. But it is not the greatest worry for people on Earth’s coasts, because the bumpy bed can also keep the ice from speeding toward the sea.

The same mechanism presents a stronger threat if it happens on an ice shelf. In very cold places, the ice flowing into the ocean remains attached but floating. These ice shelves almost always occur in protected bays or fjords. The motion of ice shelves is slowed by friction along the shorelines around them and perhaps with upward protrusions from the seafloor, where the ice locally runs aground. The shelf slows the flow of the nonfloating ice on land toward the sea.

Meltwater pours into the Greenland ice sheet, hastening its slump toward the sea—a sign of things to come in Antarctica. Credit: Konrad Steffen, CIRES and University of Colorado

Warming air can create lakes on top of the ice shelves. When the lakes break through crevasses, a shelf can fall apart. For example, the Larsen B Ice Shelf in the Antarctic Peninsula, north of Thwaites, disintegrated almost completely in a mere five weeks in 2002, with icebergs breaking off and toppling like dominoes. That did not immediately raise sea level—the shelf was floating already—but the loss of the shelf allowed the ice sheet on land behind it to flow faster into the ocean—like pulling a spatula away, allowing the batter to run. The ice flowed as much as six to eight times quicker than it had been moving earlier. Fortunately, there was not a lot of ice behind the Larsen B Ice Shelf in the narrow Antarctic Peninsula, so it has raised sea level only a little. But the event put society on notice that ice shelves can disintegrate quickly, releasing the glaciers they had been holding back. Ice shelves can also be melted from below by warming seawater, as happened to Jakobshavn.

When shelves are lost, icebergs calve directly from ice-sheet cliffs that face the sea. Although this delights passengers on cruise ships in Alaska and elsewhere, it speeds up the ice sheet’s demise. At Jakobshavn today, the icebergs calve from a cliff that towers more than 300 feet above the ocean’s edge—a 30-story building—and extends about nine times that much below the water. As these icebergs roll over, they make splashes 50 stories high and earthquakes that can be monitored from the U.S.

So far ice-shelf loss and ice-cliff calving are contributing moderately to sea-level rise. But at Thwaites, this process could make the rise much more dramatic because a geologic accident has placed the glacier near a “tipping point” into the great Bentley Subglacial Trench.

Jump the bump

On an autumn morning in 1956, Charles Bentley (who years later would be my Ph.D. adviser) defended his thesis at Columbia University. The next day he hopped a train to Panama, then caught a ship heading south, to be part of the International Geophysical Year research project that would analyze planet Earth. He spent two years in West Antarctica before returning to find that he had not graduated yet, because his thesis fee had not been paid. In the meantime, he and his team traversed more than 3,000 miles of ice, to and from the Byrd Station research base and across vast reaches of West Antarctica. (Bentley died at age 87 in 2017.)

Of the many measurements and discoveries they made, the most important for our story involved the ice thickness. They set off small explosions on the surface and used seismometers to listen to sound traveling through the ice sheet and bouncing back off the bed. These data showed that West Antarctica was not a thin drape of ice overlying a high continent, as some had expected. Instead Bentley and his team found very thick ice, and they discovered the Bentley Subglacial Trench. There the bed plunges more than a mile and a half below sea level—Earth’s deepest place not under an ocean. And the ice filling it extends more than a mile above sea level.

Bentley and glaciologists who followed him had found a tipping point. The great trench and adjacent basins underlie the vast center of the West Antarctic Ice Sheet. If the front edge of Thwaites retreated from the coast back into the trench, it could make an ice face thousands of feet high, extending from far above the trench to deep down into it. Such a cliff—much bigger than at Jakobshavn or anywhere else on Earth—could break fast, making incredibly tall icebergs that would roll over and float away through the trench outlet to the ocean, raising sea level a lot.

Decades of additional research have established just how important this mechanism is. John Anderson, who recently retired after 43 years at Rice University, and many of his graduate students tirelessly mapped the continental shelf under the ocean around Antarctica, using side-scan sonar and other tools. During ice ages, Antarctic ice spread many miles farther in all directions and withdrew as ice ages ended. The seafloor around Antarctica today was the bed under the ice sheet in the past. Telltale imprints left in seafloor sediments give us accurate stories about ice sheets.

One story is that as expanding ice sheets push forward into the sea, they drag sediment with them. The ice stabilizes when it reaches a local high in the seafloor and then builds the seafloor higher there by piling the sediment into raised moraine shoals—long, stony walls that grow where the ice ends. Ice can sit in such a position for hundreds or thousands of years, rebuffing weak efforts to dislodge it. But if enough warming occurs, the ice retreats back down the sloping bed into the valley behind the shoal. The ice rarely stabilizes again until it reaches the next high ridge, often far behind it. Meanwhile icebergs float over the abandoned moraine shoal, which is still below sea level, and out into the ocean.

This is now happening in many places around Antarctica and Greenland. Jakobshavn Glacier has “jumped the bump” of a former moraine shoal and is retreating back through its valley-shaped fjord, “unzipping” a path into the greater ice sheet. When the first European explorers visited the area that is now Glacier Bay in Alaska, it was filled with a vast glacier ending on a large moraine shoal. Since then, the ice has retreated from that ridge, or bump, more than 60 miles inland to get to the next high ground, which today is the current shoreline of the beautiful bay.

Fortunately, most such retreats have only a limited effect on global sea level. Even a big Glacier Bay–sized glacier is small compared with the world ocean. Jakobshavn is just one of dozens of major drainages around Greenland’s ice sheet, but they do not quickly destabilize their neighbors in adjacent fjords, and they end not too far inland where the bed rises again. Similarly, Antarctica is drained by a great number of glaciers flowing down into their own waffle-iron valleys. With enough warming, many of them might retreat in unison, but each by itself is not a huge influence on the global sea.

The Bentley trench in West Antarctica and a few other deep regions in East Antarctica, including the Wilkes and Aurora basins, present a different story. Retreat through one of these to the next high ground would have global importance. Models point to Thwaites Glacier as the most likely path into the Bentley trench and connecting basins. If it started unzipping into the interior as Jakobshavn has, the melting could potentially raise sea level 11 feet before it stabilizes on high ground on the other side of the trench. The East Antarctic basins by themselves could raise sea level more than Thwaites would, but they require more warming to cause those glaciers to jump their bumps.

Note that there is nothing bizarre about this scenario. With sufficient warming, ice retreats, usually to the next high ground. This has been observed over and over in the past and present. If Thwaites becomes warm enough to start acting like ice in Greenland and Alaska, then it should retreat.

A fractured future?

How fast could Thwaites go? How much warming can we cause before it goes there?

My colleagues David Pollard of Pennsylvania State University and Robert M. DeConto of the University of Massachusetts Amherst programmed an ice-flow model that uses the relevant physics and can be run fast enough on advanced computers to study big changes in ice sheets over long times. I helped them a little with the physics of calving from high cliffs after ice shelves break off, especially if surface meltwater wedges open crevasses.

Pollard and DeConto optimized this model to match data from the geologic past and to assess the impacts of different amounts of human-caused warming. They determined that we probably have a few decades even under fast warming before the collapse of Thwaites is triggered by loss of its shelf and meltwater widening crevasses. Thwaites then would take a century or so to collapse completely. They did not know how fast the ice could break, though, so they set a top rate equal to what Jakobshavn had done in Greenland. (It has already exceeded that rate briefly.) And because Thwaites is thicker, it could make much higher cliffs than Jakobshavn. Higher cliffs tend to break faster (one reason highway engineers leave slopes rather than cliffs). So we could be underestimating the worst-case scenario, but we really do not know.

This is a good model, but it surely is not the last word from Pollard and DeConto or others. Some hope remains that Thwaites could stabilize on a deeper ridge on the downslope of the trench, behind its current position, before retreating still more, for example. Or icebergs could break off and pile up for a while behind the current ridge where the ice now starts to float, helping to re-form a shelf that could lessen the ice loss.

To address these and other questions, the National Science Foundation and the U.K.'s Natural Environment Research Council, together with other international collaborators, have launched a major effort to learn even more about Thwaites’s history, how the glacier is flowing, and what the seafloor surface is that it is flowing over, which will help all of us involved to better predict its future.* The data are almost guaranteed to reduce uncertainties and to be fascinating.

Some questions may remain difficult to answer. Think of all the ceramic coffee cups you have seen dropped on a hard floor. Some bounce, some crack, some chip, some break into a million pieces. The physical processes of these fractures are well known and readily calculated, and the behavior averaged across many dropped cups is predictable. But you would not want to bet your career, or anything else important, predicting the fate of the next cup that hits the floor.

The future of Thwaites depends a lot on fractures. Will the ice shelf fracture from the ice that now feeds it, causing the ice sheet to jump the bump and retreat into the deep basins? Will huge icebergs break off rapidly if ice-shelf loss produces a cliff along the sheet’s face that is higher than any now on Earth, driving retreat faster than any we have seen? Meltwater is important, but how much of the water will run off in rivers to the sea, and how much will percolate into snow and refreeze? How fast will the air warm? I suspect that coffee cups are easy to predict in comparison.

If the world can muster the effort, slowing and stopping warming from greenhouse gas emissions will slow sea-level rise, easing the mounting costs of coastal damage. But if Thwaites is poised to retreat briskly, preventing warming by limiting the damage incurred by human activity could be vastly more valuable.

*Editor's Note (2/22/19): This sentence from the print edition was edited after it was posted online to correct an error. The original referred to the British Antarctic Survey rather than the U.K.'s Natural Environment Research Council.