Although the latest U.N. climate report significantly increases its projections for sea level rise this century, some scientists warn even those estimates are overly conservative. But one thing is certain: Predicting sea level rise far into the future is a very tricky task.

When the scientists on the Intergovernmental Panel on Climate Change (IPCC) sat down to hash out the chapter on sea level rise for their new report, which was released last month, they had their work cut out for them. 

Sea level forecasts were the most controversial part of the previous report, issued in 2007: Scientists and the public alike bristled at the lowball estimate of under 60 centimeters (nearly two feet) by 2100, which, the IPCC admitted, did not include the possibility of rapid ice flow from Greenland or the Antarctic into the sea. That was clearly important — those two ice sheets alone hold enough water to raise sea level by 65 meters, compared to 0.4 meters from all the world’s mountain glaciers. But researchers’ understanding of the ice sheets was so uncertain, the IPCC said, they just couldn’t bring themselves to put a number on it. “Some things had to be neglected,” says Don Chambers, sea level researcher at the University of Texas. “Because of that, the projections were on the low side.” 

Things today are more certain. In its latest report, released on September 27, the IPCC finally could and did put a number on ice flow from the poles. The result was an estimate of sea level rise of 28 to 98 centimeters (a maximum of more than three feet) by 2100 — more than 50 percent higher than the 2007 projections. “We have our arms around the problem well enough to say there’s a limit to how crazy things are going to get,” says Ted Scambos, head scientist at the U.S. National Snow and Ice Data Center. 

But that doesn’t mean that everything about sea level is now understood. Far from it. Big questions still hang over the fate of the ice sheets, which, the IPCC admits, could bump up the most recent projections by tens of centimeters. And there are a ton of smaller factors for researchers to come to grips with. 

“We all think that we’re committed to a meter of sea level rise. We just don’t know exactly how quickly.”

The ocean doesn’t rise steadily like water poured into a bathtub — instead there are splashes and jiggles in its rise. Weather patterns like El Niño can shove tens of centimeters of water up onto shores for months at a time, as they did in California in 1998. Floods in Australia in late 2010 strangely resulted in water piling up on that continent, robbing the oceans of enough water to lower global sea level by 7 millimeters for more than a year. While the ocean grows, the land also shifts: The ground rises where it was once pressed down by glaciers, and river deltas sink as loose sediments compact. What looks like sea level rise in one place might really be the result of the land falling. 

All this means that unravelling what the oceans are doing today is a heinously complicated task. Extrapolating their behavior is even trickier. “Predicting that into the future is very problematic,” notes Chambers. Says Steve Nerem at the University of Colorado, “We all think we’re committed to a meter of sea level rise. We just don’t know exactly how quickly.” 

Some facts are well established. Researchers can say that global ocean levels have risen about 19 centimeters in the last century. And the rate of rise has sped up. The 20th-century average is about 1.7 millimeters per year; since 1993 the average rate has nearly doubled — to about 3.2 millimeters per year. Those sweeping statistics about decades-long trends haven’t changed much since the last IPCC report in 2007. The devil, of course, is in the details. 

One problem has been attributing what, exactly, has caused the rise seen so far. Since the 1970s, for example, it is thought that about 40 to 50 percent of sea level rise was caused by ‘thermal expansion’ — the fact that water simply takes up more room as it gets warmer; 35 percent by melting glaciers; 5 percent because people have been extracting groundwater, using it, and pouring it into the ocean; and the remaining amount probably from melting ice at the poles. The primary reason this accounting is tricky is spotty data: Satellite measures of ocean height only go back to 1993, for example, and of the world’s more than 100,000 glaciers, there are only 17 with melt records going back 30 years or more. “We have to make huge assumptions,” says Chambers. 

Another problem is untangling short-term from long-term trends. The rate of sea level rise has mysteriously slowed down in the most recent decade, for example. The leading theory is that this blip is due to heat being sucked up by the deeper, colder parts of the ocean; cold water simply doesn’t expand so much on heating as warmer water does, so the sea level rise is less, says Nerem. But the slowdown isn’t expected to last. 

The biggest question remains how fast and how far the polar ice sheets will melt.

By far the biggest question remains how fast and far the polar ice sheets will melt. For this, researchers have one key helper: the Gravity Recovery And Climate Experiment (GRACE). Launched in 2002, these two satellites detect the mass of the ground beneath them, and so can be used to monitor the changing weight of the ice caps. “GRACE was a game changer,” says Jerry Meehl, a climate modeler at the U.S. National Center for Atmospheric Research in Boulder, Colorado. “Before that they were guessing.” 

Even GRACE is not clear-cut, however. The satellites detect the end result of a combination of effects, including the change in mass caused by the shifting crust and mantle, increased snowfall, and loss of ice from melt or iceberg calving. So results are open to interpretation: The 2012 GRACE estimate for ice loss from the Antarctic was just half the best guess from 2006

In Greenland, researchers have seen the rate of ice melt double since the 1990s, and warm water licking at the edge of the island has increased glacier calving into the sea. More snow is falling, but overall the island is losing weight and is expected to continue to do so. “For Greenland we can be confident now we really know what’s going on. All the methods are converging,” says Philippe Huybrechts, an ice modeler at the Brussels Free University. Worryingly, the IPCC expects there is some global threshold — as low as 1 degree C, or as high as 4 degrees C above pre-industrial temperatures — beyond which Greenland will irreversibly melt away over hundreds of years. We’re already 0.85 degrees C warmer than 1880. 

The picture of the Antarctic is far fuzzier: The error bars on IPCC projections mean that the panel cannot even say for certain that the continent will lose mass by 2100; it may actually gain a bit in the short term. The IPCC expects to see much more snowfall, particularly in the east, and it should remain too cold for the ice to simply melt away. But the continent is also losing ice from its edges as warmer water causes ice shelf collapse. There is a chance that this ice outflow could cause runaway collapse of the entire western Antarctic ice sheet. This could add several tenths of a meter to global sea level rise by 2100. “There is a lot of ice there,” says David Vaughan with the British Antarctic Survey. “If it is knocked out of balance, Antarctica could quite easily become the dominant contribution.” 

When ice experts were surveyed, their best guesses for polar ice loss were all over the map.

In the face of all these uncertainties, some have taken a different predictive approach. Instead of trying to model the physics behind every process contributing to sea level rise (from thermal expansion to melting ice), they argue, why not instead simply look at how sea levels have corresponded to temperature over hundreds of years and extrapolate? These so-called ‘semi-empirical’ models tend to top out twice as high as the ‘process-based’ models, making 2 meters of sea level rise feasible for 2100 — enough to flood the homes of 187 million people. But the IPCC says it doesn’t have much confidence in these results. “They’re interesting,” says Chambers, “but I don’t think they should be given as much weight as the process-based models.” 

A few scientists disagree, including Stefan Rahmstorf of the Potsdam Institute for Climate Research, who works on semi-empirical models. “We have two different approaches, and they give different results,” says Rahmstorf. “I don’t know which one is closer to the truth. But I object to the IPCC selecting one type and dismissing the other.” Other reports, notes Rahmstorf, including a 2012 assessment by the U.S. National Oceanic and Atmospheric Administration, give more pessimistic sea level predictions, going up to 2 meters by 2100. 

Another way to capture the range of expert opinions is to do a formal “expert elicitation” — a structured way of simply surveying experts and asking them what they think. When two dozen ice experts were surveyed in 2012, their best guesses for polar ice loss were all over the map — but their average was higher than the IPCC’s estimates, and their range easily put sea level rise over a meter by 2100. For a middle-of-the-road emissions scenario, for example, this group guessed at a sea level rise of 33 to 132 centimeters; the upper bound of that is roughly in line with semi-empirical models, but nearly double the IPCC’s most recent estimate (which allows, to be fair, tens of centimeters of wiggle room for possible dramatic ice sheet flow from the Antarctic). “The [IPCC’s latest] AR5 process-based projections appear optimistic and over-confident when compared with views of ice sheet experts,” writes glaciologist Aslak Grinsted of the University of Copenhagen on his website. There is “no good reason” for this, he adds. 

This debate about whether there will be 1 or 2 meters of sea level rise by 2100, however, pales in comparison to the numbers for the long-term outlook. The last time the planet was steadily 2 degrees C warmer than pre-industrial times, some 120,000 years ago, sea levels were 5 to 10 meters higher than today. It’s likely we’ll hit 2 degrees C of warming by 2100, unless we take extreme measures to mitigate emissions. “The bigger concern is the longer term,” agrees Scambos. “By the end of this century the rate of change in Greenland will be so high that the next hundred years will be dialed in for significant sea level rise.” 

And any given city may have to contend with worse. While 70 percent of the world will see local waters rise within 20 percent of the norm, others will see extremes. In China, the Yellow River delta is currently sinking so fast that local sea levels are rising by up to 25 centimeters per year, nearly 100 times the global average. Places that were once covered by kilometers of ice, like northern Canada, are now rebounding upwards — which means local sea levels are actually falling in some parts of Alaska. But that upward-moving land is hinging nearby areas, like the U.S. East Coast, downward by millimeters per year — adding millimeters per year to the local sea level rise there. The U.S. East Coast has another problem too: Climate change is weakening the Gulf Stream current, and that is allowing water to slop back towards shore. Overall, the U.S. East Coast is seeing rates of sea level rise that are 3 to 4 times the global average. The tropics, meanwhile, are seeing extra sea level rise thanks to a strange gravitational effect. As high-latitude ice melts, there is less mass at the poles to pull ocean water towards them; instead, the water slopes more towards the equator. 

No matter which way you look at it, the result is cause for concern. “I always tell people if they live under 3 feet above sea level, they should be worried about the next 100 years,” says Chambers. “We probably can adapt to a certain extent. The problem is that we’re not planning for it.”