SKAFTAFELL, Iceland -- Just north of here, on the far side of the impenetrable Vatnajokull ice sheet, lava is spewing from a crack in the earth on the flanks of Bardarbunga, one of Iceland's largest volcanoes.
By volcanologists' standards, it is a peaceful eruption, the lava merely spreading across the landscape as gases bubble out of it. For now, those gases -- especially sulfur dioxide, which can cause respiratory and other problems -- are the main concern, prompting health advisories in the capital, Reykjavik, 150 miles to the west, and elsewhere around the country.
But sometime soon, the top of Bardarbunga, which lies under as much as half a mile of ice, may erupt explosively. That could send plumes of gritty ash into the sky that could shut down air travel across Europe because of the damage the ash can do to jet engines. And it could unleash a torrent of glacial meltwater that could wipe out the only road connecting southern Iceland to the capital.
All of that could happen. Then again, it may not.
Such are the mysteries of volcanoes that more than four months after Bardarbunga began erupting, scientists here are still debating what will happen next. The truth is, no one really knows.
Volcanic eruptions are among the Earth's most cataclysmic events, and understanding how and when they happen can be crucial to saving lives and reducing damage to infrastructure and other property. Scientists have several powerful tools to help, but in the end, they are often reduced to analyzing possibilities within possibilities, chains of potential events that could unfold in multiple ways.
"Volcanoes are really difficult to predict because they are so nonlinear," said Pall Einarsson, a geophysicist at the University of Iceland. "They can suddenly decide to do something very different."
Einarsson studies the earthquakes that usually accompany volcanic activity, caused by hot rock, or magma, rising within the earth and creating stresses and fractures. Seismic monitoring is essential for helping to determine if and when an eruption will occur and how it will proceed, but scientists also study the deformation of a volcano's surface -- a sign of increasing pressure within -- using GPS units and satellite-based radar, and they also monitor gases and other indicators like the melting of snow or ice.
"Ideally it's a nice combination of data from many disciplines," said Stephanie Prejean, a research geophysicist with the U.S. Geological Survey at the Alaska Volcano Observatory. "And it's easiest if the things are all escalating together, and escalating dramatically."
Over the past decade, Prejean said, the observatory has successfully forecast eruptions about two-thirds of the time for the more than two dozen volcanoes that are seismically monitored (of 130 total). In Iceland, home to 35 active volcanoes, scientists have had about the same success rate, Einarsson said.
The improvements in monitoring and studying volcanoes, as well as the fact that Earth's growing population has put more people in harm's way, may contribute to the sense that more volcanoes are erupting now than in the past. But scientists see no real trend.
Eruptions that come out of the blue can be particularly deadly. In September, 57 hikers were killed in central Japan when Mount Ontake suddenly started spewing hot ash, cinders and rocks. Volcanologists think rising magma hit groundwater, which turned instantly to steam and caused the explosion. There were no significant earthquakes or other signs that might have suggested an eruption was imminent and prompted the authorities to close the popular mountain trails.
Scientists can sometimes have a general idea that a volcano is due to erupt, but a sudden event dictates the precise timing. At Mount St. Helens in Washington in 1980, scientists knew for months that an eruption was likely -- for one thing, the north side of the mountain had started to expand like a balloon as rising magma increased the pressure inside.
Then on May 18, an earthquake caused the north face to collapse in a massive landslide. The weight of all that rock had helped to keep the magma contained; once it was gone, the volcano erupted immediately, killing 57 people, some more than 10 miles away.
Mount St. Helens was the most destructive eruption in U.S. history -- it also wiped out nearly 200 miles of roads -- but others could be worse. Mount Rainier, for example, is just 50 miles from Seattle, and an eruption there would probably produce devastating floods of volcanic ash, water and rock. (The huge Yellowstone volcanic system in Wyoming has received a lot of attention for its potential to cause a climate-altering disaster, but scientists say the probability of such an event is exceedingly small for a given year.)
In Iceland, scientists knew in mid-August that something was happening at Bardarbunga, which had last erupted in 1910. Seismometers began recording a swarm of small earthquakes, eventually numbering in the thousands, on the north side of the volcano. This was a clear sign that magma was beginning to intrude into a fissure perhaps 5 or 6 miles below the surface.
Although this was happening in a part of the volcano covered by the glacier, scientists could tell that the magma was moving horizontally and mostly to the northeast along the fissure, because the centers of the earthquakes were moving, too. Until Aug. 29, the magma was underground, but on that date reached the surface on Bardarbunga's northern flank. The magma -- which is called lava when it is above ground -- spewed out in red-hot fountains.
The eruption, which is off limits to nearly everyone except researchers, has continued since then. As of the end of the year, it had involved close to 2 billion cubic yards of lava -- enough to fill about a thousand large football stadiums -- that had spread out across 30 square miles.
According to a paper published in mid-December in the journal Nature, the spreading underground magma -- creating what volcanologists call a dike -- extended more than 27 miles before erupting.
Freysteinn Sigmundsson, a University of Iceland geophysicist who coordinated the study, said the seismic information, as well as extensive deformation data, showed that the dike grew in fits and starts through the fissure, which although deep was less than two yards wide. The magma would hit a barrier -- essentially a narrowing of the fissure -- which would cause the pressure to build up until it was great enough that the magma would overcome the barrier and keep moving.
"Think of it as a subsurface stream that comes to a dam," Sigmundsson said. "Eventually, it simply breaks the dam."
For now, the eruption remains what volcanologists call an effusive one -- the lava, consisting primarily of molten basalt, is thin enough that the gases bubble out with little explosive force. And the amounts of sulfur dioxide and other gases, while a concern locally, are nowhere near the amounts produced by an eruption at a fissure called Laki in the 1780s. In that event, the gases poisoned livestock across Iceland, leading to a famine that killed about a quarter of the country's population and had other effects in Europe and elsewhere.
One possibility is that the current eruption will eventually peter out as the source of magma is depleted.
"Maybe the most likely scenario is something similar to what we've been seeing," Sigmundsson said. But that could take a while; although the volume of lava has declined, it has done so only very gradually, he said, suggesting the eruption could continue for many months.
But there are many other possibilities. Bardarbunga sits at the heart of a complex system of volcanoes and "has a history of affecting its neighbors," Einarsson said. Were the dike to continue moving to the northeast, he said, it could set off an eruption at the nearby Askja volcano, although that seems less likely.
Of greater concern is what is happening at Bardarbunga's caldera, the wide, deep valley at the top of the mountain that is filled with hardened magma from past eruptive activity. Earthquake data and GPS measurements show that this hardened magma, which acts like a plug, is sinking, probably as the hot magma below it escapes through the fissure to the north. The subsidence is astonishingly rapid, about a foot a day, and the question is how much more of this the plug can take before it breaks up.
"As of now, the system seems to be relatively stable," Einarsson said. "But it's almost certain that this can't last very long, and that's what people are worried about. Because this plug is bound to disintegrate as it moves so much."
If the plug cracks apart, the hot magma below would have a new, easier path to the surface -- straight up -- where it would combine with ice to cause a steam-magma explosion. Such an eruption could create a large plume of ash that could disrupt air travel, as the eruption at another Icelandic volcano did in 2010. Its effects on the surrounding region could be catastrophic as well, with glacial meltwater collecting in the caldera until it overflows, causing a vast flood.
That has happened countless times in Iceland's geological history, and it is what created the eerie skeidararsandur, the vast delta west of Skaftafell that resembles the surface of the moon, as floodwaters brought huge quantities of black volcanic sand down from the mountains.
The skeidararsandur could take the brunt of a flood again, although it would depend on precisely where the eruption occurred. A short distance this way or that, and the floodwaters might flow to the north, or even to the west -- an especially troubling possibility given that several hydroelectric dams responsible for much of Iceland's electricity could be damaged or destroyed.
"One can never be absolutely certain about predicting," Einarsson said. "So we have to line up all the possible scenarios and stretch our imaginations to figure out what could possibly happen."
"If these things were nicely behaved," he added, "you'd just pour out this lava up there and the volcano would run out of pressure and that would be it."