Between Ice and Fire

Benjamin Franklin was the first to connect volcanoes and climate change. During the summer of 1783, he noticed "a constant fog" over Europe and parts of North America.

snowy cliffs with long, wide crack in the snow

This fog was of a permanent nature; it was dry, and the rays of the sun seemed to have little effect towards dissipating it. . . . They were indeed rendered so faint in passing through it that, when collected in the focus of a burning-glass, they would scarce kindle brown paper. Of course, their summer effect in heating the earth was exceedingly diminished.

Hence the surface was early frozen.

Hence the first snows remained on it unmelted, and received continual additions.

Hence perhaps the winter of 1783-4 was more severe than any that had happened for many years.

The cause of this universal fog is not yet ascertained. . . . Whether it was the vast quantity of smoke, long continuing to issue during the summer from Hekla, in Iceland, . . . is yet uncertain.

Franklin's observations, reported the American Geophysical Union in 1992, were "overlooked by scientists until this century" and were, most probably, quite right.

Except for the name of the volcano.

The volcano that produced in 1783 "the greatest flood of lava on land in historical times" (according to a recent vulcanological handbook), opening a 27-km-long string of 10 craters that erupted continuously for eight months, issuing gases that killed 11,000 cattle, 28,000 horses, 190,000 sheep, and thereby causing the deaths by famine of 10,000 Icelanders—one-fifth of the population—was not Hekla, but Grimsvotn. In 1991, Anna Maria Agustsdottir, a graduate student in geochemistry at Penn State, and her adviser, Susan Brantley, joined the yearly Grimsvotn expedition of the Icelandic Glaciological Society.

The expedition was planned that year to sample the water chemistry of the lake that forms in the volcano's crater, under a 250-meter-thick cap of ice. Agustsdottir, who would do the analysis for her master's thesis, wanted first to dissect the structure of the volcano—how fire, water, and ice interact—and second to compare the lake's chemistry to that of the periodic floods, the jokulhlaups, that rush from it. If lake and 'hlaup matched, as was suspected, she could then calculate the volcano's yearly release of carbon dioxide, chlorine, and fluorine—the so-called greenhouse gases—and sulfur, the suspected source of Ben Franklin's chilling fog. Such data from the lake at Grimsvotn might clarify not only volcanoes' but humans' ability to effect climate change.

Vatnajokull, the "Water Glacier," is the blue-white thumbprint on the southeast corner of the map of Iceland, close by the coast.

At 8,400 square kilometers, it claims 8 percent of the country. It would blot out Philadelphia from Wilmington to Levittown, Valley Forge to Barnegat Light. Its ice is in some places 1,000 meters thick. It would be flat, but for the volcanoes beneath it, volcanoes that carve and suck its surface into gentle ice valleys and frozen whirlpools, or that thrust their bedrock up in rough black nunataks. The largest of the glacier's volcanoes, its signature depression a 35-square-kilometer bowl in the ice, is Grimsvotn.

Based on the crater lake's temperature (the first direct measurements were made on the 1991 expedition), which rates a barely thawed 0 to 4 degrees C, Agustsdottir calculated the volcano's power output at 4,000 to 5,000 megawatts, confirming earlier estimates.

"This coal plant here—" Brantley gestured out her office window to the University's power source; she is lanky and ebullient; her office is graced by an elephant foot and a buffalo skull, both saved from a museum's incinerator. "This coal plant is an 8 megawatt reactor." City-sized coal-fired power plants run about 1,000 megawatts. Volcan Poas, the Costa Rican volcano she had studied previously, varied from 200 to 800 megawatts. "So Grimsvotn is a very active volcano."

"You don't think about it," said Agustsdottir over cappuccino one afternoon in the local coffee shop. She is smaller and fairer than her adviser, with a steeliness to her glance and deliberation in her speech. A native Icelander, she had monitored the chemistry of several glacial rivers before turning her attention to Grimsvotn. "It's sort of scary when you're there, in that depression, and there's nowhere to go," she added softly, then glanced up. "But you don't think about that. Grimsvotn is a very special place. We're always interested in knowing what's going to happen there."

In past ages, men from the north of Iceland were said to cross the glacier, on skis or horseback, to fish off the south coast. In 1875, a Scotsman on skis spent 12 days in the crossing. A Dane crossed on horseback in 1912, as did two Swedes seven years later. According to the Icelandic book Ice and Fire,

The weather on Vatnajokull can be very changeable. There may be frost and snowstorms at any time of the year, as well as sleet and rain. But there are also days of warm sunshine . . . so that there is great temptation to strip for sunbathing. . . . When fog covers the glacier, there is, of course, no real sunshine, but full daylight can still remain even though no shadows can be seen. This is called whiteout. . . . There is no difference any more between the sky and the glacier, all contrasts vanish, and it is impossible to spot the horizon or any irregularities or holes in the snow. It is said that in such weather one can easily walk over the edge of a cornice.

In 1934 Grimsvotn erupted, as it has at least 50 times in the last 1,100 years. "The first expedition which was made to the scene of the eruption," an Icelandic scientist noted in 1936, ". . . chose the post route from Reykjavik east. . . . Motor cars were used most of the way there. Then the expedition continued on horseback. . . . The course was set by the cloud of vapour which was visible at night and in the mornings."

The post route from the city winds high over a rock-strewn pass, where paths marked by cairns lead now to blaze-orange rescue huts for motorists caught in sudden storms. A turnout on the western slope grants a grand view of the south: the steam rising from the Hveragerdi hot springs (tamed now to heat greenhouses); lush pastures and hayfields dotted with moss-grown lava blocks; hump-backed Hekla, once called "The Mouth of Hell" (last eruption 1991); the off-shore Westman Islands, blue in the distance (Surtsey joined their number 30 years ago, exploding out of the sea); the vast black sands scoured loose by glacial runoff; and, rising above it all, like inverse images of clouds, the glaciers themselves: Hofsjokull to the north, Myrdalsjokull to the south, and, between them, mighty Vatnajokull.

To reach it, before 1974 when at last a bridge was built, one had to cross a braided series of swift-flowing glacial rivers, their depths at the fords well over the wheels even of Land Rovers and heavy trucks.

And that was when the glacier was quiet. At 10-year, and lately five-year, intervals, whether or not there is an eruption under the glacier, it is wont to "burst," a phenomenon known to scientists by its Icelandic name jokulhlaup (literally, "glacier's run" or "leap"). The farmer at nearby Skaftafell kept a journal of the 'hlaup of '34:

March 23rd. At 15 o'clock the postman crossed the river and said that it was increasing and thick with glacier mud. At 18 o'clock the river began to flood beyond its bed. There was a great deal of snow on the sand, and the river took a long time to wet it in order to clear its way. By the evening the river had become unfordable on horseback.

March 24th. The river had considerably increased during the night. . . . It was estimated that there was about three times as much water in the river as usual. . . .

March 28th. In the morning it was estimated that the volume of the water had increased to 10 times the usual summer flow. . . . On this day the river began to break down the telegraph poles. . . .

March 31st. Had increased a great deal the preceding night. The rushing sound of the water and the rumbling noises were so great that the people at Skaftafell could scarcely sleep. . . . Icebergs were carried east to Fagursholmsmyri and out to sea. . . . At 17:30 o'clock they began to be left stranded aground and continued increasingly to be so stranded until the evening. By midnight large islands of land showed above the water.

April 1st. At about the time when people began to get up the volume of water was little more than normal. By about midday the river was considered fordable.

The bursts come about, scientists surmise, when the meltwater in Grimsvotn lake rises enough to lift its thick ice cap. The water spills over the lip of the lakebed, rushing 50 kilometers under the ice before gushing into the rivers. When the lake drops a hundred meters or so, the ice claps back like a lid on a pot.

Since 1954, Icelandic scientists have monitored the rate of flow and chemistry of the jokulhlaups.

Since 1953, the Icelandic Glaciological Society has arranged yearly expeditions onto the glacier, mapping the ice, sounding its depth, examining its chemistry, looking for signs of eruptions and pending bursts.

Once the train of snowmobiles, snocats, skidoos, and trailers has been winched and hauled over the jagged landward edges of the ice and onto its smooth surface, the hardest part of the expedition—if the weather cooperates—is over. The society keeps a hut on the bare cliff above the depression that marks Grimsvotn lake. The hut sleeps 28. "There's an excellent long ski-slope," according to Ice and Fire, "from the hut . . . down to Grimsvotn, but sometimes crevasses near the top have to be avoided."

"I had never worked on a scientific expedition before where you had so many people with so many talents," said Brantley of the 1991 expedition, which enjoyed an unbroken string of sunny days and bright nights. "If something broke, someone fixed it. All we had to do was think."

"There's a competition to get on this trip," Agustsdottir explained. "People spend their summer holiday there—architects, doctors, stewardesses, mechanics—helping out, cooking, fixing things. Nowhere but in Iceland," she added, "would people use their vacation time—and spend their own money—to help scientists carry out research in such a hazardous and difficult area." As a geochemist working for the Science Institute of the University of Iceland, Agustsdottir had also gone on the 1990 expedition, the first time the Institute had tried to pierce the ice and determine the chemistry of the crater lake. (The weather that year did not cooperate. Agustsdottir recalled one sudden blizzard in particular that wiped out the skidoo tracks, stranding her and her coworkers far from camp, with no clear notion which way to go.)

"It's difficult to get on the list," affirmed Brantley, whose geochemical specialty is water-rock interactions. At Poas volcano in Costa Rica from 1987-90, she and graduate student Gary Rowe had mapped the circulation of water through the volcano based on the chemistry of the crater lake and that of the various briney springs and seeps around the volcano's base. Even before Agustsdottir entered Penn State, Brantley had written Helgi Bjornsson, the scientists in charge of the Vatnajokull expeditions, requesting a chance to do a similar study of Grimsvotn. "He was very nice, but not encouraging," Brantley noted. "They try to keep Icelanders doing the work."

But the 1990 expedition was unable to sample the lake. "There were ash layers in the ice," explained Agustsdottir, "and we had trouble drilling. We didn't make it through the ice." The water chemistry studies were put on hold; the drill sent back to the University of Iceland's technicians for further refinements.

When Agustsdottir came to Penn State in the fall (having chosen the University's graduate program over others in the United States, Great Britain, and France partly because of Brantley's interest in Grimsvotn), her "open invitation" to join the 1991 expedition was extended to Brantley.

Then there was the interminable "fishing" for samples: dangling the "bailer" (a standard American water-sampling device that could leak, if shaken, but had a long enough rope to almost reach the lake bottom, 390 meters down) or the "sampler" (made by the Icelandic Science Institute to shut automatically, but which had a shorter reach) down the borehole. In 10 days on the glacier, Agustsdottir bottled 18 samples from various depths down two holes, one at the lake's center and one at the site of the latest eruption (in 1983). She also took samples at the cliff edge, where the ice had left exposed a sliver of lake, and in ice caves on the cliff, where gases wisped out in streams called fumaroles (it took hours to get a sample with enough chemical concentration to measure, "which was good," Agustsdottir noted, "otherwise gas would have been lost from the system and my estimates would be too low"). Other members of the expedition sampled the ice itself; and, after leaving the glacier, Agustsdottir tested the Skeidararsandur river to prove "that the chemicals that show the geothermal effect are not in the river water when there's no jokulhlaup.

"We were taking samples of every end of the system," she said.

Only two of the lake samples, Agustsdottir noted, smelled of sulfur when they came up, others of "some undefined odor." Several bubbled like champagne: carbon dioxide escaping before it could be measured. Overall, the lake water was fairly dilute. "I was expecting something more to be in there," Agustsdottir said. It is, indeed, a place of mythic richness: According to the creation story in Scandinavian mythology (as translated by Jean Young), the world itself was formed in just such a void between fire and ice: That part of Ginnungagap which turned northwards became full of the ice and the hoar frost's weight and heaviness, and within there was drizzling rain and gusts of wind. But the southern part of Ginnungagap became light by meeting the sparks and glowing embers which flew out of the world of Muspell. . . . Ginnungagap was as mild as windless air, and where the soft air of the heat met the frost . . . it thawed and dripped.

"But it makes sense when you think about it," Agustsdottir added, "there's so much glacier on top of the water diluting it." Dilute or not, the lakewater chemistry revealed first that the lake was stratified, chemically richer at the bottom than at the top; and second, since the lake's chemistry was very similar to the river's during a jokulhlaup, that the fire-water-ice system was closed: The volcanic gases stay trapped under the ice, gradually increasing in concentration until released by a jokulhlaup.

The analysis Agustsdottir made of the water between fire and ice is of a type called a "mass balance."

Based on the law of conservation of mass—that the total amount of carbon in the world, for instance, does not change, though a single atom may travel from a volcano to a river to the air to the leaf of a plant on the far side of the globe—any set of components in the world can be defined as a system of inputs and outputs, of sources and sinks. "Mass balance is a way of looking at complicated natural systems," explained Brantley. "It's something geochemists use all the time. It's a very simple tool."

The computer models of the world's climate, on which predictions of global warming (an expected average rise of 2 to 4.5 degrees C by 2035) are based, rely on the same technique, on weighing inputs and outputs, balancing sources and sinks. But a global climate model is by definition the largest mass balance in the world, and its numbers of sources and sinks are proportionally multiplied.

One small source is the world's volcanoes. "For the people who look at the carbon budget," said Brantley, "the least constrained flux is the natural flux from the subsurface, from volcanism and metamorphism. We know more about the anthropogenic flux of carbon dioxide than about the volcanic flux." The anthropogenic—or manmade—carbon flux is roughly 6.6 billion tons per year. "The best estimate I know of for carbon from volcanoes is 18 teragrams per year"—roughly 20 million tons, said Brantley, or three-tenths of a percent of what humans exhaust. To balance the natural carbon budget, however, and begin to look for the signal of global warming, climate modelers must credit volcanoes with more than 100 million tons.

"Those numbers are fairly discrepant," said Brantley. "There's just all sorts of things we don't know.

"For instance, do most volatiles come out during eruptions or through passive degassing? Anna was looking at passive degassing—how much gas gets out when the magma is still buried.

"We never see magma in the subsurface, and at the surface the gas comes out like it's blowing the cork off a champagne bottle. What we'd like to do is measure the carbon dioxide in the champagne without taking out the cork. Anna's numbers may be a way of thinking about that.

"People have been really interested in her work," Brantley added. "It's created a lot of discussion, a lot of argument. The data force you to think about the system in new ways."

When the Grimsvotn magma is still deep underground, according to Agustsdottir's thesis, most of its fluorine and sulfur stay bottled, while up to 88 percent of its chlorine and up to 100 percent of the carbon escape. Because this volcanic breath is condensed and stored in the ice-covered lake—between jokulhlaups—and since the water chemistry of the outbursts has been charted since 1954, Agustsdottir could also calculate the volcano's average yearly release of the four gases over a near 40-year span. The numbers she came up with—5.3 x 107 kg of carbon per year, 5.3 x 106 kg of sulfur, 6.0 x 105 kg of chlorine, and 1.5 x 105 kg of fluorine—are equal to or lower than those recorded by other scientists studying other volcanoes, even though Grimsvotn, per its heat output, is phenomenally active.

"This may imply that the published release rates for other volcanoes," Agustsdottir notes in her thesis, "are overestimates, since they are not usually integrated over time" but instead are extrapolations from "isolated data points." If so, then volcanoes are contributing even less carbon to the global system than climate modelers have been counting on to balance the carbon budget.

Not that Agustsdottir considers her numbers exact. She calculates her possibility of error at a full 25 percent (15-20 percent from the difficulties in measuring the volume of water in a jokulhlaup, plus a 3-5 percent error in the chemical analysis). But according to Brantley, "A 25 percent error is pretty close." Error terms of other methods used to calculate volcanic releases have ranged from 13 to 45 percent. "A lot of people who calculate these fluxes don't even try to put error terms on them," Brantley added, "because there are so many ways that errors can come in.

"We think of these numbers as order-of-magnitude estimates." At one conference, she recalled, "We had some of the best gas chemists standing around arguing. One of them thinks Anna was losing carbon dioxide from the lake, but he doesn't see how. Another one thinks Anna's estimates might be the best ever made."

She shrugged. "In geochemistry, we're looking at some of the most complicated systems in existence, and trying to figure things out with a dearth of data."

Last Updated December 01, 1993