Lurking deep below the surface in California and Wyoming are two hibernating volcanoes of almost unimaginable fury. Were they to go critical, they would blanket the western U.S. with many centimeters of ash in a matter of hours. Between them, they have done so at least four times in the past two million years. Similar supervolcanoes smolder underneath Indonesia and New Zealand.

A supervolcano eruption packs the devastating force of a small asteroid colliding with the earth and occurs 10 times more often--making such an explosion one of the most dramatic natural catastrophes humanity should expect to undergo. Beyond causing immediate destruction from scalding ash flows, active supervolcanoes spew gases that severely disrupt global climate for years afterward.

Needless to say, researchers are eager to understand what causes these giants to erupt, how to predict when they might wreak havoc again, and exactly what challenges their aftermath might entail. Recent analysis of the microscopic crystals in ash deposits from old eruptions has pointed to some answers. These insights, along with improved technologies for monitoring potential disaster sites, are making scientists more confident that it will be possible to spot warning signs well before the next big one blows. Ongoing work hints, however, that supervolcano emissions could trigger alarming chemical reactions in the atmosphere, making the months following such an event more hazardous than previously suspected.

Almost all volcano experts agree that those of us living on the earth today are exceedingly unlikely to experience an active supervolcano. Catastrophic eruptions tend to occur only once every few hundred thousand years. Yet the sheer size and global effects of such episodes have commanded scientific attention since the 1950s.

Early Awe
One of geologists' first discoveries was the existence of enormous circular valleys--some 30 to 60 kilometers across and several kilometers deep--that looked remarkably similar to the bowl-shaped calderas located atop many of the planet's most well-known volcanoes. Calderas typically form when the chamber of molten rock, or magma, lying under a volcanic vent empties out, causing the ground above it to collapse. Noting that these calderalike valleys sit close to some of the earth's largest deposits of volcanic rocks laid down during a single event, those early investigators realized they were seeing the remnants of volcanoes hundreds or even thousands of times larger than the familiar Mount St. Helens in Washington State. From the extreme scale of the calderas and the estimated volume of erupted material, researchers knew that the magma chambers below them had to be similarly monstrous.

Because the thick continental crust and heat sources needed to create such massive magma chambers are rare, supervolcanoes themselves are also uncommon. In the past two million years, a minimum of 750 cubic kilometers of magma has exploded all at once in only four regions: Yellowstone National Park in Wyoming, Long Valley in California, Toba in Sumatra and Taupo in New Zealand. The search for similarly large eruptions continues in other areas of thick continental crust, including in western South America and far eastern Russia.

By the mid-1970s, investigations of past events revealed some ways that the chambers can form and become dangerous. Under the surface of Yellowstone, the North American tectonic plate is moving over a buoyant plume of warm, viscous rock rising through the mantle, the 2,900-kilometer-thick layer of the earth's interior that is sandwiched between the molten core and the relatively thin veneer of outer crust. Functioning like a colossal Bunsen burner, this so-called hot spot has melted enough overlying crust to fuel catastrophic eruptions for the past 16 million years. In Toba, the source of the chamber is different. That region lies above a subduction zone, an area where one tectonic plate is slipping under another; the convergence produces widespread heating, mainly through partial melting of the mantle above the sinking plate.

No matter the heat source, pressure in the magma chambers builds over time as more magma collects under the enormous weight of overlying rock. A supereruption occurs after the pressurized magma raises overlying crust enough to create vertical fractures that extend to the planet's surface. Magma surges upward along these new cracks one by one, eventually forming a ring of erupting vents. When the vents merge with one another, the massive cylinder of land inside the ring has nothing to support it. This "roof" of solid rock plunges down--either as a single piston or as piecemeal blocks--into the remaining magma below, like the roof of a house falling down when the walls give way. This collapse forces additional lava and gas out violently around the edges of the ring.

Fingerprinting Eruptions
Yet mysteries remained. Notably, as researchers soon realized, not every large magma chamber will necessarily erupt catastrophically. Yellowstone, for example, is home to three of the world's youngest supervolcano calderas--they formed 2.1 million, 1.3 million and 640,000 years ago, one nearly on top of the other--but in the gaps between these explosive events, the underlying chamber released similar volumes of magma slowly and quietly. Why magma sometimes oozes slowly to the surface is still uncertain.

A look at the composition of tiny crystals trapped inside erupted lava and ash at Yellowstone has suggested a partial answer, by providing new insight into how magma forms. For decades, geologists assumed that magma sits as a pool of liquefied rock for millions of years at a time and that each time some of it pours out onto the earth's surface, new liquid rises up from below to refill the chamber immediately. If that conception were correct, one would expect many more catastrophic, voluminous eruptions, because it is mechanically and thermally infeasible to keep monster magma bodies in the crust without emptying them frequently.

The old idea was based largely on so-called whole-rock analyses in which researchers would obtain a single set of chemical measurements for each fist-size piece of volcanic rock they collected. Those data provided important general patterns of magma evolution, but they were insufficient for determining the age of the ejected magma and the depth at which it formed.

Every chunk of rock is actually made up of thousands of tiny crystals, each with its own unique age, composition and history. So when technological advances made it possible in the late 1980s to analyze individual crystals with good precision, it was like being able to read individual chapters in a book rather than relying on the jacket blurb to explain the story. Investigators began to see that some crystals--and thus the magmas in which they originally formed--arose much earlier than others, for instance, and that some formed deep underground, whereas others formed near the earth's surface.

During the past 10 years, geochemists have been paying particular attention to an especially durable type of volcanic crystal called zircon. Knowing that zircons can withstand extreme changes in heat and pressure without compromising their original composition, a few researchers--among them John W. Valley of the University of Wisconsin-Madison--have been using them to study the early evolution of the earth's crust [see "A Cool Early Earth?" by John W. Valley; Scientific American, October 2005]. When I joined Valley's team as a postdoctoral fellow in 1998, we used Yellowstone zircons to trace the history of their parent magma--which in turn revealed important clues about how the volcano may behave in the future.

The first step was to measure the ratios of different forms of oxygen in zircons from the youngest Yellowstone supereruption--which after exploding 640,000 years ago gave rise to the Lava Creek tuff, a fossilized ash deposit 400 meters thick in some places--as well as younger deposits that were expelled during milder eruptions since then. When I finished my initial analyses, Valley and I were both surprised to see that oxygen composition of those zircons did not match that of deep, hot mantle, as would be expected if drained chambers always filled from below. Zircons born of mantle-derived magmas would have had a distinctive signature: as elements that are dissolved in magmas come together to form a zircon, that crystal takes on a notably high proportion of oxygen 18, a heavy isotope of oxygen that has 10 neutrons in its nucleus instead of the usual eight.

Valley and I saw immediately that the magma must have originated in rock once near the earth's surface. The zircons we studied were depleted in oxygen 18 relative to the mantle, and such depletion occurs only if the crystals formed from rocks that interacted with rain or snow. We thus suspected that the collapsed roof rock from one of the two oldest Yellowstone supereruptions must have melted to form the bulk of the magma that was ejected during the younger Lava Creek catastrophe and smaller eruptions since. This hypothesis gained strength when we learned that the ages of the zircons from post-Lava Creek eruptions span the entire two-million-year duration of Yellowstone volcanism. Such old zircons could exist in the youngest ash only if they originated in material that was ejected during the oldest eruptions and if that material later collapsed back into the magma chamber and remelted to help fuel the youngest eruptions.

Our findings mean that scientists can now expect to make certain predictions about how the Yellowstone supervolcano, and possibly those elsewhere, will behave in the future. If a new round of small, precursor eruptions begins in Yellowstone--and they usually do so weeks to hundreds of years before a catastrophic explosion--testing the oxygen fingerprint of those lavas and the ages of their zircons should reveal what type of magma is abundant in the chamber below. If the next eruption is depleted in oxygen 18, then it is most likely still being fed by stagnant remnants of the original magma, which by now is probably more of a thick crystal mush than an explosive liquid. On the other hand, if the new lava carries the fingerprint of fresh magma from the mantle and does not contain old zircons, then it very likely came from a large volume of new magma that has filled the chamber from below. Such findings would imply that a new cycle of volcanism had commenced--and that the newly engorged magma chamber had more potential to explode catastrophically.

Immediate Aftermath
Tiny crystals and their isotopic signatures have also revealed surprises--good and bad--about the aftermath of supereruptions. One of the best-studied examples of supervolcano aftermath is the Bishop tuff, a volcanic layer tens to hundreds of meters thick that is exposed at the earth's surface as the Volcanic Tablelands in eastern California. This massive deposit represents what is left of the estimated 750 cubic kilometers of magma ejected during the formation of the Long Valley supervolcano caldera some 760,000 years ago.

For decades, many geologists assumed that a series of distinct eruptions over millions of years must have occurred to produce the extensive Bishop tuff. But careful studies of microscopic, magma-filled bubbles trapped inside tiny crystals of quartz tell a different story. The rate at which magma leaves a chamber depends primarily on two factors: the magma's viscosity, or ability to flow, and the pressure difference between the chamber and the earth's surface. Because the pressure inside a bubble matches that of the chamber where the magma formed, the bubble acts like a mini version of the chamber itself.

Aware of this correspondence, Alfred Anderson of the University of Chicago and his colleagues studied the size of the bubbles under a microscope to estimate how long it took the magma to leak out. Based on these and other experiments and field observations from the 1990s, geologists now think that the Bishop tuff--and probably most other supererupted debris--was expelled in a single event lasting a mere 10 to 100 hours.

Since that discovery, investigators have had to modify their reconstructions of supervolcano eruptions. Here is what they now generally expect from an event the scale of those that struck Long Valley and Yellowstone: Instead of a slow leak of red-hot lava as is seen creeping down the sides of Kilauea Volcano in Hawaii, these eruptions feature supersonic blasts of superheated, foamlike gas and ash that rise buoyantly all the way into the earth's stratosphere, 50 kilometers high.As the land above the magma chamber collapses, immense gray clouds called pyroclastic flows burst out horizontally all around the caldera. These flows are an intermediate stage between lava and ash, so they move extremely rapidly--up to 400 kilometers an hour, some sources say; cars and even small airplanes would have no chance of outrunning them. These flows are also intensely hot--600 to 700 degrees Celsius--so they burn and bury everything for tens of kilometers in every direction.

As bad as the pyroclastic flows are, the ash injected into the atmosphere can have even more far-reaching consequences. For hundreds of kilometers around the eruption and for perhaps days or weeks, pale-gray ash would fall like clumps of snow. Within 200 kilometers of the caldera, most sunlight would be blocked out, so the sky at noon would look like that at dusk. Homes, people and animals would be buried, sometimes crushed. Even 300 kilometers away, the ash could be half a meter thick; mixed with rain, the weight would be plenty sufficient to collapse roofs. Less ash than that would knock out electrical power and relay stations. As little as a millimeter, which could well dust the ground halfway around the globe, would shut down airports and dramatically reduce agricultural production.

Only gradually would rain (made acidic by volcanic gases) wash away the thick blanket of ash. And because volcanic rock and ash float, it would clog major waterways. Transportation along big rivers could grind to a halt. Indeed, recent oil drilling in the Gulf of Mexico struck a surprisingly thick layer of supervolcanic debris near the Mississippi Delta--more than 1,000 miles from its source in Yellowstone. Only by floating downriver and then sticking to sediment that sank to the ocean bottom could that amount of debris have accumulated from a volcano so far away.

The Long Haul
Investigators have reason to believe that other consequences, arising from the great volumes of problematic gas expelled into the upper atmosphere, would also transpire and could persist for many years. New work suggests that some of these outcomes may not be as bad as once feared but that others may be worse. Once again, looking at the composition of small by-products from past eruptions has been illuminating.

Of the varied gases that make up any volcanic eruption, sulfur dioxide (SO2) causes the strongest effect on the environment; it reacts with oxygen and water to produce tiny droplets of sulfuric acid (H2SO4). These droplets are the main sun-blocking source of the dramatic climatic cooling that would grip the planet in the wake of a supereruption. Knowing that the planet's hydrological cycle takes months or years to fully wash away the acid droplets, many researchers made apocalyptic estimates of "volcanic winters" lasting decades, if not centuries. But in recent years other investigators have uncovered evidence that drastically reduces that calculation.

Almost always, traces of the sulfuric acid produced after large volcanic eruptions are trapped in snow and ice as the acid precipitates out of the contaminated atmosphere. In 1996 investigators studying ice cores from Greenland and Antarctica found the sulfuric acid peak that followed the supereruption of Toba 74,000 years ago. That eruption ejected 2,800 cubic kilometers of lava and ash and reduced average global temperatures by five to 15 degrees C. The consequences of such a chill were undoubtedly severe but did not last as long as once thought: sulfuric acid in the ice rec-ord disappeared after only six years; some researchers suggest that it vanished even earlier.

That volcanic winters are probably shorter than expected is the good news. But a new method developed in the past five years for studying the composition of the oxygen atoms in the volcanic acid rain is revealing an entirely different, alarming sign about the long-term effects of sulfur dioxide in the atmosphere. For SO2 to become H2SO4, it must be oxidized--in other words, it must acquire two oxygen atoms from other compounds already existing in the atmosphere. Exactly which compounds play the key role is a hotly debated topic of current research, so when I started working with John M. Eiler as a staff scientist at the California Institute of Technology in 2003, he and I looked for evidence in my samples of ashes from the prehistoric Yellowstone and Long Valley eruptions.

We began our analyses with a focus on a particularly efficient oxidant, ozone. Ozone is a gas molecule made up of three oxygen atoms best known for shielding the earth from the sun's dangerous ultraviolet rays. Because of rare chemical transformations that certain gases undergo in the presence of that intense solar radiation, ozone is characterized by an anomaly in its so-called mass-independent oxygen isotope signature, which in simple terms can be thought of as an excess of oxygen 17.

When ozone or any other oxygen-rich molecule in the stratosphere interacts with SO2, it transfers its oxygen isotope signature to the resulting acid--that is, the oxygen 17 anomaly persists in the new acid. In 2003 geochemists at the University of California, San Diego, found the first evidence that this signature is also preserved in the oxygen atoms of the acid that later falls as rain and in the sulfate compounds that form as the acid rain reacts with ash on the ground.

The oxygen 17 excess and other chemical patterns that we found in sulfate from the Yellowstone and Long Valley ash samples thus implied that significant amounts of stratospheric ozone were used up in reactions with gas from the supereruptions in those regions. Other researchers studying the acid layers in Antarctica have demonstrated that those events, too, probably eroded stratospheric ozone. It begins to look as if supervolcano emissions eat holes in the ozone layer for an even longer period than they take to cool the climate.

This loss of protective ozone would be expected to result in an increased amount of dangerous ultraviolet radiation reaching the earth's surface and thus in a rise in genetic damage caused by rays. The magnitude and length of the potential ozone destruction are still being debated. Space observations have revealed a 3 to 8 percent depletion of the ozone layer following the 1991 eruption of Mount Pinatubo in the Philippines. But what would happen after an event 100 times larger? Simple arithmetic does not solve the problem, because the details of atmospheric oxidation reactions are extremely complex and not fully understood.

Scientific techniques for studying and monitoring volcanoes of all sizes are developing with all deliberate speed. But no matter how much we learn, we cannot prevent an eruption. And what can be said about the aftermath of the most catastrophic occurrences is still speculative at best. The good news, though, is that researchers now know enough about the sites of possible eruptions to predict with reasonable assurance that no such catastrophe will happen anytime soon.