Soon after the space shuttle atlantis launched a new observatory into orbit in 1991, Gerald Fishman of the NASA Marshall Space Flight Center realized that something very strange was going on. The Compton Gamma Ray Observatory (CGRO), designed to detect gamma rays from distant astrophysical objects such as neutron stars and supernova remnants, had also begun recording bright, millisecond-long bursts of gamma rays coming not from outer space but from Earth below.
Astrophysicists already knew that exotic phenomena such as solar flares, black holes and exploding stars accelerate electrons and other particles to ultrahigh energies and that these supercharged particles can emit gamma rays—the most energetic photons in nature. In astrophysical events, however, particles accelerate while moving almost freely in what is essentially a vacuum. How, then, could particles in Earth's atmosphere—which is certainly nowhere close to being a vacuum—be doing the same thing?
Early data initially led us and other experts to believe that these so-called terrestrial gamma-ray flashes originated 40 miles above the clouds, but we have now determined that they are produced much farther down by electric discharges inside garden-variety thunderclouds. Meanwhile increasingly sophisticated theories devised to account for the freakish gamma rays have struggled to keep up with observations: time and again, experiments have detected energies that were previously thought impossible in the atmosphere. Even antimatter has made a surprise appearance.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Twenty-one years later researchers have a good idea of what might create these terrestrial gamma-ray flashes, although uncertainties remain. Adding to the urgency of this fascinating puzzle are its possible implications for human health: if an aircraft travels too close to the sources, the gamma rays could pose a radiation hazard for people riding inside.
Two birds with one stone?
At first, scientists wondered if the gamma rays could be related to another type of atmospheric marvel discovered only a few years earlier. Cameras trained above thunderclouds had photographed bright, brief flashes of red light, 50 miles above the ground and miles wide, that looked like giant jellyfish. These impressive electric discharges were whimsically named “sprites.” Because sprites almost reach the edge of space, it seemed plausible that they might shoot out gamma rays that an orbiting probe could see.
Soon theoretical physicists made the first attempts to explain how sprites could produce space-bound gamma rays. Sprites are thought to be side effects of ordinary lightning occurring in clouds far below. Lightning is an electrically conducting channel that temporarily opens through the air, which is otherwise an electric insulator. The bolt carries electrons between regions of the atmosphere or between the atmosphere and the ground. It is caused by an imbalance of electrostatic charge and is triggered by the resulting electric fields, whose potential differences may exceed 100 million volts.
The violent rush of electrons partially restores the electrostatic balance. Yet just as tamping down a bump in a rug often causes another bump to spring up elsewhere, a discharge inside a cloud often causes the field to spring up elsewhere, including on the ground—where it may later lead to upward lightning—or near the bottom of the ionosphere—where a sprite may result.
In 1992 Alexander V. Gurevich of the Lebedev Physical Institute in Moscow and his collaborators calculated that such secondary electric fields near the ionosphere might produce avalanches of energetic electrons, which, bumping into atoms, would unleash high-energy photons—x-rays and the even more energetic gamma rays—in addition to the sprites' characteristic red glow. The mechanism they proposed derived from a suggestion made by Nobel Prize–winning Scottish scientist C.T.R. Wilson back in the 1920s. At low energies, electrons being pushed by an electric field act like drunken sailors, bouncing from molecule to molecule and losing their energy with each collision. At high energies, however, the electrons travel in a straighter line, picking up even more energy from the electric field, which makes any collisions even less effective at disturbing their path, and so on—a self-reinforcing process. This sequence differs from our everyday experience, in which the faster we go, the more drag force we suffer, as any bicyclist can attest.
These “runaway” electrons could conceivably accelerate up to nearly the speed of light and travel for miles before they stop instead of the few feet an electron might usually move in air. Gurevich's team reasoned that when a runaway electron finally did bump into a gas molecule in the air, it could kick another electron free, and that electron could then itself run away. The result would be akin to a chain reaction: an avalanche of high-energy electrons that grew exponentially with distance and could go as far as the electric field extended. The avalanche effect, Gurevich and his collaborators calculated, could increase the production of x-rays and gamma rays by many orders of magnitude. For a while, this picture seemed very compelling because it unified two separate atmospheric phenomena: gamma-ray flashes and sprites. As we will see, reality turned out to be more complicated.
The innocence of sprites
Over the next several years, from 1996 onward, increasingly refined versions of the theory were developed that modeled sprites as a manifestation of runaway-electron avalanches that produced gamma rays. One piece of evidence that supported this sprite model was the energy spectrum of gamma rays. Higher-energy gamma rays go farther through air than lower-energy ones do, so they are more likely to make it to space. By counting how many gamma-ray photons arrive at a spacecraft at each energy level, scientists can infer the altitude of the source that produced them. The first examinations of the gamma-ray energies seen by CGRO pointed to a very high source altitude, consistent with sprites.
Then, in 2003, things took an unexpected turn. While working at a lightning-research facility in Florida and measuring the x-ray emissions reaching the ground from rocket-triggered lightning, one of us (Dwyer) and his collaborators detected a very bright burst of gamma rays that emanated from the thundercloud overhead and washed over the terrain around us [see “A Bolt out of the Blue,” by Joseph R. Dwyer; Scientific American, May 2005]. On our instruments, this burst looked exactly like one of the terrestrial gamma-ray flashes that everyone thought originated much higher: the rays had the same energies and the same duration of about 0.3 millisecond. At the time, everyone assumed that the flashes came from much too high up to be seen on the ground. The similarity implied that perhaps lightning bolts inside thunderclouds might be direct sources of the gamma rays reaching CGRO, but at the same time, the idea seemed kind of crazy: the flash would have had to be unbelievably bright to get enough gamma rays out into space through all that atmosphere.
Soon, however, other developments would undo the purported link between sprites and gamma rays. In 2002 NASA had launched the Reuven Ramaty High Energy Solar Spectroscopic Imager, or RHESSI, to study x-rays and gamma rays from the sun. But RHESSI's large germanium detectors were perfect for measuring gamma rays coming from the atmosphere as well, although they would have to do so through the back of the spacecraft, while the observatory faced our star. One of us (Smith), an astrophysicist and solar physicist, was on the RHESSI instrument team and recruited Liliana Lopez, then an undergraduate student at the University of California, Berkeley, to comb through RHESSI's continuous, years-long stream of data to look for evidence of gamma rays from below. At the time, terrestrial gamma-ray flashes were thought to be very rare. Instead Lopez found a treasure trove: RHESSI was detecting a flash once every few days, about 10 times the rate of CGRO.
RHESSI measured the energies of the gamma-ray photons in each burst much better than CGRO ever did. Their spectrum looked just like what would be expected from runaway electrons. Yet by comparing it with simulations, we deduced that the gamma rays had gone through a lot of air, so they had to originate at altitudes between roughly nine and 13 miles—typical of the tops of thunderstorms but far below the nearly 50-mile height where sprites live.
Further independent evidence quickly accumulated favoring a lower-altitude origin of gamma rays rather than a connection with sprites. Radio measurements made by Steven Cummer of Duke University of some of the lightning associated with the RHESSI events found that these lightning flashes were much too weak to make sprites. Moreover, the RHESSI map of gamma-ray flashes around the world looked very much like the map of normal lightning, which is concentrated in the tropics, and very little like the map of sprites, which sometimes cluster at higher latitudes in such spots as the Great Plains of the U.S.
One remaining argument in favor of sprites as the origin, though, was that the energy spectrum from the CGRO events seemed to point toward a high-source altitude, more consistent with sprites than thunderstorms. Many of us started to believe that there might be two kinds of gamma-ray flashes, low- and high-altitude ones. But the final blow to the sprite idea came when we realized that terrestrial gamma-ray flashes were much brighter than previously thought. In fact, working with then graduate student Brian Grefenstette in 2008, we determined that they were so bright that CGRO was being partially blinded by them and could not measure their full intensity. (This saturation also affected RHESSI, though to a lesser extent.) When researchers at the University of Bergen in Norway reanalyzed the data in 2010, they found that taking instrument saturation into account made the results consistent with lower-altitude sources.
In less than two years, then, the putative altitude where gamma-ray flashes form plummeted more than 30 miles. It is rare in science to witness a paradigm shift happen so rapidly. This change is ironic, given that when we became involved in this field of research a decade ago, sprites were the one shining example of how energetic radiation can be produced in our atmosphere. Now, 10 years later, just about everything—thunderclouds, various kinds of lightning, laboratory sparks—seems to make detectable high-energy radiation but apparently not sprites. The consensus now is that the low energy of sprites' radiation implies that they are not responsible for gamma-ray flashes after all.
Bring on the antimatter
So if it is not sprites that produce gamma-ray flashes, what does? And does the process still involve runaway-electron avalanches? As it turns out, the avalanche mechanism as modeled by Gurevich and company, though too energetic to have anything to do with sprites, is not powerful enough to generate the large luminosities seen by RHESSI or the newly analyzed CGRO data. Calculations by Dwyer, however, had shown that a supercharged version of the electron avalanche mechanism could release trillions of times more energy than previously envisioned and could do so inside a thundercloud. Astoundingly, such a mechanism would also involve the production of copious antimatter.
If the electric field inside a thundercloud were strong enough, runaway electrons—assuming they form somehow—should accelerate to nearly the speed of light and, when they bumped into atomic nuclei in air molecules, could emit gamma rays. In turn, the gamma-ray photons could interact with atomic nuclei to produce pairs of particles: electrons and their antimatter twins, positrons. The positrons would run away as well, gaining energy from the electric field. But while electrons move upward in the field, the positrons, which have opposite electric charge, would move downward. When the positrons reached the bottom of the electric field region, they would bump into air atoms and knock out new electrons that would again run away toward the top.
In this way, the upward-going electrons would create downward-going positrons, which in turn would create more upward-going electrons, and so on. As one avalanche led to others, the discharges would quickly spread over a broad area of the thundercloud, up to several miles wide. The numbers predicted by this model—known as the relativistic feedback discharge model—perfectly matched the intensity, duration and energy spectrum of the gamma rays seen by CGRO and RHESSI.
The positive feedback from positrons is analogous to the annoying screech we get by holding a microphone up to a speaker. Of course, if we want a loud noise, we could just as well shout into the microphone. That logic is behind another possible explanation, albeit one that has not yet been fully worked out mathematically: that gamma-ray flashes are more energetic versions of the bursts of x-rays emitted by lightning as it approaches the ground. For several years researchers at the Florida Institute of Technology, the University of Florida, and the New Mexico Institute of Mining and Technology have been measuring these x-rays, both from lightning that is artificially triggered with rockets and from natural lightning that strikes the ground. X-ray “movies” from a fast x-ray camera in Florida show that the bursts emanate from the tip of the lightning channel as it travels from the cloud to the ground. Most scientists think that the x-rays are generated by runaway electrons, accelerated by strong electric fields in front of the lightning. Perhaps, for reasons we have yet to figure out, lightning that moves through the electric field inside a thundercloud does a better job of making these runaway electrons. If this idea is correct, then the flashes seen by spacecraft from hundreds of miles away could be just a version—amplified through some still unknown mechanism—of the lightning-generated x-rays seen on the ground by detectors a few hundred feet from the bolt.
Out of the blue
By the end of 2005 we were confident that most terrestrial gamma-ray flashes stemmed from inside or near the tops of thunderclouds, regardless of whether antimatter or souped-up lightning bolts were involved. Before we could get too cozy with that new paradigm, however, something seemed to put our understanding into question again: one of the events picked up by RHESSI was smack in the middle of the Sahara Desert—on a sunny day with no thunderclouds in sight.
We and our students spent months struggling over this one. It turns out that thunderclouds did form that day—just not where the spacecraft was looking. The storms were several thousand miles to the south, over the horizon from RHESSI. Their gamma rays, which, like all forms of light, travel in a straight line, could not have reached the craft.
Charged particles such as electrons, on the other hand, naturally travel in trajectories that tightly spiral around the curved lines of Earth's magnetic field. The storms were precisely at the other end of the magnetic field line going through the spacecraft. Electrons that reached very high altitudes could have circumnavigated the planet and smashed into RHESSI's detectors, forming gamma rays in the process. It seemed impossible, though, that electrons unleashed inside a thundercloud could make it through many, many miles of atmosphere to an altitude in space where they could hitch a ride around the field lines. The new observation seemed once again to require a high-altitude source.
Last year, moreover, the Fermi Gamma-ray Space Telescope observed more of these circumnavigating beams and made a startling discovery: that a sizable fraction of the beams consist of positrons. Thus, it appears that atmospheric phenomena can blast not only electrons and gamma rays into space but also antimatter particles. In hindsight, we should have expected to see these positrons, given how energetic the gamma rays are. Yet considering how unusual it is to observe antimatter in nature, Fermi's finding was astonishing.
The explanation for the Sahara finding, our team soon realized, was not that the gamma rays were coming from a high altitude but rather that they were produced inside thunderclouds in more copious numbers than had been thought possible. Some of those headed for space, ran into the occasional air molecule above 25 miles of altitude or so and created secondary electron-positron pairs, which then hitched a ride on the magnetic field lines around the globe. Next time you see a tall thundercloud, stop to remember that it is capable of shooting high-energy particles into space that can be detected on the other side of the planet.
New outliers
The appearance of positrons was not to be our last shock. Later in 2011 the Italian Space Agency's AGILE observatory found that the energy spectrum of terrestrial gamma-ray flashes extends up to 100 mega-electron-volts, a value that would be amazing even if it came from a solar flare. If correct, these observations cast doubt on our models because it seems highly unlikely that the runaway mechanism could generate such energies by itself. In fact, it is not clear what could possibly accelerate electrons to such energies inside thunderstorms. At this point, we need more observations to help guide the theory. Fortunately, teams from the U.S., Europe and Russia are now beginning to launch the first space missions dedicated to detecting terrestrial gamma rays.
Meanwhile, to get closer to the action, we and our collaborators have built an aircraft instrument designed to measure gamma rays from thunderstorms. Worry about the dangers of gamma-ray exposure prevents us from flying straight into a storm. But on an early test flight in which Dwyer took part, the plane inadvertently took the wrong turn. The feeling of terror was quickly supplanted by elation as our detectors suddenly lit up. Subsequent analysis showed that the region was accelerating runaway electrons of the same kind that we expect to make gamma-ray flashes. Fortunately, the emission stayed at a low level and did not undergo the explosive growth of the events seen from space. From these flights, we have found that thunderstorms most often emit a relatively harmless, continuous glow of gamma rays.
Preliminary calculations, however, show that if an airline flight happened to be struck directly by the energetic electrons and gamma rays inside a storm, passengers and crew members could—without feeling anything—receive up to a lifetime's natural radiation dose in a fraction of a second. A bit of good news is that we do not need to warn pilots to stay away from thunderstorms, because they already do so; thunderstorms are very dangerous places to be, with or without gamma rays.
In a way, the study of terrestrial gamma-ray flashes is completing the work of Benjamin Franklin, who purportedly sent a kite into a thunderstorm to see if it would conduct electricity and thereby showed that lightning was an electric discharge. Surprisingly, two and a half centuries after his kite experiment, scientists still have an incomplete understanding not only of how thunderclouds make gamma-ray flashes but even of how they make simple lightning. Both of us have spent much of our careers studying exotic objects far from the solar system, but we have been pulled back to Earth by the lure of this research. Perhaps even Franklin did not realize that thunderstorms could be so interesting.