6
The Cosmic Wake-Up Call
I should study Nature’s laws in all their crossings and unions;
I should follow magnetic streams to their source and follow the shores of our magnetic oceans. I should go among the rays of the aurora, and follow them to their beginnings, and study their dealings and communions with other powers and expressions of matter.
John Muir
“It was the biggest thing any of us had seen,” said scientist JoAnn Joselyn. “Before it was even visible you could see a depression on the limb of the Sun. It sent a chill down your back. It was just plain ugly.”
“It” was sunspot region 5395. On March 4, 1989, observers at the U.S. Air Force’s Ramey Solar Observatory in Puerto Rico detected a surge of activity just over the Sun’s eastern edge, or limb. X-class flares, the most intense the Sun can produce, were shooting off into space from a region that scientists could not yet see. But what they could see, as Joselyn noted, was a notch in the side of the Sun where a sunspot was compressing the surface. When the gnarled patch of black spots rotated into full view early on March 6,
observers got a view of the most complex sunspot region—both magnetically and structurally—that any of them had ever seen. More than 43,000 miles across, the sunspot group was 54 times the size of the Earth. It was pulsing and roiling at latitude 34 degrees, an unusual place for sunspots in the middle of solar maximum.
“There had never been a sunspot region like that in our lifetime,” said Joselyn, a space weather researcher and onetime forecaster at the National Oceanic and Atmospheric Administration’s (NOAA) Space Environment Center (SEC). As it came around to the front of the Sun and started its march across the face “we were just waiting for something big to happen,” Joselyn said. “We were dumbfounded, and we felt a little helpless. Something was going to happen that we had never seen before.”
From March 6 to 19, sunspot region 5395 exploded with at least 195 solar flares, 11 of them of the most intense “X-class” variety, and another 48 of the next highest “M-class” of severe flares (see Figure 11). The telescope on NASA’s Solar Maximum Mission (SMM) satellite detected 36 coronal mass ejections over the course of those two weeks.
The opening salvo came early on March 6, when region 5395 flashed with one of the three most powerful flares ever observed.1 The stream of radiation from the explosion lasted 10 hours—the norm is perhaps 30 minutes. Solar physicists estimated that the temperature inside the flare reached 20 million degrees Celsius, and more energy was released in those moments than humans have consumed in the entire history of civilization. The radio signals emitted by the flare were 2,000 times more intense than the normal background noise of the Sun. X rays from the explosion arrived at Earth in 8 minutes, 20 seconds (the time it takes light to travel from Sun to Earth), and protons from the Sun swept by a few minutes later like a swarm of angry hornets. The X-ray and proton detectors on the U.S. Geostationary Orbiting Environmental Satellite 7 (GOES-7) were flooded, as readings went off the normal scales for 27 minutes. It was labeled an X-15 flare.
Equipped with satellite data, ground-based telescope images of the Sun, and computer models, the scientists and forecasters at
FIGURE 11. This time series of images reveals the evolution of sunspot group 5395 as it moved around the Sun in March 1989. The upper sequence, taken in white light, shows the usual representation of the solar surface. The lower images are magnetograms, where black and white regions show opposite magnetic polarity. Note that changes in the white-light images coincide with changes in the magnetic structures. Courtesy of NOAA/AURA/NSF.
SEC watched the biggest space weather event of their lives begin to unfold in real time. All six phones in the SEC forecast center were ringing constantly. Prophetically, lead SEC forecaster Joe Hirman told The Washington Post in a March 6 interview: “There is definitely a chance of more flares because the chances are that this spot will keep doing something.”
The Sun continued to seethe and snarl with smaller flares for three days, and Earth’s magnetosphere was bathed in a shower of high-energy protons. That stream intensified on March 9, when sunspot region 5935 unleashed an explosive flare that reached the peak of brightness (brilliance) and size on the scales solar physicists had devised. The next day observers witnessed a rare white-light flare, much like the one Carrington had seen 130 years before. The SMM satellite’s coronagraph/polarimeter detected a large halo coronal mass ejection; the halo meant that the cloud was directed at Earth. The SEC predicted that there would be “high activity” on March 12 and 13.
Late on March 12, the biggest blast of energetic particles from the Sun started to reach Earth, as the rain of high-energy protons increased to 100 times the norm. By the middle of March 13, the magnetosphere, which normally stretches out 34,000 miles toward the Sun, was pushed down inside geosynchronous orbit (altitude 22,000 miles). Geosynchronous satellites that usually fly inside the protective cocoon of Earth’s magnetic field suddenly found themselves passing out into the solar wind, as the magnetosphere shrunk to half its normal size (some estimate as low as 14,000 miles).
Around the world, sensors in magnetic observatories went off the top of their scales for five or six hours. The auroral electrojet— an electric current system swirling in the ionosphere about 60 miles above Earth—fluctuated wildly as it slid down the globe from its normal northern latitudes and flowed right over the heartland of the United States. U.S. Navy observers in Bay Saint Louis, Mississippi, watched their magnetometer get pinned to the top of its scale as auroras danced in the skies.
Vivid red auroral rays glowed over all over North America, stretching as far south as Arizona, southern California, Mississippi, and Texas. Police stations were flooded with phone calls about “funny red clouds” in the night skies. A group of backpackers in the hills of North Carolina recalled seeing the red glow in the sky and thinking that an extremely large fire or perhaps even a nuclear war had started somewhere over the horizon. Across the Atlantic, reports of aurora came in from brightly lit cities in Portugal, Spain, The Netherlands, Britain, and Hungary. In the southern hemisphere, sky watchers saw the aurora australis over New Zealand, Australia, and South Africa. The light show stretched to the tropics, where mystified observers watched the aurora in Cancún, Mexico, Grand Cayman Island, Honduras, and Dominica. One observer in Cuba said: “It was like night turning into day.”
The aurora borealis was big news in Florida. A newspaper reporter from Miami described how “an eerie red-orange glow danced across South Florida’s skies in what may have been a performance of the northern lights . . . [an occurrence] rarer than snow.” Tom Printy of the Central Florida Astronomical Society
told one reporter that “many of those who saw an aurora for the first time were baffled, thinking that the sky activity might be related to the space shuttle that was launched that morning.”
That 13-day surge of solar activity caused one of the most powerful magnetic storms ever recorded.2 In the two weeks that sunspot region 5935 marched across the face of the Sun, the forecast group at SEC issued 37 rapid alerts—urgent warnings of threats to technological systems on Earth—and 415 normal alerts. The words of caution helped, but not every industry has a practical solution for getting out of the path of a space weather storm.
In the northeastern United States, a manufacturer of computer microchips shut down operations because the magnetic storm was disturbing sensitive instruments. All around the world, navigators and surveyors noted that their compasses were distorted by as much as 10 degrees. Several North Sea oil companies stopped drilling because magnetic instruments that guide the drills were way off course, and most magnetic surveys and oil-prospecting missions came to a halt for hours to days.
Up in the thick of the storm, satellites felt the full force of the Sun. Heated and electrified by the solar protons and the compression of the magnetosphere, the density of the upper atmosphere of Earth increased by five to nine times the norm, increasing the friction—atmospheric drag—on satellites in low-Earth orbit. The U.S. Air Force Space Command at the Cheyenne Mountain Operations Center typically tracks 8,000 objects in orbit around the Earth— everything from bus-sized satellites to space junk such as floating nuts and bolts. By March 14, 1989, the Air Force had lost track of 1,300 of those objects, many of which had dropped by more than 1 kilometer in their orbits. NASA’s SMM satellite was said to have “hit a brick wall,” dropping nearly a 1/2 mile in one day and a total of 3 miles from its normal orbit due to the increased drag. A classified U.S. military satellite in low-altitude orbit began an uncontrolled tumble through space.
Aside from the drag, some satellites took a more direct hit from the solar radiation. Bombarded by radiation, NOAA’s GOES-7 weather satellite suffered communications circuit problems and outages for much of March 13, and Japan’s CS-3B communications satellite lost half of its computer brain. Many satellites endured phantom switching and tripping of circuits in the electronic controls, and seven commercial satellites in geostationary orbit had trouble staying in place. Operators of the seven satellites fired thrusters at least 177 times in two days to keep their spacecraft on track, more adjustments than are typically made in an entire year. Other satellites had communications problems when the excited ionosphere interfered with radio signals sent to and from the ground. In a few instances, signals from the Global Positioning System were severely degraded or gave outright erroneous positioning information.
The most spectacular effect of the March 1989 storm—and the effect that has many scientists and engineers concerned as we live through the current solar maximum—was the burnout and blackout of several electric power systems around the world. Stray electric currents from the storm—known as geomagnetically induced currents (GICs)—disturbed at least 200 different components of electric power systems in Maryland, California, New York, New Mexico, Arizona, and Pennsylvania.
In Sweden, five power transmission lines were tripped by large voltage fluctuations, and fire alarms were set off. At another Swedish power company, Sydkraft, operators detected a 5 degree C increase in the temperature of the rotors of a nuclear plant. At Public Service Electric & Gas’s nuclear power plant in Salem, New Jersey, a $10 million transformer was damaged beyond repair (see Figure 12). The extra current coursing through the system heated thick metal coils—bathed in cooling mineral oil—until they burned up. Typically, 500-kilovolt transformers like the one in Salem can take a year to replace, but PSE&G was fortunate enough to find an
FIGURE 12. Charred wires reveal the damage wrought by the March 1989 magnetic storm. This section was part of a large transformer at a nuclear power plant in New Jersey. Such hardware can cost tens of millions of dollars to replace, and it can take as long as a year to have a new one manufactured. Courtesy of John Kappenman.
unused spare from another canceled nuclear plant. The new transformer and plant were back in service within six weeks, but not before losing $400,000 per day in power sales.
Five hundred miles to the north, the Hydro-Quebec power company was not so fortunate. The Canadian utility suffered through a chain reaction that collapsed an entire power grid in 90 seconds. There was hardly time for the operators to assess what was happening, no less fix it. The press release from Hydro-Quebec’s managers asserted that, “the March 13 blackout was caused by the strongest magnetic storm ever recorded since the 735 kilovolt power system was commissioned.” At 2:45 a.m. the storm tripped five power lines from James Bay and caused a loss of 9,450 megawatts of power. “With a load of some 21,350 megawatts at that moment, the system was unable to withstand this sudden loss and collapsed within seconds.”
More than 6 million people in Quebec City, Montreal, and surrounding areas lost power in the middle of a frigid winter night. The morning rush hour was snarled as traffic lights went out and the subway system shut down. It took nine hours to restore power to most of the region—through purchases of power from other utilities. But about 17 percent of customers went without power for much of the rest of the day, some for several days.
The cascade of problems began in the James Bay region shortly after the onset of the magnetic storm. In a total of 59 seconds, seven voltage-regulating devices at the Chibougamou, Albanel, Nemiscau, and La Verendrye substations tripped and stopped the flow of electricity. Voltages in the lines fluctuated by as much as 15 percent. With the sudden loss of voltage, all five major transmission lines to Montreal tripped, some of them exploding into flames. The loss of 9,450 MW of power overloaded the rest of the system. Two transformers blew out in Chibougamou, and the Churchill Falls and Manicouagan-Outardes power plants automatically shut themselves down as demand for power overwhelmed the system. Transmission lines from Sherbrooke failed, cutting off the export of energy to New England.
The Hydro-Quebec blackout resulted in a loss of some 19,400 MW of power in Quebec and 1,325 MW of exports to New England and other parts of Canada. The restoration of power took nine hours because much of the essential equipment, particularly on the James Bay transmission network, was made unavailable by the blackout. Operators had to use power from isolated stations that normally export their electricity or buy it from other power companies in Ontario and New Brunswick.
By the time the blackout was over and the fried equipment was repaired or replaced, Hydro-Quebec had lost at least $10 million. The cost to its customers was estimated to be in the hundreds of millions of dollars. In the months following the event, the Northeast Power Coordinating Council and Mid-Atlantic Area Council power pools—electric power cooperatives that serve the northeastern United States from New England to Washington, D.C.— acknowledged that they nearly suffered a cascading system collapse
due to the stress of the storm and the loss of Hydro-Quebec’s power exports. The whole northeastern United States almost tumbled into a blackout.
“The 1989 event is the most significant space weather event for the power industry,” noted David Boteler, a researcher and electrical engineer who studies the effect of geomagnetic storms on power systems for the Canadian Geological Survey. “It was the event that changed people’s opinion from ‘space weather effects are just an academic curiosity’ to ‘this is a real problem that needs to be looked into.’ Before 1989, believing in space weather effects on power systems was regarded by some as equivalent to believing in little green men from outer space.”3
In high school physics, you learn that the way to generate electric current is to vary the magnetic field around a conducting wire. Move a bar magnet back and forth near the wire, as scientist Michael Faraday once proved, and the ammeter will show electricity flowing. That’s essentially what happens when a power system is hit by a magnetic storm. From the perspective of space, electric power lines, railroad tracks, oil pipelines, and communications cables look a lot like long, thin, conducting wires. When space weather stirs up the great electric current in the sky—the auroral electrojet—the magnetic field at the surface of the Earth starts to vary. The result is a surge of extra electric current into power systems and every other sort of cable.
“Magnetic disturbances are causing induced currents to flow through the conducting networks that mankind has stretched across the Earth’s surface during the last 150 years,” David Boteler says. As electric potentials on the ground reach 1 to 10 volts per kilometer, the currents flow along Earth’s surface. In the case of electric power systems in North America, direct current (DC) is pumped into lines that typically carry alternating current (AC). The DC saturates one-half of the AC cycle, causing power lines to be overloaded and transformers to try to compensate. This leads
to a chain reaction. Regulating devices in some transformers may cause that part of the system to shut itself down. The rest of the system has trouble synchronizing and regulating the flow of energy, causing other devices and transformers to shut down in a spiraling effect. According to electrical engineer John Kappenman: “You just don’t know what protective systems are going to shut off when you need them most.”
The problem is exacerbated in Canada and the United States. “North America is the most profoundly affected land mass in the world because of igneous rock geologies which cover large regions,” says Kappenman (who proudly notes that the power systems in Minnesota, which he helped oversee in 1989, did not fail on his watch). Igneous, or volcanic, rock tends to resist the flow of electric current more than other types of rock, forcing currents to flow closer to or at the Earth’s surface. When a magnetic storm begins, GICs enter and exit power systems through ground wires attached to transformers, which would normally dump some of that current into the ground. Compared to igneous rock, however, a power line is the path of least resistance. Since the crust of Earth beneath North America is rich in igneous rock, and since the continent has so much infrastructure located at high magnetic latitudes (the magnetic north pole tilts toward Greenland and Canada), North American power systems are much more susceptible to GICs.
Statistical research confirms the geophysics. Kappenman and other researchers at Minnesota Power and Electric analyzed the failure rate of transformers in the United States from 1968 to 1991. They found that certain types of transformers failed much more often in GIC-susceptible regions; in fact, failure rates in the northeastern United States were 60 percent higher than in the rest of the country. The researchers also found that transformers had shorter life spans in those GIC-prone areas. Perhaps most compelling was the finding that transformer failures followed a periodic cycle that, as Kappenman notes, “virtually mimics” the solar cycle.
In 1990 researchers at the Oak Ridge National Laboratory studied a hypothetical space weather event just slightly more severe than the March 1989 storm. In their imaginary event, engineer Paul Barnes and economist James Van Dyke wiped out most of the tightly linked electric power grid from the Middle Atlantic up to New England in order to assess the economic impact. Their hypothetical magnetic storm began by sweeping across Canada and the United States at a time of day “when power import levels to New England and New York are near their limits and the capacity margins are at low levels.” First, a Canadian utility cut off its exports to stabilize its own network. Then several voltage regulators were tripped and a few transformers were damaged. The extra load on the system and the loss of voltage in some of the transmission lines caused the “tie lines” from the Midwest and South to become overloaded and unable to supply the extra power needed for the Northeast. The cascade of problems flooded and wiped out large portions of the power grid. In the simulation it took 16 hours to restore half of the power and 48 hours to get the whole system back online.
Barnes and Van Dyke tallied the damage—the cost of purchasing power from outside utilities to cover the lost power generation, the cost of replacing fried equipment, and the revenues lost from the loss of power sales. They did not even begin to assess the economic losses to the local economies that would be halted by the blackout. Their total cost: $3 billion to $6 billion. The assessment by Barnes and Van Dyke, coupled with reviews conducted by the North American Electric Reliability Council, rated the hypothetical 1989 storm as an electric power disaster comparable to the damage caused by Hurricane Hugo.
That scenario, according to Kappenman, is not as improbable as it seems. Since 1965, electric power consumption in the United
States—particularly in the Northeast and California—has steadily increased, while very few power plants have been added to the infrastructure because of environmental protests and economic concerns. The result is that power grids are running so close to the operating margin that electricity is regularly being imported from Canada and the Midwest over “transmission lines operating near their limits.” With less spare capacity on the lines, the risk of saturating the system during a magnetic storm rises substantially. In addition, most power companies across the continent have computerized their systems and linked them together to share power. Such power sharing helps utilities ensure that an isolated event does not bring down an entire power grid. But that also means there is a wider network that can get pulled down when faced with a widespread problem such as a global geomagnetic storm.
“The industry is more focused on making money than on protecting power,” David Boteler notes, “more focused on accounting than engineering. The power-sharing strategy breaks down when there is a large event. By being interconnected, the companies are effectively making even longer transmission lines through which current can flow. Instead of 100 extra volts across 100 miles, you get 1,000 volts across 1,000 miles.”
It would take billions of dollars “to plug the GIC sieve that the network has become,” John Kappenman notes. The cost of rebuilding the existing power systems or retrofitting with devices that could block GICs is too prohibitive when weighed against the perceived risk. So Kappenman and several colleagues have been promoting specialized forecasting as the solution. With 30 to 45 minutes of advance warning, he notes, power companies could reconfigure the flow of power or shore up certain parts of the system to withstand the onset of a magnetic storm in the same way that they brace for a hurricane.
Monitoring conditions from the ground does not help, because by the time the ground-based observatories can detect magnetic fluctuations, the storm has already arrived. But by monitoring conditions in space on the sunward side of Earth, and by modeling what the varying solar wind will do to the magnetic field, power
grid operators can steal 30 minutes of preparation time, Kappenman says. While he has been developing a computer model and software to provide such a forecast of ground-induced currents, Kappenman also has been lobbying industry and the government for years to invest in a reliable solar wind monitor to provide the data the power industry needs. In 1998, Kappenman’s wish was fulfilled. With the launch of NASA’s Advanced Composition Explorer (ACE) research satellite, Kappenman got the real-time data he needs to run his GIC modeling system. “Powercast,” as he calls it, predicts when space weather might become dangerous to a power grid. That system—the world’s first space weather prediction system for national electrical power grids—was put into operation in England and Wales in January 2000. During testing, according to Kappenman, the system was 95 percent accurate in alerting users to potentially hazardous current-inducing geomagnetic storms. Time will tell if it works in the real, sunny world.
But his efforts and those of several colleagues are being met with mixed interest. “Very few people have a real understanding of what GICs will do to a transformer,” Kappenman says. “There are subtleties that no one fully appreciates.” Furthermore, the life cycle of an electrical engineer working in system operations is just a few years, so there are not many people monitoring the power lines who remember 1989.
“An event like 1989 is not going to happen very often, and there is a danger that the risks get overstated,” Boteler notes. “There are a lot of other things that can go wrong in a power system. And shutting down exports or turning on reserve power stations to prepare for a predicted magnetic storm can cost a lot of money. The question is whether you design power systems for the ‘one-hundred-year event,’ as you do for earthquakes and floods,” Boteler adds. “It is amazing how often these ‘one-hundred-year events’ come around. Do you buy insurance, or do you hope it is not going to happen?”
For Kappenman and other colleagues now selling their own brand of electric power insurance, another 1989-style blackout could be very good for business.
