On July 4, 1999, a rare and terrifying storm swept through the Boundary Waters Canoe Area Wilderness in northern Minnesota. What began like a standard-issue thunderstorm soon turned strange and fierce, generating green clouds and strong winds reminiscent of a tornado. In fact, the storm was a cousin of the tornado: a derecho (pronounced "de-RAY-cho"), a type of storm so infrequent and fast-moving that only in recent years have meteorologists begun to understand how to recognize and forecast it. This feature tells the story of this unusual event and the efforts of scientists to understand it.
Read these related articles.
A derecho is a damaging, straight-line windstorm that arises from a line of thunderstorms. Over the years such events have been called "wind rushes," "blowdowns" (when they strike forests), and "plough winds" by the people who have experienced them. These monikers are apt: the derecho causes considerable destruction along a long path (400 kilometers long or more, by definition), and it races forward with train-like speed and force. The straight-line windsthe "derecho" part of the stormappear at the front of the system, plowing everything aside like a cowcatcher on an old-fashioned locomotive. Once a derecho gets moving, it can keep moving--sometimes for as long as 24 hours--before it eventually runs out of steam.
Thunderstorms thrive on warm, moist air. As a result, many derechos occur during the hot, humid days of summer. "They seem to be most common towards the last of June and the beginning of July," says Robert Johns, a former meteorologist with the National Weather Service's Severe Storm Prediction Center, and a leading expert on derechos. "That's when moisture levels peak in the lower atmosphere. In fact, many derechos, including the one in 1999, occurred on July Fourth." On the day of the Boundary Waters storm, Johns recalls, temperatures had reached into the 90s, and the dewpoint temperature, which is a more sophisticated measure of humidity, was near 80°F"which is very high. It was a nasty, sticky day. And this was true all the way across U.S. to New York City, where it got up to 96 degrees."
A derecho grows in roughly four stages, Johns explains. First comes the rainstorm. On the hot, sticky day in question, warm air near the ground begins to rise, carrying moisture with it. As it rises, the warm air cools and the moisture within starts to condense into clouds. The process of condensation releases heat, which warms the surrounding air and causes the air to rise even farther and faster. Eventually the rising moisture condenses into droplets too heavy for the cloud to carry, and rain begins to fall.
(In general, given two equally sized packets of air, one moist and one dry, the moist air will rise farther than the dry air. Even though the two air packets start out at the same temperature, the moist air gains additional heat from its condensing moisture, so it contains more inherent energy than the dry air. Scientists sometimes describe moist air as being more "buoyant" than dry air, since it has a tendency to rise higher. The buoyancy of moist air plays an important role in the late stages of a derecho's formation.)
The second critical phase is the formation of a "downburst." As rain falls from the storm, it pulls some upper-level air with it toward the ground, creating a downward movement of wind known as a downdraft. If conditions are rightif enough dry air is available higher up, between about 3 to 6 kilometers above groundthe downdraft can become a violent downward burst, or downburst, of wind. As the downdraft falls, it draws dry air behind it into the storm. The dry air evaporates some of the raindrops in the downdraft, which cools the downdraft. This causes the downdraft to sink quickly, transforming it into an intense downburst of wind that hits the ground and spreads out like fast-moving pancake batter. Though it lasts only a few minutes, a downburst can be 50 kilometers wide and produce winds of up to 200 kilometers per hour. A small-scale downburst is called a "microburst." These are isolated events, and are typically no more than a kilometer wide. Especially around airports, however, they pose a severe hazard: the winds from a microburst are capable of pushing an airplane from the sky during takeoff or landing. Microbursts caused several commercial airline crashes in the 1970s and early 1980s. The strongest microburst ever recorded, with winds approaching 200 kilometers per hour, occurred at Andrews Air Force Base on August 1, 1983, just six minutes after President Reagan had landed there safely. That close call was instrumental in the federal government's decision to install advanced Doppler radar equipment at major airports around the country.
Under the right conditions, solitary downbursts can combine to form a large, fast-moving "macroburst." This is the third critical phase in the formation of a derecho. Two ingredients are necessary: a moderate to strong wind pushing the storm system from behind, and a rich supply of moist air in front of the storm. If the high-elevation winds are sufficiently strong, they provide the storm with a running jump-start. Meanwhile, the winds from the downburst hit the ground and begin to spread out, most strongly in the forward direction (because the storm is being pushed from behind). If there is moist air ahead of the storm, it is pushed aloft by the winds at the storm's front edge. The moisture condenses and new rain-clouds form. This is now the front edge of the storm: the clouds produce new downbursts, which scoop yet more moisture into the storm, and the cycle continues. The downbursts have evolved into a macroburst, an atmospheric locomotive powered by what scientists call a "convective engine." Gulping down moist air, the convective storm system advances with gathering speed and force.
At a certain point, the storm systemwhich by now has made itself apparent to forecasters by the bow shape of its radar echohas moved far enough and reached sufficient strength for meteorologists to call it a derecho. "To be classified as a derecho, a bow-echo must produce wind damage for at least four hundred kilometers," Johns says. The number is somewhat arbitrary, he adds, "but this qualifies it as being a very large-scale wind event." The winds in a derecho can reach 200 kilometers per hour; the storm itself can advance as quickly as 150 kilometers per hour, many times the typical speed of an advancing tornado.
Johns has found that it takes at least three hours for a large-scale "convective system" to evolve to the point where it produces a full-blown derecho. The derecho that hit the Boundary Waters Canoe Area Wilderness at midday on July 4, 1999, actually formed at about six o'clock that morning as a group of severe thunderstorms near Fargo, North Dakota. "As they were crossing Minnesota, they became a series of bow-echoes that accelerated up to eighty or a hundred kilometers per hour. By the time they reached the Boundary Waters region they had become a full-fledged bow-echo complex and were producing a lot of devastation. It continued moving very rapidly across Ontario and Quebec during the rest of the afternoon and night-time hours, reaching Maine by the early morning hours of the next day, creating devastation along the whole route."
Once it gets moving, a derecho follows a path dictated in part by the moisture content of the air ahead of it. Moist air is more buoyant than dry air, so it is more easily pushed aloft by the advancing storm and fed into the convective engine. A derecho may travel for hours, sometimes as long as a day, providing there is sufficient moist air to fuel it. When the supply of moist air runs out, or the pushing winds weaken, the derecho dies out. The July 4, 1999 derecho was particularly long-lived, lasting for about 24 hours before it moved out to the Atlantic.
In tracking the development of storm systems, a meteorologist relies on two essential observational tools: satellite imagery and weather radar. Satellite images reveal the formation of storm fronts over broad areas: the whole state, several states, entire nations. From above, the satellite can detect rising columns of warm air, which often herald – again on a broad scale – the development of a severe storm. Satellites are particularly useful for tracking tropical storms and hurricanes, which cover thousands of square miles and typically form far out over the ocean, beyond the range of other observing methods.
For tracking smaller, fast-moving storms over the mainland, however, radar is the tool of choice. In use since the 1950s, radar enables the forecaster to monitor storms in more detail and to track their motions on a minute-by-minute basis. At its simplest, weather radar works by detecting the precipitation in incoming storms. (It can also pick up other small objects, from dust motes to migrating butterflies and birds.) The radar unit emits a radio wave at very high frequency, then measures how long the signal takes to return after bouncing off of raindrops, hail, snow or whatever else the coming storm has to offer. Since the radio wave travels at a fixed rate (the speed of light, more than 350,000 kilometers per second), the length of the echo delay reveals how far away the storm is. Most radar units send out about 1,000 radio pulses per second and aim them at several altitudes simultaneously; the result is a rich, real-time view of the storm.
Though a marvelous advance at the time, the first-generation weather radar pales in comparison to the technology in use today. In the late 1980s, the federal government began building an advanced, nationwide radar network called the National Weather Service Next Generation Weather Radars, or NEXRAD. The new radars make extensive use of computers; among other advantages, this means they can be programmed to sound an alarm when weather patterns are beginning to appear dangerous, and they can be operated by automation. (In the old days, a forecaster had to watch the screen continuously; taking readings of high-altitude activity involved pointing the radar upward with a hand crank.) More important, NEXRAD radars gather critical data using so-called Doppler technology. Imagine a distant car: as it advances, its sound registers at a slightly higher frequency than it would if it remained stationary. Likewise, precipitation that is blowing toward a radar antenna shifts toward a slightly higher frequency, while precipitation that is blowing away from the antenna shifts to a slightly lower frequency. From these Doppler shifts, the radar's computer calculates the direction and speeds of the winds surrounding the droplets. The result is a detailed map that reveals the location, intensity and direction both of wind bursts and precipitation across dozens of square miles.
Doppler radar has drastically improved the ability of forecasters to track the development of severe storms like tornadoes and derechos. When a tornado takes shape, its winds blow raindrops in a circular pattern; on a radar screen, that pattern is as distinctive as the whorls of a fingerprint. A derecho also leaves a unique radar fingerprint. When a broad, fast-moving line of thunderstorms begins to take on the characteristics of a derecho, a wide band of high winds, all blowing directly forward, develops at the front edge of the storm. This band of wind produces a spray of raindrops that, when it appears as a radar echo, typically forms the shape of a plow or a bow; this distinctive pattern is called a "bow echo." With these patterns in mind, and with a deep knowledge of the kinds of storm systems and radar patterns that often precede the formation of tornadoes and derechos, today's meteorologists can provide the public with remarkably accurate storm warnings. Of the two storm types, tornadoes are easier to monitor: though they pack strong winds, tornadoes themselves move relatively slowly (30 to 50 kilometers per hour), so their paths can be predicted with reasonable confidence. Derechos are trickier: they are both faster and more amorphous than tornadoes, and since only a dozen or so occur annually in the U.S., their vicissitudes are less familiar to atmospheric scientists and forecasters.
Radar does have its limitations. Due to the curvature of the Earth, an individual radar station can't "see" effectively beyond about 200 kilometers. For that reason, to track storms accurately over long distances, meteorologists rely on a network of widely spaced radar stations. In addition, although an individual radar unit can measure precipitation and winds that are moving toward or away from it, it can't detect winds that are moving directly perpendicular to it. With tornadoes, that's not a problem: their winds move circularly, so there is always some wind moving toward the radar and other wind moving away from it. Once it develops, a tornado generates a distinctive echo pattern that remains visible to the radar regardless of which way the tornado itself moves.
Derechos, however, can present a unique challenge to the meteorologist. Their winds blow largely in one direction: forward. If the derecho is moving toward a radar unit, even at an angle, its distinctive wind-pattern – a bow-shaped front – will become visible, enabling the forecaster to follow the storm as it moves and develops. But there are times, right when the derecho passes the radar station (often at a distance of several miles), when the storm and its accompanying winds are moving exactly perpendicular to the radar. At this point the winds produce no radar signature at all, and the storm front dwindles on the radar screen. This can be a distressing moment for the forecaster, who suddenly (and for several minutes) suffers an acute lack of information.
That's what happened on July 4, 1999, shortly before a derecho struck the Boundary Waters Canoe Wilderness Area in northern Minnesota. Forecasters in nearby Duluth, Minnesota, had been monitoring the advance of a large, severe storm from North Dakota. Just as the storm passed to their north, barreling toward the Boundary Waters region, the wind signature diminished on the radar screen. Meteorologists could still view the bow-echo reflection of raindrops well above ground, so they still had a rough idea of where the gust front was located at the surface. But they could no longer determine actual wind speeds at ground level. As it turned out, this was precisely the moment when the storm evolved into a full-blown derecho. Typically, a forecaster could compensate for the loss of wind data by checking data from another radar station. As fate would have it, however, Duluth was the last radar station along the storm's path to northern Minnesota--and the last radar to see the storm before it turned into a derecho.