The Calculators of Calm

Just how far out of their way will airlines go to give you a smooth ride?

The DC-8 lost its left outboard engine and 19 feet of wing and fell 500 feet in 10 seconds, but landed safely. NCAR

It’s a typical January day in Chicago—overcast skies, snow on the ground, a high in the 20s—but Terry O’Toole is far more interested in some menacing thunderstorms boiling above the eastern Pacific. O’Toole stares at weather data displayed on a desktop computer screen in United Airlines’ flight operations center, near O’Hare International Airport. A dispatcher, O’Toole devises and assigns flight plans, which include routes and altitudes. Like United’s passengers and crews—and, perhaps most of all, the company’s accountants—O’Toole wants United flights to avoid turbulence as much as possible.

Thunderheads can, of course, be easily seen by both pilots and radar, but not the violent winds that sometimes swirl for miles downwind of a storm. In slightly less than two hours, United flight 52 will depart Honolulu for Los Angeles. O’Toole clicks up a satellite image of the Pacific on his other monitor. A box superimposed on the image warns him of the area in which airliners can expect to encounter turbulence between 28,000 and 38,000 feet. The box lies directly across every possible route O’Toole can assign UAL 52.

Though the storm may dissipate before the flight enters the area, residual turbulence can linger for a long time, causing choppiness. The best O’Toole can do is find an altitude either above or below the “chop.” It’s the start of a complicated numbers game.

“Rarely do we fly below 28,000 feet,” says O’Toole. Ideally, an airliner will climb to a higher altitude, where the air might be smoother. But to minimize fuel consumption, airliners typically fly as high as weight limits allow, so they’re already at their ceiling. To escape turbulence, most airliners have to descend; that causes them to burn more fuel, which eats into company profits.

O’Toole clicks on icons of United flights already in the alert area and keys a message to a UAL aircraft flying at 37,000 feet: “How’s the ride so far?” He hits the return key, and the message is transmitted via satellite to a computer screen in the airborne jet’s cockpit.

As O’Toole waits for a reply, he inputs weather and fuel parameters into a computer that will plot the most economical flight plan for UAL 52, a Boeing 777 that will carry an almost-full complement of passengers and cargo. Minutes later a dot-matrix printer cranks out the results: the recommended altitude and route, and a wealth of weather information and figures relevant to other routes and altitudes UAL 52 could be assigned.

“At 33,000 feet, we’ll be at five hours and 32 minutes—which is considerably higher than our target time. And it’ll require 60,000 pounds of fuel,” says O’Toole, looking over the computer’s calculations. By contrast, “if I run him at 24,000 feet, we’ll be on target time because the tailwinds are so strong there. But you’re also looking at burning 8,000 pounds more fuel”—costing about $3,000. “It’s nice to make it in less time, but instead of that 8,000 extra pounds of fuel, you could be carrying 8,000 more pounds of cargo or passengers…. But if the air is bad, we go down lower,” he assures me. “We don’t give it a second thought.”

A response comes in from the United flight crew already in the area: “Light chop on and off most of the way; seat belt sign on.” According to the printout, if UAL 52 flies at 37,000 feet, it will actually burn slightly more fuel than it would at 33,000 feet, because the tailwinds aren’t as strong higher up. But the air seems to be smoother up there. O’Toole decides to take the hit on fuel and schedules UAL 52 for a flight at 37,000 feet. Chances are it’s less than the hit—or hits—the airline could take if the flight runs into severe turbulence.

According to statistics compiled by the National Transportation Safety Board, between 1987 and 2000, only two fatal accidents (involving one death each) aboard U.S. commercial airliners were attributed to turbulence, and the phenomenon is believed to have caused the crash of just one U.S. airliner—some 40 years ago—mostly because the pilots failed to respond properly when severe winds struck.

While the majority of encounters with turbulence are not lethal, rough air wreaks havoc on the airline industry. “Probably the least of our worries is that the airplane is going to fall out of the sky,” says Lou Andelmo, a dispatcher for U.S. Airways. “It’s the injuries we’re most worried about.” Turbulence is the leading cause of nonfatal passenger and crew injuries, which result in work time losses as well as overtime paid to other crews to fill in. Passenger injuries can also result in lawsuits and settlements. And if a turbulence encounter is severe enough, Federal Aviation Administration rules mandate the airline conduct an immediate, unscheduled (i.e., costly) inspection of the aircraft for damage or stress before the airplane can return to service.

Turbulence can result in so many kinds of financial losses that exact numbers are hard to come by, but the Commercial Aviation Safety Team, a government-industry partnership, has been trying to ascertain and understand costs associated with turbulence as part of an overall mission to study air travel safety. According to Sherry Borener, an analyst at the U.S. Department of Transportation’s Volpe Center, CAST found that, among other things, an unscheduled inspection coupled with one day of out-of-service costs totals about $24,000 per incident. A diversion to another airport because of turbulence costs anywhere between $25,000 and $150,000, depending on the airline and the number of passengers affected. Estimates of losses due to delays and cancellations run as high as $866 million a year.

Flight attendants, who spend most of their time on their feet, are most vulnerable to injuries. Northwest Airlines has even produced a training video based on a recent incident in which one of its flights approached an area of reported severe turbulence: Following a request from the cockpit, passengers and crew were returning to their seats to buckle up when a flight attendant noticed that a door in the galley had an open latch. Behind the door was a rack that could have spilled onto another crew member if the turbulence was rough enough. Just as the attendant stood to lock the latch, a violent downdraft slammed the aircraft; the attendant was knocked to the ceiling, and injured her head and arm.

Candace Kolander, the Association of Flight Attendants’ coordinator for air safety, health, and security, says that in 1996, the most recent year for which she has data, in one airline, flight attendants reported 310 turbulence-related injuries, resulting in more than 3,500 lost work days. These are only reported injuries. The CAST study estimates that for every reported injury, more than 15 go unreported; reported injuries alone cost the industry approximately $26 million a year.

And then there are the costs incurred when a passenger sues because of an injury. Darryl Jenkins, a visiting professor at Embry-Riddle Aeronautical University in Prescott, Arizona, who has researched flight-related insurance and litigation issues, says that while airlines will contest some claims, court costs and the chance of bad publicity often force them to settle. In the mid- to late 1990s, he says, “it cost about $30,000 to settle the average claim.” CAST’s summary of findings states that litigation costs have “dramatically increased,” with one recent settlement reaching $10 million.

According to Borener, CAST has estimated that from 1988 to 2001, turbulence cost the industry $31 billion. That number doesn’t include costs nearly impossible to calculate: After a bad encounter with turbulence, people who are generally fearful of flying may get on airplanes even less; other passengers, angry that they got no food service while the aircraft was bouncing and lurching, may refuse to fly the airline again. Occasionally, however, the consequences of a rough flight can be known precisely: In 1999, American Airlines shelled out $2 million to a group of 13 passengers who convinced the court that their flight crew’s failure to take steps such as lighting the “Fasten seat belt” sign in advance of storm-related turbulence caused “psychological distress,” because the passengers thought they were going to die.

Turbulence comes in two forms: convective and clear air. The first involves updrafts and downdrafts created by hot rising air and cool, moist falling air—both of which you find in and around thunderstorms. Fortunately, moisture reflects on weather radar, making convective turbulence visible and somewhat predictable. Storms can also cause clear air turbulence, but CAT, as meteorologists call it, usually manifests itself in the jet stream—the west-to-east winds that gust just below the tropopause (the separation between the troposphere—the lower portion of the atmosphere where “weather” happens—and the stratosphere, at roughly 30,000 to 35,000 feet over the continental U.S.). One of the big causes of CAT is the phenomenon known as mountain waves: surface winds that hit mountains and then swirl upward, sometimes for miles, in powerful gusts.

Garry Hinds, manager of United’s meteorology department, likens the jet stream to a stream of water. “If the stream is straight and moving quickly, you can get in it and there’s really no problem,” he says. “But put rocks in that stream, causing white water, and that’s what mountain wave is—the atmosphere running into the rocks and getting pummeled and spun—and you get wind shear, which is simply a change in wind speed and direction. That generates breaking waves of air, just like breaking waves of water. The difference between the waves in the atmosphere and the waves in the ocean is you can’t see waves in atmosphere.” Well, for the most part.

“ ‘Clear air’ is kind of a misnomer,” says scientist Larry Cornman of the National Center for Atmospheric Research in Boulder, Colorado. “People use that because pilots fly around and get hit by something they don’t see. You can have a mountain wave, which has water vapor that condenses, and so you can actually see the lower part of the wave structure.”

Avoiding those waves is one primary job of the dispatcher, who begins working on finding a route even before passengers arrive at the airport. When U.S. Airway’s Andelmo starts a shift at the airline’s operations center in Pittsburgh, the dispatcher who is about to go off duty briefs him on air traffic control issues, the weather in general, and turbulence in particular. Before planning routes, Andelmo reviews weather information provided over the Internet by the National Weather Service, Weather Services International, and the National Oceanic and Atmospheric Administration (forecasts, turbulence alerts, and pilot reports—“pireps”—can be seen at

These sources show thunderstorms and provide the dispatcher with a map of all recent pilot reports of moderate or worse clear-air turbulence. Trouble is, CAT is so mercurial that pireps may be obsolete after only 20 minutes. “Turbulence is like a secondary atmospheric effect,” says United meteorologist John Goldman. “You can forecast wind shear, you can forecast atmospheric stability, but it’s a combination of those things that causes turbulence, and within an area where there is turbulence, it’s not going to be observed at all locations. It’s very random.”

Indeed, flights often smack into turbulence in areas that pireps had said were smooth. In 2000, a U.S. Airways flight sailing uneventfully over Chattanooga, Tennessee, at 35,000 feet got banged hard by an encounter with turbulence. “People’s heads hit the [ceiling] and cracked it,” Andelmo says. “We had some serious injuries.”

The airlines and the government also work together to prevent rough rides. “We have regular conference calls throughout the day with the FAA and the other airlines where we share any information we have,” says United spokesman Jeff Green. “In the case that our dispatchers need to get or share any information, we do have the ability to contact the operations control centers at other airlines.”

The airlines and the FAA characterize turbulence as light, moderate, severe, or extreme. Respectively, the categories are defined as (1) causing slight, erratic changes in altitude or attitude and rhythmic bumpiness; (2) same characteristics but greater intensity and rapid bumps and jolts, with passengers straining against seat belts; (3) large, abrupt changes in altitude or attitude and large variations of airspeed, with aircraft temporarily out of control; and (4) violent jolts making control of aircraft nearly impossible and structural damage possible.

But what feels like light turbulence in a 747 might seem severe in a 737. Turbulence even varies along the length of an aircraft: Pilots feel some bouncing, but the center of the aircraft shakes more than the cockpit, and the rear sections sway back and forth more than the center. Says Goldman, “Turbulence is the only meteorological parameter that’s a subjective report.”

NOAA, in conjunction with the National Weather Service, regularly issues clear-air turbulence forecasts based on calculations of things like temperature, wind speed, and wind direction in different parts of the country and at different altitudes. But some in the industry find the forecasts too conservative. So a few airlines, like United, have invested in their own meteorology departments.

“To put out a CAT forecast that we’re confident in is very labor-intensive,” Goldman says. “We need to be evaluating a lot of things. Even then, we wait until we start to get verification from flights in the area. And if their reports correspond to what the data are telling us, we then put out a warning.” Otherwise United might impose unnecessary deviations, resulting in extra costs. Exactly what those extra costs are is difficult to say, but United’s willingness to spend more than $2 million every year just to have a meteorology department is some indication.

Some parts of the country are known for having either convective or clear air turbulence almost all the time, such as above and just east of the Rocky Mountains, particularly around Denver. (United refers to a circle defined by a 50-mile radius around the city and up to 25,000 feet in altitude as “the Denver cylinder,” in which pilots can always expect rough rides.) Other usually turbulent areas include the Gulf Coast, southern Florida, the area around Cape Cod, Massachusetts, and the Montana-Canada border.

Along with the National Center for Atmospheric Research and other organizations, NASA’s Aviation and Security Program is currently working on what it calls a Turbulence Prediction and Warning System, which is intended to help reduce turbulence-related injuries. According to Jim Watson, the TPAWS project manager, two very promising technologies are being developed to help detect turbulence. The first is a software upgrade of the Doppler radar systems that many airliners already use to detect wind shear. With new algorithms, these radars will be able to look for and process weak radar reflections of moisture or ice crystals typically found in convective turbulence far away from storm clouds.

Laser-based radar, or lidar, is the second TPAWS technology under development (see “How Lidar Detects Turbulence,” next page). Much more sensitive than Doppler radar, lidar can show air motion at a very high resolution, a capacity that gives Watson and his colleagues hope that—pending more refinement and development—it will be able to reflect off minute dust particles blown around by clear air turbulence. But after the attacks of September 11, 2001, as airlines lost passengers and were forced to spend money on security, it became clear that developing an expensive technology like lidar would be difficult. “One of the challenges for a laser-based system is to make it more affordable,” says Steve Hannon, chief scientist of CLR Photonics, a Colorado firm that is working with NASA to develop turbulence warning sensors.

According to Hannon, most research has focused on 10- to 20-centimeter-diameter “pencil beam” lidar, which stays fixed straight ahead of the aircraft and merely alerts crews to the presence of turbulence, rather than more expensive scanning lidar, which sweeps across the path of the aircraft and returns an image of the air. It’s still unclear whether pilots need to know the shape of turbulence, as opposed to just the fact that it is looming ahead.

“In reality, I think all [airlines are] doing is trying to have a more reliable, robust seat belt warning light, basically to get people buckled down, and to secure the cabin with enough advance warning,” says Hannon.

The Doppler upgrade is closer to deployment than lidar, but both should give pilots and crews about 90 seconds of warning. NASA staff scientist Rod Bogue, a former manager of TPAWS, says that, based on recent experiments conducted on the length of time it takes to get passengers and flight attendants seated and buckled, 90 seconds is just about the right amount of time.

Another promising technology unrelated to TPAWS is the use of Next Generation Radar—a nationwide network of extremely powerful and sensitive Doppler radars—to detect turbulence. The National Weather Service, NOAA, the U.S. Air Force and Navy, and the FAA operate nearly 160 NEXRAD radars. Cornman is confident that the FAA will fund an NCAR project for this year in which NEXRAD data from around Chicago will be quickly posted online for United’s dispatchers and meteorologists. “The other aspect of the program that we’re also talking to United about is to uplink this data to the cockpit,” says Cornman. Printers in the cockpit could spit out alerts that would inform pilots about turbulent conditions along their approach paths.

“Basically we tell our students that whenever possible, just avoid turbulence,” says Mike Corradi, chief flight instructor at Embry-Riddle. “That way, you’ll never have to prove your superior talents as a pilot.”

Even moderate turbulence poses unwelcome piloting challenges, a point made soberingly clear when Corradi straps me into the left seat of a flight training device. Unlike a flight simulator, the device doesn’t mimic the sensation of motion, but the 220-degree wrap-around screen showing computer-generated flight in real time provides an extremely realistic illusion of it.

After a relatively smooth takeoff, I level off at 8,000 feet, gazing at the detailed scene around and “below” me. The instruments on my control panel are all steady, calm, and easy to read. Then Corradi switches on the turbulence input, dialing up to 3 on a scale going up to 10. The horizon starts to bounce, pitch, roll. My instruments—when I can read them—indicate constant yawing. The scenery dances. It’s all I can do just to maintain a semblance of level flight, which gradually slips from my grasp as I try to right the aircraft according to the instruments. Finally Corradi freezes the screen image and tells me to look up, saying, “Now, this is a picture you never want to see.” I’m locked in a bank of more than 30 degrees, a mountain range is at eye level and about a mile dead in front, and I’ve been steadily losing altitude.

I’m not a pilot—and the flight training device is simulating a small aircraft—but it’s not much easier for professionals flying big jets. “I can tell you, it’s very fatiguing to be in turbulence for a long period of time,” says Terry McVenes, an Airbus A320 pilot for U.S. Airways. Fatigue can undermine concentration, and in areas of high turbulence, pilots must constantly be at the controls; a good jolt to the aircraft can disengage an autopilot system. More recent and sophisticated autopilot systems can handle it, but many pilots prefer to fly manually through turbulence because it gives them a better sense of control.

When a flight hits turbulence that is moderate or worse, pilots follow certain procedures and then pursue options. First, they turn on the seat belt sign in the cabin, perhaps even make an announcement. Then they slow the aircraft down, just as a driver brakes when hitting a bumpy road. “All aircraft have a turbulence penetration speed,” McVenes says, referring to a manufacturer’s recommendation of an optimal speed for flying through bad air. “If you’re in the passenger cabin, you’ll hear the engines back down.”

After alerting the cabin, says Hank Krakowski, a veteran pilot for United, “I talk to air traffic control to see what they know about turbulence in the area and if changing altitude makes sense.” Air traffic control usually has some knowledge of any recent reports about turbulence at various altitudes. If the turbulence is on the downwind side of a thunderstorm, Krakowski, per United policy, has to double the wind speed at his altitude to determine the number of miles he must keep between him and the storm. For instance, if the wind speed is 23 miles per hour, he must fly at least 40 miles away from the lee side of the storm. “Once I’ve gotten into smooth air,” Krakowski says, “I then let United dispatch know I’ve changed altitude and then see the effect it’s going to have on my fuel.”

In other words, the numbers game begins again. If the lower altitude is going to eat up so much fuel that Krakowski may find himself in a dangerous situation should he have to enter a hold at his destination—or worse, have to divert—he may later climb back up and stay there, provided the turbulence is not severe. “While you want to create a great experience for the customer, the most important thing we do for them is get them to their destination and connections on time,” he says. “So sometimes they have to live, as we do, with the foibles of the atmosphere.”


Sidebar: How Lidar Detects Turbulence

Like radar, lidar (light detection and ranging) can calculate objects’ distances, speeds, and rotation rates by directing electromagnetic pulses at them and measuring the pulses that are reflected back. In the case of turbulence detection, the objects are tiny atmospheric particles. But unlike conventional radars, which send radio waves, lidar uses laser light, with wavelengths 10,000 to 100,000 times shorter.


The advantage of lasers is that laser light rays travel parallel to one another in a tight beam, as opposed to a radar’s radio waves, which diffuse in all directions. The concentration increases the odds that the laser light will hit and reflect off of dust and other minute particles—known as aerosols—that lie directly in a lidar beam’s path.

Lidar is thus ideal for detecting clear air turbulence, which has only tiny particles, not large water droplets, to reflect radiation. But for the very same reasons, lidar cannot help in examining the interior of a storm; the laser light would be reflected entirely by the outermost layers of clouds or rain. To see through moisture, conventional radar works better.

To detect CAT, lidars shoot laser pulses into the air ahead of the aircraft, where aerosols are being carried in the same direction and at the same velocity as the wind. The speed of the aerosols is measured by observing the Doppler shift of the laser reflections. If the aerosols are moving away, the returning light waves will have a lower frequency and longer wavelength than those of the original laser light; the shift will be toward the higher frequencies and shorter wavelengths if the aerosols are approaching the laser. By comparing the relative motions of aerosols in a beam’s path, computers on aircraft can predict when CAT is imminent.


The DC-8 lost its left outboard engine and 19 feet of wing and fell 500 feet in 10 seconds, but landed safely. NCAR