AS USAir Flight 1016 began its final approach to Charlotte Airport in North Carolina on 2 July last year, the weather erupted. Light drizzle became torrential rain and mild winds turned into violent gusts of up to 60 knots. As the McDonnell Douglas DC-9 approached, the captain decided that poor visibility and unpredictable winds made a landing too risky. With the plane 1.1 kilometres from the runway and only 60 metres above the ground, he ordered the first officer to fly around for a second approach. As the plane nosed upward, disaster struck. Suddenly the 35-knot headwind reversed into a 25-knot tailwind. The plane鈥檚 airspeed dropped from 160 to 115 knots causing an immediate loss of lift. The crash was unavoidable.
The plane broke into three pieces as it hit the ground, bursting into flames. The accident killed 37 passengers and seriously injured 21 others. Later, the captain and first officer told accident investigators that the plane just fell out of the sky.
This is the hallmark of windshear 鈥� a violent shift in wind speed and direction. The International Civil Aviation Authority, based in Montreal, says 565 people died in accidents caused by windshear between 1970 and 1992. In the US, more airline passengers die in accidents of this type than in any other.
Advertisement
Now scientists at the Lincoln Laboratory of the Massachusetts Institute of Technology in the US have worked out a way of anticipating windshear so that pilots can take evasive action. Their method relies on a powerful computer to combine weather data from several sources and then look for telltale patterns of temperature, humidity and air movement that indicate that windshear is imminent. The Federal Aviation Administration, which regulates aircraft safety in the US, hopes to employ the system at 47 American airports. It is currently installing a new generation of weather radars that will supply a crucial part of the data. These systems could have prevented the USAir crash, says Matthew Hampton who monitors the FAA safety programmes for the US Congress. 鈥淚t would have given the pilots and the controllers an edge.鈥�
Windshear occurs most commonly in a phenomenon known as a microburst in which a column of air drops from the sky, fanning out in all directions as it hits the ground. An aircraft close to the ground would experience a headwind as it flew into the microburst but a tailwind as it flew out. Directly beneath the microburst, the wind direction reverses. This is the dramatic change that caused Flight 1016 to crash (see Diagram).
Microbursts form most often in the warm, moist air found in thunderstorms. In 1991, researchers from the Lincoln Lab spent three months studying microbursts that formed in thunderstorms around Orlando in Florida, an area renowned for this type of weather. They discovered that microbursts occur when warm, moist air rises above a storm, where it cools causing the moisture to condense and even freeze. The ice, water, and cool air then drop, rapidly reaching speeds of up to 40 knots before hitting the ground. This all happens in as little as two minutes. And compared with most weather systems, microbursts are relatively small, forming at heights of up to 6000 metres and covering only a few square kilometres when they hit the ground. This makes them difficult to spot.
Microbursts are common in the US and Australia where there tends to be large amounts of thermal energy in the atmosphere, according to Alan Woodfield, an aviation research consultant based near Bedford. This can result in levels of microburst activity which are a hundred times higher than in Europe. The problem is more acute in America where the high density of air traffic increases the risk of an accident.
More than a hundred airports in the US, including the one at Charlotte, are equipped with the first generation of windshear detection systems. They are known as the Low Level Windshear Alert Systems (LLWAS) and consist of a number of anemometers positioned around the airport to measure wind speed at ground level. The data are fed to a central computer which determines how the wind patterns are changing. If it spots any sudden changes in wind speed or direction, the computer issues a windshear warning. While this is useful for pilots approaching the airport who have time to take avoiding action, it is no help to aircraft like Flight 1016 that find themselves in a microburst as it forms. At Charlotte, the LLWAS sounded 22 seconds after the plane hit the ground.
The new generation of weather radars should help by measuring the movement of air above the airport and in the surrounding atmosphere. The system is called Terminal Doppler Weather Radar (TDWR) and, in theory, it can detect windshear up to 70 kilometres away and tell whether it is moving towards the airport 鈥� unlike LLWAS which can only spot windshear within the immediate vicinity. TDWR works by bouncing radio waves off airborne particles such as dust, pollution, rain drops and even insects, and measuring the frequency of the reflected signal. This frequency depends on whether the particles are moving towards or away from the radar and how fast. This is known as the Doppler effect. An analogous phenomenon is the change in pitch (frequency) of a passing train whistle. The signals are fed to a computer, which constructs a three-dimensional picture of wind patterns in the atmosphere around the airport. A rapid change in wind speed or direction triggers a windshear warning.
Doppler radar is not new. Weather forecasters commonly use it to measure the speed of rain clouds, for example. TDWR has a much more difficult task, says Bob Polese, an electrical engineer in charge of weather programs at Raytheon, a defence company based in Lexington, Massachusetts, which is building the system. Unlike meteorological radars, he says, TDWR scans the atmosphere close to the ground which is crowded with objects such as trees and buildings that create unwanted reflections. To minimise the number of objects on the ground that the beam hits at any one time, the TDWR operates with an unusually narrow beam less than half the 1 or 2 degree width used in most meteorological radars. Also, the computer is programmed with the positions of buildings and trees so that it can strip out their reflections before analysing the remaining signal.
Danger signs
Nevertheless, TDWR, like its predecessor, can only spot a microburst in action. Predicting microbursts before they happen can only be done by combining the radar data with measurements of temperature and humidity in the atmosphere, says Marilyn Wolfson, a researcher at the Lincoln Lab. She and her colleagues think they know how they can get these data. US airliners continually register the outside air temperature and humidity and relay the measurements to the ground every 45 minutes or so. In addition, the data can be gleaned from weather balloons launched by meteorologists. The researchers are currently developing software that combines this information with radar data and looks for the weather patterns that herald a microburst.
The software works by looking for areas in the atmosphere where the amount of liquid and ice is increasing rapidly 鈥� a condition that could lead to a sudden downdraught. It estimates the amount of water using radar measurements and the size of water droplets in the air and the amount of ice by analysing radar measurements at altitudes where the temperature is below freezing. It also calculates where the centre of mass of the storm lies (by estimating the entire mass of water) and monitors its position 鈥� if it begins to drop, a downdraught could follow.
Last summer, Wolfson and her colleagues tested the program at airports in Memphis, Tennessee, and Orlando, Florida. They said it performed well, predicting over 70 per cent of the microbursts that later developed. Wolfson says the FAA plans to incorporate the program in a project that will link TDWR with LLWAS and other weather instruments such as rain gauges. The network will be called the Integrated Terminal Weather System, and will provide ground controllers with an up-to-date three-dimensional map of the atmosphere around the airport. The maps will also be broadcast to aircraft. At present, pilots rely on ground controllers to tell them about weather conditions. The FAA plans to start development of the new system later this year.
But the project may fall at the first hurdle if the FAA is unable to set up the TDWRs. At the moment it is having trouble finding land for the radars, which have to overlook an airport from a distance of at least 12 kilometres. 鈥淣ot many people want a 200-foot radar in their backyard,鈥� notes Hampton. So far, only two TDWRs are in operation at Houston鈥檚 Intercontinental airport and at Memphis, Tennessee. The FAA says it has chosen the sites it needs and may have to use legal measures to seize the land. The agency claims the TDWRs will all be in place by the end of 1996.
Fatal flaw
Even when that happens, many smaller American airports will not have TDWR and aircraft will have to rely on less accurate on-board windshear detection systems which all American aircraft are required to carry. Known as reactive windshear systems these measure the speed of air over the plane鈥檚 wings. If this changes suddenly, a warning sounds in the cockpit. However, moving the flaps also causes the airflow over the wings to change. To avoid false alarms, the device switches itself off when the flaps are moved. As Flight 1016 nosed upwards to begin its fly around, the pilot retracted the flaps and the windshear detector was immobilised for 7 vital seconds. The warning sounded only 3 seconds before the plane hit the ground. The devices have since been altered so that they can detect windshear while the flaps are being moved.
Now a new generation of on-board detectors is set to make reactive systems obsolete. The most advanced of these are essentially portable versions of the TDWR: a doppler radar that can detect windshear in the atmosphere in front of the planes giving the pilot time to take evasive action. In 1992, NASA researchers tested the system by intentionally flying through microbursts near Denver, Colorado and Orlando where they are common on summer afternoons. They discovered that it worked well in wet microbursts but had trouble spotting the 10 per cent of windshear events that occur in dry conditions when there is no rain to reflect the radar. 鈥淲hen it gets dry, the radar system sees right through a microburst,鈥� says Bill Weist, an engineer at AlliedSignal, an electronics manufacturer based in Fort Lauderdale, Florida, that has developed an on-board radar system.
One alternative is to use light waves rather than radiowaves to monitor the movement of smaller particles known as aerosols, which are present in dry air. The reflected signals can be used to measure the speed and direction of these particles using the Doppler effect. The disadvantage is that this system, known as lidar (light distance and ranging), does not work in the wet because the rain simply absorbs the light. Lockheed, an aerospace company based in Sunnyvale, California, has developed such a system and says it would complement the radar systems. Russel Targ, an engineer at Lockheed, says lidar has other applications because it can deduce wind speed at high altitudes where there is little moisture to reflect radar signals. Pilots could use this information to choose a cruising altitude with the most favourable wind for reducing fuel costs. 鈥淥n long-haul flights lidar would have paid for itself within a year,鈥� he says. However, lidar has yet to win approval from the FAA.
Radar systems, on the other hand, were approved by the FAA last year. In November, a Continental Airlines Boeing 737 became the first passenger aircraft to use such a radar windshear detection system. All Continental planes will be equipped with the system by the end of next year. Built by AlliedSignal the radar can detect wind shear up to 5 nautical miles away, which provides ample opportunity for the plane to fly around. The systems are not cheap, costing $70 000 or more, because of the electronics needed to process and analyse the radar data. On-board windshear radar will be useful for landings at airports not equipped with TDWR, but is also useful at airports that do have it, says Bruce Matthews, an engineer at Westinghouse Electronic Systems, which has also developed an on-board windshear radar. He points out that the data from the on-board radar are updated every 5 to 10 seconds compared with once a minute for a TDWR.
For the moment, most US airliners carry neither lidar nor the radar systems. But the scramble has begun to fit them.