Ever since the Wright Brothers flew the first powered airplane, low-altitude encounters with atmospheric turbulence have posed daunting challenges to the in-flight safety of aircraft of all types, including commercial airliners. Weather is typically a cause, direct or indirect, of about a third of all commercial aircraft accidents and over a fourth of all general aviation accidents.
One form of turbulence, wind shear, has been particularly catastrophic. Between 1964 and 1985, wind shear directly caused or contributed to 26 major civil transport aircraft accidents in the U.S. that led to 620 deaths and 200 injuries.
Aircraft often experience a shift or drop in winds while in flight. But they usually have enough speed and altitude to compensate for the resulting loss of airflow over their wings. During takeoffs and landings, wind shear is a more serious threat because the aircraft is close to the ground and its speed is slower (Fig. 1).
Due largely to the development of ground-based and airborne wind-shear detection and alert systems, airplane flying is safer than ever. This safety record can be traced to the work performed by the National Aeronautics and Space Administration's Langley Research Center in Virginia. NASA's Airborne Wind-Shear Detection and Avoidance Program (AWDAP), a seven-year research and development effort, built the foundation for present commercially available wind-shear radar systems.
Today, nearly 4000 commercial airliners worldwide use wind-shear detection and alert systems based on NASA's AWDAP. "Statistically, the rate of wind-shear-related accidents today can be calculated as one incident in every 10 years, versus one each year before the AWDAP system was developed," says the program's director Roland L. Bowles.
Development actually began in the 1980s when alarm over commercial airliner crashes related to wind shear moved the U.S. government into action. On July 24, 1986, the Federal Aviation Administration (FAA) and NASA signed a memorandum of agreement formally authorizing the start of the AWDAP, and a wind-shear program was created in the Flight Systems Directorate of NASA's Langley Research Center.
The agreement mandated NASA to work with industry to develop a test-bed AWDAP for eventual technology transfer. In September 1988, the FAA ordered all commercial aircraft to have forward-looking airborne wind-shear detection and alert systems by 1993. Three airlines—American, Northwest, and Continental— received exemptions until the end of 1995. However, solving the problem was far more complex than originally estimated, and nearly another decade passed before commercial aircraft began to adopt the technology in earnest.
Many of the commercially available wind-shear radar systems continue to evolve with better features like more advanced warning time, greater detection precision, and fewer false alerts. Ever since wind-shear detection systems have seen commercial service, the record has been spotless. There have been no reported incidents related to wind shear.
The FAA had developed the ground-based Low-Level Wind-Shear Alert System (LLWS) in 1976 as well as a ground-based technology called Terminal Doppler Weather Radar (TDWR). These systems detected and automatically reported low-altitude wind shears to pilots. They also let air-traffic controllers change landing and takeoff runways to avoid the wind shears. But they weren't sensitive enough to detect all wind shears. It wasn't until an airborne system was developed that the technology of wind-shear detection and alert significantly contributed to aircraft flight safety.
A THREE-PRONGED APPROACH
To get a good handle on the wind-shear problem, NASA researchers working on the AWDAP divided the project into three main elements: hazard characterization, sensor technology, and flight management. Five years into the program and after intensely studying various weather phenomena and sensor technologies, the researchers needed to validate their findings in actual flight conditions.
They chose an extensively modified Boeing 737 flying laboratory, dubbed the Transport Systems Research Vehicle (TSRV), for the task. The aircraft came equipped with a rear research cockpit in what would have been the forward section of the passenger cabin (Fig. 2). Many flight tests were performed primarily at two locations: Orlando, Fla., where "wet" microbursts containing rain droplets were common, and Denver, Colo., where "dry" microbursts were observed.
In trying to understand the wind-shear phenomenon, Langley researchers used supercomputers to develop meso-scale numerical weather models. They also employed extremely high-fidelity fluid-dynamics models.
In time, researcher Fred Proctor came up with a detailed model known as the Terminal Area Simulation System (TASS). This 3D, time-dependent model included representations of liquid and ice microphysics with the effects of condensation, evaporation, freezing, and sublimation in the atmosphere. Their impact on atmospheric winds could then be used to simulate the time-dependent life cycle of a convective storm, including microbursts. Data sets generated from the TASS model were eventually used by the FAA to certify onboard wind-shear sensors.
After studying the work of meteorological researcher Tetsuya "Ted" Fujita at the University of Chicago, Bowles developed an F-Factor that was key to understanding the wind-shear phenomenon. Fujita developed a severity scale for tornadoes that bears his name. He subsequently coined the term microbursts after extensively studying weather-related aircraft crashes. His work played a key role in understanding the effects of wind-shear microbursts.
Fujita theorized that a concentrated strong 3D air outflow associated with a wind downdraft was the real hazard in airplane encounters with wind shears, particularly during takeoffs and landings. Radar meteorologists have since refined his explanation, defining a microburst as a divergent low-level wind field with a velocity change of at least 5 knots over the wind's surface for distances between 1 and 4 km.
Bowles' F-Factor quantifies the loss in aircraft performance during a specific wind-shear condition. This nondimensional metric considers the aircraft's weight, thrust, and drag as well as the effects of wind-shear velocities on a specific excess thrust (thrust minus drag divided by weight) required to maintain steady flight conditions due to wind variations in a microburst. Bowles, Michael S. Lewis (who was the AWDAP deputy program director), and David A. Hinton later refined this metric to include the length of time over which the aircraft is exposed to the wind shear averaged over a distance of 1 km.
A typical commercial aircraft traveling at 150 knots over 1 nautical mile and encountering a wind shear with an F-Factor of 0.15 or higher (24 seconds of flight time) will lose 511 feet of altitude if the pilot doesn't take immediate recovery action. In fact, F-Factors of 0.15 and higher can exceed the maximum performance capability of the aircraft. Today, the F-Factor is an FAA regulations parameter. The FAA requires wind-shear warnings to be provided for F-Factors of 0.13 and higher.
Langley's researchers needed to develop a simulation model that gave pilots enough warning time before entering a microburst. Therefore, the aircraft would have more altitude, energy, and speed with which to combat the effects of wind shear. That time was determined to be at least 30 to 40 seconds. The warning would give pilots the option of either preparing to fly through a wind shear or steering around it if wind-shear conditions were very high and air-traffic conditions were favorable.
Langley's simulation efforts involved integrating information about wind-shear and microburst conditions with information about aircraft energy states. Researchers concluded that even 20 to 25 seconds of advanced warning (as little as 17 seconds) would suffice for a pilot to take corrective action.
PICKING THE RIGHT SENSOR TECHNOLOGY
Three technologies were investigated for the AWDAP's forward-looking airborne wind-shear detection: microwave Doppler radar, Doppler light-emitting and ranging (LIDAR) systems, and infrared radiometry. NASA evaluated all three types, working with companies such as Lockheed Missiles and Space, Coherent Technologies, the Research Triangle Institute, and Turbulence Prediction Systems Corp.
A huge challenge remained, though. The researchers needed to devise a system that could detect both wet and dry microbursts. LIDAR technology performed well in wet and dry microburst detection but was less effective in heavy rains, where signals tended to be absorbed. And, infrared radiometry wasn't reliable enough for wind-shear detection due to lack of development at that time.
Many operational Doppler weather radars operate in the S-band (1 to 3 GHz), C-band (3 to 8 GHz), and X-band (8 to 12 GHz). The higher frequency provides higher resolution (higher sensitivity and smaller cell resolution) at the cost of some increase in signal attenuation. Ku-band (12 to 18 GHz) radar also is useful for wind-shear detection where signal attenuation levels are still acceptable.
With the assistance of researcher Emilio M. Bracalente, who helped evaluate radar sensor technology, the NASA team settled on X-band Doppler radar. It best combined high resolution and low attenuation levels. Besides, most airport and aircraft weather radars were X-band, making the choice easy to integrate within existing avionic systems.
A major problem was overcoming the effects of "ground clutter." Not only does Doppler radar look ahead, it also looks down, reflecting off objects on the ground. As a result, it generates spurious returns as an aircraft descends for a landing, so the system must distinguish between wind-shear data and ground clutter. To deal with this, the researchers experimented with a minimum flight altitude of 750 feet above the ground for wind cells exhibiting an F-Factor greater than 0.1 and a minimum air speed of 10 knots.
NASA's research showed that lower clutter-to-signal ratios occur at short ranges in front of the aircraft where the radar pulse in the main beam hasn't touched the ground. Here, ground clutter comes primarily from the radar signal's side lobes, which can be kept sufficiently low to suppress the clutter. For a 3°-beam-width antenna pointed down at a 0° tilt angle and at a 5-km aircraft range, the −3-dB point of the main beam first touched the ground at a range of 3.5 km, and the first side-lobe null point occurred at about 2.7 km.
TWEAKING A COMMERCIAL TRANSCEIVER
A modified Rockwell Collins model 708 X-band ground-based radar unit was used in the AWDAP experiments. It consisted of a stabilized antenna pedestal and a flat-plate slotted-array antenna. The system was modified so that its size, antenna, and power requirements were compatible with the 737 TSRV aircraft being used. NASA modified the receiver subsystem and the signal- and data-processing subsystem. Also included were a control computer, a data-recording subsystem, and displays.
The basic hardware was carefully matched and tuned to provide 3 dB more output power. Rockwell Collins incorporated several pulse widths and pulse-repetition frequencies as well as a means of selecting various configurations. An optional traveling-wave-tube (TWT) high-power amplifier with 2000 W peak was incorporated, along with the means of switching it in and out of the system.
Also provided were outputs of the first intermediate frequency (prior to the automatic gain control, or AGC, and calibration circuitry), the local oscillator, and system timing signals. Researchers Steve D. Harrah and Les Britt came up with algorithms to run a simulation program that detects microbursts among radar and clutter signals.
The radar receiver subsystem uses a fast-acting (feed-forward per bin) AGC circuit. (A range bin is a discrete element along a single radial of radar data at which the received signals are sampled.) Prior to the I/Q detector, a log detector is sampled, and the return power level is estimated at each range bin. These samples are digitized using an 8-bit analog-to-digital converter (ADC) and averaged over several pulses (fewer than 128). The distribution of attenuation across the three attenuators is set to optimize the signal-to-noise ratio at the detector and minimize the amount of intermodulation distortion in the signal (Fig. 3).
Outputs of the I/O detector are digitized by a 12-bit ADC, driven by the clock range. The digitized data is stored in FIFO buffers for later inputting to a data-recording system and for processing by a Wind Shear Radar Signal Processor (WSRSP). The WSRSP was located in the radar experimental pallet at the rear of the 737 TSRV. It included two high-resolution graphics boards as well as a 1280- by 1024-pixel, 14-in. color monitor that displayed four separate tiled windows.
The real-time radar processor system used in the 1992 wind-shear flight experiments was a VMEbus-based system with a Motorola 68030 host processor and three DSP boards. Each board contained three Texas Instruments 320C30 33-MHz DSP chips rated for 33-MFLOPS operation.
In the trials, the WSRSP accepted data directly from the radar system or data generated by one of the three DSP boards. This board processed the data according to predetermined algorithms (which could be modified in-flight) that were developed by Harrah and Britt. Researcher Rosa Osegurn performed in-situ recording and data validation.
The result of all this processing produced values for the reflectivity, velocity, and spectral width of the weather located ahead of the aircraft. This data was sent to the 68030 host, which processed it using hazard algorithms also developed by Harrah and Britt. A 1024- by 768-pixel, 14-in. color monitor located in the aft cockpit displayed wind-shear information on two separate tiled windows. Windows on this monitor as well as on the radar-pallet monitor could be configured independently to display aircraft or radar parameters or various types of radar maps or line plots.
A SUCCESSFUL CONCLUSION
On September 1, 1994, Allied-Signal/Bendix's (now Honeywell's) model RDR-4B became the first predictive wind-shear system to be certified for commercial airline operation. That same year, Continental Airlines became the first carrier to install an airborne, forward-looking, wind-shear detection system on its aircraft.
By June 1996, Rockwell Collins and Westinghouse's Defense and Electronics Group (now Grumman/Martin) also produced FAA-certified, forward-looking, wind-shear detection systems. To date, over 4000 such systems have been installed on aircraft of both domestic and foreign carriers. All U.S. military aircraft have had the technology for some time.
These modern wind-shear detection and alert systems now routinely provide pilots with both visual and audible warnings of wind-shear conditions ahead, giving them the choice of preparing to fly through a wind shear, depending on weather severity and airline traffic in the area (Fig. 4).
By using a commercially available 737 aircraft (the TSRV), NASA played a significant role in expediting and validating forward-looking airborne detection and alert systems. That wouldn't have been the case if only computer simulations were used. The NASA/FAA AWDAP development, cited as "NASA at its best" by NASA's Aeronautics and Advisory Committee, was nominated for the nation's prestigious Robert J. Collier Trophy in recognition of the most significant aerospace accomplishment. But AWDAP came in second to the repair of the Hubble Space Telescope, which nabbed the top spot.
Because of space limitations, this article was not able to name every individual who contributed to AWDAP's success, but we heartily congratulate the entire team.
Further, we wish to acknowledge, in particular, the assistance of the American Institute of Aeronautics and Astronautics (AIAA) and Katherine Barnstorff of NASA's Public Affairs office in Langley, Va., in helping to obtain information for this article. Thanks also go to Joe Chambers for his article "From Concept to Reality" detailing the NASA Langley Research Center's contributions to civil aviation during the 1990s, which provided the basis for many of the technical details related in this feature.