Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Low-Altitud~e Wind Shear The Nature of Low-Altitude Wind Shear Wind variability is a perennial and inescapable problem for aviation. Meteorological circulations or terrain-induced airflows can on occasion induce large and rapidly changing variations in air velocity over small distances. These variations produce correspondingly sudden changes in the relative flow of air over an aircraft's wings and other lifting surfaces, with attendant changes in an aircraft's flight path. Thus, small-scale wind variations and turbulence can pose hazards to aviation, particularly when they occur in the lowest few hundred feet of the atmosphere, the zone that aircraft must penetrate while landing or taking off. To eliminate--or at least appreciably reduce--the hazards posed by low-altitude wind variability, it is necessary to understand the sources of wind changes and the risks they represent. It is also necessary to know how to detect, measure, and predict them and how to communicate useful information on wind variability to air traffic controllers and pilots in a timely fashion. The wind changes not only with distance but also with time. As a result, the term ~ is sometimes used when considering low-altitude flight hazards. Technically speaking, wind shear is the local variation, at a particular time, of wind velocity with distance. It is measured by dividing the velocity difference at two points by the distance between them. Strong wind variations over horizontal distances of 1 to 10 miles can cause particular difficulties for aircraft. Most often, in this report, we refer to wind variations as wind shears. In circumstances where time variations are important, they will be identified. The three-dimensional airflow in the lower atmosphere, and the associated wind shears and turbulence, vary from place to place by season and by meteorological conditions. Based on experience, rough estimates can be made of the degree of hazard, frequency of occurrence, and difficulty of detection of various types of wind shear. 19
Atmospheric turbulence is generally defined statistically in terms of scale and intensity. Its effects are seen in an aircraft's ride and handling qualities and are taken into account in aerodynamic, flight control, and structural design criteria. However, patchy small-scale turbulence need not be present at low altitudes withi layers of air with strong wind shear (Lee and Beckwith, 1981~. The most serious ef fects of wind shear are those that cause an aircraf t to lose lift and altitude. This is particularly hazardous when an aircraft is close to the ground, either landing or taking of f, when an aircraft unexpectedly flies from a region of headwinds into a region of strong tailwinds, and especially ~ f the transition occurs in a s bong downdraf t . Turbulence and heavy rain, when occurr ing in association with wind shear, can contribute to flight hazards and increase the chances of an accident. The following paragraphs describe the types of wind-shear situations and the risks each poses to aviat ion. Convective Outflows Thunderstorms and other convective clouds are critically important sources of low-altitude wind variabil ity. Many produce strong downdrafts that transport air downward, which then spreads out rapidly over the ground. The size and strength of the downdraft depend on the properties of the thunderstorm and on the humidity and temperature structure of the atmosphere. As shown in Figure 1, thunderstorms occur most frequently in Florida, along the Gulf of Mexico coast, and over the central parts of the United States. Whenever there is a thunderstorm or precipitating convective clouds, hazardous low-altitude wind shear can be present. Some experts believe, however, that strong downdrafts and associated flight hazards are more likely when the thunderstorm cloud bases are high and the surface humidities are low. Microbursts. Following the crash of Eastern Airlines Flight 66 at New York City's Kennedy Airport on June 24, 1975, the term downburst came into use to describe a strong downdraft that induces an outburst of damaging winds on or near the ground. Subsequently, studies of the EAL crash and of Continental Flight 426 at Denver on August 7, 1975, and Allegheny Flight 121 at Philadelphia on June 23, 1976, concluded that each of these accidents was related to downburst-induced wind shear (Fu jita and Byers, 1977; Fu jita and Caracena, 1977) . _ Results of the above studies indicated that the downbursts that contributed to these accidents were of small size and short life, and the term microbursts has been used to describe them. Microbursts are small downbursts, less than 2.5 miles in outflow size, with the peak winds lasting only 2 to 5 minutes. Some microbursts reach the ground, while others dissipate in mid-air and are not detected by ground-based anemometers. ~ critical point is that microbursts can come from convective clouds that are not accompanied by lightning. 20
~-to-~[o ~ - of ~ :,l: J~^~ a:' ~~::-- 1~ ~ ,,% a- I ~ - ~ lo Ad \ ~ q'~ 1--~- \, ~ . .. .. 1 \m ~~- · ' '; ~ ~ i:-': ::::::':::::::~:i`XI::: I.-:: t - ~--2:2 ~ 21':"" ~ I%~ By ~ I.- 5 ~ ~ ~ k 3; r. . . . i A / , ., . I ^2 . ~ 2 - ~2./2 '. 'a 2 .~ B. . . ~ J ,. w?.2 ~ . 2 2 If . . . ~2. 2. - - ~ ~ ~ - t.;;- . . Jar,, ~ . . ~ I. ~ ~ ~ ,,,~,,.,, , ,,,.~ ~.,,.,,,~-,,., E'''' \~ /. ,`...~. I. .. . . .k '... . I' - . , . {,., , .,.2. '':: ~ a. - t .. - . . , ·, ·, - -,,, ., . .. ~ ~. o ~ o a ::. :: ~ .. :: : :: .\ ~J .. : : $ : :: ~ ~ O .: .............. .: : :: ; . W.: ,,.~.\ ~ 1 ,,., ,~ ~; ~ ,% ~ : - ~ O :, ~ .~ : . - . :.: .-.- : ·. . : ~ ., ~ ,,,~. '.- :: . ~ .- 2 - 2: ,.: 2 8, :: 2 2. :: ,.:'.f ~ O ~ .. : : r ~ ~,~0 ,. .. . : ... ~ : .- - ` Y - - .. ... , :,~ , .~ ; \,... ::....................... X,;, .... ..... .. :... ~ ............ ... .... ,? _. _ . y / -., Y ~ / ~In _ PA C / F I C O C E A N 21 .: J oo ! U · - · - C~ S~ o ·. C~ o cn _~ U~ 1 U~ ·^ ~q S~ o U S" U · - S" S~ ~q :^ Q 4- o S~ a) z "a). ;~; ,.,. ,, _ 1 E4
The most recent wind-shear-related accident, the crash of Pan American Airlines Flight 759 in New Orleans, was reported by the NTSB (1983) to be a result of microburst-induced wind shear. Climatology of Microbursts. Data on which to base averages of the frequency and intensity of microbursts are very limited. During a 43-day period in May and June 1978, ground-based anemometers associated with NCAR's portable automated mesonet (PAM) were deployed as part of the NIMROD Project near Chicago. PAM automatically measures wind speed and direction, dry- and wet-bulb temperatures, and pressure and rainfall at 27 surface stations (wind measurement is at 12 feet above the ground). Fifty microbursts were detected, 32 with rain (wet microbursts) and 18 without rain (dry microbursts) (see Table 3~. Wind shear is apparently not related to rainfall intensity. TABLE 3 Frequency of Microbursts Detected by PAM NIMROD JAWS - Number of operational days 43 86 Number of microburst days 11 49 Number of wet microbursts* 32 31 Number of dry microbursts** 18 155 Total number of microbursts 50 186 Number per operational day Number per microburst day 1.2 2.2 4.6 3.8 * Equal to or greater than O.O1 inches of rain during the period of peak winds. ** Less than 0.01 inches of rain between both the onset of high winds and the end of the microburst winds including the calm period if any. The PAM deployed in the Joint Airport Weather Studies (JAWS) Project near Denver detected 186 microbursts on 49 days of an 86-day observation period from May to August 1982 (Fujita and Wakimoto, 1983~. This amounts to 2.2 per day, nearly twice the rate found in the NIMROD Project. Of the total 186 microbursts, 155 were dry and only 31 were wet. Most were not associated with active cumulonimbus clouds but rather occurred under streaks of evaporating precipitation (virga) from dissipating cumulonimbus or dissipating cumulus congestus clouds. Because of their association with convective clouds, microbursts tend to occur most often between noon and midnight, but, as shown in Figure 2, the diurnal pattern differed at the NIMROD and JAWS locations. 22
40 30 In cry 20 m o - 10 IS6 Microbursts i n ~ AWS (86 days DRY WET fig 3p.m. ~ ,. f. 6 p m it. 00 03 06 09 1 2 15 18 21 DO MDT 50 Microbursts in NIMROD (43 days) 10 . _ _ " ~ ~ 1 I, 00 03 06 09 15 18 21 OOCDT FIGURE 2 Diurnal Variation of Surface Microbursts of All Intensities Measured at the 27 PAM Stations. ~ Source: Fuj its and Wakimoto, 1983 ~ . Aircraft passing through the center of a microburst will experience a change in wind velocity that can be specified as the vector difference of the headwind and the tailwind along the flight path. The maximum wind difference of the JAWS microbursts, measured by ground-based anemometers, exceeded 95 knots on one occasion (see Figure 3~. 23
20 _ co cn a:, z) 1 m o Cat - 11 ° IC IJJ m me 5 15 -~ _. _ 5 10 15 20 25 30 As 40 4s m/see I i I I I I I I I 10 20 30 40 50 60 70 80 90 kts ~636 Surface Microbursts REDRY WET at 27 JAWS PAM Stations MAXIMUM WIND SPEED DIFFERENCE FIGURE 3 Frequency of Surface Microburats as a Function of the Maximum Wind-Speed Difference Measured at the 27 PAM Stations. (Source: Fujita and Wakimoto, 1983~. The JAWS Project used Doppler radar primarily to observe microbursts. With three wind-measuring Doppler radar systems, three-dimensional pictures of microbursts were obtained. Figure 4 shows the horizontal and vertical velocity profiles of a particularly strong microburst. The classical profile of headwind, downdraft, and tailwind are clearly seen in this figure. Because the Doppler radars were used to concentrate on specific cases of interest and because they were not operated continuously, their data cannot be used to establish a microburst climatology. Nevertheless, the Doppler radars detected and observed 75 microbursts on 33 of the 86 operational days (McCarthy et al., 1983~. The Doppler radar data allowed for an examination of the time history of JAWS microbursts observed near Denver. Figure 5 shows the percentage of microbursts that reached maximum velocity differential as a function of time from the detection of the initial velocity divergence near the ground. Approximately half of those observed reached maximum intensity within 5 minutes, and nearly all reached maximum velocity within 10 minutes. This figure dramatically illustrates the extremely short duration of these events. 24
HORIZONTAL16-33 AGL ~_~ 40 Knots `, ~ , , ~ ~ · ~ · ~J , , ~1 .L ~ i, ASH ~ .L i,~,1~1 ~ · ~ ~ · ~ ~ ~ ~ ~ ~ 1 ~ ~ ~ . i] ~ ~ ~ ~ ~ ~ ~ ~ - ~ ~ ~ ~ ~ ~ ~ ~ ~~ _ _ _ ~ - ~~` 1 L~ ~ ~ 1 ~ ~ ~ ' ~ ~ I) ~ ~ ~ ~ ~ ~ - 00000 1 i~ ~ 1 ~ ~ .'~ - ~~` ~ ~ ,`40, ~ ~ ., - ~--~d,-~` - - J~ _ i; ~ ~ ~ I---- - ~ _ ~ 1 ~ .' A,,, ,.--_ ~ O- 1- c~ 11 OI o u' LL Cal He ~ 2 ~ 1 ~ ~ ~ ~ ~ ~~ - ~ ~ ~ ~ i ~ ~ b lo' lo' ~ ~t'~ ~ ~ a' at' ~ ~ ~ ~ d ~ ~ 30 ~ . <. , . . ~ ~ . ~ · .;, . . ~ ~ C a\ ~ ~ ~ '. ~ . /~ ~ ~ ~ ~ ~ ~ l) ~ 'a u I, ~4; ~ ~ ~- 30 ~ ~ id ~` ~ 'a ~ ~ ~ ~ _ _ tar ~ _ _ _ ~ ~ _~ __. ._. ~ ~, ., , j _ · ~ ~ ~ ~ ~ ~~` ' 1 - ~ ~ ~ ~ `~` 'd~ '~ -/ ~'i `~"'~ ~ ~'~`'~F i ~ , ~ ~430 ~ ~ \~ \ .~"~.~.~.~. =, ~ ~ *\ . ,~ ,' , ~ ~ ~ 7 8 9 10 DISTANCE EAST OF CP-2 (miles) (b) VERTICAL 5000 cL, 4000 a, ',, 3000 < 2000 .. LL I 1 000 1 >130 Knots , . , , ~, ~ t ~ ~ ~~ ~\ _. ~ `1 1 I .! ,, ., , ; . ~ ? ~ ~ 1` ~ ~ . ~ , 1 , ~ 71 ~ ~ ~ . ~ ~ ~ ~ ? ~ ~ ,, 1 l4 `_~ i _ _ , , 7 8 9 10 DISTANCE EAST OF CP-2 (miles} (CP-2, S-Band Doppler Radar Site) FIGURE 4 Velocity Fields With Respect to the Ground,Based on a Dual Doppler Analysis for a Microburst Occurring at 1452 MDT on July 14, 1982. Contours Are Radar Reflectivity Factors(dBze). (Source: Wilson and Roberts, 1983~. 25
- ~ ~ ~ loo - I ~ ~ ~ 75 LD ~ ~ Z lo m cr ~ > Z Us > ~ X llJ 50 25 o / / 1 1 1 1 1 0 5 10 15 20 25 TIME FROM INITIAL DIVERGENCE (min) FIGURE 5 Percent of Microbursts Reaching Maximum Velocity Differential as a Function of Time from Initial Divergence. (Source: Wilson and Roberts, 1983~. Gust Fronts A gust front is the leading edge of a mass of cool air that has recently descended from a thunderstorm or convective cloud. There is a large amount of literature on this subject, which was examined for the Thunderstorm Pro ject (Byers and graham, 1949 ) and has since been much studied at the NSSL and by many researchers (Brandes, 1977; Charba, 1974; Goff, 1976, 1977; Sasaki and Baxter, 1982~. As shown in Figure 6, the cool air near the gust front, which may be up to 1 mile in depth, is characterized by strong turbulent winds. The cool air sinks, while the warm air rises. The depth of the gust front, the associated wind shear and turbulence, and its speed of advance over the ground depend on the nature of the parent cloud and the wind distribution through the layer in which the cloud is Embedded. Distortion X Z= ~ 2.4 ~~: ~ ~ : ~ 8 7 6 5 4 3 2 l V' '' 1'"~- }~ I 13 12 11 10 9 DISTANCE FROM GUST FRONT (miles) Gust Front 1 1 0 1 2 3 6,000 5,000 4,000 LLl 3,000 IL 2,000 1,000 FIGURE 6 Squall Line Thunderstorm Outf] ow ~ schematic) . (Source: Goff ~ 1980) 26
the winds are light and change little with height, the gust front is almost symmetrical around the storm that produced it. By the time the storm has dissipated, the gust front may have moved tens of miles away from the parent storm and weakened substantially. When there is strong vertical wind shear through the atmosphere and a severe long-lasting convective storm, the associated gust front tends to be maintained at the leading edge of the parent storm. The pattern of its advance can be very asymmetric with strong outward-blowing winds in those sectors coinciding in direction with the strongest winds aloft in the cloud layer. Gust fronts, as they sweep over the ground, can usually be detected by radar, by a suf f ic lent ly dense network of ground-based anemometers (such as LLWSAS ), and by microbarographs ~ Once detected , it is possible--largely through extrapolation--to predict their position anywhere from a few minutes to perhaps one-half hour in advance. Also, radar echoes often indicate the presence of deep convective clouds and the possibility of a gust front. Gravity and/or Solitary Waves As a gust front moves away from its parent s form, the temperature contrast across it is gradually reduced. The resulting circulation, when it exists in a shallow surface layer capped by a temperature inversion, can persist as a strong clear-air circulation that usually moves at speeds of 15 to 40 knots over long distances. Known variously as gravity waves or solitary waves, these motions are caused not only by gust fronts but also by dowxlslope winds and by sea-breeze fronts. Solitary waves occur most often during the night and early morning. Over northern Australia, where they have been studied extens ively, they occur throughout the year but are mos t frequent during the late winter and early spring (August to November). Corresponding climatic information for the United States is almost totally lacking. In vertical cross section, a typical solitary wave can have a horizontal dimension of several hundred feet to 6 miles with a 15-knot updraft at the leading section and a downdraft of similar strength on the trailing part. Because of the relatively long-lasting nature of solitary waves, their distinctive wind shifts, and associated pressure patterns at the ground, it should be possible to detect, track, and predict their arrival over an airport, given a surface-observing network of sufficient density. Sea-Breeze Fronts A sea breeze is a local wind that blows from sea to land. It is caused by the temperature differences that occur daily between the sea surface and the adjacent land. It usually occurs on relatively calm, sunny summer days. Often, the onset of a sea breeze occurs suddenly as a sea-breeze front, separating the cool air from the warm air, 27
moves inland. Sea-breeze fronts cause a sudden change in wind velocity, from near calm to a brisk cool breeze. At the onset, the sea breeze flows across the coastline, but as time goes on it turns to its right (in the northern hemisphere) and has a component along the coastline. Wind shear associated with sea breezes can prove to be a hazard at airports that are located along coastlines, such as Logan International Airport in Boston and the John F. Kennedy International Airport in New York City. For example, the occurrence of thunderstorm downdrafts and outflows when a sea breeze could have been expected caused some confusion about the wind velocity over the approach to runway 22 at Kennedy Airport when EAL Flight 66 crashed on June 24, 1975. In some areas, sea breezes occur regularly during the summer and can be predicted with a fair degree of accuracy. Ground-based anemometers and airborne detectors should be able to detect wind shears, so that pilots can take appropriate action. Air-Mass Fronts Separate air masses do not mix readily when they come into contact if they have different temperatures and humidities. Instead, the colder, more dense air mass passes under the warmer, less-dense air mass. The zone of transition between the two air masses is called a front. When the cold air advances, forcing the warm air to retreat and pass over the wedge of cold air, it is called a cold front. When the warm air advances, the frontal boundary moves toward the cold air and a warm front is said to exist. All fronts have some degree of wind shear across the zone of transition between the air masses, but the narrower the zone the stronger the wind shear is likely to be. When the transition zone is Perhaps 300 or more feet deep, the wind-velocity change with height is gradual and turbulence is weak or nonexistent. Nevertheless, an aircraft taking off or landing through such a zone experiences changes in wind velocity. There usually is a sharp change of wind velocity across fronts. Figure 7 shows a cross section of a warm front just south of O'Hare International Airport in Chicago. The data were recorded aboard an instrumented Boeing 747 during its approach to O'Hare's runway 32L (Sowa, 1974~. The effective tailwind-to-headwind change was 22 knots in this case. Had this landing been attempted on runway 04R (oriented 80° clockwise), the wind variability would have been 59 knots. The front had an extremely narrow transition zone and was approximately 300 feet thick. The change of wind velocity was abrupt and had moderate, or more severe, turbulence associated with it. 28
~- _Warm Front oPoo /0~/ <BY /.5 L Drift ~ ~_J_ He i= FIGURE 7 Wind Shear Across a Warm Front During a Landing Approach. (Source: Sowa, 1974) The following pilot report is typical of a flight through a frontal zone with high wind variability. It was recorded at the Madison, Wisconsin, airport. Have report on wind shear forecast for Madison. Took off on Runway 18 on 0233Z. Hit base of front at 1500 ft. Rate of climb pegged with excessive pitch-up, and moderate turbulence lasted through 5000 ft. (Sowa, 1974) 29
One U.S. air carrier has been forecasting low-altitude wind variability associated with frontal conditions since 1962 (Sowa, 1974~. The forecasts give wind direction and speed on either side of the front, tell whether the front has an abrupt or gradual transition zone, and give the intensity of turbulence, if any. This technique is now used by many airlines. The NWS provides forecasts for low-altitude wind variability for warm and cold fronts, low-altitude jet and nocturnal inversions, cold surface inversions, friction-surface slowing, inversions, and sea-breeze fronts. Weather forecasters cannot now specify the altitude of the base of the layer of strong wind shear beyond 4 hours. Techniques need further refinement to improve such forecasts. Frontal conditions and the accompanying wind shears occur in all parts of the United States, but they are most frequent over the middle latitudes during the colder months of the year. While Hawaii averages about two shears per winter, the central and northeast portions of the United States average four to five per month during the fall, winter, and spring. The southern states and those east of the Rocky Mountains average one significant frontal passage per month during the same seasons. Terrain-Induced Wind Shear Mountain terrain can cause significant low-altitude wind variability, depending on the nature of the large-scale wind field. Airports located close to mountains, near breaks in mountain ranges (known as gorges), or on hills with sharp dropoffs near the ends of runways are subject to steady-state winds that break down into chaotic gusts that are constantly changing. The presence of turbulence, often severe, can compound the problem of operating aircraft in or out of these airports. The following is a typical pilot report of this condition. It came from a Boeing 707 airplane that had just taken off from the Anchorage, Alaska, airport: Take-off on Runway 06R. Light turbulence right after lift-off. At 500 ft. turbulence changed to what I can only describe as massive bursts. The aircraft pitched, yawed, and slipped. The cockpit became a capsule of ricocheting manuals, log books, and debris. Aileron control was stop-to-stop, air- speed changes of plus or minus 50 knots. I made a slow left turn, established a slow climb attitude and, at 4000 ft. the shearing action subsided and turbulence became moderate.* *Sowa, 1977, Copyright American Institute of Aeronautics and Astronautics 30
Because of the pulsing nature of terrain-induced winds, another air- craft on the same approach several minutes later might encounter winds much different from those described above. This type of low-altitude wind variability may or may not have clouds or precipitation associated with it. The synoptic weather patterns that cause terrain-induced wind shear and turbulence are known and their occurrence can be predicted, but it still is not possible to adequately predict precise wind values at specific altitudes and locations (Sowa, 1977~. Wind shear induced by terrain occurs most of ten in late fall, winter, and early spring. It may happen only three to four times per season in some places but as often as 20 days per month in places where the terrain and the prevai ling weather patt ems are favorable . Mountain Waves Mountains induce high-amplitude undulations or waves in air currents flowing over them (Lilly, 1978; Lilly and Zipser, 1972) . They are associated with strong shears and turbulence, and their influence can extend from near the ground to very high altitudes. Mountain waves are typified by descending air over the lee side of the mountain range. The descending air has acquired distinctive names in various parts of the world--"Santa Ana" in the Los Angeles Bas in; "chinook" along the Rocky Mountains of western Canada and the United States; and "foehn" in Switzerland, Norway, and Sweden. These strong, gusty winds at the earth's surface produce low-altitude wind shear and turbulence at airports located in the lee of mountains. It is not unusual to have gust velocities double that of the steady-wind values. In extreme cases these gust velocities can exceed 100 knots. Under certain meteorological conditions, a strong temperature inversion existing near the ground restricts the downslope winds from reaching the surface. Instead, the air glides along the top of the inversion, producing wind shear and turbulence about 300 to 1,500 feet above the ground . Mountain waves can be predicted. It is not possible, however, to precisely forecast the steady-state wind velocity, the gust factor, or the turbulence intensity of such winds. These have been observed during every month of the year in Alaska and in the western mountainous regions of the United States and Canada. Mountain waves are most frequent in the fall, winter, and spring. In an average year, to the lee of the Rocky Mountains, in Montana and southern Canada, there are 15 wave days a month. In Colorado and the more southern states there is an annual average of 7 wave days a month. The mountains in the eastern United States usually do not produce strong downslope winds because their lee slopes are not particularly steep. 31
Low-Level Jet Streams The strength of the wind near the ground is tightly linked to diurnal processes in the lower atmosphere. During daytime the earth's surface is heated by the sun, and the planetary boundary layer is marked by vertical air motions. This process causes the frictional influence of the ground on the wind to be transmitted through a deep layer of air. Thus, wind velocities near the ground tend to be relatively high in the form of a concentrated current called a low-level jet stream. The formation of such a jet stream depends also on the distributions of heating and cooling and their daily variations over sloping terrains (see, for example, McNider and Pielke, 1981~. Figure 8 shows a low-level jet stream observed at the NSSL in Norman, Oklahoma. 5000 4000 3000 LL up lo 2000 1 ,000 o 57 I I I I I I I ~ I I I Isol ines of Horizontal 54 ' _ Wind Soeed in ft/sec ~ ) j j it 1(~( <~ ~J~J~ V~ ]'~' ~ 271 267 1200 27 18002724 240024 24 21 0600 1200 June 26 CENTRAL STANDARD TIME June 27, 1965 FIGURE 8 Low-Altitude Jet Stream Near Norman, Oklahoma, Observed by Means of Doppler Radar. (Source: Lhermitte, 1966~. In a typical low-altitude jet stream situated over an airport, the wind at the surface tends to be light and to come from the same direction as the stronger flow immediately above the airport. Consequently, an aircraft that is landing will typically approach the runway into the jet stream wind. As the aircraft descends below the jet stream, headwinds decrease, often substantially, as the aircraft nears touchdown. The sudden loss of headwind can be a serious problem if the pilot is unaware of the situation. In a typical low-altitude jet stream situation, as described above, the event occurs often in clear air but at night when the visual perspective of the pilot may be inhibited. 32
Tornadoe s No discuss ion of low-altitude wind shear would be complete without at least a mention of tornadoes and other high-speed atmospheric vortices, such as waterspouts and dust devils. Most often dust devil s, o f the type commonly seen over desert areas during the hot, dry Sumner months, are too small to pose serious risks. Because of the very s bong wind shears that characteriz e them, tornadoes should always be avoided. Most of the time they can be seen by pilots flying below the bases of the parent thunderstorms. An appropriate radar should be capable of detecting existing and incipient tornadoes in sufficient time to allow pilots to fly around them. Ground-Based Sensing of Low-Altitude Wind Shear Low-Level Wind Shear Alert System Following several airline crashes during the mid-1970s, the FAA developed the Low-Level Wind Shear Alert System (LLWSAS) (Goff, 1980~. Initiated by NOAA's NSSL, the system consists of an array of wind-velocity measuring ins truments located on the ground at or near an airport . * LLWSAS typically consists of a centerfield wind sensor and S outlying sens or s normal ly about 2 miles from the center s ite, located on the basis of meteorological factors, terrain considerations, logistical constraints, and to favor the Instrument Landing System (ILS). The sensors have propeller vanes on standards that rise about 10 to 60 feet above the ground as necessary to obtain clear air flow above terrain or other obstructions. Each site is polled once every 10 seconds. The centerfield site is considered a reference site, for which a 2-minute running average of wind velocity is maintained. LLWSAS is controlled by a central miniprocessor (usually located in the control tower), which maintains the 2-minute running average of the centerfield wind. This information is continuously displayed in the tower, is used by controllers, and is relayed to pilots. In addition, once every 10 seconds, the miniprocessor compares the 10-second wind at outlying sites to the 2-minute average at the center site. A vector dif ference computation is made and, i f a 15-knot threshold is reached or exceeded, an alert is given to tower *The FAA estimates the cost of the present LLWSAS system to be approximately $200,000 for each airport installation. 33
controllers. The computer normally displays only the centerfield average wind velocity, plus a gust factor,* if appropriate. But ire the wind shear threshold is exceeded, the wind velocity at the appropriate peripheral anemometer also is displayed. Controllers may, however, choose to display any or all sectors at one time. The wind-shear calculation is des igned to detect the sudden onset of a gus t front at an outlying site by comparing the wind discrepancy at the outlying site with that at the centerfield site. The center- field site, because of its long averaging period, cannot effectively detect wind shears. LLWSAS has been installed at 59 airports and is scheduled to be ins tailed at 51 others by 1985 ~ Table 2 ~ . LLWSAS data were recorded during the JAWS Project at Stapleton International Airport in Denver in 1982 but are not usual ly recorded at other times or at other airports. LLWSAS is the only operational means currently in use for detecting wind shear. Nevertheless, the system has several limitations. These include the following: o The system cannot measure winds above the sensors on the ground. This may not be a big problem for gust fronts or sea breezes, but it limits the detection of wind shear that may not be present at the surface. o There are temporal and spatial resolution limitations that may present serious problems for detecting the smallest-scale events. Although LLWSAS sensors are located an average of about 2 miles between the centerfield and remote sites, the effect ive wind-shear resolution is near 4 miles because of the long averaging period at the centerfield site. Likewise, the temporal resolution is compromised by the long averaging at the centerfield site; a brief high-wind encounter at center geld would probably not be identified . This ef fectively eliminates the centerfield site as a high-resolution wind-shear sensor. o Surface wind events outside of the perimeter of the anemometer field would not be detected. O Vertical wind motions are not sensed directly; only horizontal ones are detected, and these, of course, may have been inn' fated by downdrafts . 0 Sensors do not directly measure wind along flight paths and are thus susceptible to reporting events that may not reflect wind shear or lack of one on an airplane's flight path. PA ~ ~st factor is also calculated and displayed for the centerfield site, this factor being the difference between the peak and mean values of the wind velocity during the 2-minute averaging period. 34
During an 86-day observation period of the JAWS Project, the LLWSAS data at Stapleton International Airport were recorded. Figure 9 shows that there were nearly 4,000 triggering wind-shear events (defined as a 10-second sample with a vector wind-velocity difference equal to or greater than 15 knots, between any remote site and the centerfield site). Once a trigger occurs, the wind-shear warning light remains on for three consecutive samples, or about 30 seconds or longer if another stronger shear occurs. Hence, multiple triggers may be seen as a single LLWSAS alarm. 107 ~ 1 1 1 1 1 1 1 1 _ 1o6 an LL c: Z 105 cat o LU > cut 104 103 1o2 1o1 10° _ ~ & N l I I I I I I I I I ~ 0 10 20 30 40 50 WIND VECTOR DIFFERENCE (knots) FIGURE 9 Cumulative Occurrences of Triggering Wind-Shear Events for All LLWSAS Remote Stations During JAWS. (Source: McCarthy et al., 1983~. During the JAWS Project, 466 alarms were detected by LLWSAS, distributed by day as shown on the right side of Figure LO. In comparison, Fujita (private communication) tallied the number of microburst events that occurred during the same period within 8 miles of Stapleton Airport detected by the NCAR's PAM* system, as shown on *PAM (portable automated mesonet) automatically measures the speed and direction of the wind at each of its sensors 12 feet above the ground. Because the sensors were located over a wide area during the JAWS Project, PAM was able to measure the wind speed and direction on opposite sides of microbursts occurring within its field and thus provide evidence of the existence of the microburst. 35
the left side of Figure 10. Notice that although the microburst frequency was high on certain days, comparison with the LLWSAS identification of comparable wind-shear events indicates an erratic correlation. For example, on some days, notably July 14, 15, and 18, both counts seem comparable. However, on other days, particularly May 19 and June 29 when a number of microbursts occurred, there was no clear identification of a wind-shear event on the LLWSAS. On yet another type of day, the LLWSAS shows many wind-shear alerts with no corresponding microburst indications, as seen on June 5, 22, and 30; July 8; and August 1 and 7. An examination of this latter set of days by Bedard et al. (1983) indicates that a number of LLWSAS alarms were associated with weak gustiness situations that barely reached the 15-knot vector difference threshold. It must be clearly understood that the microburst daily frequency statistics shown on the left side of Figure lO do not include other wind-shear events, such as gust fronts. Consequently, the data presented here illustrate only the inadequacy of the LLWSAS as a microburst detector. MICROBURSTS FROM PAM DATA 1 MAY JUNE JULY AUGUST o ~ ~ l 20 25 , BO , 5 10 15 20 2S i 10 IS 20 2S 50 _ LLWSAS ALARMS ~ o 10 20 30 40 -- 1 1 1 In' . FIGURE 10 Daily Frequency of Microbursts within Eight Miles of the Stapleton Runways and the Daily Count of LLWSAS Alarms During JAWS. (Source: Fujita, private communication). 36
From the JAWS LLWSAS analys is and from a more gener ic s tudy o f the system, it is concluded that the LLWSAS system, as presently configured with large sensor spacing, performs adequately as a detector of larger-scale wind-shear phenomena occurring near the earth' s surface, such as fronts, thunderstorm gust fronts, sea breezes, certain orographic situations, and some larger-scale downbursts. LLWSAS will also detect the wind shear caused by a microburst if it occurs in the vicinity of the sensors. However, there is substantial evidence that because of the small horizontal scale of microbursts the present operational LLWSAS system does not detect some microbursts that occur between or beyond the sensors. Furthermore, the system detects wind- shear events that, on careful inspection, are found not to represent a hazard to aircraft. The JAWS analysis makes it clear that substantial improvements to the LLWSAS, such as increas ing the sensor density and improving individual sensor responsiveness, would significantly enhance the LLWSAS's capability to detect a broader range of hazardous low- altitude wind-shear events near the earth's surface. An improved LLWSAS system is being developed for installation at New Orleans International Airport. This upgraded system, to be operationally tested in early 1984, should provide the basis for modification of current LLWSAS installations and for improved system performance for future installations. Pressure Sens or s It has long been known that active thunderstorms sometimes cause the atmo sphere to undergo distinctive pressure changes related to the vertical motions of the air in the storms. Also, as the cool out- rushing air in the gust front moves away from the storm that produced it, a "pressure jump" is co ~ only seen in association with wind shifts associated wi th the gus t front . Several tests have been conducted on a system of sensors designed to detect changes in atmospheric pressure. At Dulles International Airport, outside Washington, D.C., a network of microbarographs was coupled with ground-based anemometers to detect low-altitude wind shear. The microbarographs were found to give warnings up to 3 minutes (see Figure 11) earlier than would anemometers alone (Bedard et al., 1979). A 3-month test at the Hartsfield International Airport in Atlanta showed that pressure and wind sensors could be successfully integrated into a single system. During this brief test, the downdraft air in several small microbursts impacted the ground between the sensors, and the resulting gust fronts passed over the sensor array without triggering the pressure sensors. The Atlanta investigations did not adequately test the effectiveness of a combined wind-pressure sensor system for the detection of low-altitude wind shear. 37
110 (A 1 00 In 90 c: 80 111 LU cat en As 111 6 to , O ) _ 70 _ 60 _ 50 _ 40 30 20 -200 -150 - 100 -50 0 50 100 150 200 Microbarograph Lags ~ I| Micro barograph Leads Tl ME I N SECONDS O - _ i A A A Too} A Oo° 80 80 0 0 oSo80 00 0 ~ O O O O - 10 1 . I A A FIGURE 11 Relative Arrival Times of Pressure Jumps and Gust Surges. (Source: Bedard et al., 1979~. Ground-Based Microwave Radar Rain showers and thunderstorms, common sources of hazardous low-altitude wind shear, are routinely detected by conventional radars of the type operated by the NWS and the FAA. The radar echo intensity and its dimensions, shape, and motion can be used to judge the likelihood of wind shear, although not with great certainty. For traffic controllers to use such information, the data must be displayed to them in an appropriate and timely fashion. When mature, intense thunderstorms with heavy precipitation, which causes strong radar echoes, occur over or near an airport, hazardous 10w-altitude wind shear and turbulence are likely~to be present. In such cases, suspension of takeoffs and landings should be considered. Air traffic controllers do not have direct access to the information from the network of weather radars operated by the NWS, but they do receive weather advisories from the NWS. It needs to be reiterated, however, that hazardous low-altitude wind shear often occurs without precipitation. The absence of a radar echo over an airport does not mean that wind shear does not exist. Nor does the presence of an echo guarantee that wind shear is present. 38
Microwave Doppler Radar Pulsed Doppler radar can effectively detect low-altitude wind variability (see Figure 4~. Research at the NSSL, the NCAR, and the University of Chicago (JAWS Project), as well as field experiments, have demonstrated that an appropriately designed, pulsed Doppler radar can detect tornadoes, downbursts, microbursts, gust fronts, solitary waves, and sea-breeze fronts. It can also identify the boundaries between distinct air masses (e.g., Wilson et al., 1980~. Wind shear has been observed by Doppler radar in many regions of the United States and during all seasons. Highly sensitive research radars have detected echoes from clear ~ to the human eye) air as well as from precipitation. Clear-air echoes are commonly detected in moist atmospheric boundary layers in spring, summer, and autumn but are rare in cold, dry, winter air. These discrepancies are usually explained in terms of differences in the refractive index inhomogeneities of air, quantities that depend on the distributions of temperature and water vapor in the air. Moreover, insects sometimes produce clear-air echoes, especially during warm months. Since most wind shears, particularly those associated with convective storms, occur during the warm months, the wintertime clear-air performance limitation does not appear to be a major problem. A Doppler radar measures directly the velocity of the wind. toward or away from the radar (parallel to the radar beam). As a consequence, cross-beam velocities are not measured directly and vertical velocities cannot be measured directly, except when the antenna is pointing vertically. In research programs, several Doppler radars have been used together to examine the same scattering volume from different aspects. From such spatially independent measurements, it is possible to reconstruct a complete three-dimensional vector wind field. Figure 4 shows an analysis of a microburst based on the use of multiple Doppler radars. Because of their cost, multiple Doppler radar systems are unlikely to become operational at airports in the foreseeable future. Much has been done with single Doppler radar to deal with the problem of wind shear. In the case of features possessing certain symmetries, such as small-scale cyclones and some microbursts, the Doppler velocity "signature" is sufficiently independent of viewing angle to allow operators to infer the cross-beam components from the spatial pattern o f the radial velocity alone. Shears associated with gusts and other frontal systems can be measured accurately using the gradients of radial velocity in the vicinity of the front along with cant iguous radial velocity measurements on both sides of the front . Thus, a s ingle Doppler radar located away from an airport can ef fectively detect and quantify many hazardous phenomena over the a irport . 39
But most microbursts are not symmetrical, because air descending toward the earth's surface carries the velocity characteristics of winds aloft. In such cases an off-airport radar might easily miss a hazardous case or overestimate its actual threat. In these cases it would be best to have airport radar with beams parallel or nearly parallel to the runways. Alternatively, it is important that the space over the airport be carefully monitored, since precursors of microbursts on and near the ground are most 1 ikely to be found through the examination of Doppler signatures aloft. These can most effectively be monitored from an of f-airport s ite. Next Generation Radar (NEXRAD ~ The FAA, NOAA, and DoD are working together to develop an advanced pulsed Doppler weather radar. Referred to as the next generation radar, or NEXRAD, it will address the common need among the principal government users of meteorological data for information on the current location, severity, and movement of such weather phenomena as tornadoes, severe thunderstorms, heavy precipitation, tropical cyclones, hail, high wind shears, and severe turbulence. Each NEXRAD installation will provide weather information for ranges exceeding 200 miles and heights of 60,000 feet at a rate of once every 3.5 minutes. Because of the large number of meteorological phenomena, the size of the coverage volume, and the magnitude of the resultant data rate, NEXRAD, whether manned or unmanned, will rely heavily on automation to perform the required observations and to communicate the results to users. Several factors make the NEXRAD system an unlikely choice to measure low-altitude wind shear over airport terminals. One serious consideration is the minimum scan update time of 3.5 minutes for the NEXRAD network. Such a rate is too slow to optimally detect microbursts. Radar siting criteria also makes NEXRAD's use unlikely in terminal areas. NEXRAD will require an unobstructed view for 200 miles. In addition, the height of the NEXRAD antenna and tower will force some NEXRAD systems to be located sufficiently far away from some airports as to eliminate any possibility for low-altitude coverage of the terminal area. However, because NEXRAD can outperform and suppress ground clutter better than the radars used in research to detect microbursts, a stand-alone NEXRAD devoted strictly to an airport's terminal area might be useful. However, this would be difficult to accomplish because of frequency overcrowding. NEXRAD uses a 2700- to 2900-MHz frequency band, which is used by many other radar systems in many areas of the continental United States. NEXRAD's frequency will require additional design or operational measures to be employed in some areas. Adding approximately 100 NEXRAD systems for terminal use woul d, there f ore, appear impractical. 40
A Terminal Radar for Wind-Shear Detection The FAA has proposed that an unmanned Doppler radar be dedicated to the effective detection of low-altitude wind shear and the reduction of hazards to aviation. The committee concludes that the concept of a terminal radar can be justified and that FAA should aggressively support its development. But such a program must not interfere with the development of NEXRAD. To effectively reduce flight hazards over airports, radar observations are necessary over and in the vicinity of airports. Information from a terminal radar and other sources must be communicated to both pilots and air traf fic controllers in a timely, concise, reliable, and easily understood manner . If a Doppler weather radar is to serve airport terminal operations, it must detect and warn of the most serious aviation hazards with a high reliability. An analysis of wind-shear-related accidents shows that the turbulent airflows associated with convective clouds and thunderstorms represent major hazards. The design of a Doppler radar system must take into account the nature of microbursts, specifically their small dimensions, short durations , and the fact that the s tronges t winds may be at altitudes of a few hundred feet. Furthermore, such a radar should also be able to detect the other types of shears described earlier. A terminal radar must be able to measure low-altitude wind shear. It must be able to encompass a wide range of velocities and radar reflectivities in both clear and cloudy air, to eliminate or greatly reduce ground clutter, and to observe phenomena located very near the ground. It will also need a resolution capability of approximately 500 feet to enable it to identify a downdraft and an antenna that scans the airspace surrounding the terminal every 1 to 2 minutes. Although the scan strategy depends on the location of the radar relative to the airport, scans at approximately 6 to 8 elevation angles are necessary to properly monitor the airspace. Since each Doppler radar scan produces three fields of data-ore fleet ivity, mean velocity, and spec trum variance--the average time available to scan and analyze one field varies from 2.5 to 6.7 seconds. Microbursts, downbursts, and gust fronts can be reliably detected in research programs us ing manual techniques and highly skilled observers. There is considerable uncertainty, however, that it would be possible in day-to-day operations to extract and communicate wind-shear informal ion in a timely fashion us ing manual detection techniques at the very high scan rates that appear to be necess ary . A high degree of automation is needed if a terminal radar is to be effective given the very short lifetimes of some microbursts and the even shorter time available for making decisions. At present, 41
however, there are no automated techniques for detecting lo~altitude wind shear. For an unmanned Doppler weather radar to be viable, a systematic s tudy must be made of algorithms needed to detect wind shear for representative climatically different areas of the country. System Design Considerations. Ground clutter presents a significant problem to using data from radars used to observe meteorological targets near the ground. All available measures will be needed to mitigate clutter in an airport's vicinity. The combination of a low sidelobe antenna design and electronic cancellation should be able to reduce clutter by at least 40 dB compared with existing research radars . A microwave Doppler radar operating at a 10-centimeter wavelength proposed for NEXRAD can ef fictively measure wind shear around an airport. What is not known is whether a 10-centimeter wavelength is best for Doppler radars used at airport terminals. A 10-centimeter wavelength is needed for NEXRAD in order to "see through" severe storms and heavy prec ip itat ion to measure re f lee t ivi ty at long ranges. This is not necessary for terminal radars that need to detect s ignals out to distances of only 20-30 mi les . A narrow beam (approximately 1°) requires an antenna about 25 feet in diameter at a 10-centimeter wavelength. The antenna diameter needed for a fixed beamwidth is directly proportional to wavelength. There fore, a 1° beamwidth at 5- or 3-centimeter wave lengths can be achieved with antenna diameters of 13 and 8 feet, respectively. A small antenna costs less and is more easily sited on a busy airport than a large one. On the other hand, a larger antenna and shorter wavelength would provide a narrower beam and improved ground-clutter red ect ion . There is a strong possibility that the signal-to-ground clutter ratio can be increased further by using shorter wavelengths. This is true when the objects that scatter the radar beam are water droplets or insects in the clear air. In this case the power backscattered by small particles is inversely proportional to the fourth power of wavelength, while the ground-clutter return is only weakly dependent on wavelength. In determining the most appropriate wavelength for detecting low-altitude wind shear, it is necessary to consider inhomogeneities in the atmosphere that may be caused by water vapor and temperature variations, because observing dry microburs ts depends on detecting such radar targets. Range and Velocity Ambiguities. Because there is very little time to make decisions, the processing and signaling scheme must eliminate range and velocity ambiguities for airport radars. Many techniques 42
address this problem, but none will be described in detail here. Shorter wavelengths tend to make this problem more severe; however, the re latively short range of the radar somewhat counteracts this effect. Therefore, range and velocity ambiguity considerations alone should not rule out the use of 5- or 3-centimeter wavelengths. Siting of Doppler Radar Antenna. Ideally, a Doppler radar antenna should be located such that it can measure wind velocities along the takeof f and departure corridors, including curved approach paths that will be permitted by the microwave landing system (MLS) in the near future. For a variety of reasons, such a goal is virtually impossible to achieve because major airports usually have a number of runways with different orientations. Airport buildings complicate this problem. The question is further complicated by the nature of wind shear. Some experts have concluded that radar antenna should be installed on airport grounds, while others think it would be best to have it off the airport. Studies are needed at each airport where a terminal weather radar is to be installed to ascertain the best location f or the antenna. Remote Sens ing by Means of FM-COO Radar Some experts have suggested that a frequency-modulated, continuous-wave (FM-OW) Doppler radar might be used instead of a pulsed Doppler radar to detect low-altitude wind shear. FM-COO Doppler radar has many of the characteristics of the pulsed Doppler radars described in the preceding section. For the same average power, antenna aperture, and bandwidth, both types have the same sensitivity and resolut ion. The key di f ference between them is the t a pulsed Doppler radar transmits high-power, low duty-cycle pulses, while an FM-COO Doppler radar transmits a low-power, full duty-cycle signal with linear frequency modulation. Because of this, an FM-COO Doppler radar can transmit high average power over a broad signal bandwidth and there fore achieve high sensitivity and resolution at short ranges . The high resolution contributes to significant ground-clutter rejection. Wind measurements have been made at ranges as short as 40 feet. An FM-COO clear-air radar might also locate wingtip vortices, another type of low altitude wind variability that is potentially hazardous to small aircraft. Research should be continued on the use of FM-COO radar for wind shear and turbulence detection. Longer-Wave leng th Profiling Radar Vertically pointing Doppler radars can measure a vertical profile of the wind above the radar throughout the troposphere when operating at long wavelengths of 1-10 meters. The measurements can be made at a rapid rate. These radars can measure winds above the earth's boundary layer better than microwave systems because of the nature of clear-air turbulence. On the other hand, a VHF or UHF band radar cannot ~3
effectively detect microburs ts because ache beam is not eas fly scanned and the beamwidth will tend to be wide. With adequate clutter cancellation, however, such a radar could be a superb tool for measuring vertical shear of the horizontal wind in all parts of th country and during al 1 seasons . The NWS is inves ligating the use of VHF or UHF wind profilers to supplement its rawinsonde network. Such systems, if equipped with ground-clutter cancellers designed to measure the lowest 3,000 feet of the atmosphere, should yield useful data on the vertical shear of the horizontal wind at or near airports . It does not appear, however, that such systems are necessary components of a low-altitude wind-shear detection system. Doppl er Lidar Lidar (light detection and ranging) is the optical equivalent of microwave radar. A suitably designed pulsed Doppler lidar operating at a wavelength of about 10 micrometers can measure winds in optically clear air during al 1 seasons by detecting the backscattered power from small atmospheric particles. The principal advantage of lidar is its ability to make clear-air measurements that are virtually free from ground-clutter contamination. A disadvantage is that the laser signal can be absorbed by rain, clouds, and fog, which can strongly attenuate the signal over ranges of 1 to 2 miles. Doppler lidar should be investigated as a technique for detecting wind shear, especially in the western United States, where many clear-air microbursts occur. In the JAWS Project, an infrared Doppler lidar detected and measured some microbursts more clearly than did the radar. Acous tic Sounding Doppler acoustic sounders can measure profiles of the horizontal wind in the boundary layer. The beams cannot be steered easily, however, and the signals can be severely contaminated by natural and man-made audio noises Using such a system at airports Is, therefore, most appropriate for detecting areawide wind shear, such as that associated with low-level jet streams. Such sounders are already in use at the Hong Kong and Calgary, Canada, airports . Airborne Remote Sensing of Wind Shear Microwave Doppl er Radar An airborne Doppler radar would be an effective tool for detecting and warning of hazardous wind shears if it had capabilities equivalent to those available in ground-based systems. Unfortunately, because of size, space, and power limitations, the Doppler radars now under development for airborne commercial use appear to be insuf ficiently sensitive and provide inadequate spatial resolution to address the low-altitude wind-shear problem. 44
Some airborne Doppler radars used in research programs where space is not a problem (e.g., that operated by NOAA on its Lockheed P-3 aircraft and NASA on its Skyvan aircraft) offer capabilities similar to those of ground-based systems. It is beyond the scope of this study to determine whether a suitable airborne system can be developed. However, a thorough study should be conducted to determine the feasibility of such a development. If the results are positive, a prototype radar should be developed and tested. Such a prototype might have to use a wavelength between 1.5 and 3 centimeters, a flat plate antenna to achieve very low antenna sidelobes, a ground-clutter canceller, and substantial transmitter power (probably between 25 and 100 kilowatts of peak power). Doppl er L idar The British Royal Aircraft Establishment and the Royal Signals and Radar Es tab lishment have collaborated to develop an airborne cant inuous-wave (COO) Doppler lidar that is f town on a smal 1 je t aircraft (HS 125 ~ and was tested during the JAWS Pro ject. The system is focused to measure the wind at a range of approximately 1000 feet ahead of the aircraft. The system has been flown reliably for extended periods of time, is rugged and lightweight, and requires minimal main- tenance and calibrations. Pulsed lidar technology should be explored to see if a practical system can be developed to sense wind shear an order of magnitude farther ahead, say to 2 miles . The CW system can measure velocities accurately but only in the region o f focus . Combined with the true airspeed measurement on the aircraft, the system can warn of impending shear about 4 seconds in advance o f an encounter during approach or departure. These short CW lidar warnings are not unambiguous because of the lack of spatially contiguous data but are certainly of value i f for no other purpose than to alert pilots when to abort landings . On takeof f, ~ warning after the aircraft is airborne may be of little value. As with other optical systems, airborne lidar is severely attenuated by clouds and precipitation. Such systems deserve careful scrutiny for use by the aviation industry but at the present time remain in the research ca tegory . Passive Infrared Radiometry NASA has been conducting test flights with a passive IR radiometer. Devices of this type detect the IR radiation emitted by gases and particles and thus detect temperature differences in the field o f view. Gus t fronts are accompanied by a decrease in temperature, and according to NASA an IR radiometer can give as much as a 70-second warning of the arrival of a gus t front. An IR radiometer is inexpensive, small, lightweight, and easily installed and flown on aircraft. 45
Unfortunately, there is little correlation between measured temperature differences and the occurrence of wind shear caused by microbursts. In some instances, temperature increases; in others, it decreases; and in yet others, no temperature change occurs. Even for gust fronts there is no s bong correlation between the degree of temperature change and the intensity of the wi nd shear resulting from it. The IR radar beam can also be absorbed in clouds, fog, and rain, causing ambiguous interpretations. Thus, IR radiometry cannot be recommended as a primary sensor for wind shear, but it deserves add it tonal re s earch . Interpretation and Communication to Air Traffic Controllers and Pilots The previous sections have described various types of wind shear and a variety of sensing techniques be used to detect them. Equally if not more important to the overall aviation system is the need to successfully interpret and effectively communicate wind-shear data to air traffic controllers and pilots. The detection of low-altitude wind shear requires application of sophisticated technology. Unless this technology produces a useful product, its ultimate value will fall short of expectations. A useful product is one that is easily interpreted without ambiguity, provides definitive and quantitative information on hazards, and does not have a large number of false alarms. Any detection system must meet these criteria or its usefulness will be limited at best. An effective warning system has several essential elements: (1) sensing, (2) computation, (3) display, (4) interpretation, and (5) dissemination. Microbursts impose particularly severe requirements on such systems because of their short lifetimes and small size. All too frequently, plans for detection and warning systems place too much emphasis on basic sensors and too little on the other ingredients, which are equally important. Indeed, if any of the links is missing, the system will be ineffective. For example, a Doppler radar is a sensor, an essential ingredient for detecting wind shears. Computation is also essential and in a Doppler radar system would provide the signal and data processor to compute the fields of velocity and various other derived information. In our era of high-speed integrated circuits and rapidly advancing technology in computers, computation is the strongest link in this chain. Today' s computational systems are fast, inexpensive, and accessible. The principal ques Lion is what computations are necessary to accomplish the desired result. 46
Display and interpretation do not necessarily occur in that order. Automated sys tems perform the interpretation in machines and then display the result. Interpretation by humans requires, first, that the data fields be displayed in a fashion suitable for viewing. Rapid and reliable interpretation is the most challenging task in a microburst detection and warning system. Automation is highly desirable because machines are fast, vigilant, and promise low operational costs. Unfortunately, machines are far inferior to humans in pattern recognition, which is crucial to an ef fective system. Machines also do only what they are programmed to do. An effective detection and warning system must make the best use of machines and people. Machines mus t be used for rapid real-time calculations and image interpretation. Display of hazardous situations mus t be provided in relatively simple and unambiguous formats for controllers and other users. Meteorologists in Center Weather Service Units (CWSUs ~ should also be able to view the data f ields from which automated interpretat ions have been made in order to resolve uncertainties that surely will occur in real time. With such a process, false alarms will be minimized while retaining a high probability for accurate detection of hazardous conditions. Dissemination is the last link. Dissemination of information in various forms is obvious ly involved throughout the chain, but mos t important are the links from controllers to pilots or from machines to controllers and pilots. Techniques exist to transmit virtually limitless data to users, but this capability alone is no more than a tool. Effective dissemination will occur only if technologists, meteorologists, controllers, and pilots work together to design the final products that are to be disseminated. Air traffic controllers and pilots are very busy and cannot be burdened with excess ive data. Neither can they afford to work with data that are insuff icient for them to take effective corrective action in real time. The system must provide just enough and no more. There also will be a good deal of training needed. Users must be well schooled in what the phenomenon is, what the system is trying to do, why it was designed as it was, and what its limitations are. It is only through a s bong education program for all operations and flight staff that a system will realize its full potential. In summary, the system's developers must give serious attention to the interpretation and communication of information needed by the user. In fact, this link is as important and as difficult as the development of basic sensors. 47
Wind- She ar Pred i c t ion The discussion to this point has concentrated on the nature of low-altitude wind shear and its detection. I t should be clear, given the highly localized and transitory nature of most wind shears and the speed with which they develop, that effective monitoring techniques must play a significant role in providing for aircraft safety. It is also important to develop the capability to forecast weather conditions and systems that are conducive to generating wind shear. Such forecas ts can in turn be used to develop a corresponding hierarchy of wind-shear alerts. Forecasting strategies need to be developed us ing techniques that identi By precursor condit ions to microbursts, rather than the more rigorous techniques of numerical weather prediction. As noted earlier, some types of low-altitude wind shear (e.g., shears caused by fronts, low-level jet streams, and sea breezes) can be accurately predicted hours in advance, but this is not the case with other types of shear. The hazardous shears caused by the downdrafts associated with convective clouds cannot yet be predicted successfully by either numerical or operational techniques. However, in the latter area, some researchers have made limited progress in using characteristic signatures of radiosonde temperature and humidity profiles to identify days with a high potential for the occurrence of microbursts (e.~., Caracena et al., 1983~. Others have developed a model of downburst formation that, if verified, could provide some basis for shorter-term microburst predictions (Emanuel, 1981~. The primary objective of operational prediction is to make possible the warning of conditions conducive to generating microbursts or severe downdrafts, modeled after the severe thunderstorm or tornado watch currently produced by the NWS. Such a wind-shear watch would provide controllers and pilots with an early indication that conditions are prevalent for hazardous low-altitude wind shears. Because such conditions are relatively rare, such a watch would be unusual and would heighten awareness in the several hours before a possible encounter. Finally, the watch concept could be implemented by the NWS. National Weather Service Interaction with the Federal Aviation Administration The NWS provides a wide variety of forecast services, including those provided to the FAA and to aviation community. It has often been stated that the ability to forecast any meteorological phenomenon is directly related to the ability to observe it. Development of an adequate capability to observe low-altitude wind shear still requires a lot of work . 48
Thunderstorms present the greatest danger to aircraft. Along with the usual hazards to aircraft, thunderstorms also produce low-altitude wind surges or gust fronts that move out ahead of the storm. Dowobursts and microbursts are closely associated with the gust fronts but on a smaller temporal and spatial scale. Radar and satellite pictures are used mainly to detect thunderstorms . Unfortunately, we are not yet able to observe with any degree of consistency which thunderstorm cells have significant low-altitude wind shear associated with them. Meteorologists can predict wind shear associated with cold and warm fronts and wi th low-level jet streams with greater accuracy than those induced by convective activity. However, the low frequency and wide spacing of upper-air observations inhibits the meteorological monitoring of the parameters that cause the events. The turbulence portion of the Aviation Area Forecast indicates the likelihood of low-altitude wind shear. The forecast is prepared three times each day by the National Aviation Weather Advisory Unit in Kansas. In addition, 52 Weather Service Forecast offices issue, at the same frequency, 498 site-specific terminal forecasts, including the likelihood of low-altitude wind shear. These aviation/weather notices are transmitted to FAA and NWS of fices as well as to those in the aviation community having the appropriate communications equipment. Progress has been slow in the development of techniques for forecasting low-altitude wind shear associated with convective clouds and thunderstorms. Thunderstorms are assumed to have the potential for Great ing local titude wind shear, but only on some days do convective clouds cause hazardous wind shears . Recent research at the NOAA's ERLs indicates some skill in forecasting low-altitude wind shear caused by convective clouds. Potential hazards can also be identified by certain precipitation radar echoes or satellite cloud images known to be associated with low-altitude wind shear. These short-duration (less than 2 hours) features are useful in specifying areas where hazards exist, but they do not yield the type of specific indications of low-altitude wind shear needed to plan flights. The National Severe Storms Forecast Center in Kansas City issues daily forecasts of thunderstorms that may approach severe limits. Severe-thunderstorm watches are then issued as necessary. However, as recent investigations in the JAWS Project and analyses of the Pan American Airlines crash in New Orleans have shown, accidents can occur in storms that are not considered severe. Dissemination of Weather Advisories While disseminating forecasts of the more persistent nonconnective wind shears can be handled adequately, it is the convective low-altitude shears that pose the greatest danger to aircraft. An estimated 90 percent of significant operational low-altitude wind 49
shears are connectively induced (NWS, 1982 ~ . frequently dealt with after the fact. These ins Lances are While weather forecasts and in-flight advisories admonish that forecas ts of thunderstorms imply the existence of low-altitude wind shears, pilot reports (PIREPs ~ are the most effective warnings of the actual existence of shear. In fact, until more definitive airborne and remote-sensing systems are instituted, PIREPs may be the single most important safety item in identifying most hazards to aircraft operations. Thus, aircraft encounters with severe low-altitude wind shear should be reported in a timely and accurate manner. Current reporting procedures are considering how much Current cumbersome and time-consuming, particularly pilots have to do during takeoff or landing. _ - Compounding the problem, many PIREPs are not disseminated beyond the air traf fic controller receiving them. There are several weaknesses in the pilot reporting system that mus t be improved before pilots can depend on being consistently advised by other pilots of the existence of low-altitude wind variations. The NWS and the FAA currently have several arrangements for communicating and disseminating weather information within the National Airspace System. These arrangements need to be examined and improved. One shortcoming Is that most ~ntormat~on ~ s communicated via telephone. The FAA is currently developing equipment that will enable meteorologic ts at its CWSUs to transmit weather messages automatically to towers and control facilities within any center 's area of responsibility. 50