Wind and the Built Environment: U.S. Needs in Wind Engineering and Hazard Mitigation (1993)

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.

2 The Nature of Wind A WIND PRIMER Wind is air in motion relative to the earth. It is, in general, three dimensional, with both horizontal and vertical components. However, the vertical component normally is small, so the term wind is usually taken to refer only to the horizontal component. Winds near the earth's surface—as opposed to upper-air flows—impinge most directly on human activities. The prevailing wind direction at a site may suggest the orientation of a building to enhance natural ventilation and prevent the development of undesirable winds at street level. Strong wind gusts buffet buildings and can lead to occupant discomfort and eventual fatigue of building components. A knowledge of probable, extreme wind speeds is required so that buildings may be designed to resist these wind effects. Wind direction information is used to select runways for aircraft takeoffs and landings. To ensure that chemicals are not missing the target field, farmers must consider both strength and direction of the wind before spraying crops. Provided they are not too strong, winds also represent an energy source for windmills and wind generators. The Near-Surface Wind Near-surface wind is the most variable of all meteorological elements. Due to friction, the wind speed always vanishes at the ground. Approximately 50 percent of the transition in wind speed due to frictional effects takes place in the first 6 ft (2 m) above the surface. The adjustment in wind direction takes place over a 1- to 2-km depth (0.6 to 1.2 miles). These adjustments give rise to the wind profile; the profile of wind speed in particular must be taken into account when tall buildings or stacks are being designed. At elevations greater than ~ or 2 km (0.6 or l.2 miles) above the surface, the wind on most days wit] be more or less uniform in height and will blow steadily at moderate speeds (~20 knots or 12 - 23 mph; ~ knot = 1.15 mph = 0.5 m/s). Wind at these elevations is usually almost normal to the local pressure gradient (with lower atmospheric pressure to the left in the Northern Hemisphere) and so tends to flow parallel to the height contours of isobaric surfaces. There are narrow bands of high-speed air very high in the atmosphere, at 30,000 to 40,000 ft. These jet-stream winds can have speeds up to 200 mph (90 m/s). Under most conditions, wind varies continuously in the near-surface layer. However, it usually can be expressed by a superposition of high- frequency oscillations of small amplitude in both speed and direction around a much more slowly varying sustained speed and a prevailing direction, 19

20 Hued and the Built Environment respectively. Both high- and low-frequency variations play an important role in mixing aIld dispersing pollutants in the atmosphere. As wind speeds increase, some of the variations can become more abrupt and of greater amplitude these are called wind gusts. Measured wind speed is a function of anemometer height, averaging time, and terrain upwind of the measurement site. The variations of wind with respect to height and time make it difficult to oblong values that are representative of the conditions over large regions. Near-surface winds are conventionally measured at a point some distance above the ground in flat, open terrain Apical of an airport exposure. Although the recommended height is 33 ft (10 m), which is also the reference height for wind speed designated in building codes, actual measurement heights vain widely due to the peculiarities of each observing site. Temporal variations must be smoothed out by averaging speed and direction over time. The averaging can be done in a variety of ways, depending on the application. In the United States, -minute averages are used to specie the sustained wind, whereas extreme or peak wind speeds are averaged over 2 to 5 seconds. From a practical point of view, this average time may range from ~ to 5 seconds depending on the response time of the anemometer and/or the recorder. Internationally, lO-niinute averages are used to specify the sustained wind; however, adherence to this standard is not universal. To ensure comparability of readings of the mean wind from different locations, heights of measurement and averaging procedures must be standardized or the data from each station must be "adjusted" to standard conditions. If characteristics of gusts are to be compared, then the response characteristics of the measuring instruments must also be matched. In the United States, climatological wind speeds are usually adjusted to lO-m, -minute averaged sustained winds. However, adjustment procedures usually entail assumptions about the variation of wind with time and height; these can lead to biases in the climatological values. The Wind Climatology of the United States The mean annual wind speed for the contiguous 48 states is ~ to 12 mph (4 to 6 m/s). In sheltered areas, mean speeds tend to be lower; on mountain ridges, in passes, and along coasts they are considerably higher. At most locations throughout the United State, winds speeds of 50 mph (22 m/s) occur frequently. As shown in Figure 2-l, nearly every area of the counts ~l occasionally experience wind speeds over 70 mph (31 m/s). Many coastal regions can expect speeds of more than 100 mph (45 m/s). In the central one- half of the United States, most of these high wind speeds can be associated with frontal passages or thunderstorms (the latter either isolated or in a grouping such as a squall line). In the western mountain states, downsIope winds account for many extreme reports (e.g., gusts to 135 mph (60 m/s) in

of W; En ~ o c.-~, l - : : - ~ ~ ~ ~ - - - ~ ~ ~ ~ lo_, A I _ ~ O f 7ii, ad\ 21 o ~ ~= ~ ~ Ct = Ct ~ ~ ~ ~O Ct = ~ ~ cn V ~ ~ Ct =.° :S ~ ~ Ct o ;`,~Z V o ::.= · ~ ~ ~ =-.= ~ l ~ Call ~ ~ ~ ~ . . _ O., ~ ~ ~ V, C) 8 ~ ~ ~ · - ~ ~ ~ is >~ o -I =- ~ ~ ~ .3 ~ Ct C).~4 Cd == ~ Z o .= ~ .~ ~ o Cot _ C) _ ~ "S ~ ~ no , ~ ~ o s.O ° ° o ° ~ ~ ~ V,.o ._ ~ 5 i=) Vat ~ .= ~ ~ Ct - - \0'~0 =~<,,,~ ~ - ~ ~ ~ - ~ '~ ~s (L) E-' ~ s ~ ~ ~ ~ ~ oo · , c: · :, , ~ [.VL, ~ ~ ~ ~

22 Mind and the Built Environment Boulder, Colorado). Along the eastern and Gulf coasts, thunderstorms and hurricanes are the main producers of strong, near-surface winds. Although the highest, near-surface wind speeds likely occur beneath thunderstorms and in hurncanes, very high wind speeds can also be associated with deep extratropical cyclones developed from winter low-pressure systems. In these weather systems, strong winds may be found both swirling around the center of lowest pressure and in association with the steep gradients of temperature and pressure that characterize the attendant cold frontal zones. The strongest officraDy recorded winds observed in the United States, listed In Table 2-l, have all occurred at the summit of Mt. Washington, New Hampsh~re. These record values illustrate what the atmosphere is capable of producing and provide estimates of the maximum speeds that on rare occasions may be encountered elsewhere in the United States (see Riordan and Bourget, 1985, for additional discussion of extreme winds). TABLE 2-1 Extreme (nontornadic) Wind Speeds in the United States All Set at Mt. Washington, New Hampshire (elevation 4800 ft. mean sea level) PEAK GUST SPEED Highest Peak Gust: 231 mph (103.6 m/s) April 12, 1934 (also the world record) SUSTAINED WIND Highest 5-Minute Speed: 188 mph (84.2 m/s) April 12, 1934 (also the world record) Highest 24-Hour Speed: 128 mph (57.2 m/s) April 11—12, 1934 Highest Monthly Speed: 70 mph (31.1 m/s) February 1939 Highest Annual Speed: 35 mph (15.6 m/s) 1934 to 1983 Source: Riordan and Bourget, 1985 Thunderstorms About 100,000 thunderstorms occur in the United States each year. Most develop in the spring and summer months, though they can occur every month of the year. Figure 2-2 shows that thunderstorms occur in every state

c-- '-a ' Also : ~':~ ;' l ~ t~ Am ,. it. L ( I t li- \' / / ~ ,I' ~ d' TO 23 O A: V Can O ~ :> Ct Can ~ ~ ~ ~¢ o o o Can Z Ct ~ ~ 5 0 V: ~ ~ an?= em Can I_ ~ ~ U: no-= O en: . - -.= E ~ ~ a Ct ~ Ct U: ~ ~ E° .~ ~ ¢ Can ~ _ ~ Can ~ _ At) ~ Ct Ct Em .C) V, O ~ ~ ~ C) .= Eo O- o ~ ~ ,' o Ct = can Z V: ~ ~ o._ c, Can ~ Cd 5 U: c: - ,5= A:: O ~ o 3 (_) . 1 of ~ en 0 = Z =.. ~ ~ 8 ~ so As, ~ o Car ~ U) ~ or ~ Ct ~ U: so ~ o

24 Wind arid the Built Environment but are concentrated In the central Great Plains and the southeastern coastal states. Thunderstorms are most frequent in Florida and most infrequent in upper New England, the western states, and along the West Coast (Easterling and Robinson, 1985~. A varieW of mechanisms give rise to thunderstorms; therefore, storms behave somewhat differently in the venous cTimatological regions of the nation. All types can occasionally become severe. A thunderstorm is officially considered severe if it produces a tornado, winds in excess of 58 mph (26 m/s), or surface hail greater than 0.75 inch (20 mm) in diameter. About 10,000 severe thunderstorms are reported in the United States each year, resulting iI1 properly losses exceeding $1 billion annually. On occasion, thunderstorms occur as organized elements of larger systems. Such systems can persist for hours, travel distances up to 625 miles (1000 km), and produce a variety of severe weather phenomena. The squall line, a linear pattern of thunderstorms, has long been recognized. More recently, mesoscale convective systems large areas of interacting thunderstorms-have been identified. Thunderstorms can produce large variations in surface wind over a small region. In a severe thunderstorm, air flowing into the storm base can reach speeds of 45 mph (20 m/s), while less than half a mile away, air flowing from the region of heavy rain can reach speeds greater than 145 mph (65 m/s). At the same time, winds in unaffected regions around the stow can be nearly calm. An observer on the surface experiences rapid changes in wind speed and Erection as a system of thunderstorm winds moves past. As implied in Figure 2-3, gusts exceeding 50 knots are frequently reported in nontornadic, severe thunderstorms (Kelly et al., 1985~. In the exceptionally severe thunderstorm, one or more tornadoes and downbursts may be produced. (A complete discussion of winds in and around thunderstorms can be found in Kessler (1986~.) Outflows A thunderstorm produces tremendous quantities of rain. The falling raindrops induce a strong downdraft within the storm by aerodynamic drag and evaporation in the dry air that enters the storm at niidievel. A downdraft, as it approaches the earth's surface, begins to spread out beneath the storm. The leading edge of this outflow, the thunderstorm gust front, can contain moderately high sustained winds (speeds of 3~50 mph (13 - 22 m/s) are common). It is a region of strong wind shear (where the wind changes rapidly in speed or direction or both) and so constitutes a hazard to low-flying aircraft. If the parent thunderstorm is moving rapidly, then near-surface, peak gusts can be expected that are roughly equal to the sum of the sustained outflow winds and the storm speed. The flow in the gust front and in the Laurent behind it has been investigated by Charba (1974), Goff (1976), and Wakimoto (1982~. A conceptual mode] based on these findings is shown in Figure 2-4.

- -~ of\\ ~ '. ~ ~ - !\ i Is - ~57 ,/ - ~ - - ~ us on . - - ~4 · ~ c) o v: ce ;^ so so so At 1 ~ 4< 1 ~ , Got 25 C) C) o S: ·_4 3 o - V: so :> U: Cal o sly V

26 Wnd and the Built Environment l ~ Dissipating Thunderstorm - ~ atlv0 sta~0 Radar refletibvely detects ~e advancing rain at lowlevels .3 it_ Early - ~e ~~e Precipiftabon Foci late Burt Stem No Thunderstorm Dissipating STAB FIGURE 2-4 A conceptual mode] for the evolution of the flow in a thunderstorm gust Font and the following current. Stipling denotes rain (and In later stages, dust and small debris) carried along with current. Source: Wakimoto, 1982

The Nature of Hand 27 Small tornadoes can form along a gust front. These small, gust-front vortices have recently been investigated by Wakimoto and Wilson (1989~; they may account for the majority of all weak tornadoes reported. Downbursts and Microbursts Fujita (1976, 1981, 1985) was the first to describe the strong, localized downdrafts, termed dowobursts, that are produced by some thunderstorms. Although similar to and resulting from the same processes that produce the thunderstorm downdraft, downbursts are smaller and much more concentrated. They can induce an outburst of damaging winds near the ground, with near-surface speeds up to 125 mph (56 m/s). The strong winds tend to flow radially outward from where the descending current strikes the earth. Very strong wind shears occur at the leading edge of the downburst. The Spicy area of damaging winds is 3.8 to 5 miles (6 to ~ km) across. At a point beneath the thunderstorm, these strong winds may persist for up to 30 minutes. Beneath a traveling mesoscale convective system, the individual outflows of the component storms may merge to produce a derecho (Johns and Hirt, 1987~. As the combined outflow passes a location, very strong, gusty winds can occur for several hours. In addition, such a system can produce a widespread pattern of strong downbursts, both as downburst families (produced in sequence by the same thunderstorm cell) and as downburst clusters (produced by adjacent thundersto~s in the same general area). Microbursts are small, very concentrated downbursts, about 2 miles (3 Ian) or less in horizontal extent. They can contain very high winds; the record observed wind speed in a m~croburst is 150 mph (67 m/s) reported by Fujita (1985) from an analysis of a m~croburst at Andrews Air Force Base, Maryland, on August 1, 1983. A m~croburst may last only 5 to 7 minutes, yet these events pose an especially serious hazard to aircraft during takeoff and landing because of the exceptionally strong wind shears that occur. Both "wet" and "dry" m~crobursts have been identified: the former appears as a descending shaft of rain and so may be avoided, whereas the latter is nearly invisible and so may not be recognized by pilots until too late. Tomadoes A tornado can produce the highest wind speeds known. This swirling column of air is usually associated with a severe thunderstorm, though its location beneath the storm can vary. A worse phenomenon, tornadoes occur most often in the United States. From 1953 through 1989, about 27,000 tornadoes were documented in the United States. The distribution of these events by year is shown in Figure 2-5. On average, tornadoes were reported in the United States on 169 days per year during this period. As indicated in Figures 2-6, 2-7, 2-S, and 2-9, tornadoes are concentrated in the central half

'e - ~ :T TV r _ ~ . . 1 \ : _ - . . ~ . __~ : : I: 2 it_ : : ~' ; : : : .1 ._ ~ ~ _ 2' ~ OF 1_ a, GO _ ~ - Ct o _ CO - W a, O~O Ct ~' C} 0 _ ~ O of ~ O CO ~ ~ _C ~ _ _ __ Cal ~0 ~ CO _ ~ O O O O O O O O O O O O C~ O ~ CO ~ C~ 28 o o c t ~t ,_, _ ~ V, ~ .= V . ~s c: ·~ ~Q ~ ·= Ct ~ ~ o o Ct Cd Ct ~: ~ o th4 ·CD Ct ~ ~_ V: _ — C,0 o ~ ~ 3 C'0 o o V, - ~ - o ~ _ ~ ~ '~ D 0,= Ct C) C'0 . 1 Ct ~ 3 L ~ as 00 0N C~ ~ ~ V) 04 I_ O

O c~ 0 o · o < · to · · to · ~ es ~ o ) - .0 o co \ {D O ~ O ~ IE ~ \` x~ ~ of ~ to · ~ ~ ~.,~i. ~ ~~ - ~ en ~ _ Lo) a, · to ~ 0 . to. ~ - - n ~ <_5~ ~- a' ~ )~ Y ,1~_\- Hi 8~N';~: Z ~C`1 (D \ Al <I) ·1 A · A · ) ~ a) 'D ~ ~ Cat · n - I_ )m _ CD Cat CO ~ O an _~ lo ~ 0 of - : ~ 0 ~ 29 1~ ~ V Q ~ a: 7 ~ ~ ~ / ~ o) Cat Q o ~ C: ~ _ ~ :) _ ~—— ~o ·0- · O Ce Ct C) O ,, ", o Ct U: V: ~ C~ I_, ._ _ _ V: _. V .w ~ '~ ~V ~S E~ · _ oo 0N 0 ~ ._ O C 0N s~ ~ V - O ~ s~ . . V, ~ o o ~V) ~ C~ O Ct ~) — C;, Ct ~ O V, _ ;~. '' ~ = ~ 0N C~ ~ ') =~? ·' ~ ~ ~ ~ o v: $-~ ~ ~' ^ 2 ~ Ct V, C) C) V: o ~ Ct Ct ~ ~ o — _ ~o ~ ~o V-~ ~ ~ ~ V

~ cd em o 'd en ~7' r~ ~ ~ ~ 1 r_ 1 ~ <_ ~ - ~ - we co and Co o ~ eo ' ~ - l" 1 0 0 - ~ - Ln l ~ ~ - ~ - ~ ID In ~ o o ~ Loo ~ — 1—~— ~ ~ lo, - ~ ~ A o J— O a ~ - e, - ° ~ ~ e ~ ~ _ ~ ~ _ _ ~ _ O ~ O ~ e A it: -~_/ / ~ In of / 30 At; ¢S oo o c~ d c~ ~ ~ ~ - ~ ~ c~ c~ r" t v' - ~ cd u: ~ ~ 'e 'e 3 t: ~ ~ ._ _ _' ~ Ct ~ 3.° ~ _ ~ ~ ~Z; 5 cd V, S~ 0'_, ~ ~ V s~ ~ Ct c: ~ C) Ct ~ ~ U. ~ m0 ~ ~ V, ~4 ~ O O .= ~ U: Ct _ C~ - - l ~.cn Q cd ce ~ V' o —~ ~V ~ o ~ Z ~ s~ ~ - C) ~ O C) 0 ~ r~ ~ ~_ O E~ _1 Ct . ~ ° tV 0 ox ~ - `' ~ ~ 0 0 V' 1 c~l <~~ cn ~ 0 ~: _ =O V -,~ i_ ~ ._ ~ o~

~ c;, o ~ v: ~ ~~ To o x ~ I: .> At 'e ~ · - At - = At ~ ce cd -= o ce , 4 o ~ ~ hi o ~ ce .= ~ And ~ · o itch JO ~ o c': g ~ Act x .= .= ~ ~~ - :,.O - ~ ce o it' I U. ~ it_ 31 ( W~ ->v~y

o o Hi /~<om to ~1,, ,,:N ~ J l i In - ce 1 v, o cd o ct o fib - cd N Ct 32 - o Ct - Ct - Ct D O - 1 A; O A: _% Ct

The Nature of Wnd 33 of the county and are most prevalent in parts of the states of Texas,Oklahoma, and Kansas (though on a whole-state basis, Indiana ranks ahead of Kansas). During 1953 1989, tornadoes claimed 3550 lives—an average of 96 deaths per year. Figure 2-10 shows tornado-related deaths to be concentrated in Mississippi, Arkansas, and Indiana (Alabama and Ohio also have had large numbers of deaths due to tornadoes). Inspection of Figure 2-5 shows that three years contribute disproportionately to the total number of deaths: 1953 (516 deaths), 1965 (298 deaths), and 1974 (361 deaths). The deaths in 1953 were due to strong tornadoes striking three urban areas: Waco, Texas on May If, 1953 (~14 dead, 597 injured); Flint, Michigan on June 8, 1953 (~16 dead, 867 injured); and Worchester, Massachusetts on June 9, 1953 (90 dead, 1288 injured). The high death counts in 1965 and 1974 were the result of an extensive tornado outbreak in each of these years. A tornado outbreak is defined as the occurrence of several tornadoes over a region, usually all by thunderstorms embedded in the same extratropical cyclone. Although infrequent, outbreaks strongly affect all tornado statistics. Tornado wind speeds are especially threatening to human activities because the speeds apparently maximize only a few tens of meters off the ground. Current estimates of tornado wind speeds are based on postevent analyses of damage to engineered structures, on photogrammetric determination of the motion of objects (such as dust packets, pieces of vegetation, and building debris) documented on film and videotape, and on measurement of speeds of objects with Doppler radar. The observation of a tornado with Doppler radar is a fortuitous event, because until very recently, the only such radars available were large, f~xed-station systems. This meant that the tornado had to occur within 25 - 30 miles (4050 km) of the radar site for it to be observed. Each of the w~nd-speed estimation techniques has its shortcomings, but taken together, they produce results suggesting that the _ , _ . ~ ~ ~~ ~ · . ~ . . . . maximum probable net speed occurs JUU to Ib5 tt (30 to 50 m) above the ground and is in the range of 250 to 310 mph (~10 to 135 m/s), with some likelihood that the actual value is near the lower end of this range (Dav~es- Jones, 1983~. Vertical speeds may be as high as IS0 mph (80 m/s), whereas radial speeds have been estimated to reach Il2 mph (50 m/s). The speed of translation of a tornado—roughly the same as that of its parent thunderstorm- can exceed 56 mph (25 m/s). It should be emphasized that all of these values are based on indirect measurements as described above. Very little detail is actually known about the airflow in tornadoes. Tornado wind speeds are categorized by using the Fujita Tornado Scale (commonly called the F-scale) given in Table 2-2. This scale, which was established by Fujita (1971, 1981; also Fujita and Pearson, 1973), was based pnmanly on the study of damage produced by the 1970 Lubbock, Texas, tornado. A subjective scale value is assigned based on visual assessment of damage severity; this in turn fixes a speed range (note that the ranges in the scale do not overlap). Table 2-2 shows that extreme speeds are usually attained in only a few tornadoes each year, in what are referred to as 'T5"

0 ~ ~ o to to ~ · · o · ~ ~ · ~ a ~ - N O V N O (V) C! %` `% \ ~ O ~~;~ =~ O O O · O <1 5 N lo CV) ~ P =0 0 0 Z. 0 - 1 · art\ N i~ ~~ ~ _ Z · O - )~L -C- 1~-~: _ . , ~ 2_ O $ | · O · ) 1 O 1 ~ O V 1 ·0- ~ .~ Cat ~ O ~ / At, ~ O' _ O O O O ·0 ~ O /- oT7~ / · ° · 1 moo ~ 0< 34 - ce o · - ct z · e C) o V, . o 0N o ~0 0N ~ 1 - O V: O ~ ~0 7 - O ,C) ~ 2 C~ ~ Ct Ct '~ :> _ V, ~ o ~ .~ ~ _ Ct 7 O ^ Ct ~ U' O Ct o [I] ~ ~ U: V o V,

The Nature of Hand 35 TABLE 2-2 Fujita Tornado Scale Wind-Speed Average Rangea Description Number per Scale of Year Value (m/s) (mph) Damage (1953—1989) O IS 33 4072 Light 218 (demo) ~ 33 50 73-~12 Moderate 301 (demo) 2 5070 113—157 Considerable 175 (23~o) 3 7~92 158-206 Severe 43 (6%) 4 92 - 117 207 - 260 Devastating 10 (my) 5 117 - 143 261 - 318 Incredible 1 (0.002~o) Total 748 a Wind speeds in the ranges given for the scale values are defined hv Flliita to he the. "f~t~.~t · v c7 ~ ~ ~J ~~~ ~ ~ ~ ~ ~~ ~ ~~~ ~ ~ ~~ I ~ At e ~~ —~ I_ ~ · 1 t ~ ALA ~ . ~ ~ . ~ e e ~ ~ 1/+mue winds. For an L4 wind speed of ZW mph, the duration of the damaging wind at a point Would thus be about 4 seconds. Source: Fujita, 1971. tornadoes (Fujita, 1987~. In 98 percent of all tornadoes, speeds near the ground are considerably less than 200 mph (90 m/s). Also, since flow configurations vary considerably with overall intensity, extremes in vertical speeds may not occur in conjunction with extremes in tangential speeds. In 50 percent of all tornadoes, speeds are less than 100 mph (45 m/s) and should cause only minor damage to structures. As discussed in Doswell and Burgess (1988), the F-scale is essentially a damage severity scale and not a true "tornado intensity'' scale. In particular, the w~nd-speed ranges of the F4 and F5 categories have not yet been calibrated by independent observations near the surface. Indeed, w~nd-speed assessments based on analyses of damage to engineered structures have consistently indicated wind speeds lower than the ranges given by Fujita. These later assessments also suggest that there should be an overlap of wind speeds between the F4 and F5 categories to allow for variability in construction materials and practices (Minor et al., 1977~. Results from laboratory modeling coupled with results from photogrammetric measurements of actual events show that the distribution of wind in a tornado is highly complicated (e.g., Snow, 1987~. These findings suggest that in small, generally low-intensity tornadoes, the maximum tangential speeds are contained in a toroidal volume very close to the ground. Maximum vertical speeds are contained in an upward jet through the central

36 Wnd and He Built Environment hole of this toroid, whereas maximum radial speeds occur in an ingrowing surface layer beneath the toroid. Large, generally high-intensity tornadoes appear to have a more complicated structure, with embedded secondary or "suction" vortices circling a clear, perhaps nearly stagnant central core. These secondary vortices are very intense and usually short-lived, and give rise to the maximum wind speeds cited above (Blechman, 1975; Forbes, 1978~. In addition to suction vortices, other asymmetric flow configurations have been documented in tornadoes (e.g., an accelerating radial inflow appearing as a dust band has been desc ibed by Golden and Purcell (1977~. Pressure falls associated with tornadoes are generally fairly small, probably reaching around 100 millibars (mb) in extreme events. Noah sea- leve} pressure is about 1013 mb (29.91 inches fig). The local time-rate-of- change of pressure experienced at a point on the surface depends on several factors; for an intense, fast-mov~ng, small vortex, this may approach 100 mb/s. Although this change is large, recent engineering studies have shown that the "explosion" of structures due to an extreme pressure difference developing between inside and outside is an unusual occurrence. Most structures are sufficiently leaky that even with the high rates of pressure change encountered in tornadoes, interior pressures can adjust quickly enough to prevent an explosion. Parallel studies have shown that almost all of the devastating damage caused by tornadoes can be attributed to w~nd-induced forces tearing structures apart from the outside. Evidence gathered during postevent damage surveys also suggests that a "domino effect" (i.e., large debris elements, such as roofing materials, becoming airborne and then impacting nearby structures) plays a key role in much tornado damage. Extrairopical Cyclones The passage of a midiatitude low-pressure system (perhaps over a period of several days) and its associated fronts produces a characteristic pattern of shifts in wind direction and variations in wind speed. Such systems (extratropical Cyclones) tend to be strongest in late winter and early spring when the north-south temperature contrasts are greatest. There is some tendency for low-pressure systems to follow certain preferred paths across the globe. Thus, in many areas, systematic changes in the wind have long been recognized and used to formulate rules of thumb for forecasting local weather. Patterns of wind that have unusual characteristics and that occur in the same region with certain frequency are often given descriptive names by local people. For instance, in New England one may hear reference to a nor 'Easter. This wind blows onto the New England coast from the northeast with moderate to strong force and is generally wet and cold. Topography can compress and funnel the strong air flows accompanying extratropical cyclones to produce very high wind speeds locally. Examples of this effect are provided by the extreme winds cited in Table 2-1 for Mt.

The Nature of Wnd 37 Washington, New Hampshire; the records for April ll 12, 1934, were set during the passage of a strong extratroDical cyclone. Winds are stronger at the ~ ~ ~ ~ r - ~ ~ ~ . . ~ . ~ . _ _ top ot this peak-as they are on the summits ot most mountains-than they are at the same elevation in the free air because of the air being forced over the mountain, which presents an obstacle to the wind flow. Another example of a topographic effect is the chinook or foehn wind of the eastern slopes of the Rocky Mountains. This is a dry, warm wind that blows down the mountains and has been known to produce near-surface winds in excess of 100 mph (45 m/s) accompanied by temperature fluctuations of up 22 C (40 F) in half an hour (Lilly and Zipser, 1972; Bedard and LeFebvre, 1983~. Gusts may reach 140 mph (62 m/s). As examples of extreme winds that can be produced by these systems, consider that during the storm of November 24-26, 1950, a fastest-m~le wind speed of 82 mph (37 m/s) was measured at Newark, New Jersey, and a 110- mph (49-m/s) gust was observed in Concord, New Hampshire. This storm caused 200 deaths and $] million in property damage. Damage throughout New Jersey, eastern Pennsylvania, and interior and southeastern New York was more severe than in the hurricane of 1938 that struck these same regions. More recent examples of the high winds that can be produced by extratropical cyclones are provided by the violent storms that struck the United Kingdom and Western Europe on October 16, 1987, January 25, 1990, and February 27, 1990; these storms produced near-surface winds well over 100 mph (45 m/s), resulting in 47 deaths and over $1.5 billion in damage, with thousands of large trees uprooted. Tropical Cyclones A few times each year, from early June through October, low-pressure systems termed tropical cyclones form over warm ocean waters in the tropics. Occasionally, where sea-surface temperatures are greater than 22 C (72 F) and when winds high in the atmosphere are supportive, one of these systems win become more organized and intensify to become a hotcake (w~nd speeds exceeding 73 mph (64 knots) or more (Anthes, 1982~. When fully developed, a hurricane has a calm, central core, or eye, surrounded by very strong winds concentrated in a doughnut-shaped region of heavy rain, termed the eyewall. A pattern of rainbands, each consisting of many thunderstorm cells, spirals inward to merge with the eyewall. These violent tropical cyclones tend to be self-sustaining until they move either into a region of sea-surface — O O temperatures less than 26.5 C (80 F) or over land. In either case. the system is cut off from its main energy source, the warm sea surface. ~ . ~ ~ ~ ~ . ~ ~ ~ ~ . . _ _J _ _ _ line eastern and QU11 coasts ot the umteo states are threatened by tropical cyclones that form in the tropical Atlantic and in the Gulf of Mexico (Figures 2-~1, 2-12, and 2-13~. From 1931 through 1987, 551 tropical storms were observed in these areas; 314 of these reached hurricane status. In an average year, about four tropical storms will make landfall in the continental United States; on average, two of these wall be hurricanes. The actual number

o 1 f z z to to 11 - `' 1 1 1 1 z to LO y at Hi ~ ~ - z z lo cO cM 38 ~( 1 ~ z to CM / lo LO o CO LO CO o U. U: Ct V) ·s a: C) U: Ct "S: V: ~ O ·O Z - .O o V) o C ~ 3 Ct C) _ ~ A: o o ~ ._ ._ U. . - Ct ~ V i' ~ .,, Ce o sly — ~; ~ o ~ ·~ ~ ·= 'e v: o ~ - · - d z ,) o ~ o~' · — c~ o u,

i\ - ~'1:~ ~~ ~ '' o cr. v: To ~ ~ - at v: 'e v, ~ ~ o ~ z o a, - 1 Ce or ~ ~ 3 Ct Cal _ Ct `: ~ 1 o ~ ._ ~ z ._ A ~ 1 l_' Ct ~ ~ V ,,, Ce o .O o ~ .= so ~ Cat o ._ ._ z ~ .. A: ~ ~~ o _ =\ ~ , /1oo-~1 Ic o C9 CD }/~y ~ - ~~ it's z z z O LO O cO cM cM 39 o LO CO to Z ° to

z to ~ 1 ~ '7 )~ art ~o.- ,5 ~~ l \ ' z to In to co ~- 'P' - 1 - - TO z to GO 3( z CM 40 o V2 - Ct - V, - - . - - C) - ~n . V, ~ ~ o C) ·7 . ~ '. s ~ . o U) TV as - o ~ Ct Cot Ct ~ A: o o .~ ~ C;, .sZ V, ~ 1 ~ Ct ~ V - _ l_ C) O . - - ~; ~ - o ~ _ ~ .= Ct ~ . O Ct z ~ .. 3 U' ;~

The Nature of Hand 41 of hurricanes per year in the subtropical northern Atlantic, Caribbean, and Gulf of Mexico has shown wide variability, ranging from 2 (1931, 1983) to 12 (1969~. Tropical cyclones and hurricanes also pose a serious threat to Puerto Rico arid the other islands of the Canbbean. Hurricane Donna passed close to Puerto Rico on September 5, 1960, causing 107 deaths and millions of doDars of property damage. The eye of Hurricane Hugo passed directly over St. Crow and northeast Puerto Rico early on September I8, 1989, with a lowest pressure of 946 mb and peak gusts to 120 mph (104 knots; 52 m/s). The West Coast and the islands of Hawaii are occasionally threatened by tropical cyclones forming off the western coast of Mexico. These seldom produce strong winds along the California coast, but can lead to very heavy rains along the Pacific coast and in the southwestern states. Hawaii receives moderate to strong winds from a tropical cyclone about once even four years; it was affected by hurricanes only five times from 1904 to 1972. The only storms to pass directly over one of the major islands were Hurricane Dot, which passed over Kauai in August 1959, and Hurricane Twa, which passed over Kaua~ in November 1982. As suggested in Figures 2-~l, 2-12, and 2-13 hurricanes make landfall most often in Florida. However, historically these tropical systems have come ashore all along the coast, from Texas to Massachusetts. The northerly landfalls are a consequence of the Gulf Stream that transports warm water northward along the eastern seaboard. Over this current, these cyclones can maintain their tropical characteristics much farther north than they can over the central North Atlantic. When a humcane makes landfall in a populated area, the potential is great for significant loss of life and enormous property damage from both high winds and the storm sure. The storm surge is the rise of the sea in advance of an approaching tropical storm. The decree of rise is related to ~ , _ , ~ note the central low pressure and the overall wind held in the storm; it also depends on coastline bathymetry. The sea experiences its maximum rise in the storm's r~ght-front quadrant taken relative to the storm's track. The Saffir-Simpson Scale (Simpson and RiehI, 1981) shown in Table 2-3 has been developed to express the severity of these storms in terms of their maximum wind speed (usually estimated from ship and aircraft data) and depth of the storm surge. Of the 138 hurricanes to reach the continental United States in the period l899~1980, 82 had maximum sustained winds from 74 to 110 mph (33 to 49 m/x)--categories ~ and 2 on the Saffir-Simpson Scale. Of the remainder, 54 were major hurricanes, falling into categories 3 and 4. Only two category 5 hurricanes have hit the mainland this century: the Labor Day hurricane at Matecumbe Key, Florida (September 2, 1935), and Hurricane Camille on the coasts of Louisiana and Mississippi (August 17, 1969). -

42 Hued and Me Built Environment TABLE 2-3 Sa~r-Simpson Hurricane Scale Category Wind-Speed Range (sustained) (m/s) (mph) Storm Surge Depth Above Normal Tide (m) (ft) Storm Description 1 33~3 74 - 95 1.2- 1.6 4-5 Weak 2 4309 96~110 1.7-2.5 6-S Moderate 3 4058 111 - 130 2.6 3.8 012 Strong 4 5809 131 - 155 3.05.513-18 Very Strong 5 > 69 > 155 ~ 5.5 > IS Devastating Source: Simpson and Retell, 1981. Hurncane Hugo was the second most costly—though not the most deadly-hurncane in U.S. history. The total number of deaths attributed to Hurricane Hugo during its n~ne-day life (September 13-22, 1989) is estimated at 49, with 21 on the mainland. It inflicted direct damage estimated at nearly $10 billion, with about $7 to ~ billion occumng on the mainland. Paths and changes in intensity of tropical cyclones have proved difficult to predict. In the Atlantic and the Caribbean, these systems usually move slowly westward initially, gradually increasing in strength. Often they recurve, turning first to the north and then to the northeast. Detailed studies using aircraft reconnaissance and satellite observation (the latter has been available since the m~d-1960s) confirm that hurricane movement is often erratic: a hurricane on the move can slow down, remain nearly stationary, speed up, or change direction. The track of Hurricane Hugo, shown in the map of Figure 2-14 for the 10 days from 1800 GMT September 12, 1989 to 1800 GMT September 22, 1989, illustrates many of these features. However, large-scale mappings of storm-center motion as derived from a mixture of aircraft and satellite data can be misleading because many details are smoothed out. As illustrated in Figure 2-15, the actual hour-by-hour track of Hurricane Hugo's eye, when observed by radar, is much more complex and illustrates the problems of predicting landfall time and location. Although satellite observations have proved invaluable in continuously monitoring the overall development and motion of a storm system, most of our detailed knowledge of the inner workings of hurricanes has come from aircraft reconnaissance flights by the National Oceanic and Atmospheric Administration (NOAA) and the military services. These flights penetrate the swirling clouds to collect data on wind, temperature, and pressure from about 1500 ft (450 m) to 25,000 ft (7500 m) above the sea surface. Over the ocean, expendable packages are dropped from the aircraft to measure sea-surface temperatures and vertical profiles of winds and other atmospheric parameters.

The Nature of Wnd 43 aO 70 an ' 30e an 60 An _ . _ ~= flu 40 r ~ ~ A i:. re Rugo ~ lo' oooo sn2~s , `~ ~ c' And / ._ 00009/19A _ _ ~= ~~ _: , ~ . . 00009118/89 _ 00009/19A 9 ~ . _ of_ . . . , _ ~ En BO 70' 60 ~ r~s/l' ~9 ~ . r b T— O(K>09/16/89 ooOO 9/15/8S r 0000 9/13/89 (00009/14/89 ~~~~ 1 1 1 50 40 FIGURE 2-14 The track of Hurricane Hugo over a ten-day period, 1800 GMT September 12, 1989 to IS00 GMT September 22, 1989, as determined from composited aircraft, satellite, and other observations. Most frequently cited extreme wind speeds in hurricanes were measured at aircraft altitude. Using aircraft data, Riehl (1979) has estimated that over the open ocean, the strongest winds are found at about 1000 ft (300 m) above the sea surface in the eyewaB region of the storm. The annulus of strong winds extends upward to more than 10,000 ft (3500 m). Aircraft and radar studies have shown that some of the more intense hurricanes have contained a double concentric eyewaD structure and a related double maximum in wind speeds, as was the case with Hurricane Alicia in 1983 (Willoughby et al., 1982). Aircraft measurements provide limited spatial and temporal coverage of weather systems as large and as long-lived as hurncanes. Most seriously, airborne instruments provide no direct measurements of the quantity of greatest interest: near-surface wind speeds as a storm comes onshore. As a consequence, various indirect methods have been devised for estimating near- surface speeds from composite aircraft (usually taken at 10,000 ft (3500 m)) arid satellite data. Because of the physical relationship between the winds circling the eye and the pressure gradient, an inferred central pressure is often used to 40 30 20

44 W=d and ~ By Eat 4~r/C / - N - - \ \ OCEAN \ INTO RICO \~ (UNITED STATES) \ \ K~ \N Magaquez ~ \ it_: ISLA DE CULEBRA V _ :~' ~,Z' 0~ if\ as. an, \ \ \ i ~ , _ _ \ FIGURE 2-15 A detailed mapping of the track of the center of the eye of Hurricane Hugo as it swept past the Virgin Islands and Puerto Rico September 18, 1989. This track was determined by continuous monitoring of the storm system by the NWS radar at San Juan, Puerto Rico.

The Nature of Hind 45 estimate near-surface, peak wind speeds (neglecting surface friction). By this means, crude estimates of maximum sustained winds of about 200 mph (90 m/s) were made from the two strongest hurricanes to make landfall in the United States in modern times, the Labor Day Hurricane of 1935 (26.35 inches Hg/892.3 mb; McDonald, 1935) and Hurricane Camille in 1969 (26.73 inches Hg/905 rob). ~ . ~ . ~ In a nurncane, high near-surface winds affecting land occur as the major rainbands, and especially the eyewall, come onshore. Although wings have been measured at the ocean shore during hurricane landfalIs, most observations have been made some distance inland from the coast. Because of poor mast design or inadequate maintenance, anemometers positioned at the shore have usually failed before peak wind speeds are suspected to have occurred, particularly in the extreme storms of most interest. Figure 2-16 shows a trace of wind speed Apical of the passage of a hurricane eye by a near-shore station, in this case from Hurricane Hugo, October 22-23, 1989. The active storm clouds in the rainbands can produce tornadoes (NovIan and Gray, 1974; Gentry, 1983; McCaul, 1987~. Although most of these are small and short-lived, occasionally one produces significant damage. Hurricanes Carla (September I - 13, 1960), Beulah (September 20, 1967) and Gilbert (September ~ ~ 19, 1988) are known for the destructive tornadoes that accompanied them. It is unportant to recognize that wind speeds collected by weather reconnaissance aircraft are estimated values applicable at aircraft altitude. The lower limit for safe flight is around 450 m (1500 ft). Knowledge of the vertical distribution of wind speed in hurricanes, particularly as they are making landfall, is very limited (e.g., Black et al., 1988~. Speeds near ground are modified by the roughness of the terrain and trees. The increased friction encountered by the swirling flow after it moves inland quickly reduces the speed of the winds at near-ground level. For example, a wind of 160 mph (71 m/s) at 500 ft (150 m) wall Epically translate into a fastest-~rule wind speed of 120 mph (54 m/s) at 33 It (10 m) over open seas and into just over 100 mph (45 m/s) in open terrain near the coast. Winds 300 to 400 ft (100 to 125 m) above ground may be little changed for some distance inland. Very strong (though generally less than hurricane force) near-surface winds can persist well inland, particularly in fast-moving storms. On October 15, 1954, Hurricane Hazel, with a forward speed of 50 mph (22 m/s), produced 100- mph (45-m/s) winds in Buffalo, New York. In 1985, Hurricane Gloria, also moving at 50 mph (22 m/s), produced wind gusts to 70 mph (31 m/s) ir1 New Hampshire. As shown in Figure 2-17, Hurricane Hugo (1989) produced winds of humcane or near-hurricane force over a large inland area; wind damage was reported as far as 150 to 250 miles from the coast. For reasons not yet understood but probably related to decade-Ion" trends in the Earth's overall climate, yearly hurricane activity has varied extensively over the last century. For example, since 1950 there have been 39 AtIantic hurricanes in categories ~ and 2 of the Saffir-Simpson Scale. Thirty- three, or 85 percent, of these occurred during the "very active period" of 1950 to 1969; only 6, or 15 percent, occurred during the "quiet period" (1970 to

SPEED (kts) o o o o ° o CI5 o o I CD Cal O a_ _ ~ in _ ~ — CD, Cal O o O O O O ~ ~ ~ N , 46 o o O lo ~ , ,,, ,, . ~ it,, O ce ~ ~ 5 ~ V 5, I_ 4~ CJ O O" O ~—hi'= .~8 ~ ~~ o ~ =: ~ ~ A= ~ ~ Ce' I ~ ~ g 3 o ~ ,~ ~ ~— o A, 3 ~ _- ._ 3- - 5 ¢; E ~ o = ~s t_ ~ ~ ~ ~ n ~ ~ 0 ° 0 ~ E '' , .a ~' 3=Q 3~° ,= -5 ~ ~ 5 o ~ C) ~ _= <~ ~o ° —~ <., ~ .~4 3 ~i O ~ u: ~ ~—~ o =i z ~ ~ o ~ V: ~ ~ oo <: ~ ~ O ~ E ~ 0 ~ ~ ~ 0 ~ ~ :~ ~ 3 ~ p~ v = ~, [L, CQ ~ O O C~,

—'—i''—' \—-~\0 ~EC\B-R~ am- I —~ Jim _ _ ~ Al '% i ~ v The Nature of Hand 47 HURRICANE HUGO ( 1989) , UNION I ~ - -- 10 23 30 40 50 60miles ~ \ I ARSON ~ ~ ~ ~ ,,~ ~ I | _ · · —2-: W: =~20—- ~~ su5' ,_-11 mNt - ~~(J - ~. \ FAIRFIEL) i| 4\ /~ \~-~l ;`U; 1, I ^ '9 2J 32 40 ~ GO ~ RO 90 I~ km N. / R 0 8 E S O \ / ~ ~ 8L40EN .~ ! ~ ~ ~ // N. . ~ ~ , - it. C4 s" · \ii ~ · ~ ~ H O R R Y ~C`( - ( _ ' .. COLUM 9 US r~~ rut- - ' S · / BRUNSWICK By' i 34- LEXINGTON ~N V~: a W~—' ,' ally I ~1 ~rj!/ ~ ~tumPi' / ~ ~A~ 1 9Ah19ERG - ~~ . \ ~ ~ GENTLE _3,3 my\ ~ ~. \< \ ~~` , ~ ~ A; ) r F1AMPTOK ~ 5 / ~~ - \~ O 3 \ \~ E f F ~ NGHA:~' ~ B E ~ U F O R T NI '' SAVER ;, '~ Ah \3 999A ~5 \ a? ,.Q~\~ at\ it'\ FIGURE 2-17 Estimated distribution of peak gust speed along the track of Hurricane Hugo as this rapidly moving storm came indand. Contours in miles per hour. The track of the center of the storm is shown by the broad solid line. Note that the highest peak gusts, probably in excess of 130 mph (38 m/s), occurred to the right of the track at the sea coast. It is in this location where storm motion and circulation combine and surface frictional effects are minimized. Source: T. T. Fujita, University of Chicago.

48 Wnd and the Built Environment 1988~. It is unclear whether the next decade wall see a continuation of this quiet period or a return to the more vigorous activity of the 1950s and 1960s (Gray, 1990~. The pane} notes that the last two years of the "quiet" 1980s featured two very, intense storms: Hurricanes Gilbert (1988) and Hugo (1989~. It has been speculated that the projected global warming may result in an increase in the overall intensity of hurricanes (Emanuel, 1987~. IMPROVING THE WIND CLIMATOLOGY OF THE UNITED STATES Surface wind has traditionally been measured by instruments mounted on masts and towers extending upward from the surface of the earth. Measurements of surface wind direction using wind vanes were first made by the ancient Greeks and are among the oldest meteorological measurements. The effects of the wind on surface features have been the basis of qualitative estimates of wind speed from antiquity; quantitative observations of wind speed using anemometers began in the middle of the fifteenth century. Use of such instruments, albeit greatly improved by several centuries of experience, continues as part of the routine collection of weather data. Currently, near-surface wind data are measured at approximately 750 stations across the Uruted States. Approximately 225 of these are NOAA National Weather Service facilities; others are operated by the Federal Aviation Administration, the military services, and other cooperating organizations, both public and private. The standard measurement height is 33 ft or 10 m. However, in practice a wide range of heights are in use. Upper-air data are obtained at fewer stations. The main observing system for winds from the surface to around 100,000 ft (30 km) consists of balloon-borne raw~nsondes (short for "radio wind sounding") launched from 70 National Weather Service facilities scattered across the 48 states. The primer purpose of these observations is to provide the data necessary to initialize numerical weather forecast models. Upper winds are also estimated by tracking clouds observed from satellites; these data are used to supplement those obtained from rawinsondes, particularly over the oceans. Supplemental reports from commercial aircraft provide information on turbulence in the upper air. Special U.S. Air Force and NOAA aircraft are also used to observe upper-level winds and are particularly valuable for observing winds within hurricanes. However, such special aircraft observa~ic~n~ arc ~~H nrim~r;~, per atmospheric research. These traditional techniques suffer from a number of shortcomings. For example, observations from many observing stations are available only hourly and from some stations for only selected hours of the day. Rawinsondes are launched only twice daily (7 a.m. and 7 p.m. EST). Manv surface nh~ervntinuc rely on human observers visually estimating a sustained speed and a prevailing direction. Where continuous data are available, they are often in analog or strip chart form and therefore labor intensive to reduce. Few data are available from many remote sites. As a consequence of these and other ~ ~~~~~~~ I ~ ~~~~ r ^~.~~ eve · ~ ~ , ; · .

The Nature of prod 49 problems, the current, best-available wind climatology of the United States is known to contain many biases. The ongoing modernization of the operational observation program of the National Weather Service and the concurrent development of new research instrumentation present opportunities to greatly improve the wind climatology data base (Golden, 1989~. The three, key, wind-measuring systems being deployed In the 1990s by the National Weather Service are the Next- Generation Radar (NEXRAD), the Automated Surface Observing System (ASOS), and the profiler. These win be supplemented and complemented by similar systems being deployed by the Federal Aviation Administration, other federal agencies, and the military services. NEXRAD is a Doppler radar capable of measuring winds (with respect to the radar location) up to il2 mph (50 m/s). The National Weather Service will deploy Il4 NEXRAD systems throughout the United States, probably by 1996. The Federal Aviation Administration wall deploy a complementary system of shorter-range Terminal Doppler Weather Radars to watch for the occurrence of downbursts and wind shears near airports. The ASOS wall record both 10-m mean wind speed and direction, and peak gust speed at 1-m~nute intervals. Approximately 1500 of these units wall be installed throughout the United States, mainly at airports. Thirty tropospheric profiler systems have been installed in the central one-half of the United States in a demonstration network. These are vertically pointing Doppler radars that provide high-resolution profiles of the horizontal winds from about 1500 ft (500 m) to 4S, 000 ft (16,000 m) above the ground. A wind profile is available every 5 minutes. At represent these Bra he.ina llC~ only for atmospheric research. rid ~ ~ _ It must be stressed that these systems are being deployed to support the current operational requirements of the deploying agencies. In the case of the National Weather Service, these are weather forecasting and severe weather warning. Current plans call for very small, selected sets of data to be archived. There is no plan to archive data routinely for national climatological purposes. Yet archiving is critical to issues of levels of accent~hle rick and corresponding structural design requirements. ~ ~ ·_ _^ ~ A 1~ 1 , , t . ~ ~ ~,^ ~ 111~; ammo; (l~erence between Inese state-ot-tne-art systems and the traditional instrumentation that they are scheduled to replace is that they we provide nearly continuous digital data streams of wind information. Many features of these data streams are complementary; if used together, they could provide a much improved picture of winds from near surface to the tropopause in both stormy and clear weather. However they still `1n not ~ ~ _ ~ ~ 1__ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ J ~ plUV1~C Ll1C Gnu ana quaml~ or mna-speea cata needed by wind engineers. EXPANDING THE WIND-ENGINEERING DATA BASE The prediction of wind speed and its micrometeorology are critical in the application of research to practice. l he micrometerology of wind includes its gustiness, its variation with time and space, and gust sizes.

50 Wind and the Built Environment The American Society of Civil Engineers (ASCE) standard ANSI/ASCE 7-~8 (formerly American National Standards Institute standard ANS! A58.~) gives w~nd-Ioad requirements for design of buildings and structures in the United States. The standard is developed by using a consensus process wherein it is bigoted by committee members with backgrounds in consulting engineering, research, the construction industry, education, government, design, and private practice. The load requirements provided in the standard are intended for use by architects, structural engineers, and those engaged in preparing and administering local building codes. Prediction of wind speeds for design given in the current ANSI/ASCE 7-~8 standard are based on data collected at 129 National Weather Service stations around the country. These stations number less than three per state on the average. Some states contain only one station that provides data usable for the prediction of wind speed. To predict wind speeds for the lifespan of a building or longer, it is necessary to have long-term, w~nd-speed records at a closely spaced network of stations. With the development of the ASOS, there is the potential to gather data at many more stations because the instrument package can be deployed in large numbers and its data can be retrieved by a centralized location. However, standardization in data recording, continuation of historical records, and organized archiving are critical to ensure that the data are usable to the wind-engineering profession. There are very few w~nd-recording stations located in coastal areas that are in the path of hurricanes. Moreover, w~nd-recording instruments are often damaged or without power during the passage of severe hurricanes. As a result, recorded data of hurricane winds are sparse. Design wind speeds in coastal areas given in the ANSI/ASCE 7-~8 standard are based on the Monte Cario numerical simulation of hurricanes. Closely spaced stations along the coastline are needed to obtain better definition of hurricane winds. Another option is to develop a deployable wind-recording station that can be placed in the path of an approaching hurricane. It is extremely important to improve current methods for collecting hurricane w~nd-speed data if reliable predictions of wind speeds are to be obtained. Wind speeds are also affected by wind climate and by topography. Mountains and valleys significantly affect the wind speed at various locations. Deployment of ASOS packages in a closely spaced grid network can provide data in topographically difficult terrain. Good examples of this type of terrain are Puerto Rico and the islands of Hawaii. Data gathered at many stations can help regional zonation of design wind speeds. Calibration of the anemometer sites is critical for improving w~nd-speed assessments for design and other purposes. It is recommended that a systematic calibration of the w~nd-speed variation due to roughness characteristics and nearby obstructions be undertaken for all existing anemometer sites. Standardization of wined data collection and archiving is critical. Wind data are used for a varied of applications including assessing wind forces on buildings, predicting atmospheric diffusion, extracting wind energy, and

The Nature of Wnd 51 · ~ Improving agricultural efficiency, and in the aviation industry. To meet these applications, a set of recommendations was set forth at the Workshop on Wind Climate held In Asheville, North Carolina, in 1979. These recommendations can provide a starting point in standardization of wind data collection and archiving (see Table 24~. In general, then, the wind data base can be expanded by establishing a closely spaced network of data-collecting stations, by developing rapidly deployable instrument packages in hurricane-prone areas, and by standardizing collection and archiving of wind data. RECOMMENDATIONS Measurements of Extreme Wind Speeds There is a need to confirm the maximum wind speeds that can occur in tornadoes, downbursts, hurricanes, and other high-speed winds. Technology in the form of portable Doppler radar (Bluestein and Unruh, 1989) and Laser-Infrared Doppler Atmospheric Radar is now available to directly measure speeds in tornadoes and downbursts. Small, easily transported balloon sounding systems (e.g., the Cross-Chain Loran Atmospheric Sounding System (CLASS) developed by the National Center for Atmospheric Research; also see Bluestein et al., 198S, 1989) are also available to explore the thermodynamics of the parent storm. Further, the research community has demonstrated the feasibility of intercepting such phenomena with both ground vehicles and light aircraft (Colgate, 1982~. Thus, in principle, it is now possible to obtain actual measurements of dangerous winds and related, in-cloud conditions and so refine the current, max~mum-speed estimates. Research on the development of instruments for probing tornadoes, downbursts, and their parent storms should be continued, and field programs to apply these instruments should be encouraged. It should be recognized that to collect data from several events watt require a multiyear effort. There is also an urgent need to establish the characteristics of the near- shore surface winds produced by hurricanes. Valuable data can be obtained by equipping existing installations, such as those operated by the Federal Aviation Administration and the U.S. Coast Guard, with recording systems for use in extreme events. Rapidly deployable, storm-resistant surface instruments are needed to establish detailed pressure and near-surface wind fields beneath these storms as they make landfall. Many of the research tools developed for exploring tornadoes and downbursts could be adapted to investigating hurricane winds, particularly during the critical landfall period. Singularly, the storm chase operations pioneered by the NOAA National Severe Storms Laboratory and researchers at the University of Oklahoma serve as models for the development of techniques for deploying such instruments in the path of advancing hurricanes. Further, to increase the probability of obtaining data from hurricanes, cooperative international

52 Hued Id the Built Environment TABLE 2.4 Meteorological Vanables Required for Wind Engineering Applications Surface wind data (10 m above ground) Variable Averaging time Primary 'acre ._ ~ O ~ ^~e ~ 1. Wind speed and direction (derived from components) 20 min Wind energy, pollutant dispersion, forces on structures, agriculture**, aviation Forces on structures Wind turbine, forces on structures 2. Peak wind speed/direction 3. Fastest one-minute speed/ direction 4. Fastest-mile wind speed/ direction 5. Standard deviation of 20 min Pollutant dispersion wind fluctuations 2 s 1 min Continuity with previous data Low-level wind data (at levels of approximately 100, 300, and 500 m above ground)*** Variable Averaging time Primary uses 1. Wind speed and direction 20 min Wind energy, pollutant dispersion, and their fluctuations forces on structures, aviation, numerical and physical modeling * ~ Associated meteorological variables At levels Averaging Variable above around time 1. Air 20 min Temperature 100, 300, and 500 m 1 mm 2. Barometric pressure 3. Relative humidity 3 m 3 m 100, 300, and 500 m 20 min 20 min 1 min 3/h Frequency Uses 3/h 8/day Wind energy, pollutant dispersion, agriculture, forces on structures Wind energy, pollutant dispersion, agriculture 3/h WiDd turbine icing, 8/day agriculture, forces on structures * These data should be obtained three times per hour (20 min module) for every hour of the day and every day of the year. ** Wind speed and direction data at 3 m above ground are also desired for agriculture. $*# These data should be obtained at three-hour intervals every day. Source: KC. Mehta, Proceedings of the Workshop on Wind Climate, Asheville, North Carolina, November 12-13, 1979. projects could be initiated with nations having high frequencies of tropical cyclone occurrences: the eastern coast of India, the southern coast of China, the eastern coast of Taiwan, the northern portion of the Philippines, and certain South Pacific islands.

The Nature of Wnd 53 There is a pressing need to deploy in the Caribbean a network of reliable, upper-air stations based on the CLASS balloon system and the profiler technology. These would replace the currently degraded balloon network and provide the data needed to develop improved models for prediction of humcane movement and intensity changes. There is also a need to install Doppler radars in the Caribbean (presently there are only three conventional radars to cover this region: one each at the National Weather Service ounces in Key West, Flonda, and San Juan, Puerto Rico, and the third in the Bahamas). Extreme Wind Climatology There is a need to better establish the distnbution of the occurrence of extreme wind speeds in the United States. Not only are there requirements for more data, but there are urgent needs for programs to ensure the quality of these data and to archive them in ways that facilitate analysis. Grazulis (1990) has made several recommendations for revising the national data base on tornadoes (maintained by the National Severe Storms Forecast Center, Kansas City, Kansas). These include documenting the basis for assigning an intensity value, standardizing assessment of damage (particularly to mobile homes), and providing training on damage assessment to National Weather Service personnel and others contributing material to the data base. Violent tornadoes and tornado outbreaks merit special efforts since these pose great threats to life in the central half of the United States and are the most difficult wind hazards to forecast. Violent tornado winds present the greatest challenges in building design and construction. For such cases, damage surveys should be made both to verity forecasts and warnings and to obtain data on building performance. These surveys, conducted by a multidisciplinary team, should be systematically carried out using both low- flying aircraft and ground inspections during postdisaster investigations: the survey team should be on call for immediate response. ~ , Cirazubs has also recommended the development of tornado climatology on a state (Schm~lin, 1988) or regional basis (Anthony, 1988) to allow for the strong influence exerted by local effects. Historical research should be undertaken to extend the data base to include all events that have been documented only in local newspapers and insurance records, with due recognition being given to biases in such data. With the deployment of the complementary NEXRAD and Terminal Doppler Weather Radar systems, there will be the opportunity to develop a data base on the frequency of occurrence of mesocyclones (the precursor circulation of many tornadoes and other strong, thunderstorm-related winds), to monitor the strong winds accompanying extratropical cyclones, and to observe the low-level (but not near-surface) winds in hurricanes. Routine measurements with these systems should provide additional data on the likelihood of damaging winds in remote areas. A program for establishing a

54 Wind Id the Built Environment climatology of strong winds using data derived from these Doppler radars should be developed and implemented in parallel with the deployment of the NEXRAD and Terminal Doppler systems. One of the problems frequently encountered in obtaining wind data during extreme events is that many facilities rely on commercial electricity service with limited, if any, provision for backup power. Dunng an extreme wind event, commercial service often fails before the extreme wind speeds occur, with consequent loss of data. Particularly in hurricane-prone coastal regions, NEXRAD and Terminal Doppler Weather Radar systems should be provided with full backup power capability to ensure that full sets of data become available.