Ralph J. Donaldson, Jr. methods lor Identifying Air Force Cambridge Research Laboratories severe thunderstorms Bedford, Mass. by radar: a guide and bibliography'
1. Introduction Thunderstorms have always been prominent among the meteorological targets observed by radar. They command attention because of their greater precipitation intensity and height. Moreover, some of the more severe and hazardous thunderstorms pose an im- portant short-range forecasting problem to which radar may be profitably applied. This paper is intended as an aid to the forecaster of severe thunderstorms. An attempt will be made to sketch the historical development and evaluate the current success of con- ventional radar techniques which are sensitive to the identification of severe thunder- storms. The bibliography includes most of the pertinent references by Canadian and United States authors and a few citations representative of significant work in this field in other countries. Radar meteorologists quite early learned how to distinguish between echoes from con- vective and stratiform precipitation. With convective precipitation, in which vertical and horizontal air speeds are comparable, the echoes are cellular, rather similar in verti- cal and horizontal scale, and display a tendency to form into linear groups. Convective echo intensities were qualitatively observed to be greater, in general, than the intensities of stratiform echoes, as indicated by echo brightness on the oscilloscope display. Further- more, convective echoes never, except in their decaying stages, displayed the melting-layer "bright band." Some of these early discoveries were reported by Bent (1946) and Wexler (1947a) and were confirmed and organized by Byers and Braham (1949) in their compre- hensive report on the Thunderstorm Project. Convective storms may be classified conveniently into four categories of increasing dis- turbance to human activity: showers, ordinary thunderstorms (showers accompanied by lightning but no hail or damaging wind at the ground), hailstorms, and severe thunder- storms (usually these contain large hail as well as damaging wind). Many of the radar investigations of convective storms have been concerned with techniques for distinguish- ing among the various categories. However, the four types of storms are not clearly sepa- rated from one another: they form a continuum with class boundaries defined only by conveniently observable events. For example, the presence of small hail at the ground depends on surface temperature and the height of the wet-bulb OC isotherm as well as on favorable hail growth conditions within the convective cloud. The definition of a severe thunderstorm is quite arbitrary. In this discussion, a thun- derstorm will be classified as severe if it produces a tornado or straight-line wind capable of inflicting major structural damage to dwellings and snapping off tree trunks and large limbs. An additional criterion of severity is the appearance of hailstones with major di- mension greater than 3/4 inch, whether or not tornadoes or damaging winds are also released. Large hail, tornadoes, and other damaging winds may be regarded not only as a severe weather hazard, but also an indicator of the intensity and persistence of convection in a thunderstorm. Large hail requires sufficient time to grow in an environment rich in wa- ter droplets at temperatures below OC. This implies an adequate supply of low-level moisture and an unusually intense updraft to retard the fallout of the growing hailstone. Long hail swaths deposited by some severe storms require a persistent as well as intense i Under the title, "Radar as a Severe Thunderstorm Sensor," an earlier version of this paper was presented at a Radio Meteorology Symposium during the XIII General Assembly of the International Union of Geodesy and Geophysics, Berkeley, Calif., 24 August 1963. 174 Vol. 46, No. 4, April 1965
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society updraft; therefore, the storm must be organized in such a manner that its updraft draws upon a continuing supply of moist, buoyant low-level air. The initiation and mainte- nance of a tornado is in itself a highly organized event, and it seems to occur in a storm in which some degree of organized air motion persists during the life of the tornado. The identification of a severe thunderstorm by radar, then, depends on the observation of storm features which reflect either the intensity or the persistent state of organization of the convective process. Not surprisingly, the most successful of the conventional radar methods involve measurement of echo top heights; echo intensity or reflectivity; and un- usual, suggestive, and generally persistent echo configurations. A high echo top, and especially an unusually reflective echo at high altitudes, indicates an updraft of sufficient intensity to carry radar-detectable particles to the observed height. The intensity of a storm echo is a direct indication of the size and concentration of precipitation particles, and hence an unusually intense echo indicates an updraft magnitude and organization favorable for growing large particles and for converting the supply of water vapor to pre- cipitation at a high rate per unit volume. This change of phase of water will also release a great deal of latent heat over a limited region, which will contribute significantly to the buoyancy of the storm and raise its top. Finally, a persisting echo configuration suggests an organized pattern of convection. Later discussion will show how one type of organ- ized convection can produce long swaths of large hailstones one inch or more in diameter. Evidently tornadoes also occur frequently with this same pattern. The study of the radar characteristics of severe thunderstorms has revealed a few sur- prising discoveries, however, and has contributed to new understanding of these storms. For example, carefully measured storm echo tops show penetrations into the stratosphere by as much as 20,000 ft. A highly reflective echo observed in some severe storms has led to studies of the scattering properties of hail of various sizes and states and inferences re- garding the hail growth condition in these storms. Recent observations of hooked echoes and echo-free vaults have provided clues to the form of the air flow in storms with these unusual echo configurations. The intensive coverage of radar over a wide area provides an opportunity to locate storms and, hopefully, to recognize the hazardous ones. Unfortunately, a unique, reli- able, and objective radar echo signature for severe thunderstorms has not yet been found. However, research during the past decade has furnished a guide to certain radar echo characteristics which frequently attend very intense convection. The experience with ob- jective measures, such as echo height and reflectivity, is now sufficiently extensive at cer- tain locations that tentative working thresholds may be specified which divide frequent from rare occurrence of severe thunderstorms. The most successful methods are listed in the summary, following a discussion of their discovery and development. Doppler radar techniques have scarcely been applied as a tool for observing thunder- storms. However, recent developments in Doppler radar offer great promise, furnishing an attractive capability of direct measurement of wind velocity, and hence an unambigu- ous method for identification of tornadoes and other severe windstorms. A comprehen- sive discussion of the potential use of Doppler radar in severe storm identification was given by Lhermitte (1964). Also, the reviews of severe storm analysis and the entire field of radar meteorology by Atlas (1963, 1964) incorporate extensive, detailed surveys of Doppler radar techniques. 2. Arrangement and Severe weather occasionally appears with isolated thunderstorm echoes, but most of it motion of a group occurs in echoes forming a part of a line. In a case study, Bigler (1955a) prepared com- of convective echoes posite echo patterns from nine radars in the central plains of the United States, and he noted that tornadoes and hail occurred during the organization of echoes into a line pat- tern which eventually extended over a length of 1000 miles. Further study of many cases by Inman and Bigler (1958) showed that 82 per cent of tornado-producing situations could be classified as an echo line, even though some of the most destructive major tornadoes occurred with isolated echoes. A wave-like pattern in an echo line was first described by Wexler (1947a, b) and related to a synoptic pressure feature. Tepper (1950) noted the occurrence of a tornado at a bend in an echo line which he felt to be compatible with the intersection of two pressure- jump lines. Nolen (1959) described a characteristic line echo wave pattern, or LEWP, which appeared in near proximity to about three-fourths of the tornadoes he studied. 175
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 An example of the development of this kind of pattern is given in Fig. 1. Stout, Black- mer and Wilk (1960) made a detailed study of two cases of distortions in squall lines. They found that die LEWP, that part of the line which was distorted in the direction of line movement, corresponded closely with the areas of most hail and damaging winds. The upper wind field suggested a mid-tropospheric jet at 4-5 km in the vicinity of both LEWPs. An evaluation of LEWPs by Cook (1961) revealed that 40 of a total of 49 of them had severe weather, most of it in the more advanced section of the wave pattern. Although this is impressive evidence, which suggests a preference for tornadoes to occur in a mesoscale disturbance of sufficient intensity to cause distortions in an echo line, an observation of a LEWP is not entirely reliable because of its dependence on radar char- acteristics and range and because it may be mistaken for small irregularities in which severe weather is not especially likely. A study of echo motions in severe storms in the midwest United States by Stout and Hiser (1955) showed a high degree of association of echo convergence and vortex motion with severe winds and hail. This motion is revealed by echo mergers, curling motion of an appendage, echo rotation, or line rotation. Staats and Turrentine (1956), in their study of the devastating Blackwell, Oklahoma, and Udall, Kansas, tornadoes, also noticed strong convergence of echoes toward the tornado-producing storms, but Hiser (1958) later noted that echo convergence in sub-tropical Florida is not generally accompanied by tor- nadoes. Bailey (1955) reported an indication of tornado occurrence in a band of maxi- mum vorticity of echo motion, and most active convection in the area of maximum echo speeds. The association of high echo speeds (30 to 60 knots) with severely-damaging
FIG. 1. Development of line echo wave pattern (LEWP) observed from Oklahoma City on 22 April 1957. Range marker inter- val is 20 miles. Tornado occurred at arrowhead in panel d. (From Nolen, 1959.)
176
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society winds, even on occasions with surprisingly low echo tops, is agreed upon by many investi- gators, for example, Stout and Hiser (1955), Staats and Turrentine (1956), Hiser (1958), Fujita and Brown (1958), and Changnon (1960). Arrangement of echoes in wavy lines and convergent, rapid motions of individual echoes serve as an alerting mechanism for a forecaster, but these echo characteristics have not been tested sufficiently to provide the sole basis for severe storm identification. 3. Individual echo A remarkable echo pattern associated with an Illinois tornado was photographed in 1953 shape: horizontal by D. A. Staggs and reported by Stout and Huff (1953), and later analyzed in detail by configuration Huff, Hiser and Bigler (1954). A projection from the right rear side of the echo length- ened and curled cyclonically to form a hooked or "figure 6" echo, illustrated in Fig. 2. With further development the hooked echo closed upon itself and filled in, leaving a large bulge projecting from the main echo body. A visual observation of a hooked echo with a tornado had first been reported by G. Austin (1945), but unfortunately the echo was not photographed and the observation was not disseminated widely. However, this distinctive echo shape was reported frequently after 1953. Within a few months after the Stout and Huff observation, other investigators recognized the hooked echo pattern and noted its association with tornadoes in Waco, Texas and Worcester, Massachusetts (Freeman, 1953; Blackmer, 1954; and Penn, Pierce and McGuire, 1955). Fujita (1958) made a detailed analysis of echo alignment and motions in the Illinois tornado reported by Stout and Huff, and combined this with a mesoanalysis of the sur- face pressure and wind field. He found a tornado cyclone, identical to the type discov- ered by Brooks (1949). The analysis just before the tornado touched down is given in Fig. 3. The echoes around the tornado cyclone resembled hurricane spiral bands though on a much smaller scale. The tornado was located one to two miles south of the cyclone center, at the ring of maximum wind. The "eye" at the pressure center of the tornado cyclone was coincident with the encircling "figure 6" echo in the early stages and later was located in a small area of weak echo surrounded by more intense echo. Fujita also found a convergence toward the tornado cyclone of any small echoes which appeared within 15 miles of the circulation center. The impression of a rotating hook echo in the Stout and Huff tornado, suggested by Fujita's analysis, was illustrated very clearly by the observations of echo motions in a Kansas tornado by Garrett and Rockney (1962). Successive photographs of the storm
FIG. 2. Hook echo of Champaign, Illinois, tor- nado photographed on APS-15 radar scope by Don- ald A. Staggs of Illinois State Water Survey, Ur- bana, 111., on 9 April 1953. This is the first occasion in which a tornado hook echo was recognized and photo- graphed. At this time, 1715 CST, the tornado was about 10 miles to the NNE of the radar, in the southern end of the hook. (Photograph courtesy of Mr. Glenn E. Stout.)
177
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 echo, when viewed as a time-lapse movie, revealed an inward-spiralling and cyclonic mo- tion of fine-scale irregularities at the edge of the hook echo. The hooked echo has also been observed when no tornadoes were present but the storm was severe in other ways. Das, De and Gangopadhyaya (1957) found the typical hooked echo with destructive thunderstorm squalls but no tornado, and Lintner and Atlas (1956) and Schaefer (1960) found the hooked shape revealed in severe hailstorms, which also had some high winds but no tornadoes. All of these studies suggest rather strongly that the hooked or "figure 6" echo is a characteristic of a well-organized tornado cyclone in which large hail is likely to be grown and in which tornadoes frequently, but not always, occur. Large thunderstorm echoes with echo-free notches or echo fingers protruding, or a clearly scalloped edge, were found by Harrison and Post (1954) to be associated quite reliably with hail of 1/2 inch or greater diameter. They believed that these echoes iden- tified hail shafts falling on the fringes of the storm. Sudden changes in the protuberances, sometimes occurring within seconds, indicated that the peripheral hail shafts often fall in sudden bursts. Stout and Hiser (1955) also noted a peculiar notched echo shape in a damaging hailstorm, and Sarrica (1958) showed how the long echo fingers moving toward the right of the parent echo motion in an Italian hailstorm were consistent with the fall of small hail (up to 1 cm diameter) in strong shear. By way of tempering unbridled enthusiasm for tornado hooks, fingers, etc. as severe storm indicators, Bigler (1955b) pointed out that most of the tornadoes that he surveyed did not have the characteristic hook shape protruding from the echo obtained on search- type radars having poor elevation resolution. Chances of observing an unmistakable tornado hook are better at short ranges and with radars having a narrow antenna beam in both azimuth and elevation, but even so some major tornadoes which were remarkably devastating (e.g., the Blackwell, Oklahoma tornado) displayed the hook echo at ranges up to 80 miles. It has been the experience of many observers that hooks, notches, and the like some- times occur on the periphery of a convective echo without a report of a tornado, damag- ing winds, or large hail. Sometimes there is uncertainty whether more complete report- ing of surface events might have revealed severe weather manifestations, but often there is reasonable assurance that no severe weather has occurred. However, there are few, if any well-documented cases in which no severe weather occurred when the projecting hook
FIG. 3. A mesoanalysis of the wind and pressure field and radar echo of the Champaign, Illinois, tor- nado cyclone at 1710 CST, 9 April 1953. (From Fujita, 1958.)
178
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society echo and its development presented the appearance of a spiral band wound about a cy- clonic vortex, on the 5- to 25-mile scale of a tornado cyclone. (The observation of cyclonic rotation in a seemingly non-violent thunderstorm in Southern Rhodesia by Soane (1957) is a possible exception.) Because of the weight of past evidence, weather forecasters with a radar at their disposal are virtually unanimous in regarding the characteristic cyclonic tornado hook as sufficient basis for issuing a tornado warning to communities in its ex- trapolated path. It is, of course, realized by forecasters that the hook echo does not guarantee a tornado. What is really required is a method for direct measurement of dangerously intense vortex motion. Doppler radar, therefore, should eventually be a most helpful tool in identifying tornadoes. 4. Individual echo The vertical structure of radar echoes may be studied by moving the antenna in a vertical shape: vertical plane. The echoes are usually displayed on an RHI (range vs. height) scope. Vertical configuration echo structure may also be deduced by some variation of the "CAPPI" technique devised by Marshall (1957) in which the antenna is caused to scan about a vertical axis over a series of elevation angle increments, and constant-altitude planes are synthesized from the scanned volume. A word of caution is necessary about the errors possible in the resolution of vertical echo structure. Aoyagi (1963), Probert-Jones (1963), and Donaldson (1964) discussed the errors in echo top heights caused by power returned in the side lobes and outer regions of the main lobe of a radar antenna. If the core of the thunderstorm is unusually intense, over- estimates of the top in excess of 10,000 ft are possible when the product of range (in nau- tical miles) and antenna half-power beam width (in degrees) is equal to 100. Even larger errors are possible, and have been recorded, at greater ranges or beam widths. In the following discussion, the echo heights reported by Douglas are likely to be rela- tively free of overestimates, since he used a radar with a 3/4-degree vertical beam width and restricted his measurements to ranges within 80 miles. My measurements, using the CPS-9 radar with 1-degree beam width, are not as accurate, but most of them were taken well within 100 miles and are not likely to be in serious error. a. Echo top measurements. The height of echo tops was recognized as an indicator of the intensity of convection from the very earliest days of radar meteorology. Bent (1946) suggested that a low echo top height, less than 15,000 ft, was the best radar evidence avail- able at that early time that a shower had not developed into a thunderstorm. Later re- search, particularly in the Soviet Union, confirmed the idea that showers and thunder- storms may be distinguished on the basis of echo top heights, but the critical height was found to be much higher. One of the more comprehensive attempts to distinguish show- ers and thunderstorms by radar was made by Kotov (1960). His definition of echo top, the maximum vertical extent of an echo intensity corresponding to a rainfall rate of 1 mm hr~\ is independent of range and radar characteristics, which contribute significantly to the variability in the height of the conventional or minimum detectable echo top. Kotov found the altitude of the — 22C isotherm to be a rather sharp critical height di- viding shower and thunderstorm echo tops. Nearly all, or 93 per cent, of the tops of the thunderstorm echoes corresponding to 1 mm hr_1 rainfall rate extended above the height of — 22C, and the same large percentage of shower echoes failed to reach the — 22C altitude. Echo tops in hailstorms are generally taller than in thunderstorms which do not deposit hail on the ground. The relationship of hail production to echo top height, unfor- tunately, is not as clear-cut as the distinction between showers and thunderstorms. Don- aldson (1958, 1959) found that the probability of hail occurrence somewhere at the surface in New England thunderstorms increased from zero at echo top heights below 20,000 ft to 48 per cent for echo tops above 50,000 ft, with a slight secondary maximum of hail oc- currence at echo tops between 30,000 and 35,000 ft. (See Fig. 4.) Median echo top heights of 43,000 ft and 38,000 ft were found for hailstorms and thunderstorms without hail, respectively. In the high plains of Alberta, in the lee of the Canadian Rockies, Douglas and Hitschfeld (1959), supplemented by later analyses by Douglas (1960a, 1961), found the same sort of relationship between echo top height and the probability of hail occurrence, shown in Fig. 5. Alberta hailstorms, however, occur with echo top heights about 10,000 ft lower than New England hailstorms. Also, the secondary maximum in hail probability in Alberta, located about 5000 ft below the tropopause, is more promi-
179
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965
FIG. 5. Hail occurrence probability in Alberta thunder- FIG. 4. Upper curves: Frequency distributions of New Eng- storms, as a function of maximum echo top height above land thunderstorm echo top heights classified according to terrain (3000 ft MSL) during the lifetime of a storm (left observation of hail or absence of hail at the ground. Lower curve), and environmental temperature at that height (right curve: Hail occurrence versus echo top height derived from curve). Number of storms are given in parentheses. (From upper curves. (From Donaldson, 1959.) Douglas, 1961.)
nent than the same feature in New England. Douglas found that hailstones with a diameter of 4 cm or more rarely appeared unless the echo tops exceeded 30,000 ft above terrain, which is near the median tropopause height in Alberta. Geotis (1963) also found, during his 1961 study of New England hailstorms, that the largest hail sizes gen- erally appeared in the storms with the tallest echoes, though 1-1/2-inch hail was reported on one day with echo tops as low as 25,000 ft. Schleusener and Marwitz (1963) discovered that the mean echo tops of hailstorms and thunderstorms without hail in Colorado, in the lee of the U. S. Rockies, were very similar to their counterparts in Alberta measured by Douglas and Hitschfeld. In Texas, where the tropopause is generally higher than in New England and much higher than in Alberta, Inman (1961) reported a corresponding increase in height of the distribution of echo tops, averaging a few thousand feet higher than New England thunderstorms, and with increasing occurrence of larger hail with higher echo tops. Douglas (1963) summarized the Alberta, New England, and Texas hail probability as a function of echo top height. If echo tops reach 40,000 ft the probability of hail is negli- gible in Texas, substantial (about 40 per cent) in New England, and almost certain in Alberta. It is evident that penetration of an echo into the stratosphere is a more universally-applicable measure of the intensity of convection than height alone. Donald- son, Chmela, and Shackford (1960) demonstrated this by comparing echo tops with the best available estimates of tropopause heights in New England. Of those thunderstorms which did not produce hail, only one-third extended above the tropopause and half of them failed to reach an altitude 4000 ft below the tropopause. However, most of the hailstorms penetrated the tropopause; the larger the hail, the greater the penetration. For example, 80 per cent of the storms which released 3/4-inch or larger hail had echo tops above the tropopause. The median and maximum tropopause penetrations of these large-hail thunderstorm echo tops were 5000 and 15,000 ft, respectively. Thunderstorms which produce tornadoes tend to have the highest echo tops of all. Blackmer (1954) noted the tall echo top associated with the Worcester tornado; it ex- tended beyond the 50,000 ft limit of the RHI scope of his radar. Donaldson (1958, 1959) observed echo tops of major tornadic storms to reach 47,000 to 57,000 ft during their life- times, and Donaldson, Chmela, and Shackford (1960) found that some form of severe storm invariably occurred on the five days in their study when echo tops exceeded the tropopause by 10,000 ft or more, with tornadoes occurring in the tallest storms on four of these five days. Inman (1961) showed that echo tops of Texas tornadoes and damaging
180
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society windstorms, which average over 50,000 ft, are about 5000 ft higher than hailstorms, but Arnold's (1961) case studies of selected severe storms in Texas did not show any temporal relation between tropopause penetration and destructive activity. Browning, Donaldson, and Lamkin (1963) studied a family-type tornado outbreak in Oklahoma in which the echo tops of the tornadic storms towered above the tropopause by more than 20,000 ft. The tallest echo at 62,000 ft was measured at a range of 20 miles with a radar beam width less than 1 degree in elevation, so an error of more than 2000 ft is highly improbable. Farther south, in tropical Panama, Crow (1961) has noted that destructive winds arise from the few convective storms with unusually tall echoes of 25,000 to 45,000 ft. High echo tops in India also accompany the more severe thunderstorms. De and Sen (1961), for instance, related stronger squalls from cumulonimbus with greater vertical growth. Pautz and Doloresco (1963) have initiated a study of the relation of severe weather oc- currence in the United States east of the Rockies to tropopause penetration by echo tops. Their preliminary results, in regions where data are not too scarce, indicate that thunder- storms which produce tornadoes or hail greater than one inch in diameter have echo tops extending on the average about 10,000 ft above the tropopause. The trend of evidence accumulated so far suggests that, although hail and damaging windstorms may occur with relatively low echo tops, severe weather is increasingly prob- able as echo tops rise above the environmental tropopause. In particular, most thunder- storms with echo tops penetrating the tropopause by 10,000 ft or more are likely to be severely destructive, with an excellent chance of producing a major tornado. b. Organized convection indicated by vertical echo structure. Browning and Ludlam (1962) made an intensive study of a severe hailstorm which occurred in southeast Eng- land in July 1959. The largest hail was about two inches in diameter, falling near Wokingham. Scope photographs of five radars and nearly 2000 reports of surface weather were available for the investigation. Hail fell almost continuously over a swath 130 miles long, and during a period of about 40 minutes the maximum hailstone diameters exceeded one inch. Coincident with this period, the radar echoes showed a rather unusual con- figuration in vertical structure which changed only in minor details. The right forward quadrant of the storm was echo-free below an altitude of 12,000 ft, with an overhanging echo above this clear space. At the rear of the overhang the echo descended to the ground in a nearly vertical, sharp, intense wall. Just ahead of the wall, in the overhang region, a vault or clear space penetrated upwards to 15,000 ft over a depth of about 2 miles in the direction of storm travel. The echo top of the storm reached its greatest
FIG. 6. RHI photographs illustrating the similarity of the echo structure of the Wokingham, England, and Geary, Oklahoma, storms during their most intense phases, with a sketch illus- trating the key features. Storm motion in both pho- tographs is from right to left. (From Browning and Donaldson, 1963.)
181
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 FIG. 7. Simplified Brown- ing-Ludlam model of air- flow in organized storm and environmental winds, pro- jected into the vertical plane of storm motion. Dotted paths indicate tra- jectories of hailstones fall- ing out of top of updraft and re-cycling into foot of updraft for a second trip. (Sketch courtesy of Dr. Keith A. Browning.)
height of 45,000 ft during this period and remained as a single large dome almost directly above the vault. Fig. 6 shows a sample vertical section through this storm which illus- trates these echo features. Browning and Ludlam deduced a model of the airflow through the storm during its intense phase, which was consistent with the echo structure as well as other meteorological observations. A simplified two-dimensional projection of the airflow in the plane of the storm motion is sketched in Fig. 7. They proposed the entry of the updraft air, coming mostly from ahead and partly from the right of the storm, into the overhang region where it helps to prevent the fall of the moderately intense precipitation there. The vault is near the most intense axis of the updraft, which is tilted toward the rear and left side of the storm as it rises, opposing the wind shear. The updraft air leaves the storm at high altitudes in the general direction of storm movement. Downdraft air enters the storm at moderate altitudes and is cooled by evaporation of precipitation from the over-lying updraft as it descends. In the overhang region some of the descending small hailstones are caught into the updraft for a second trip, where they may readily grow to large hail. This advanced state of organization, which is capable of continuous production of large hail, is regarded as quasi-steady. The Browning-Ludlam model gained further support when Donaldson (1962) discov- ered a remarkably similar vertical echo structure in a tornadic storm in Oklahoma, illus- trated in the middle panel of Fig. 6. Surface observations of hail in this storm were sparse and unreliable as to time, but Ward (1961) made accurate visual observations of the time and location of several tornadoes near Geary, Oklahoma. Further analysis of the Geary storm by Browning and Donaldson (1963) placed the tornadoes beneath the vault, which endured for an hour and was echo-free to a maximum height of 24,000 ft. Above this height, the vault was still apparent as a roughly cylindrical region of low re- flectivity up to about 40,000 ft. A most revealing demonstration of the identity of the wall echo and the typical tor- nado hook echo is provided by Fig. 8. The echo photograph on the left was obtained in a low-altitude PPI (or horizontal) scan of the WSR-57 radar. The picture on the right was put together in this same low-altitude horizontal plane from a series of FPS-6 RHI (or vertical) scans, closely spaced in both time and azimuth angle. (One of these vertical scans, approximately in the direction of travel and across the hook, is reproduced in the middle panel of Fig. 6.) The similarity between the two patterns is striking when allow- ance is made for the loss of detail in the synthesized picture. It now becomes clear why the tornado hook echo has only rarely been detected by radars with wide vertical beams and can scarcely be distinguished at ranges much beyond 50 miles even with narrow-beam radars: if an appreciable part of the radar beam intersects the overhanging precipitation forward of the hook, the precipitation-free area below is filled in with echo and the hook (as well as the vault and wall) is masked. The vault-hook echo structure was again detected in severe thunderstorms near Okla- homa City by Browning, Donaldson and Lamkin (1963). These storms produced a fam- ily of tornadoes and released hail up to four inches in diameter. Browning (1964) found
182
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society conclusive evidence in these storms that the vault region is an area of strong updraft: A curtain of precipitation from the left flank of one storm was drawn into the low-level inflow of its neighbor immediately to the north. The base of the curtain tilted upward as it approached the hook and vault area of the neighboring storm. It was possible to discern the curtain, acting as a natural tracer of the airflow, ascending into the vault. The Wokingham, Geary, and Oklahoma City storm analyses provide limited but con- vincing evidence of a preferred configuration for severe thunderstorms in temperate lati- tudes. Under proper conditions of buoyancy, wind shear, and low-level moisture, con- vection may proceed to an advanced state where the updraft is intense, persistent, and organized with the downdraft into a continuously self-regenerative mechanism, in which large hail is produced by the re-cycling of small hail. Such a form of organization evi- dently is favorable for the generation of tornadoes, though the manner in which this comes about is not yet known. This state of organized convection may be recognized by means of radar as a tornado hook echo or completely-enclosed echo hole in plan view, or a vault and wall configuration in vertical section, provided the pattern is not distorted by too wide an antenna beam or by excessive attenuation. Most present-day radars are not capable of satisfying the stringent resolution requirements except at ranges well under 50 miles for all except unusually large vaults and high overhangs, which explains in part why this type of organization has not been recognized earlier.2 Evidence of the relationship between vaults and tornadoes may be traced back several years before Browning and Ludlam proposed their model, although the early observations were not interpreted in terms of an organized convective structure. The association of a tornado with an echo-free region extending up to great heights within a thunderstorm echo was first noted by Garrett and Tice (1957), and an RHI photograph presented by Schuetz and Stout (1957) in a tornado situation also shows a vault near the tornado loca- tion. Bigler (1958) obtained a number of excellent echo photographs, both in elevated plan view and in vertical (RHI) section, which demonstrated the close proximity of tor- nado damage on the ground with a persistent vault which penetrated up to a height of 36,000 ft. Bigler and his colleagues suggested that the absence of precipitation within the vault was a consequence of centrifugal ejection of particles from the center of a rotating tor- nado cyclone. Indeed, the observations of Fujita (1958) and Garrett and Rockney (1962) of cyclonic rotation of echo features in the hook echo suggest particle generation in a ro- tating updraft. However, Browning and Donaldson noted that the creation of an echo- free region by centrifugal force alone would require unreasonably high tangential veloci- ties over a rather extensive area, considerably larger than the usual tornado vortex. They preferred to explain the vault in terms of its intensity and persistence. Cloud particles forming within the vault simply do not have sufficient time to grow to radar-detectable size until they are carried to great heights in the high-velocity core of the updraft. Also, recycling hailstones which attempt to enter the vault will penetrate the peripheral regions of the updraft and eventually encounter vertical velocities of sufficient magnitude to carry them above or around the vault. The maintenance of a vault, then, can be explained most reasonably as a product both of the rotation and the persistent high intensity of the updraft. Consequently, a clear observation of a persistent vault is reliable evidence of a well-organized storm in which large hail is likely, with favorable conditions for the appearance of a tornado or other damaging wind in or very near the vault. 5. Measurement of Some of the more useful techniques for identifying and studying severe thunderstorms Storm reflectivity ^ave involved measurement of the echo reflectivity, or sum of the back-scattering cross sections of the irradiated particles per unit volume. It is convenient to express reflectiv- ity in terms of equivalent radar reflectivity factor, Zc, which is defined as the summation per unit volume of the sixth power of the diameters of spherical water drops in the Ray-
2 In May 1964, on two more occasions of severe thunderstorms, vaults were observed from an Oklahoma City FPS-6 radar by W. E. Lamkin. A tornado occurred in one storm and "baseball" hail was reported in the other. The vault in the hailstorm must have been very broad, for it was located at a distance of 110 nautical miles from the radar and yet was clearly defined. 183
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 leigh scattering region (drop diameters less than one-tenth radar wavelength) which would scatter back the same power as the measured reflectivity. Conventionally, Ze is expressed 6 3 in units of mm /mm . In case the scatterers are indeed small water drops, Zez=.Z and it may be related to rainfall intensity by an empirical relationship. For thunderstorm rain- fall, Jones (1956) found the relationship Z = 486.R137 (R in mm hr1), which, for Z of 103, 104, and 105, gives rainfall rates of 2, 9, and 50 mm hr"1, respectively. When the particle diameters are larger than one-tenth of the radar wavelength, the back-scattered power is no longer proportional to the sixth power of particle size. Such large particles are in the Mie scattering region, where the back-scattered power is a com- plex function of particle size with numerous maxima and minima and an overall increase at a rate considerably less than the sixth power of particle diameter. The scattering properties of hail in the Mie region were studied by Atlas, Harper, Ludlam and Macklin (1960). They obtained experimental measurements of the back-scattering cross sections of large ice spheres. Their measurements were confirmed and extended by the computa- tions of Herman and Battan (1961). It is important to note that a particle may be in the Rayleigh scattering region for one radar wavelength and in the Mie scattering region for a smaller wavelength. In this case the value of Ze computed from the power returned at the smaller wavelength would be less than the value of Ze given by the larger wavelength. This discrepancy is not at all troublesome in rain or snow with radar wavelengths of 3 cm or greater, but it is a serious problem with the larger particle sizes that are possible with hail. Still another difficulty in interpreting measurements of Ze in hail are the complexities of the scattering properties of non-spherical shapes and mixtures of water and ice which frequently occur in hailstones. Despite these difficulties, it is best to express radar measurements of storm echoes in terms of Ze, since this is far more conservative with respect to radar wavelength than re- flectivity. A measurement of Ze may be made by use of Probert-Jones' (1962) revision of
FIG. 8. Comparison of the 0-degree full gain WSR-57 echo, during the most intense period of the Geary, Oklahoma, tor- nadic storm, with the 0-degree echo synthesized from a series of full gain FPS-6 RHI photographs. The region marked "©" in the FPS-6 echo synthesis was free of echo when ob- FIG. 9. Estimated occurrence probability of hail and severe served by the WSR-57, giving a hooked appearance to the weather in New England thunderstorms, as a function of echo projection. (From Browning and Donaldson, 1963.) maximum reflectivity measured at an altitude of 30,000 ft.
184
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society the radar equation, modified so that variables are in commonly-used units, assuming com- plete filling of the beam and no attenuation:
KK (i)
In this equation, Pr and Pt are received and transmitted power, respectively, expressed in watts; G is antenna gain; 0 and
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 tive. For example, there is no sense in attempting to find a threshold echo with greater precision than its expected error. Therefore the minimum interval of the steps in re- ceiver gain should be 5 db. Yet, in view of the variation of maximum thunderstorm reflectivity over four or five orders of magnitude, a measurement precision of half an order of magnitude should be sufficient to separate the strong ones from the weak ones. Also, one cannot expect 3-cm measurements of heavy thunderstorm rainfall to be mean- ingful, with the handicap of severe attenuation at this wavelength. On the other hand, 3-cm attenuation imposes much less difficulty at altitudes above the OC level where ice predominates and large concentrations of liquid water are found only in relatively small regions in cores of updrafts. Compared with water, ice of nearly all sizes has a consid- erably lower ratio of total attenuation cross section to back-scattering cross section. Therefore 3-cm reflectivity measurements of thunderstorms should be restricted to alti- tudes above OC, but 10-cm radar may be used with confidence at all altitudes. The foregoing points are illustrated by the relative degrees of success attained by various methods in classifying thunderstorms according to their measured reflectivity. Donaldson (1958, 1959), using a simple stepped-gain threshold technique with a 3-cm radar, failed to find significant low-altitude reflectivity differences between New England thunderstorms with hail reported at the ground and those without hail. However, Cope- land (1958), using integrated contours presented by a 10-cm radar, found a good relation- ship between hail occurrence during one stormy day and regions of abnormally high re- flectivity at the ground. Geotis (1963) followed this lead with a systematic study of New England hailstorms during 1961. He found that 10-cm values of Ze greater than 3 X 105 mm6/m3 near the ground are almost always present with hail. In every case, when the maximum reported hail diameter exceeded 2 cm, the maximum Ze of the storm was above 10® mm6/m3. This is convincing evidence of the capability of radar to identify hailstorms on the basis of their reflectivity, when care is taken to minimize measurement errors and there is assurance that attenuation is negligible. Nevertheless, the simple threshold technique for measuring thunderstorm reflectivities with 3-cm radar has proved its value at altitudes above the OC level. In 1956, I meas- ured maximum reflectivities at various heights in a line of thunderstorms (reported in 1958). This sort of thing had been done many years earlier by Langille and Gunn (1948), but in the 1956 observations one of the storms produced several tornadoes and much hail. In contrast to the other storms, which generally displayed the greatest reflectivity near the ground, the tornadic storm showed a rather sharp maximum of reflectivity at a height of 20,000 ft which was 25 times in excess of the reflectivity near the ground. Subsequent measurements by Chmela (1958), Donaldson (1959), Ludlam (1959), Wilk (1960, 1961), Arnold (1961), Atlas and Ludlam (1961), Inman (1961), and Ward (1961) provided further examples of abnormally high reflectivities aloft in hailstorms and tornado-bearing thunder- storms in various parts of the United States, Italy, and England. All of these measure- ments used 3-cm radar and were based on estimating a threshold of returned echo power. A survey of the vertical profiles of maximum 3-cm reflectivity in many New England thunderstorms by Donaldson (1961a) showed significant differences among the reflectivities above 15,000 ft of thunderstorms with rain but no hail, hailstorms, and tornadic storms, with the greatest difference at a height of 30,000 ft. A distinct maximum in reflectivity at a height of 20,000 ft or more was generally observed in the severe thunderstorms. The most sensitive indicator of thunderstorm severity in this New England sample is the mag- nitude of 3-cm Ze at an altitude of 30,000 ft. Fig. 9, derived from data presented by Don- aldson (1961b), shows the best estimate of the probability of the occurrence of hail or a severe thunderstorm as a function of this parameter. The exact form of these probability curves are dependent on the criteria for sample selection and the nature of corrections for unreported non-severe thunderstorms. However, regardless of the sampling and correc- 4 6 3 tion techniques, a value of Zc = 3 X 10 mm /m at 30,000 ft separates negligible from frequent occurrence of severe manifestations in New England thunderstorms. A survey of Texas thunderstorms by Inman (1961), also using 3-cm radar, revealed a similar set of characteristics, though the Texas storms were somewhat taller than their counterparts in New England. Wilk (1961) was handicapped by having to work with a 3-cm radar which was limited to a very few discrete settings of antenna elevation angle, and by the lack of tornadoes during his period of data acquisition. However, he found
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society that the maximum reflectivity for hailstorms in Illinois was most frequently located be- tween 15,000 to 19,000 ft, and this was the height interval showing the greatest difference between hail and rain-only thunderstorms. The vertical structure of reflectivity as a means of distinguishing between showers and thunderstorms has received the attention of several Soviet investigators. For example, Sal'man (1957) and Kotov (1960) found that 10-cm reflectivity diminished exponentially with height above the level of the maximum, with showers decreasing at three times the rate of thunderstorms. The modal value of the decrease in thunderstorms is 10"°3 per km. Markovich and Muchnik (1960), apparently using a 3-cm radar, found thunderstorm reflectivity maxima aloft, generally three to five times the reflectivity lower down. Their illustrated example shows a five-to-seven-fold decrease in reflectivity from a height of 4 km down to the ground. There are two interesting problems in connection with measurements of vertical 3-cm reflectivity structure. First, how is it possible to explain the extremely high values of re- flectivity factor that are observed occasionally? And, second, what is the meaning of a 7 8 3 persistent reflectivity maximum aloft? Values of 3-cm Ze in the vicinity of 10 mm /m have been reported as extreme cases by Ludlam (1959), Atlas and Ludlam (1961), Don- aldson (1961a), and Inman (1961). Such a high reflectivity factor could be given by cloudburst rain intensities approaching the world's record, or by surprisingly high con- centrations of large hail. However, large hail is never found on the ground in sufficiently high concentrations. Atlas and Ludlam in England had some good measurements of large hailstones on the ground, but they did not find the concentrations necessary to explain their high reflectivities. Douglas (1960b, 1964) and Douglas and Hitschfeld (1961) col- lected hail samples at the ground in Alberta and calculated a maximum water content of 3 4 g m~ averaged over the reported hailfall duration. This is equivalent to 3-cm Ze be- tween 106 and 2 X 106 mm6/m3 depending on whether the hailstones were considered to scatter as dry or wet particles. Unfortunately, Douglas had no corresponding radar measurements. One answer to the problem of high reflectivities is, of course, given by a consideration of the measurement errors. Since all of the extremely high 3-cm reflectivity measure- ments employed the method of estimating a threshold power at reduced receiver gain, these measurements are subject to an error of 5 db or more. (Some investigators estimate the error to be as high as 10 db.) A downward correction of this magnitude would re- duce the scope of the problem considerably, shifting most of the extremely high reflectiv- ities down to moderately high values. I do not believe that an overall correction for errors solves the problem completely, however, because a very few of the most intense echoes that I measured exceeded the capabilities of the radar receiver and could not be reduced to a threshold signal. Therefore it seems most likely that the extreme maximum 7 6 6 3 value of 3-cm Ze is nearer 10 rather than 10 mm /m , but such a high value occurs more rarely than the uncorrected measurements would indicate. The scattering properties of spongy hailstones provide one reasonable explanation for unusually high reflectivities. Spongy hail has a coating of porous ice saturated with liq- uid water. It is most likely to occur in the most active convection, where hailstone growth is too rapid for the complete dissipation of the latent heat of freezing of the accreted wa- ter. Experiments by Joss and List (1963) with artificial spongy hail demonstrated that ice spheres of diameter approximately half the radar wavelength, with a spongy ice-water coating which incorporates at least 6 per cent liquid water in the total mass of the sphere, back-scatter about twice as intensely as equivalent water spheres and much more intensely than ice. Atlas, Hardy, and Joss (1964) suggested that extremely high reflectivities could most reasonably be attributable to spongy hail. A moderate concentration of only 3 g m"3 of sponge-coated hailstones with modest diameters near 1 cm would have a 3-cm Ze of 107 mm6/m3, whereas any other solid or liquid particle would be required in much greater concentrations. A persistent maximum of 3-cm reflectivity high in a storm is a consequence of low- altitude attenuation in persistent heavy rain. Evidence for this is provided by the de- terioration of attenuation notches with height at 3-cm wavelength, and by the rarity of 10-cm reflectivity maxima aloft. In the severe Wokingham hailstorm, Atlas and Ludlam (1961) noted one maximum high in the troposphere with a 10-cm radar, but the general
187
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 experience of Geotis (1963), who also uses a 10-cm radar, shows a maximum near the ground in hailstorms, with little change aloft until a rather sharp decrease above 15,000 to 20,000 ft. Another contribution to diminished reflectivity below the 0C level is the melting and subsequent fragmentation of 1-cm or smaller hailstones, suggested by Markovich and Muchnik (1960). The spongy hailstones proposed by Atlas, Hardy and Joss (1964) would undergo a similar fate, but the process would proceed more rapidly and with greater ef- fect. As soon as the sponge coat melted away the excess water would be shed, leaving a thinly water-coated hailstone and some small raindrops which would scatter much less intensely than the original particle. The reflectivity could easily decrease by an order of magnitude or more. Both of these mechanisms would work in varying degrees at all radar wavelengths. A few thunderstorms, however, especially the most severe ones, do have 3-cm reflectivity maxima at altitudes up to twice the height of the 0C isotherm, with a substantial reflectiv- ity decrease taking place wholly in sub-freezing air. During the presentation of this paper I suggested that a persistent reflectivity maximum high above the 0C level might be main- tained in a well-organized thunderstorm by the growth of sponge-coated hail in a tilted updraft rich in cloud liquid water, followed by the ejection of these particles into a much drier atmosphere in which their spongy coats would freeze as they descended to the 0C altitude. The reflectivity of the falling hailstones would decrease sharply as their surface layers froze and their scattering cross sections approached the lower values of pure ice. Following this suggestion, Atlas, Hardy, and Joss (1964) performed computations which demonstrated that theoretically a reflectivity maximum well above the 0C level could in- deed develop in this manner. Attenuation in excessive liquid water at altitudes above 0C was offered by Austin (1965) as an alternative (and perhaps complementary) explanation for the increase of 3-cm re- flectivity with height above the 0C level in some of the more intense storms. She com- pared the vertical structure of 3-cm and 10-cm reflectivity in a number of thunderstorms and found evidence of appreciable 3-cm attenuation up to an altitude of 20,000 ft in some of them and even higher than 25,000 ft in a very few of the tallest, unusually intense storms. These measurements indicate that the most active thunderstorms are capable of carrying extraordinarily large concentrations of liquid water well above the 0C level in a column at least several km in diameter. Since this would also be a favorable environment for the growth of spongy hail, the 3-cm reflectivity maximum owing to attenuation may be enhanced by the presence of such particles. Donaldson (1961a) suggested storage of precipitation in sequentially-active convection cells as a contributing cause of persistent 3-cm reflectivity maxima aloft. In effect, each cell is envisioned as a prominent "first echo" in which precipitation accumulates during its early stages, as illustrated by the computations of Kessler (1961). However, the gen- eral absence of a 10-cm reflectivity maximum aloft in thunderstorms (e.g., Geotis, 1963) demonstrates that storage of radar-detectable particles aloft cannot play a significant part in a mature storm, except possibly in a very few exceptional cases in which a 10-cm reflec- tivity maximum has been noted high in a storm. In any event, there is impressive observational evidence, reported here for the first time, linking severe storms with an increase of 3-cm reflectivity with height above the 0C level. From the 233 reflectivity profiles in New England thunderstorms which were treated in a previous paper (Donaldson, 1961a), I found 155 suitable for a detailed study of the change in reflectivity in the layer immediately above the 0C level. The interpolated re- flectivity at 15,000 ft, which is normally just above the 0C isotherm during the thunder- storm season in New England, was compared with the reflectivity at 20,000 ft, or with the maximum reflectivity in those cases where the maximum was located above 20,000 ft. Profiles were not considered unless the two reflectivity values were derived from inde- pendent observations. The results, presented in Table 1, were classified according to cooperative observer reports of the occurrence of hail, severe weather, or tornado. Most, but not all, of the thunderstorms in the severe category also were reported as hailstorms. This method of identifying a severe storm is actually somewhat better than Table 1 indicates. For one thing, the one tornado case with a decrease in reflectivity above 15,000 ft was a very small vortex which caused damage of about $100 over a narrow, short
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society
TABLE 1. Occurrence of hail, severe weather, and tornadoes in New England thunderstorms as related to structure of 3-cm reflectivity above the OC level.
Occurrence (by per cent) Comparison of reflectivities Severe at 20,000 ft or higher to 15,000 ft Number of cases Hail weather Tornado Decrease or constant 118 13 3.4 0.8 Increase (up to 5 db) 24 42 33 8 Increase (more than 5 db) 13 46 38 31
path. Also some of the thunderstorms with an increase of reflectivity above 15,000 ft which were reported as non-severe or without hail at the time of their radar measurement produced hail or severe weather at some other stage in their life cycle. Finally, we have to reckon with the likelihood that very many weak thunderstorms occurred which were unreported because they failed to excite the interest of the radar operators and/or the cooperative observers. Hence the data are undoubtedly heavily biased in favor of the more spectacular thunderstorms, and a removal of this bias would very likely sharpen the division between the categories. An important observational advantage of comparing reflectivities at different heights in the same storm should not be overlooked: systematic measurement errors will cancel out. Thus, if the antenna gain has not been measured too well, or the power meas- urement technique consistently over- or under-estimates the true integrated echo power, absolute measurements will suffer corresponding errors which will not affect relative measurements. 6. Summary Many radar techniques have been tried in the effort to find a unique and reliable sensor of severe thunderstorms. Although this search has not been rewarded by unqualified success, a number of radar indicators appear frequently enough with severe thunderstorm weather to warrant serious consideration. Conventional radar can help the forecaster of severe weather in two ways. First, the appearance of certain storm echo features may serve to focus the attention of a forecaster to a particular area, at a particular time, in which thunderstorms of greater than ordinary intensity are probably taking place. This may signify nothing more than a reasonable expectation of small hail or moderate wind gusts. On the other hand, it may provide a warning of the impending development of a severe thunderstorm in the suspect echo or in one of its neighbors, which will attain maturity during the following hour or two. In the author's opinion, a meteorologist using radar should recognize the possibility for development of severe storms when he observes any one or several of the following con- ditions (listed in the order in which they were discussed previously): (1) Echoes are in a line with a wave pattern (see Fig. 1). (2) Echoes are noticeably converging. (3) Echo speeds exceed 30 knots. (4) Echo edges are sharply indented or scallop-shaped. (5) Echo tops penetrate the tropopause. (6) Equivalent reflectivity factor measured with a 10-cm radar exceeds 3 X 105 mm8/m8 anywhere in the storm. (7) Equivalent reflectivity factor measured with a 3-cm radar at an altitude of 30,000 ft exceeds 104 mm8/m3. (8) 3-cm reflectivity increases above the 0C level. Radar may also serve the forecaster by stimulating and informing his decision to issue severe storm warnings. Present experience has demonstrated that some form of severe thunderstorm is likely in the presence of one or more of the echo characteristics listed below, although severe storms may occur sometimes when none of these echo features are observed. Some of this experience, however, is limited to a few cases, and some of it is restricted to a few regions of the United States. Moreover, the subjective nature of
189
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 pattern recognition, which is a problem in the detection of tornado hooks and vaults, requires the application of experienced, intelligent judgment. Although further research may very well bring about some changes, the following severe thunderstorm indicators are intended to provide the forecaster with currently most reliable radar criteria for issuing warnings: (1) The characteristic tornado hook appears on PPI scan at low altitudes (see Figs. 2 and 8). (2) A persistent vault-wall structure beneath the highest echo top is observed on RHI scan (see Fig. 6). (3) Echo tops penetrate the tropopause by 10,000 ft or more. (4) Equivalent reflectivity factor measured in any part of a storm with a 10-cm radar exceeds 106 mm3/m3. (5) Equivalent reflectivity factor measured with a 3-cm radar at an altitude of 30,000 ft exceeds 105 mm6/m3. (6) A persistent 3-cm reflectivity maximum occurs at a height of 20,000 ft or more, at least 5 db greater than the reflectivity at the height of the 0C isotherm. The measurement in real time of reflectivities and heights in all convective echoes which appear on the radar scope is often beyond the capability of the unaided human observer. Hence modern data acquisition and processing techniques should be employed to assure more complete monitoring of the objective severe thunderstorm indicators. The reliability of any observing tool depends on well-planned research. With few ex- ceptions, radar measurements of thunderstorms have not been intensive enough to pro- vide good time resolution of more than one or two simple echo features on any occasion. A thunderstorm, even more a group of such storms, is a complex system of interrelated processes undergoing surprisingly rapid changes. One radar cannot possibly follow all the more important processes. Also, simultaneous measurements with a high degree of accuracy at two or more radar wavelengths would be helpful in resolving some of the ambiguities in the determination of precipitation particle sizes, types, and concentrations. It seems clear that significant contributions to our understanding of atmospheric con- vective processes are conditional on the employment of a magnification of space and time resolution in the observing tools. In a thunderstorm research project the resolution of reflectivity structure may be improved through a concentration of several radars and the maximum use of the data acquisition capability of each radar.
Acknowledgments I am grateful to Dr. Pauline Austin of MIT for her thorough, painstaking criticism of the paper, in particular the section on reflectivity measurements, and for her valuable suggestions. I am also pleased to thank Drs. David Atlas, Keith Browning and Kenneth Hardy of AFCRL for reading the manuscript and offering helpful comments. References Aoyagi, J., 1963: The quantitative estimation for the heights of radar echo tops. Proc. Tenth Weather Radar Conf., Boston, Amer. Meteor. Soc., 123-133. Arnold, J. E., 1961: Some characteristics of severe Texas thunderstorms. Final Report, Contract AF 19(604)-6136. Dept. of Ocean, and Meteor., The A&M College of Texas, 47-73. [Report No. AFCRL-800, Air Force Cambridge Research Labs., Bedford, Mass.] Atlas, D., 1963: Radar analysis of severe storms. Severe Local Storms, Meteor. Monogr., 5, No. 27, Boston, Amer. Meteor. Soc., 177-220. , 1964: Advances in radar meteorology. Advances in Geophysics, 10, New York, Academic Press, 317-478. , K. A. Browning, R. J. Donaldson, Jr., and H. J. Sweeney, 1963: Automatic digital radar re- flectivity analysis of a tornadic storm. J. Appl. Meteor., 2, 574-581. , K. R. Hardy and J. Joss, 1964: Radar reflectivity of storms containing spongy hail. J. Geo- phys. Res., 69, 1955-1961. , W. G. Harper, F. H. Ludlam and W. C. Macklin, 1960: Radar scatter by large hail. Quart. J. R. Meteor. Soc., 86, 468-482. , and F. H. Ludlam, 1961: Multi-wavelength radar reflectivity of hailstorms. Quart. J. R. Meteor. Soc., 87, 523-534. Austin, G., 1945: Radar storm detection. Weather Sei-vice Bulletin, 3, No. 7, Army Air Force, 12-19.
190
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society Austin, P. M., 1961: Some observations of attenuation of 3.2 cm radiation by heavy rain. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 325-330. • , 1964: Observations of attenuation of 3-cm radiation by precipitation. Proc. World Conf. on Radio Meteor, and Eleventh Weather Radar Conf., Boston, Amer. Meteor. Soc., 166-171. , 1965: Attenuation of 10 gc radiation by intense storms in New England. URSI Meeting, Washington, D. C., April 20-23. , and S. Geotis, 1960: The radar equation parameters. Proc. Eighth Weather Radar Conf., Boston, Amer. Meteor. Soc., 15-22. Bailey, R. W., 1955: On the motion of precipitation echoes in a cyclone. Proc. Fifth Weather Radar Conf., Boston, Amer. Meteor. Soc., 121-129. Bent, A. E., 1946: Radar detection of precipitation. J. Meteor., 3, 78-84. Bigler, S. G., 1955a: A comparison of synoptic analysis and composite radar photographs of a cold front and squall line. Proc. Fifth Weather Radar Conf., Boston, Amer. Meteor. Soc., 113-119. , 1955b: An analysis of tornado and severe weather echoes. Proc. Fifth Weather Radar Conf., Boston, Amer. Meteor. Soc., 167-175. , 1958: Observations of a tornado using the AN/CPS-9 radar. Proc. Seventh Weather Radar Conf., Boston, Amer. Meteor. Soc., pp. K1-K5. Blackmer, R. H., Jr., 1954: Radar scope photographs of the Worcester tornado. Suppl. Proc. Conf. on Radio Meteorology, Austin, Univ. Texas, paper X-9. Brooks, E. M., 1949: The tornado cyclone. Weatherwise, 2, 32-33. Browning, K. A., 1964: Interaction of two severe local storms. Proc. World Conf. on Radio Meteor, and Eleventh Weather Radar Conf., Boston, Amer. Meteor. Soc., 366-371. , and R. J. Donaldson, Jr., 1963: Airflow and structure of a tornadic storm. J. Atmos. Sci., 20, 533-545. , , and W. E. Lamkin, 1963: Severe local storms near Oklahoma City, 26 May 1963. Pre- sented at the Third Conference on Severe Local Storms, Amer. Meteor. Soc., Urbana, Illinois, 14 November 1963. Unpublished manuscript, Bedford, Mass., Air Force Cambridge Research Laboratories, 21 pp. , and F. H. Ludlam, 1962: Airflow in convective storms. Quart. J. R. Meteor. Soc., 88, 117— 135. Byers, H. R., and R. R. Braham, 1949: The Thunderstorm. Washington, U. S. Gov't Printing Office, 287 pp. Changnon, S. A., Jr., 1960: Climatological study of radar-depicted lines in the Middle West. Proc. Eighth Weather Radar Conf., Boston, Amer. Meteor. Soc., 73-80. Chmela, A. C., 1958: Reflectivity measurements in the vertical through a severe squall line. Proc. Seventh Weather Radar Conf., Boston, Amer. Meteor. Soc., pp. B17-B24. Cook, B. J., 1961: Some radar LEWP observations and associated severe weather. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 181-185. Copeland, R. C., 1958: A relationship between echo intensity and the observation of hail in New England. Proc. Seventh Weather Radar Conf., Boston, Amer. Meteor. Soc., pp. K22-K30. Crow, L. W., 1961: Cumulus cloud development and movement in southwestern Panama. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 70-75. Das, P. M., A. C. De, and M. Gangopadhyaya, 1957: Movements of two nor'westers of West Bengal: a radar study. Indian J. Meteor. Geophys., 8, 399-406. De, A. C., and S. N. Sen, 1961: A radar study of premonsoon thunderstorms (Nor'westers) over Gangetic West Bengal. Indian J. Meteor. Geophys., 12, 51-60. Donaldson, R. J., Jr., 1958: Analysis of severe convective storms observed by radar. J. Meteor., 15, 44-50. , 1959: Analysis of severe convective storms observed by radar—II. J. Meteor., 16, 281-287. , 1961a: Radar reflectivity profiles in thunderstorms. J. Meteor., 18, 292-305. , 1961b: Range distortion of thunderstorm reflectivity structure. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 165-174. , 1962: Radar observations of a tornado thunderstorm in vertical section. Report No. 8, Na- tional Severe Storms Project, Washington, U. S. Weather Bureau, 21 pp. , 1964: A demonstration of antenna beam errors in radar reflectivity patterns. J. Appl. Me- teor., 3, 611-623. , A. C. Chmela and C. R. Shackford, 1960: Some behavior patterns of New England hail- storms. Physics of Precipitation, Geophys. Monogr. No. 5, Washington, Amer. Geophys. Union, 354-368. Douglas, R. H., 1960a: Observations of Alberta hailstorms. Cumulus Dynamics, New York, Perga- mon Press, 145-147. , 1960b: Size distributions, ice contents, and radar reflectivities of hail in Alberta. Nubila, 3, No. 1, 5-11.
191
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Vol. 46, No. 4, April 1965 , 1961: Radar observation of Alberta hailstorms. Nubila, 4, No. 2, 52-58. , 1963: Recent hail research: a review. Severe Local Storms, Meteor. Monogr., 5, No. 27, Boston, Amer. Meteor. Soc., 157-167. , 1964: Hail size distributions. Proc. World Conf. on Radio Meteor, and Eleventh Weather Radar Conf., Boston, Amer. Meteor. Soc., 146-149. , and W. Hitschfeld, 1959: Patterns of hailstorms in Alberta. Quart. J. R. Meteor. Soc., 85, 105-119. , and , 1961: Radar reflectivities of hail samples. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 147-152. Freeman, J. C., Jr., 1953: Radar echoes in tornado situations. Proc. Conf. on Radio Meteorology, Austin, Univ. Texas, paper X-5. Fujita, T., 1958: Mesoanalysis of the Illinois tornadoes of 9 April 1953. J. Meteor., 15, 288-296. , and H. A. Brown, 1958: A study of mesosystems and their radar echoes. Bull. Amer. Me- teor. Soc., 39, 538-554. Garrett, R. A., and V. D. Rockney, 1962: Tornadoes in north-eastern Kansas, May 19, 1960. Mon. Wea. Rev., 90, 231-240. , and R. T. Tice, 1957: Some radar aspects of the May 20, 1957 Kansas-Missouri tornado. Mon. Wea. Rev., 85, 206-207. Geotis, S. G., 1963: Some radar measurements of hailstorms. J. Appl. Meteor., 2, 270-275. Harrison, H. T., and E. A. Post, 1954: Evaluation of C band (5.5 cm) airborne weather radar. Denver, Colorado, United Air Lines, Inc., 108 pp. Herman, B. M., and L. J. Battan, 1961: Calculations of Mie back-scattering of microwaves from ice spheres. Quart. J. R. Meteor. Soc., 87, 223-230. , Hiser, H. W., 1958: Radar analysis of two severe storms in South Florida. Bull. Amer. Meteor. Soc., 39, 353-359. Huff, F. A., H. W. Hiser and S. G. Bigler, 1954: Study of an Illinois tornado using radar, synop- tic weather and field survey data. Rep. of Investigation No. 22, Urbana, Illinois, State Water Survey Division, 73 pp. Inman, R. L., 1961: Summary of Texas thunderstorm observations for 1960. Final Report, Con- tract AF 19(604)-6136. Dept. of Ocean, and Meteor., The A&M College of Texas, 27-46. [Re- port No. AFCRL-800, Air Force Cambridge Research Labs., Bedford, Mass.] , and S. G. Bigler, 1958: A preliminary classification of radar precipitation echo patterns as- sociated with midwestern tornadoes. Proc. Seventh Weather Radar Conf., Boston, Amer. Me- teor. Soc., pp. 116-125. Jones, D. M. A., 1956: Rainfall drop size-distribution and radar reflectivity. Research Report No. 6, Urbana, Illinois, State Water Survey Division, 20 pp. Joss, J., and R. List, 1963: Zur radarriickstrahlung von eis-wasser-gemischen. Z. Angew. Math. Phys., 14, 377-380. Kessler, E., 1961: Kinematical relations between wind and precipitation distributions, II. J. Me- teor., 18, 510-525. Kotov, N. F., 1960: Radiolokatsionnye kharakteristiki livnei i groz. [Radar characteristics of showers and thunderstorms.] Leningrad. Glavnaia Geofizicheskaia Observatoriia, Trudy, No. 102, 63-93. Langille, R. C., and K. L. S. Gunn, 1948: Quantitative analysis of vertical structure in precipita- tion. J. Meteor., 5, 301-304. Lhermitte, R. M., 1964: Doppler radars as severe storm sensors. Bull. Amer. Meteor. Soc., 45, 587-596. Lintner, H. A., and D. Atlas, 1956: Interim procedure for severe storm identification with the CPS-9 radar. Unpublished manuscript, Bedford, Mass., Air Force Cambridge Research Center, 18 pp. Ludlam, F. H., 1959: Hailstorm studies, 1958. Nubila, 2, No. 1, 7-27. , 1963: Severe local storms: a review. Severe Local Storms, Meteor. Monogr., 5, No. 27, Boston, Amer. Meteor. Soc., 1-30. Markovich, M. L., and V. M. Muchnik, 1960: Struktura grozovikh zliv za danimi pro rozpodil intensivnosti radioluni z visotoiu. [Structure of thunderstorm showers, according to data on the intensity distribution of radar echoes with height.] Ukrains'kyi Fizychnyi Zhurnal (Kiev), 5, 259-269. Marshall, J. S., 1957: The constant-altitude presentation of radar weather patterns. Proc. Sixth Weather Radar Conf., Boston, Amer. Meteor. Soc., 321-324. Nolen, R. H., 1959: A radar pattern associated with tornadoes. Bull. Amer. Meteor. Soc., 40, 277-279.
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC Bulletin American Meteorological Society Pautz, M., and F. Doloresco, 1963: On the relation between radar echo tops, the tropopause and severe weather occurrences. Proc. Tenth Weather Radar Conf., Boston, Amer. Meteor. Soc., 51-56. Penn, S., C. Pierce and J. K. McGuire, 1955: The squall line and Massachusetts tornadoes of June 9, 1953. Bull. Amer. Meteor. Soc., 36, 109-122. Probert-Jones, J. R., 1962: The radar equation in meteorology. Quart. J. R. Meteor. Soc., 88, 485-495. , 1963: The distortion of cumulonimbus precipitation observed by radar. Imperial College, London, Tech. Note No. 13 under USAF Contract No. AF 61(052)-254, 21 pp. [Report No. AFCRL-64-249, Air Force Cambridge Research Labs., Bedford, Mass.] Sal'man, E. M., 1957: Radiolokatsionnoe issledovanie struktury livnei i groz. [Radar study of the structure of showers and thunderstorms.] Leningrad, Glavnaia Geofizicheskaia Observa- toriia, Trudy, No. 72, 46-65. Sarrica, O., 1958: Una caratteristica evoluzione degli echi radar provenienti da temporali. Atti del VII Convegno annuale, Associazione Geofisica Italiana, Rome, No. 7, 1-13. Schaefer, V. J., 1960: Hailstorms and hailstones of the western Great Plains. Nubila, 3, No. 1, 18-29. Schleusener, R. A., and J. D. Marwitz, 1963: Characteristics of hailstorms in the high plains as deduced from 3-cm radar observations. Proc. Tenth Weather Radar Conf., Boston, Amer. Me- teor. Soc., 39-44. Schuetz, J., and G. E. Stout, 1957: RHI radar observation of a tornado. Bull. Amer. Meteor. Soc., 38, 591-595. Soane, C. M., 1957: Cyclonic rotation in a radar echo. Bull. Amer. Meteor. Soc., 38, 269-273. Staats, W. F., and C. M. Turrentine, 1956: Some observations and radar pictures of the Blackwell and Udall tornadoes of May 25, 1955. Bull. Amer. Meteor. Soc., 37, 495-505. Stout, G. E., R. H. Blackmer and K. E. Wilk, 1960: Hail studies in Illinois relating to cloud physics. Physics of Precipitation, Geophys. Monogr. No. 5, Washington, Amer. Geophys. Union, 369-381. Stout, G. E., and H. W. Hiser, 1955: Radar scope interpretations of wind, hail, and heavy rain storms between May 27 and June 8, 1954. Bull. Amer. Meteor. Soc., 36, 519-527. Stout, G. E., and F. A. Huff, 1953: Radar records Illinois tornado-genesis. Bull. Amer. Meteor. Soc., 34, 281-284. Tepper, M., 1950: Radar and synoptic analysis of a tornado situation. Mon. Wea. Rev., 78, 170— 176. Ward, N. B., 1961: Radar and surface observations of tornadoes of May 4, 1961. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 175-180. Wexler, R., 1947a: Radar detection of a frontal storm, 18 June 1946. J. Meteor., 4, 38-44. , 1947b: Radar photographs of a frontal wave. J. Meteor., 4, 69-71. Wilk, K. E., 1960: An investigation of a severe local hailstorm. Proc. Eighth Weather Radar Conf., Boston, Amer. Meteor. Soc., 481-487. , 1961: Radar reflectivity observations of Illinois thunderstorms. Proc. Ninth Weather Radar Conf., Boston, Amer. Meteor. Soc., 127-132.
Meteorological Society. Board procedures require publication of names of candidates in the BULLETIN upon confirmation by the Chairman of the Membership Commission. Any person possessing information indicating that the above listed Pro- fessional Members of the Society do not meet the standards for certification should communicate with the Secretary. Mr. Baynton is associated with the National Center for Atmospheric Research, Boulder, Colorado, in the Field Ob- Board of Certified Consulting Meteorologists serving Support Facility. Colonel Smith, USAF (Ret.) is af- filiated with the Geophysics Laboratory, Research Triangle Harold W. Baynton is the 50th candidate and James Roy Institute, Durham, North Carolina. Smith is the 51st candidate to have met the qualifications of the Board of Certified Consulting Meteorologists, American (More announcements on page 218)
193
Unauthenticated | Downloaded 10/05/21 10:48 AM UTC