Intense Atmospheric Vortices Associated with a 1000 MW Fire

Home , Fire whirl, Thermal low

Christopher R. Church and intense Atmospheric vortices John T. Snow Department of Geosciences, Purdue University Associated with a 1000 MW Fire W. Lafayette, Ind. 47907 and Jean Dessens Centre de Recherches, Atmospheriques Henri Dessens, University de Clermont, Campistrous—CIDEX B47, 65300 Lannemezan, France

Abstract The special case of the fire whirl, with its central core filled with burning material, is at once impressive Observations of vortices of various types produced in a large thermal plume are described. The apparatus used to generate and dangerous, and is particularly intriguing. the plume is the M£teotron, an array of 105 fuel oil burners The dynamic structure of these and other atmo- with a total heat output of approximately 1000 MW. Three spheric vortices (tornadoes, waterspouts, and dust types of vortices have been observed: 1) large counter- devils) is not well known, primarily due to the lack rotating rolls in the downstream plume, 2) intense small-scale vortices resembling very strong dust devils seen at the sur- of direct measurements within the actual flows. Such face on the downwind side of the plume, and 3) very large measurements have been precluded by several ob- columnar vortices produced when the lower portion of the stacles. These include the design of instrumentation plume goes into rotation as a whole. Three mechanisms to survive in the adverse environment of the vortex leading to the concentration of vorticity necessary to produce core, the logistical problem of following and inter- these vortex types are discussed. These include tilting and stretching of horizontal vorticity present in the environ- cepting a moving natural vortex, and considerations mental wind field, generation of vorticity within the plume of personnel safety. Here we report on an opportunity by the action of buoyancy and drag forces, and convergence for generating and making direct measurements of of preexisting background vorticity from the environment. large man-made atmospheric vortices and discuss the It is concluded, based on these observations and physical considerations, that the generation of vortices of moderate results of a recent preliminary effort to observe and intensity is to be expected in large plumes, be their source document the vortices that formed in the plume from a forest fire or an industrial operation. a large controlled fire. It is the intent of this and future work to rectify, in part, the current lack of direct observations of intense vortical flows in a natural 1. Introduction environment. It is well established that intense atmospheric vortices occur in the vicinity of large man-made and natural 2. Experimental program fires. The literature documents the production of such vortices in forest fires (Graham, 1955), refinery ex- The work reported herein was carried out at the plosions (Hissong, 1926), gas well fires (Dessens, 1963), Centre de Recherches Atmospheriques Henri Dessens, volcanic eruptions (Thorarinsson and Vonnegut, 1964), located on the plateau of Lannemezan (^600 m MSL), area conflagrations in cities (Ebert, 1963), and at- some 30 km north of the central Pyrenees. The current tendant to a number of industrial operations involving series of experiments was begun in 1978 as a joint the large-scale release of heated gases into the atmo- research venture between the Observatoire du Puy de sphere. Such vortices vary greatly in intensity, ranging Dome and Electricite de France (EdF), the French from tornado-like flows strong enough to throw tree national electrical power company. (EdF has an trunks around and destroy small homes to much more interest in assessing the environmental impact of the modest whirls that resemble very small dust devils. release of waste dry heat in amounts of the order of 1000 MW into the atmosphere.) A number of re- 0003-0007/80/070682-13$07.25 searchers from private industry and neighboring French © 1980 American Meteorological Society universities participated in the program. The present

682 Vol. 61, No. 7, July 1980

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 683 project is the successor of two previous programs: aging 5-10 m in length and 1 m in diameter. Fuel is 1961-64, dealing with the creation of cumulus clouds consumed at a rate of 1 m3 h-1 per burner, with each (Dessens and Dessens, 1964) and the formation of burner having a calorific power output of some 10 MW. tornado-like vortices (Dessens, 1962); and 1971-73, Thus a combined output of some 1000 MW for the dealing with a quantitative study of an artificial con- array as a whole is achieved. An experimental period vective plume initiated from the ground (Benech, at full pressure normally lasts from 20 to 30 min. 1976). In the current program, experimental burns Under background conditions of even very moderate were carried out in June, July (15 burns), and No- conditional instability, the large thermal plume gen- vember (15 burns) 1978, and in May and June (15 erated by the Meteotron triggers the production of burns) 1979. The experiments were performed under moderate size cumulus clouds. These can occasionally a wide variety of background meteorological condi- grow to the cumulus congestus stage. As a consequence, tions. Two of us (Church and Snow) participated in much of the supporting instrumentation is focused on the 1979 phase for the specific purpose of investigating the downstream, nearly horizontal portion of the plume. the large vortices that have been observed in the plume. Due to the interests of the Observatoire, a more ex- The array of burners used for producing the buoyant tensive set of instrumentation has recently been in- plume is called the Meteotron. Since the debut of the stalled in the immediate vicinity of the plume base initial design by H. Dessens in 1961 (see Dessens, in order to measure the flow (convergence and mean 1962), the Meteotron has undergone several changes vorticity) in the immediate vicinity of the fire. How- in burner design, number of burners used, and con- ever, the main scientific thrust (and hence the bulk figuration of burner deployment. As shown in Fig. 1, of the instrumentation) continues to focus on the far the array currently consists of 105 fuel oil burners downstream portion. deployed in a three-armed spiral pattern within a Instrumentation for observing the base of the plume 140 m X 140 m square. A burner of the type used in during the 1979 summer program consisted of 42 in- the present series of experiments is shown in Fig. 2. strument assemblages suspended over the burner array Three diesel-powered pumps supply fuel oil to the by means of an elaborate cable assembly, supplemented burners at pressures up to 5900 kPa. Each burner is by 16 similar ground-based assemblages (see Fig. 3). equipped with a set of spark gap electrodes and Each assemblage contained a three-component ane- a step-up transformer so as to be ignited electrically mometer and a temperature sensor. The primary by striking an arc next to the fuel injection nozzle. purpose for which this extensive array was installed At full pressure, each burner produces a flame aver- was the collection of baseline wind and temperature

FIG. 1. The M£t£otron, looking to the northeast. For scale, the pylon in the far corner is approximately 280 m away and 60 m tall.

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 684 Vol. 61, No. 7, July 1980 data concerning the source region of the plume. Thus details, such as occur with the development of small the assemblages were positioned just outside the region vortices, could not be resolved in the data field. wherein the details of the individual burners were Other supporting experimental equipment in 1979 important. The 16 ground-based packages were each consisted of time-lapse camera systems, 8.6 mm and mounted on a 4 m fold-down tower, and spaced four 10 cm radar, and Doppler sodar. A B23 research to a side along the flanks of the fire area to obtain aircraft from the University of Washington, equipped a picture of the strength of the low-level convergence with an array of environmental monitoring and particle into the base of the plume. The suspended assemblages sampling equipment, provided data on the micro- were arranged in two levels, 21 to a level, at 30 and physical aspects of the downwind portion of the plume 60 m AGL. Horizontal grid spacing was 40 m. These and cloud. Frequent rawinsonde observations were served to define the main features of the vertical taken from a site 4 km to the southwest of the burner motion and temperature field in the lowest portion array. of the plume. Additional instrumentation was mounted In order to investigate more fully the small-scale on each of the four 60 m tall towers supporting the details of the low-level plume structure and to observe cable assembly. All data from this array were digitized, the vortices reported therein, several hand-held still multiplexed, and recorded on magnetic tape. However, and movie cameras were used and the observers the spacing of the instruments is such that small-scale (Church and Snow) worked in close to the base of

FIG. 2. A close-up view of one of the 105 burners in the array. The diameter of the cylindrical drum is 90 cm. Also visible is the fuel injection nozzle and the elements of the arc ignition system. The burners were spaced 4 m apart.

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 685

FIG. 3. A schematic (not to scale) of the anemometer and temperature sensor array used to investigate the base of the plume. The supporting towers are 60 m tall and are located at the corners of a square 200 m on a side centered on the burner array. The suspended assemblages are raised into position by a hoisting mechanism built into the towers. Data are collected, filtered, and transmitted to the control building by the electronics located at the base of each tower. the plume. Various types of smoke grenades were to determine magnitudes of velocities in the vortices tested as to suitability for visualizing the lower portions by means of photogrammetric analysis. of the smaller vortices. Simple pressure sensors were deployed for making surface pressure measurements and a portable, hand-held, hot-wire anemometer was 3. Vortex types associated with the Meteotron prepared for direct probing of vortices. In addition, several time-lapse cameras were positioned at various Each of the burners within the Meteotron array distances and focused on the plume base in order to produced a small, hot plume. These merged (at 10- record possible evidence of rotation in this portion of 20 m AGL) to form a single buoyant column, made the column. Some operational difficulties were en- visible mainly by the copious amounts of black smoke countered due to the high infrared output of the fire associated with the combustion. Under calm conditions and the noxious quality of the combustion products. (rarely encountered), the composite plume rose nearly The infrared heating prevented one from approaching vertically. More commonly, under the influence of the the line of burners closer than about 10 m. Determina- prevailing horizontal wind, the rising column was tion of the optimum exposure settings was difficult deflected into the downwind direction. At some height due to the high contrast between the very hot, yellow- above surface (typically in the range 1000-2000 m), white flames and the black smoke. Further, the oily the plume became nearly horizontal. The details of smoke was found to adversely affect several of the plume behavior were directly connected to both the cameras. These difficulties severely limited the freedom convective stability of the low-level atmosphere and with which one could make physical measurements the local wind field. In general, there was much en- and indicate the need for protective clothing and trainment and mixing of environmental air into the instrument covers in future work. In spite of these rising plume so that it spread with distance downstream. obstacles, many vortices were observed and photo- Three distinct types of vortices were observed within graphed in detail. We were also able to physically the fire area and thermal plume. These are: penetrate a few vortices that moved out of the main fire area—these had visually weakened as they moved 1) The largest, but perhaps least intense vortex out of the fire, so our impressions of them are likely configuration occurred under conditions of ther- to be unrepresentative of actual maximum intensities. mal instability with moderate horizontal surface Efforts are presently underway to determine more winds (i.e., 4-8 m s""1). At some distance (typi- precisely, primarily from rawinsonde data, the mete- cally 30-40 m) above the point where the separate orological conditions under which the various types plumes of the individual burners merged, the of vortices occurred. Also, an attempt is being made column bifurcated into two counterrotating roll

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 686 Vol. 61, No. 7, July 1980

vortices. An example of this phenomenon is shown in Fig. 4. The vortices were visible only from the downwind (under-)side of the plume, and had typical initial diameters of 30-60 m. The sense of rotation of the pair is such that there was inward (upward) motion at the center of the back- (under-) side of the plume; on occasion the circulation was strong enough to produce a clear hole through to the front of the plume. This vortex pair became increasingly deflected into the horizontal as the plume moved upward and downwind. The motion field within these vortices appeared to consist mainly of rotation about their internal axis, with no indication of the presence of an intense vortical core or of pronounced axial velocities. These vortices reoriented themselves so as to follow changes in wind direction with height, and appeared to be a persistent feature of the plume. This bifurcation is similar to that observed in plumes from large industrial chimneys.

2) Intense vortex cores that have a strong visual resemblance to dust devils occurred under conditions of moderate instability and light to moderate winds (1-6 m s-"1). These always appeared in preferred (downwind) sections of the fire area. They took on a variety of physical appearances, depending on the materials locally available to render them visible. Visualizing materials included dust, small stones, smoke, FIG. 4. A photograph of the underside of the plume taken steam, and droplets of unburned fuel. Numerous 600 m downwind, looking upward and upwind. This clearly erect columnar vortices of this type were ob- shows the bifurcation into a pair of large counterrotating served. An example of an intense one is shown roll vortices that act to move air up and into the plume along its centerline. Taken 21 October 1978, with surface in Fig. 5. Such vortices were always attached winds at 3-5 m s-1. to the surface, with both cyclonic and anticyclonic vortices occurring. Observations of smoke and vapors near the surface indicated that there to the production of a fire whirl, with an internal was usually significant low-level inflow spiraling laminar flame spiraling up through the core to into the visible core for a radial distance of some a height of 50 m or more. Other features worthy 5-10 core diameters out from the core. of note concerning vortices of this type are that These vortices could become quite intense. they sometimes appeared in pairs, and that the Frequently, one would move across the line of cores of the larger ones underwent transitions to burners, and in doing so would noticeably affect the multiple vortex configuration. the plumes of several adjacent burners. Occa- 3) The most intense type of vortex is also the least sionally, one or two burners would be extinguished common, and appears to be associated with by the winds in the vortex. Lifetimes varied from conditions in which at least the lower portion about 5 to 100 s, with a few events lasting for of the thermal plume was in rotation as a whole. several minutes. If the vortices drifted too far This has been observed in detail only twice from the vicinity of the fire area, they dissipated. during the nearly 20-year history of Meteotron Visible core diameters varied from 0.2 to 2 m. experiments—once in 1961 and again in July Vertical dimensions were such that the upper 1978. In the 1961 case, previously described by limits of the strong cores were buried in the Dessens (1962), a large persistent whirl some smoke plume. In a few cases, particularly where 40 m in diameter was observed after some 15 min the vortex was of large diameter, there was a of operation of the burner array. It was visible visual impression that it was an extension of for 3-4 min and centered within the array. The the surface of one of the large counterrotating winds within this vortex interacted strongly with rolls described in 1). A very dramatic form was the fire and were of sufficient intensity to ex- observed when a vortex of this type would move tinguish three burners. The rotating plume con- over a blazing pool of fuel oil. This would lead tained a "glowing" core about 1 m in diameter.

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 687

FIG. 5. Photograph of the visible core of a large vortex that formed in the right downwind quadrant of the plume. The lower portion of the vortex is made visible by dust; the upper, by steam and fuel oil vapor. The visible vortex was approximately cylindrical with a diameter of 2.5 m. It extended upward until lost from view in the main plume.

Even after shutdown of the burners, a visible rotating rolls within the downstream plume. central core (apparently of water droplets) about Dessens (1962) observed a large vortex of the 1-2 m in diameter remained for several minutes. latter type on 17 June 1961, while a large number The formation of the rotating column was closely of the former were observed by us during the linked to the development of a remarkably in- summer 1979 experiments. These all dissipated tense ascending current of air. On this day, the fairly quickly, moving no more than 200 or atmosphere was particularly unstable and the 300 m downwind of the fire area. winds light (windspeed <1.5 m s-1 to 800 m AGL). One of the photographs taken during the very 4. Physical mechanisms for vortex production similar 1978 event is shown in Fig. 6. This reveals that the rotating plume again contained The emission of hot gases by the fire produces a very a laminar-appearing core some 100 m in length, warm buoyant plume, which, in a still environment, surmounted by a broader, more turbulent region rises vertically. Upward velocities are greatest along that was apparently also part of the vortex the centerline, so that the internal buoyancy forces system. Although the central core was first seen produce vortex tubes that are circular rings concentric in this form only after the burner array was with the vertical axis. As shown in Fig. 8, the resulting turned off, analysis of earlier photographs circulation at the periphery of the plume entrains indicate rotation of the lower section of the plume environmental air. Smaller-scale turbulence mixes the for several minutes prior to the end of the ex- entrained air with the plume material. In a neutral periment. Examination of the concurrent sounding or stable stratified environment, as a result of this data, shown in Fig. 7, indicates that calm, very continual entrainment of cooler air and because of unstable conditions existed in the lowest layers, cooling through expansion, the buoyancy forces weaken a situation that seems to be required for the with height and the plume spreads into a broad cone. development of plume rotation. Since concentrated vortices form only where vor- It appears that this type of large vortex ticity of sufficient strength and proper orientation is (plume being in rotation as a whole) should be available, it is evident that such a plume rising in distinguished from both the large vortices that a still environment will not give rise to vortices of develop on the downwind side of the fire area the types observed. A relevant consideration is thus and move downwind away from the fire, and the identification of physical mechanisms capable of from those that develop out of the counter- producing vorticity of the proper strength and orienta-

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 688 Vol. 61, No. 7, July 1980

wind vector to the left. Since in the absence of dissipative processes the vortex tubes correspond to material tubes within the fluid, they are carried along by the wind. A segment of each tube is thus advected up to and entrained into the upstream surface of the rising plume. As the entrained fluid becomes mixed into the plume, it is carried aloft. This forms a loop in the attached vortex tube. Within the sides of the loop, the vorticity vector is reoriented so that a vertical component appears. Since the environmental wind speed generally increases with height, the plume progressively tilts over towards the horizontal. The rising fluid in the plume is acce- lerated downwind so that the loop portion of each tube is carried downstream faster than the portion that does not enter the plume. Thus, as shown in Fig. 9, the loop is both tilted and stretched out along the plume. This stretching increases the magnitude of the vorticity vector in the stretched portion of the tube. Because of the tilting, within the loop the vorticity vector bends back toward the horizontal as the distance downstream increases, but now pointing with or against the environmental wind rather than across it. Further, as the cooler, less buoyant air on the upstream surface slides around to the sides, the vortex tubes tend to become concentrated along the flanks of the plume. These entrained vortex tubes thus extend down to the ground on each flank, and are tied into the horizontal tubes trailing downwind on each side of the plume. These latter form the outer edges of the surface wake of the plume in the environ- FIG. 6. One of a series of photographs of the large vortex observed centered in the burner array on 8 July 1978. This mental wind field. particular picture was taken at 10:34 a.m., a few moments after the burners were shut down. The ropelike funnel was approximately 5 m in diameter and extended upward about 2) Generation of vorticity from the combined effects 100 m where it merged into a larger (rotating) cone-shaped of buoyancy forces and surface drag forces. cloud. All the smoke and steam appeared to be in slow rota- (This mechanism produces new vortex tubes.) tion about a common center, indicating that the funnel was The buoyancy forces act within the plume to the core of a single large vortex. (Photograph by Madeleine Dessens.) produce an upward acceleration whose strength is dependent on the environmental lapse rate. This buoyancy acceleration varies with height as tion. Based on our observations and physical reason- the parcels within the plume rise and cool through ing, three processes, all involving interaction of the expansion and as environmental air is entrained plume with the environmental wind, are offered as and mixed into the plume. The effect of the the most likely candidates. In terms of simple physical mixing on the vertical motions within the plume models, these are: depends upon the stability of the entrained air. The parcels also initially experience a horizontal 1) Reorientation and stretching of the predomi- acceleration due to a transfer of horizontal nantly horizontal vorticity that is present in momentum from the entrained environmental air. the lower surface boundary layer. This vorticity This effect is strongest on the windward side of is due to the vertical shear in the horizontal the plume at low levels and decreases with height wind. (This mechanism concerns itself with the as the fluid within the plume is gradually ac- tilting and subsequent stretching of vortex tubes celerated up to the speed of the environment as already present in the environment.) As shown it rises. This transfer of horizontal momentum in Fig. 9 for the case of no directional shear with from the environment into the plume appears as height, the vortex tubes all lie parallel to one a net drag force acting across the upwind plume another and perpendicular to the wind vector surface. with the vorticity vector pointing across the Entrainment and mixing are strongest on the

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 689

FIG. 7. Data extracted from a rawinsonde sounding taken from a site 4 km southwest of the Meteotron. Release time: 11:00 a.m., 8 July 1978. The dotted sloping lines represent the (dry) isentropes for the absolute temperatures indicated. Note the indication of a superadiabatic layer between surface (942 mb) and 928 mb, and also the presence of a capping inversion between 880 mb (750 m) and 830 mb (1320 m). The local surface observation indicates very sultry, nearly calm conditions (wind at 6 m was <2 m s"1) with partially overcast skies (6/8 cirrus at 7000 m, 4/8 cumulus fractus at 500-800 m). The convective condensation level was 460 m AGL.

upwind and flanking surfaces of the plume, while buoyancy effects are weakest there. Hence fluid parcels on the windward surface of the plume tend to move with the local environment. Only a portion of the horizontal momentum entrained at any level is eventually transferred into the protected center and downwind side of the plume, where the buoyancy forces are strongest. Thus fluid in the center and downstream side of the plume will initially have a stronger vertical (weaker horizontal) velocity component than fluid on the front and flank surfaces. As a consequence, as the plume rises it tilts progressively downwind, tending toward the horizontal at high elevation where the buoyancy effects become small. The distribution of the buoyancy forces leads to a horizontal shear in the vertical velocity component (production of horizontal vorticity), while the distribution of the drag forces produces a horizontal shear in the horizontal velocity component (production of vertical vorticity). The symmetric, ring-shaped vortex tubes of Fig. 8 (wherein no cross-wind drag forces were operative) are changed into the distorted rings shown in Fig. 10 cutting across the plume at an angle to the axis. These form a layer of vorticity on the plume surface that acts to match the FIG. 8. A schematic of a symmetric buoyant plume rising in still air, showing that the vortex tubes are in the form motion field within the plume to the environ- of symmetric vortex rings. mental flow. At low- and midlevels, those por-

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 690 Vol. 61, No. 7, July 1980

FIG. 9. A schematic illustrating how the rising plume acts to produce loops in the otherwise horizontal vortex tubes present in the environmental wind field. This results in a clustering of vortex lines with a strong component parallel to the plume axis along each downwind flank. This effect also produces a surface wake extending downstream from the plume.

tions of the vortex tubes lying on the windward vorticity should be observed. In the right down- side of the plume have strong vertical com- wind quadrant, this component should be posi- ponents of vorticity. Note that there is tilting tive; in the left, it should be negative. These and stretching of the portions of the vortex two regions are thus predicted to be the favored tubes that lie along the flanks of the plume. areas for the production of intense vortices with As the plume rises and takes on the speed of its cyclonic whirls on the right and anticyclonic environment, the vortex tubes become almost whirls on the left. A related process for large- completely horizontal. In the far downwind, scale clouds is described by Connell (1975). nearly horizontal portion of the plume, this An event that supports this prediction is de- reflects the fact that the drag forces are no picted in Fig. 12 and on the cover. Here the longer significant and only (weak) buoyancy photographer is looking upwind towards two effects are active. Again the strongest values of vortices horizontally separated by a distance of a vorticity appear to be along the flanks of the few tens of meters. Examination of the motion plume, with a significant vertical component at picture footage has shown that the vortex on the low- and midlevels. left is in cyclonic rotation, and the one on the right Note that both mechanisms 1) and 2) result is rotating anticyclonically. This result is entirely in the formation of layers of vorticity with a consistent with the physical discussion just pre- strong component oriented along each flank of sented. It should also be noted that during the the tilted plume. These along-the-flank com- filming of this event the anticyclonic component ponents of the vorticity cause the layers to exhibited vortex splitting in which it became a sys- roll up so that the plume bifurcates into two tem of two anticyclonic vortices rotating about a counterrotating roll vortices. At low levels, the common center. plume rises vertically so that the axes of these Further, movement of large whirls downwind roll vortices are nearly normal to the surface. away from the fire area is to be expected. As They contact the surface on the protected, was frequently observed, such whirls moved out downwind side of the fire. Far downstream, the and away from the downwind quadrant of each rolls lie nearly parallel to the surface and are flank of the plume, being steered by the environ- oriented in the direction of the wind. Thus both mental wind. The direction of rotation appeared of these mechanisms predict that on the surface, to be determined by the side from which each in the downwind quadrants of the fire area (see eddy was shed. Such vortices are taken to be Fig. 11), an enhanced component of vertical analogous to the large eddies that are shed from

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 691

FIG. 10. A schematic illustrating the production of vorticity within the plume by buoyancy and drag forces. The vortex rings are tilted so that the vorticity field matches the motion field within the plume to the environmental flow. This again results in a clustering of vortex lines with a strong component parallel to the plume axis along each flank.

the lee sides of multistory buildings and other (1970) have demonstrated the existence of small tall structures. These then constitute a portion regions within the atmosphere wherein the local of the wake of the plume in the environmental vertical vorticity is enhanced. These may be of flow. The frequency of shedding and the intensity either sign, and will have horizontal dimensions of the vortices should be determined by the of 0.5 to 1 or 2 km. Some vertical vorticity may extent to which the lower plume acts as an also be locally produced by the present arrange- obstacle to the flow. Since they moved out of ment of the burners in three spirals—this is the fire area, these traveling vortices left the suggested based on experiments by one of us region of enhanced vertical vorticity, and so (as (Dessens) with scaled laboratory models where is to be expected) they quickly dissipated. quite intense vortices are produced at the center 3) Concentration and amplification of vertical vor- of a spiral array of gas flames. If the buoyant ticity present in the lower surface boundary convection process is strong enough to cause the layer. This vorticity is the result of horizontal formation of a localized thermal low with low- shear in the horizontal wind. (This mechanism level convergence into the plume from over a concerns itself with the advection and stretching region of several kilometers extent, then such of properly oriented vortex tubes already present preexisting background vorticity will be advected in the environment.) Several sources can be sug- into the rising plume. As illustrated in Fig. 13, gested for the production of this vorticity. The the radial convergence of vortex tubes results in low-level horizontal wind field will generally con- an increased concentration of tubes in the plume. tain a small amount of vertical vorticity due, Simultaneously, the stretching of the tubes by perhaps, to variations in topography and surface the rising plume results in a local increase in the roughness in the surrounding terrain. In addi- magnitude of the vorticity vector. tion, the mesoscale background environment may Here can be seen a fundamental difference contain an enhanced amount of vertical vorticity between the first two mechanisms and 3). Whereas due to the synoptic situation. Ryan and Carroll both 1) and 2) lead to a bifurcated, double-roll

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 692 Vol. 61, No. 7, July 1980

FIG. 11. An illustration of near-surface regions on the downwind flanks of the plume where the mechanisms shown in Figs. 9 and 10 lead to an enhanced component of vertical vorticity. The directions of rotation of the whirls one would expect to form in such regions are also indicated.

FIG. 12. Photograph of a vortex pair that formed on the downwind side of the plume. The visible core of the left vortex was in cyclonic rotation and approximately 1 m in diameter. The core of the vortex on the right was in anticyclonic rotation and about 1.5 m in diameter. Both extended upward until lost from sight in the main plume. The senses of rotation are what would be expected from the vertical vorticity distribution sketched in Fig. 11. Both vortex cores were accompanied by spiral inflow patterns at the surface—these were 10-15 m in diameter (all the steam and fuel vapor around the right vortex was in slow rotation).

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC Bulletin American Meteorological Society 693

with the horizontal wind, resulting in the production of regions of enhanced positive and negative vorticity in the downward portion. Intense cyclonic and anticyclonic whirlwinds are created for a particular range of wind speed. If the winds are too high, rotation is evident only in the tilted downwind portion of the plume. If the winds are light it is conjectured that the concentration of horizontal vorticity in the con- verging inflow is occasionally sufficient to cause the entire plume to go into rotation. Conditions of high thermal instability also appear necessary for the production of the most intense vortices. Although it is evident that the kinetic energy contained in the vortices represents only a small fraction of the total thermal energy being released, the vortex energy can nevertheless become highly concentrated. As men- tioned in the Introduction, forest fire whirlwinds can be intense enough to be a cause for concern. It is conceivable th^t a new generation of power plants releasing ever larger amounts of waste heat and other industrial operations will, under certain meteorological conditions, inadvertently produce whirlwinds of dam- aging intensity. The concentration of vorticity by large power sources has been discussed by Hanna and Gifford (1975). In a more positive sense, there have been various proposals for tapping the mechanical energy of artificially created mesoscale atmospheric vortices (for example, Michaud (1975)). This is an attractive proposition because it would utilize low- FIG. 13. A schematic illustrating the conjectured arrange- grade thermal energy that would otherwise be wasted. ment of vortex lines in a thermal plume in rotation as a Finally, since it appears that heat and momentum are whole. The vortex tubes are advected into the plume from transported more effectively in vortical flows, it seems the background environment. This mechanism requires that worth investigating the feasibility of developing atmo- there be sufficient background vertical vorticity and that spheric vortices to serve as wall-less chimneys for the fire cause significant convergence over a wide area. potential industrial applications. Whether the ob- jectives are to encourage or preclude the development structure for the plume, 3) indicates that the of large atmospheric vortices, the central problems to plume may go into rotation as a whole. The key be addressed in future work are the same: to identify factor for mechanism 3) to be operative appears the physical factors that affect the fraction of input to be the development of sufficient low-level energy being concentrated in the vortices, and the convergence to concentrate the weak background factors that allow the vortices to be sustained. Our vertical vorticity. It is speculated that this immediate goal in following up on these preliminary mechanism is responsible for the very largest observations will be to evaluate quantitatively the vortices observed in the plume. It may be sig- effectiveness of the Meteotron as a device for con- nificant that these strong central vortices have centrating vorticity under optimum environmental been observed only when the winds are almost conditions. A field program intended to address this calm, so that the plume rises nearly vertically. issue is being planned for 1981. Mechanisms 1) and 2) then make only a small contribution. It should be noted that the low- level convergence may be enhanced by the formation of a small cumulus overhead. Acknowledgments. The authors would like to acknowledge the support received from the National Science Foundation through Grant No. ATM 77-16955. Dr. Ron Taylor, Director 5. Future work of the NSF Meteorology Program, was particularly helpful and encouraging. The cooperation of the personnel of the In summary, we have identified and documented Observatoire du Puy de D6me and of Electricity de France is also to be recognized. Dr. Bruno B£nech deserves special various types of atmospheric vortices that form in the recognition for his assistance. Ms. Sharon Von Tobel typed vicinity of large thermal plumes. Physical mechanisms the manuscript and Ms. Barbara Chance assisted in the have been proposed that may explain how such vortices preparation of the figures; their help is gratefully are formed. It is considered that the plume interacts acknowledged.

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC 694 Vol. 61, No. 7, July 1980

References Hanna, S. R., and F. A. Gifford, 1975: Meteorological effects of energy dissipation at large power parks. Bull. Am. B£nech, B., 1976: Experimental study of an artificial con- Meteorol. Soc., 56, 1069-1076. vective plume initiated from the ground. J. Appl. Meteorol., 15, 127-137. Hissong, J. E., 1926: Whirlwinds at oil tank fire, San Luis Connell, J., 1975: A non-thermal mechanism for forcing Obispo, California. Mon. Wea. Rev., 54, 161-164. cumulonimbus cloud. J. Appl. Meteorol., 14, 1406-1410. Michaud, L. M., 1975: Proposal for the use of a controlled Dessens, H., and J. Dessens, 1964: Experiences avec le tornado-like vortex to capture the mechanical energy M^otron. J. Rech. Atmos., 1, 158-162. produced in the atmosphere from solar energy. Bull. Am. Dessens, J., 1962: Man-made tornadoes. Nature, 193, 13-14. Meteorol. Soc., 56, 530-534. , 1963: Tourbillon de sable pr£s de l'incendie du puits de Gassi-Touil. J. Rech. Atmos., 1, 209. Ryan, J. A., and J. J. Carroll, 1970: Dust devil velocities: Ebert, C. H. V., 1963: The meteorological factor in the Mature state. J. Geophys. Res., 75, 531-541. Hamburg fire storm. Weatherwise, 16, 70-73. Thorarinsson, S., and B. Vonnegut, 1964: Whirlwinds pro- Graham, H. E., 1955: Fire whirlwinds. Bull. Am. Meteorol. duced by the eruption of Surtsey volcano. Bull. Am. Soc., 36, 99-103. Meteorol. Soc., 45, 440-444. •

announcements'

Conference on Port and Ocean Engineering under 1980. Notice of papers selected for presentation will be Arctic Conditions—Call for papers given by the end of December 1980. Final manuscripts must be in the hands of the organizers before 1 May 1981. The Sixth International Conference on Port and Ocean For detailed information about the conference, special Engineering under Arctic Conditions—POAC 81—will be postconference tours, registration, deadlines, fees, and ab- held in Quebec City, Canada, during 27-31 July 1981. The stracts and papers, contact: Prof. Bernard Michel, Depart- POAC conference is held every second year alternating ment of Civil Engineering, University Laval, Cit^ Universi- between northern countries. The original initiative for the taire, Quebec, Canada G1K 7P4 (tel: 418-656-2205; telex: conference was taken by the Norwegian Institute of Tech- 051-31-586). nology to improve the knowledge of Arctic problems, and scientists and technologists in the northern countries created a forum for discussion and the exchange of ideas about NSF opportunities for international science activities common tasks. The conference is being sponsored by the Ministfcre de l'Environment, Gouvernement du Quebec, A new program announcement in the National Science University Laval, Quebec, Canada. Official conference lan- Foundation's Division of International Programs describes guages are English and French. opportunities for support of international cooperative science The conference program will include the following topics activities involving developing countries of Africa, Asia, and will be divided partly into parallel sessions: environ- Latin America, and the independent nations of the Caribbean. mental aspects; oil spill; sea ice conditions in cold regions; The Science in Developing Countries program will consider icebergs; interaction between ice and shore; oceanography proposals for three categories of awards: dissertation im- and meteorology; wave mechanics and statistics; marine provement grants for research to be performed by a student foundations; remote surveillance and instrumentation tech- in one of the developing countries on a problem within that nology in cold regions; navigation in cold regions; harbors country; conference grants for the support of travel and in cold regions; coastal and offshore structures; and risk other expenses associated with bilateral seminars, workshops, analysis. POAC is open to scientific and technological papers and colloquia; and grants for the support of research par- on all the above topics. Authors are requested to submit ticipation—either by a U.S. scientist in an institution within a title and an abstract (two-page limit) before 15 November a developing country or a scientist from a developing country in a U.S. institution. Deadlines for submission of proposals are 1 April and 1 September. For further information, contact 1 Notice of registration deadlines for meetings, workshops, Dr. Gordon Hiebert (tel: 202-632-6545). The National and seminars, deadlines for submittal of abstracts or papers Science Foundation address is: 1800 G. St., N.W., Washington, to be presented at meetings, and deadlines for grants, pro- D.C. 20550. posals, awards, nominations, and fellowships must be received at least three months prior to deadline dates.—News Ed. Continued on page 711

Unauthenticated | Downloaded 09/23/21 06:12 PM UTC