90 BULLETIN AMERICAN METEOROLOGICAL SOCIETY Inflow Patterns of Thunderstorms as Shown by Winds Aloft * HORACE R. BYERS and EDWIN C. HULL B. S. Weather Bureau Thunderstorm Project f ABSTRACT Balloons ascending simultaneously at various points around thunderstorms are followed by radar or radio-direction finding (rawins and rawinsondes) to determine horizontal inflow or outflow of the air. It is found that in the early stages of development of a thunderstorm cell, inflow or horizontal convergence is present at the ground as well as all heights reached by the cloud. In a mature thunderstorm cell, the observations show outflow or horizontal divergence under the cloud base and convergence at all heights between about 4,000 and 23,000 feet, with divergence again in the uppermost levels. Relationships to rainfall and to cloud entrainment of environmental air are shown. NE of the most important phenomena [6] has shown that ordinary cumulus clouds "en- which the Thunderstorm Project f was train" air from the environment. His preliminary O designed to measure was the inflow or thermodynamic treatment has been expressed in outflow of air at various altitudes in and around relatively complete graphical form leading to a thunderstorms [4]. This was accomplished by new conception of convection and thermal sta- establishing ten balloon stations with radar or bility in a paper by Austin [1]. Fluid convection radio-direction-finding equipment concentrated in models designed and operated by Ference, Phillips a small area. By simultaneous releases of the and Fultz [7] show in the laboratory the devel- balloons in the vicinity of or inside thunder- opment of frictionally-driven circulations bringing storms, the horizontal divergence or convergence surrounding fluid into the warm ascending and of the winds at the various levels could be com- cold descending currents in a manner similar to puted by measuring the rate of shrinkage or that of jet streams as described by Tolmien [5] expansion of the areas outlined by the balloon and others. positions. Thus, for example, three balloons The data on the inflow and outflow of air would form the vertices of a triangle. If at any around thunderstorms have been studied from height the area of this triangle were decreasing, the observations made in the Thunderstorm Proj- horizontal convergence would be indicated; di- ect operations in the St. Cloud, Florida, area vergence would be shown by an expansion of area. during the summer of 1946 [8, 9]. The obser- In the circulation of a thunderstorm, these vations consist of rawin and rawinsonde runs horizontal inflows and outflows are only part of made by as many as ten stations within an area the complete picture, involving also the vertical of 120 square miles. The releases were made currents which are well known as being especially simultaneously into a thunderstorm cloud or into characteristic of these storms. In a recent paper its nearby area. The data show the wind direc- [3], by Byers and Braham, the complete three- tion and velocity for each 1000-foot level as well dimensional representation was shown, based upon as the location of the balloon and the time at vertical currents measured from airplanes as well which it reached each level. During all observa- as the horizontal circulation obtained from the tions, the large CPS-1 (MEW) search radar balloons. However, it is believed that the hori- was operated and photographs of the 'scope were zontal measurements are of such significance that made showing the outline and location of the they deserve separate treatment and should be storm echo. The surface data of rainfall, wind, made available in relatively complete form for temperature, pressure and relative humidity were meteorologists to examine. Other recent studies available from the micronetwork of 55 recording have emphasized the importance of and the inter- stations. est of meteorologists in this problem. Stommel Storms and upper-wind runs were selected in which there were simultaneous wind readings 151 Published report No. 4 of the Thunderstorm Project. from three or more stations on two or more f A joint project of the Air Force, Navy, National sides of the radar storm echo. Whenever pos- Advisory Committee for Aeronautics and the Weather Bureau. sible, situations were chosen that included a single Unauthenticated | Downloaded 09/25/21 10:57 AM UTC VOL. 30, No. 3, MARCH, 1949 91 FIG. 1. Diagram showing wind field about a cumulonimbus cloud, illustrat- ing deformation of wind field by presence of cloud. or isolated thundercloud, in order that all the had been made for a particular level within an effects noted could be ascribed to the single cloud eight-minute period were included. The radar and the results could be more easily interpreted. echoes showing the position of the storms at the Wind vectors for each 1000-foot level were time were plotted on the same chart. plotted with 1 inch equal to 20 mph on a chart It was immediately evident that the normal having a scale of inch to the mile. The origin smooth pattern of the wind field is destroyed in of the vector was plotted at the actual position of the vicinity of the storm (FIG. 1). The winds the balloon rather than at the position of the at a distance from the cloud continue on a con- station making the run. All wind readings that sistent and orderly course, while the winds near FIG. 2. Map of St. Cloud area showing triangles spanning radar cloud echo, used for computations of convergence and divergence aloft. Lettered stations released rawins, numbered stations released rawinsondes. Triangles used in computations are identified by the station letters and numbers from which the balloons were released. Unauthenticated | Downloaded 09/25/21 10:57 AM UTC 92 BULLETIN AMERICAN METEOROLOGICAL SOCIETY FIG. 3. Distribution of convergence and divergence with height as meas- ured for the various triangles which span storm No. 53. (Positive values indicate divergence.) or within the storm have a greatly different direc- tion and speed. This change in the wind field in the region of the storm cloud varies with time and with height and with the age and intensity of the storm. In order to compute and demon- strate the amount and direction of this change for all levels of the wind run, the following method was used. Triangles were drawn with vertices at the ori- gins of the various wind vectors and computations of convergence or divergence were made for the triangles which most nearly spanned the storm (FIG. 2). The computations were made by a graphical method proposed by Dr. John C. Bel- lamy, using the nomogram designed by him [2]. When applied in suitable situations, this method determines the average velocity convergence or divergence over the area enclosed by the triangle expressed as a percentage change in area per unit time where decrease in area corresponds to con- vergence and an increase in area to divergence. FIG. 4. Generalized convergence-divergence aloft curve Throughout this paper the divergence in the hori- for storms in which rainfall has begun at the surface. Unauthenticated | Downloaded 09/25/21 10:57 AM UTC VOL. 30, No. 3, MARCH, 1949 93 zontal velocity field is expressed as a percentage rapidly with the maximum rates occurring be- change in area for 12 minutes. A 10% increase tween 15,000 feet and 20,000 feet. In cases in area per 12 minutes is equivalent to divergence where the balloon runs are adequate, the con- of 1.39 X 10-4 sec"1. vergence is shown to continue as high as 22,000 Eight swarm ascents were computed and illus- feet. Above 22,000 feet divergence was found to trated. The graphs of the convergence and di- be present. A generalized pattern for this type vergence aloft for the various storms were of two of storm is shown in FIGURE 4. general types. Five storms, including No. 53 A previous study has shown that the surface (FIG. 3), had very similar patterns with diver- divergence is related to the intensity of the sur- gence present in the lowest levels and marked face rainfall and that rainfall in excess of 0.30 convergence aloft. There is a notable consistency inch per 5 minutes is always accompanied by among similarly-situated triangles in the same divergence in the winds at the surface (FIG. 5). storm and also from storm to storm. The divergence continues at the surface under the The divergence usually extends to 3000 feet storm, at the same time that there is convergence but may extend to 5000 feet. In all of the storms, aloft, as long as the intensity of the surface rain- the rates of convergence are approximately equal fall continues high (FIG. 6). from 4000 feet to 15,000 feet and then increase The remaining three ascents computed were FIG. 5. Scatter diagram showing relation of values of surface diver- gence to rainfall intensities. Points corresponding to rates of rainfall less than .03 inches per 5 minutes, were omitted from this diagram. Unauthenticated | Downloaded 09/25/21 10:57 AM UTC 94 BULLETIN AMERICAN METEOROLOGICAL SOCIETY made into growing cumulus before heavy rain had begun at the surface. They gave values of convergence-divergence aloft which are less con- sistent among themselves, but they differed mark- edly from the other five in that they did not show the divergence in the lower levels; instead, they showed convergence at all levels including the surface (FIG. 7). The convergence at the surface continued throughout most of the ascent (FIG. 8). The composite curve for these storms (FIG. 9) is similar above 4000 feet to the gen- eralized curve of the other five but the degree of convergence is not as great and the excessive maximum rate at 15,000-20,000 feet is not present.