THUNDERSTORMS 
where K combines a number of factors relating to the 
airplane used, a is the acceleration increment in g 
units, p and pp are air density at flight level and sea 
level, respectively (slugs ft~*), and V is the true air 
speed (ft sec™'). 
The relationship shows that a given effective gust 
velocity will produce a greater acceleration the higher 
the speed of the flight. Through substitution of the 
values of airplane characteristics contained in K, the 
accelerations produced on one airplane can be trans- 
lated into the accelerations on another of different 
characteristics. 
From 1362 traverses through thunderstorms on the 
Thunderstorm Project it was determined that if on 
any one of these traverses the mean maximum U, per 
3000 ft of fl'ght exceeded 8 ft sec and four gusts per 
3000 ft, the aircrews reported the turbulence as heavy. 
The flights were made by crews highly experienced in 
bad-weather flying. This information proved useful in 
convincing pilots and meteorologists that data collected 
by N.A.C.A. in which gust velocities rarely exceeded 
25 or 30 ft sec~!, really represented violent thunder- 
storms. The highest gust velocity recorded on any of 
these flights was 43 ft sec! which, incidentally, exceeds 
the maximum gust load that many airplanes are built 
to withstand at cruising speeds, free of maneuver 
loads. By contrast, sustained drafts of greater than 30 
ft sec! were not uncommon and on a few occasions the 
values were in the vicinity of 90 ft sect. 
A significant difference among the different levels 
flown was noted in the magnitudes of gusts and drafts 
and in the frequency of high values. The lowest values 
of both gusts and drafts were, in the mean, found at 
the lowest standard flight levels of the project—6000 
ft above ground in Florida and 4000 ft above ground in 
Ohio. A statistical treatment of the data shows a ten- 
dency toward increasing turbulence with height up to 
the highest levels flown (25,000 or 26,000 ft), but the 
differences between the various levels from 10,000 ft 
upward in this respect is slight. The draft velocities 
showed a distinct maximum at the highest levels. In 
general, the values of both types of turbulence were 
greater in Ohio than in Florida, possibly due to a more 
successful effort to get the airplanes into the thunder- 
storm before the very intense early-mature stage had 
passed. The mean of the maximum effective gust ve- 
locity above 4 ft sec in 3000 ft of traverse in Ohio 
was 9.4 ft see and in Florida 8.9 ft sec. 
By the use of a range-height-indicating radar, meas- 
urements have been made on the rate of growth of 
thundercloud tops, some of which have been observed 
to extend above 55,000 ft. These measurements show 
that the mean rate of growth, up to an altitude about 
10,000 ft below the top of the individual storm, in- 
creases with height. The rate of growth of these cloud 
tops is also a measure of updraft velocities. Therefore 
one might assume that the updrafts and, with them, the 
gust velocities, increase with height, at least during the 
growing stages, up to a level about 10,000 ft less than 
the maximum height reached by the storm cell. 
Hinally, it has been shown that areas of highest 
water concentration in a thunderstorm are the areas 
691 
of heaviest turbulence. This supports the idea that 
radar can be useful for avoiding areas of excessive 
turbulence. In flights below the cloud base, the heaviest 
turbulence will be found where the darkest rain columns 
are seen. 
Unsolved Problems and Future Research Needs 
Although the research in the three years from 1946 
to 1949 added more details to our knowledge of thunder- 
storms than had been accumulated in many decades 
previously, it also focused attention on a number of 
unsolved problems. 
Some of the questions can be settled by further de- 
tailed observations or measurements. One question of 
detail having fundamental importance for the whole 
circulation and energy problem is brought up by a 
finding of the Thunderstorm Project that the strongest 
downdrafts as well as updrafts are frequently found in 
the upper parts of the cells. No rational picture of a 
thunderstorm cell with the downdraft decreasing down- 
ward has been devised. Additional data should be 
gathered, since the number of measurements of down- 
drafts is small at the high altitudes. Because of difficulty 
in maintaining the attitude of the airplane, more than 
30 per cent of the drafts could not be evaluated. An- 
other detail that could be studied concerns entrain- 
ment. It is not known to what extent entrainment 
involves lateral mixing. This could be solved by ob- 
taining the horizontal gradient of water vapor from 
some distance outside the cloud up to its very edge by 
means of an airplane carrying a sensitive dew-point 
hygrometer. Further measurements are also needed 
concerning the tendency toward desiccation of the 
downdraft air, as indicated by the “humidity dip” at 
the ground in the rain core. 
Additional observations should be made on thunder- 
storms in arid regions, especially in those cases where 
the condensation level is very high and the rain may 
sometimes evaporate before reaching the ground. The 
water and energy budgets of such cells must be radi- 
cally different from those of humid regions. The prob- 
lem of hail has not been solved in relation to thunder- 
storms. Hail, unless in the form of stones more than a 
centimeter in diameter, is hard to detect in a fast-flying 
airplane, since heavy rain itself makes a great deal of 
noise. The Thunderstorm Project did not obtain data 
from the region of maximum hail occurrence of the 
Great Plains and Rocky Mountain states. The problem 
of hail in the generation of thunderstorm electricity 
is a critical one. In spite of the nearly 200 years that 
have elapsed since Benjamin Franklin discovered that 
lightning was a form of electricity, we still are not sure 
what causes it. 
The squall line and tornadoes in relation to thunder- 
storms are a wide-open field for investigation. Tepper 
[37] and Newton [23] have attacked these problems 
from noyel viewpoints. The dynamics of the pre-cold- 
front squall line—whether it is a hydraulic jump phe- 
nomenon as emphasized by Tepper or has other special 
characteristics—need further investigation. In this con- 
nection it is interesting to note that, from radar photo- 
graphs, it appears that squall lines crossing the isobars 
