688 
2. The maximum lightning frequency occurs at the 
same time as the cell reaches its maximum height. 
3. As the height of the cell top decreases, the 
frequency of lightning also decreases. 
4. It appears that a greater cell height (or lower 
temperature of the cloud top) is necessary to initiate 
lightning than is required to maitain it once it has 
started. 
5. In the life cycle of a thunderstorm the maximum 
frequency of lightning precedes the time of its maxi- 
mum 5-min rainfall. 
This last conclusion may have some meaning in con- 
nection with the possibility of the suspension of the 
raindrops by the electric field. 
The Thunderstorm as Disclosed by Radar 
A number of studies of thunderstorms have been 
made by radar. A large amount of useful data was 
obtained from the radars of the Thunderstorm Project. 
From the photographic records of the range-height- 
indicating (RHI) radarscope it was possible to obtain 
data on the altitude of first formation of 66 radar 
echoes from convective clouds. A graph was made of 
the frequency distribution of the differences between 
altitudes of the tops of the initial echoes and the con- 
current heights of the freezing level. The distribution 
showed a pronounced mode at 1000 ft above the freezing 
level but the mean was at +2200 ft. In convective 
clouds the echoes appear abruptly after the cloud has 
been visible to field observers for some time, suggesting 
a sudden release of great quantities of large waterdrops 
after penetration above the freezing level, in accordance 
with the Bergeron theory. 
It was found that the rate of vertical growth of the 
top of a radar cloud echo agrees closely with the updraft 
velocity measured in that portion of the cloud if a 
correction factor of about 3 ft sec is added to the radar 
growth rate to account for the relative free fall of the 
attenuating snowflakes or other particles. The use of 
radar in this manner for measuring updrafts appeared 
to have a practical application in detecting hail pos- 
sibilities, since all observed cases of hail accompanied 
strong updrafts. 
It was noted that the thunderstorm tops ascended in 
a series of steps, appearing as the growth of new protub- 
erances or “turrets,” during their growth and, as a 
matter of fact, in 32 cases studied, the maximum height 
reached was correlated with the number of turrets thus 
formed by a coefficient of + 0.67 + 0.10. Hach succes- 
sive turret was higher than the preceding one, and 
the mean lapse of time between successive turret 
peaks was 17.8 min. It is believed that each turret 
makes it easier for the following ones by increasing the 
moisture content in the cloud-top environmental air 
which must be entrained in further growth. Less 
heat of condensation is robbed from the new turret by 
entraining than would be the case if it were standing 
alone in a dry environment. Each turret underwent a 
growth period followed by an interval of subsidence. 
The growth period averaged about 16 min. The average 
LOCAL CIRCULATIONS 
vertical growth rate was about 18 ft sect and the 
subsidence rate about 12 ft sec7?. 
The mean of the maximum heights nen ghad by 199 
Ohio storms as indicated from the radar was 37,500 ft. 
There was no significant difference in this respect be- 
tween different types of thunderstorms, such as air- 
mass, squall-line, or frontal. 
The boundaries of the radar cloud echoes were found 
to agree closely, with the visible cloud limits except in 
the levels near the cloud base where non-echo-produc- 
ing “outrider” clouds were common. Also the anvils and 
other layer-type lateral extensions usually did not re- 
turn a radar echo. A positive correlation was found 
between the horizontal and vertical extents of thunder- 
storms; that is, the taller the cloud, the broader it was. 
The greatest areal coverage was at an altitude around 
10,000 ft. The cross-sectional area was slightly less at 
the lower level and tapered at the top, forming a total 
cloud echo shaped like a rosebud. 
From radar scans covering an area of over 55,000 
square miles, it was found that on average thunder- 
storm days in Ohio 10 per cent of the area would be 
covered by cloud echo, and on a day of maximum 
thunderstorm activity, 40 per cent of the area would be 
covered. It is found from indirect comparison that at 
20,000 ft the in-cloud areas would average 5 per cent ~ 
and have a maximum of 22 per cent, indicating the 
better chance of avoiding thunderstorms by flying high. 
(Other measurements showed that these high levels are 
the worst choices for flights within thunderstorms.) 
Effect of Environmental Wind Field 
The causes and effects of vertical shear, including 
effects of entrainment and momentum transfer in the 
drafts, have been investigated by Byers and Battan 
[6] and Malkus [20]. 
With the aid of long-range radar and abundant upper- 
air wind data, the movements of thunderstorms in 
Florida and Ohio were studied in relation to the wind 
fields in which they were embedded. A method was 
devised whereby the translational component of the 
motion could be separated from the growth or 
dissipative component. 
It was found that the motion of the storm corre- 
sponded most closely to the mean vector wind between 
the gradient level and 20,000 ft. The correlations were 
better in Ohio than in Florida owing to a stronger, more 
consistent wind flow in the former region. In direction, 
the correlations were 0.95 or better in both regions. It © 
was found that the speed of the cloud was less than that 
of the vector mean wind from gradient to 20,000 ft. 
When the two speeds are plotted, a nearly straight line 
is formed, but the relationship is better represented as 
Uw = 1.9 + 0.65U. + 0.020U-’, 
where U, is the vector mean wind speed and U- is the 
cloud speed. 
Preferred Areas of Thunderstorm Development 
The movement of thunderstorm cells is closely related 
to the question of preferred areas for their new develop- 
