MODEL TECHNIQUES IN METEOROLOGICAL RESEARCH 
low speed often provides a cheap solution for temporary 
installations. Although the air speed should be con- 
trollable over as wide a range as possible, peak speeds 
greater than 100 ft sec! are seldom necessary—a matter 
of considerable practical importance, since power re- 
quirements vary with only the first power of the ca- 
pacity but with the cube of the speed. 
Just as problems of liquid flow are sometimes studied 
with gas as the fluid medium in order to utilize partic- 
ular measuring equipment, some phases of gaseous 
movement may be investigated more conveniently with 
liquids. This is particularly true in the observation of 
low-velocityy convection currents in which the variation 
of the specific gravity of the atmosphere due to thermal 
effects is simulated in water either thermally or through 
admixture of ordinary salt. The use of different dyes to 
indicate the various specific gravities is far simpler 
than the use of smoke in air, and the greater density of 
water makes the measurement of low velocities more 
feasible. Under such circumstances the glass-walled 
channel of the hydraulics laboratory can become a 
useful observational tool in meteorological research. 
Aside from such specific apparatus, perhaps the most 
important requisite for the meteorological model labora- 
tory is available space in which temporary equipment 
of any desired naturemay be constructed. Head room as 
well as floor area is essential, together with the possi- 
bility of erecting light walls wherever desired to elimi- 
nate extraneous currents. The frequent requests for 
space in university field houses for such projects during 
the war are indicative of research needs; a small airplane 
hanger would be ideal for a laboratory of this nature. 
As for the specific instruments involved in model 
studies of atmospheric movement [10, Vol. I; 11], prob- 
ably the most essential is a series of anemometers of 
various types for the measurement of velocity. These 
range from indicators of spatial mean flow of the air 
stream itself, through point indicators of temporal mean 
flow varying spatially in both magnitude and direction, 
to point indicators of turbulent fluctuations. If the air 
speeds are above some 10 ft sec~!, the ordinary Pitot 
tube in connection with a sensitive differential gage of 
the Wahlen type is useful in making either point indica- 
tions of mean velocity or velocity traverses for integra- 
tion to yield the spatial mean. Directional Pitot tubes 
are also available. Duct or tunnel systems producing a 
pronounced acceleration of the flow into the test section 
permit the mean air speed to be indicated in terms of the 
accompanying pressure change by means of boundary 
piezometers at suitable points. Vaned anemometers 
with either counter or direct electrical indication are 
also useful in large passages, and midget anemometers 
of this nature have been made for either small-scale 
or low-velocity conditions. The only satisfactory instru- 
ment yet devised for the measurement of velocity 
fluctuations as well as mean values is the hot-wire 
anemometer [17], which requires for its proper operation 
a mean air speed in excess of about 5 ft sec~! and 
essentially constant temperature of the ambient fluid. 
The hot-wire technique has progressed to the point 
that not only the intensity but also the scale of the 
1251 
turbulence and the intensity of the resulting shear can 
be measured. The mixing characteristics of the turbu- 
lence may also be determined by the diffusion of heat 
or gas from a point or line source. Such other flow 
characteristics as the variation of pressure and tem- 
perature require the use of standard piezometers and 
thermocouples, respectively. 
Quite as important as the actual measurement of the 
model flow characteristics is the visual or photographic 
observation of the flow pattern [10, Vol. I; 12]. Because 
of the normal transparency of either air or water, it is 
necessary to observe the alignment of short threads or 
the motion of opaque material suspended in the flow, 
such material being so finely divided or of a specific 
gravity so nearly equal to that of the fluid that its 
motion will be essentially the same. The use of dye 
filaments, aluminum powder, or oil droplets is common 
in water. Very fine powders are sometimes used in air, 
but either an oil fog or a chemical smoke such as that 
from titanium tetrachloride is more generally satis- 
factory. In all cases a dark background with top lighting 
slightly to the rear is desirable. In both media the 
change in refractive index with a local change in density 
is sometimes utilized, rear lighting from a point source 
such as a carbon are being required. 
In addition to permitting visual observation or photo- 
graphic recording of flow patterns, the foregoing meth- 
ods are also directly useful in the preparation of motion- 
picture sequences for instructional purposes. Since many 
phenomena of atmospheric motion which are not sub- 
ject to quantitative study in the laboratory can still be 
simulated qualitatively with relative ease, graphic train- 
ing films in which the essential characteristics of these 
phenomena are emphasized may readily be prepared. 
One such film, assembled during the war by the lowa 
Institute of Hydraulic Research for the Chemical War- 
fare Service, showed at model scale in a low-velocity air 
tunnel the diffusion of smoke and gas by wind under 
various conditions of thermal stratification, boundary 
roughness, forestation, and urban development. An- 
other, prepared by the Hydrodynamics Laboratory of 
the California Institute of Technology, simulated by 
gravity currents in water the characteristics of a gas 
attack on a valley stronghold. 
Typical Laboratory Investigations 
One of the earliest scale-model investigations of at- 
mospheric phenomena is the study by Abe [1] of wind 
structure over Fujiyama. According to the general cri- 
teria for similitude discussed in an earlier section, it 
would appear that the Reynolds number should have 
the same magnitude for the model and prototype flows 
if viscous similarity is to prevail. Abe reasoned, how- 
ever, that the analogy between molecular motion on a 
very small scale and eddy motion on a very large scale 
permitted similarity to be obtained by using the mo- 
lecular viscosity (a fluid property) in the model Rey- 
nolds number and the estimated eddy viscosity (a flow 
characteristic) in the prototype Reynolds number. As a 
result, his model air speeds resulted in flows which were 
nearly, if not wholly, in the viscous range. 
