OBSERVATIONAL STUDIES OF GENERAL CIRCULATION PATTERNS 
ented solenoidal fields at coast lines. More recently a 
similar role has been ascribed to mountain barriers. 
By the use of Rossby’s now classical formula,* plus 
the normal strength of the westerly flow, theoretical 
stationary wave lengths may be computed for each 
season for North America and Asia. The smallest of 
these has a length of 2500 mi, considerably larger than 
the dimensions of the average cyclone. Also, the seasonal 
variation of theoretical and observed stationary wave 
lengths is similar over North America. However, the 
observed waves are considerably longer than the theo- 
retical waves. In view of the fact that average zonal 
wind speeds over the Hastern Hemisphere are somewhat 
lower than those over the Western, it becomes difficult 
to explain the larger observed wave lengths over Asia. 
Among the many factors responsible for these dis- 
erepancies perhaps the most obvious is that the conti- 
nents, in contrast to the oceans, interpose not merely 
one narrow obstacle in the path of the westerly flow but 
a whole series of more or less continuous obstacles of 
cana 
LT 
7 si 
5 
A 
6 
) 
WEATHER BUREAU 
NORMAL 10,000-FOOT PRESSURE 
FEBRUARY 
557 
a system of free stationary waves of constant wave 
length in such a current is to have a distribution of 
divergence such that in the lower layers there is mass 
convergence east of the trough lines and mass diver- 
gence to the west. At high levels they postulate the 
reverse distribution of divergence. Now the average 
level of nondivergence has been estimated to be at 
roughly 16,000 fi. Thus, the assumption of nondiver- 
gence at the 10,000-ft level, made in deriving the classi- 
cal wave formula, is not valid. The distribution of 
divergence postulated above has the effect of increasing 
the stationary wave length. 
Charney and Eliassen [11, 12] have shown that the 
effects of topography, surface friction, and baroclinity 
can all be expressed in terms of divergence at upper 
levels. It is therefore desirable to obtain an estimate 
of the normal fields of divergence. An attempt along 
this line has been made for the 10,000-ft level by the 
authors [37] and a sample of the results for February is 
shown in Fig. 8. 
\ 
~ SS 4 
KINGS 
Si 
WN 
a 
NX 
rs 
\ ‘SS 
Fic. 8.—Normal field of divergence at 10,000 ft for February. Thin curved lines are isobars for 5-mb intervals; heavy curves, 
divergence for intervals of 5 X 10~7 sec. (From Namias and Clapp [87].) 
varying importance all the way around the globe. Thus, 
at best, we must regard the observed flow pattern as a 
heterogeneous combination of waves set up by each of 
the numerous obstructions. Precisely this line of attack 
has been taken by Charney and Eliassen [12]. 
Perhaps the most important reason for the discrep- 
ancies is that the atmosphere is baroclinic so that the 
westerly winds increase with height. As shown by 
Bjerknes and Holmboe [4], the only way to maintain 
3. L, = 2rx/U/B, where U is the zonal wind speed (index), 
L, the stationary wave length, and 8 the northward rate of 
change of the Coriolis parameter. 
The large-scale features within the belt of strongest 
westerlies indicated in Fig. 8 show regions of maximum 
divergence to the west and convergence to the east of 
the trough lines, in partial confirmation of the Bjerknes- 
Holmboe theory [4] mentioned above. Another note- 
worthy finding (not illustrated here) is the increasing 
strength of the divergence fields from winter to summer 
accompanying subtropical highs while the fields asso- 
ciated with the circumpolar westerlies are weakening. 
The effect of topography is also apparent from Fig. 8. 
As air is lifted over the Rocky Mountain chain from 
Alaska to Mexico, it is subjected to vertical shrinking 
(horizontal divergence), and as it sinks down the eastern 
