838 
this study was aimed at ascertaiming the significance 
of mean flow representations not only with respect to 
the mean pattern itself but also with respect to the 
smaller-scale and more transitory disturbances going 
to make up the mean flow. This involved the applica- 
tion of the turbulence principles of fluid mechanics 
to the circulation of the westerlies. Similar mvestiga- 
tions [9, 24] have been undertaken in the past, but no 
attempt had been made to apply this concept through 
a more-or-less statistical treatment of a long record of 
weather data such as the forty-year Northern Hemi- 
sphere series [28]. In this treatment the smoothed-out 
fluctuations, the migratory and transient cyclonic and 
anticyclonic systems, were treated as turbulent eddies 
having the property of being able to transport heat 
and momentum across large horizontal sheets of the 
atmosphere. This analysis indicated that the very large 
amounts of heat picked up from the oceans off the east 
coasts of the continents [16] are removed from the 
westerlies downstream through lateral turbulent dif- 
fusion so that equilibrium is maintained. In addition it 
showed that in years when the north-south thermal 
gradient is in excess of normal, the lateral mixing in the 
westerlies is excessive, and the cross-westerly heat 
transport is also excessive. In connection with exces- 
sive lateral mixing there occurred an excess of meridi- 
onal flow types. 
The detailed analysis of the eddy motion in the 
westerlies indicated the existence of a spectrum of eddy 
periods with a mean of about six days in the heart of 
the westerlies and slightly higher values on either side. 
The most frequent value was three days, corresponding 
to the mean period in the passage of cyclone families 
as used in weather typing. The longer-period eddies 
even showed a tendency towards durations which were 
multiples of the three-day period. 
Eddies of very long duration were associated with 
larger-than-average distortions from the normal flow 
pattern, with respect to both wave length and ampli- 
tude. In many cases these large eddies appeared on the 
surface charts as quasi-stationary anticyclones in the 
middle latitudes, with trapped cyclones to the south 
of them. In the study these were referred to as “block- 
ing highs” because of their action in blocking the 
normal eastward progress of migratory systems. In 
cases where such large blocking eddies exist 1t becomes 
hardly proper to treat them as eddies in the time and 
space scales generally involved in the extended-range 
circulation representation schemes. The smaller and 
more frequent short-period eddies characterizing true 
migratory cyclone and anticyclone activity are steered 
around or completely blocked by these eddies. There- 
fore the large eddies merely serve to introduce large- 
scale distortions of the normal westerly flow on ex- 
tended-range charts. In effect, these eddies tend to 
lengthen the closed path of the westerlies around the 
hemisphere and there is thus provided a longer perim- 
eter along which the smaller eddies can act to trans- 
port heat and momentum across the westerlies. This 
situation corresponds to that existing in the meridional 
flow category discussed above. These types must then 
be those for which the cross-westerly heat transport is 
WEATHER FORECASTING 
at a maximum, and the zonal category corresponds to 
the case where the girdle of westerly winds has a 
minimum length, at least in the regions where zonal 
types are in existence, and therefore represents a state 
of minimum cross-westerly heat transport. 
It is often apparent that there is a certain time lag 
between the development of meridional flow types in 
various regions of the Northern Hemisphere. This ob- 
servation coupled with the known quasi-periodic alter- 
nation between zonal and meridional flow types within 
one type region suggests that under ordinary circum- 
stances the zonal flow types are incapable of providing 
the required cross-westerly heat transport needed to 
balance global radiational effects, whereas the merid- 
ional flow patterns are more than adequate for this 
purpose. Since there is no dynamically stable mter- 
mediate stage, an alternation between the two circula- 
tion types results. At times the alternation seems to be 
in phase throughout the Northern Hemisphere but at 
other times this is not true. In view of the rapid dis- 
persion of energy indicated by recent theoretical in- 
vestigations [25], one might expect the in-phase condi- 
tion to predominate but there is some evidence of 
slower modes of energy propagation in the atmosphere. 
Regardless of this, in any given region the action sug- 
gested is analogous to that of a relaxation oscillator, 
the potential, as represented by the north-south thermal 
gradient, buildmg up to critical levels from time to 
time and then relaxing as the flow of the westerlies 
breaks down into meridional flow patterns. A study 
by Schwerdtfeger [27] indicated that in a source region 
for polar outbreaks the polar air may be built up to 
critical strength from a stage of complete exhaustion 
through radiation processes within a period of some 
three weeks. Observation indicates, however, that the 
pulse period of outbreaks in North America is more on 
the order of a week (corresponding to rather large- 
sized eddies) with a longer-period fluctuation in the 
strength of the pulses on the order of a month. The 
latter would correspond to the oscillation in question 
but such pulsations are far from systematic and hence 
it seems that other events, such as perhaps variations 
in extraterrestrial radiation, may play an important 
role. 
The effect of the irregular geographic distribution of 
oceans and continents, although complicating matters, 
should not in itself account for the lack of regular 
periodicity. 
INTERACTION ON A HEMISPHERIC 
BASIS 
As was pointed out above, the most difficult step in 
extended-period forecasting is that involving the pre- 
diction of the movement and changes in form of the 
systems appearing on the extended-period representa- 
tions. In the weather-type method this amounts to the 
forecasting of the next weather type or types. This 
usually involves directly or indirectly the study of 
large-scale and long-period weather processes on a hemi- 
sphere-wide basis. Some of these general processes 
which can be used in connection with extended-period 
forecasting by the use of weather types were set down 
a 
