810 
also a persistent recurrence of certain circulation types 
gives a decided character to both mean pressure pat- 
terns and associated weather phenomena. This rela- 
tionship is rather fortunate in practical extended-range 
forecasting since it frees the forecaster from the seem- 
ingly hopeless task of giving a play-by-play description 
of the detail of the weather in order to make a state- 
ment of average conditions anticipated. Likewise the 
use of means makes it unnecessary to break up the 
period, if long, into a sequence of discrete types. 
One rather simple technique of translating circula- 
tion patterns into weather anomalies consists of con- 
structing isolines of departure from normal of the pres- 
sure or height. From the components of flow relative to 
normal, in conjunction with the distribution of normal 
surface isotherms or normal thickness charts of the 
lower troposphere, one can readily decide if a circulation 
is apt to bring warm or cold air to a locality and also 
get a rough measure of the degree of abnormality. 
Statistical refinements of this method as practiced in 
America and developed by Martin [9, 10] attempt to 
determine for various places and for different months 
the key regions where the anomaly of the contour pat- 
tern to a large extent governs the temperature anomaly. 
These studies bring into sharp focus the distant con- 
trols operating on apparently local weather. For ex- 
ample, for most stations in the eastern United States it 
turns out that surface temperature anomalies in winter 
are remarkably dependent upon the anomaly of 700- 
mb heights near southeastern Alaska! When the mid- 
tropospheric ridge in this area is strong (high positive 
anomaly) it sets in motion vast outbreaks of polar 
Canadian air which pour into eastern United States 
and produce low temperatures there. When the ridge 
is absent, a series of frontal waves move across the 
United States-Canadian border and a westerly flow of 
mild maritime Pacific air leads to warm weather over 
the east. These and other factors are now considered 
objectively so that it is possible to construct in a me- 
chanical fashion the temperature anomaly anticipated 
with a given 700-mb pattern. 
The relationship of precipitation to mean circulation 
is more difficult. Even here, however, a reasonable 
degree of success may be achieved by relating to the 
observed precipitation such factors as location and 
orientation of major trough and ridge systems of the 
mid-troposphere, anomalous components of wind, and 
curvature of isobars [7]. While a strictly objective 
method of doing this is obviously the goal, it appears 
to be difficult to design a system as effective as the 
objective temperature method. 
When it comes to giving a day-by-day weather fore- 
cast beyond two or three days in advance, the current 
state of knowledge is indeed unsatisfactory. Most of the 
skill of extended-range forecasts appears to rest in the 
ability to predict general characteristics of a period and 
perhaps broad trends. The strain imposed by the re- 
quirement of a running description of detailed weather 
in temperate latitudes for, let us say, a week is much too 
great for all systems of extended-range forecasting de- 
veloped so far. Consequently, the degree of skill when 
WEATHER FORECASTING 
verified on a day-by-day basis falls off rapidly after 
two or three days. However, there is some encourage- 
ment in the fact that even on the sixth day some fore- 
casts have demonstrated slightly greater success than 
climatological probabilities. But such forecasts can 
hardly be used to plan a picnic. 
Despite this low level of skill an attempt is generally 
made to specify day-to-day conditions through the me- 
dium of the prognostic chart. In the U. S. Weather 
Bureau, prognostic charts for six days in advance are 
prepared—not with the belief that they possess striking 
accuracy, but rather with a view of expressing the 
anticipated general lines of air-mass, frontal, cyclonic, 
and anticyclonic activity. The prognoses elucidate fur- 
ther the anticipated mean state of the circulation and 
serve as a guide to general trends in temperature, rain- 
fall, wind, etc. The charts are drawn by considering the 
mean prognosis and its steermg influence, and are 
naturally constructed so as to be logically continuous 
with the short-range (7.e., 30-hr) prognosis. Obviously, 
a healthy, vivid imagination and a firm hand on the 
part of the forecaster are helpful. 
A more objective method of preparing such day-to- 
day prognoses for periods of a week or so in advance is 
supplied by the analogue method, for here, once the 
analogous period is decided upon, one can substitute 
(perhaps with small modification) observed maps for 
prognostic maps. The advantage of greater objectivity, 
while desirable and appealing, is unfortunately largely 
offset by inability to find in the archives of maps proper 
analogues, not only before the events, but many times 
even after the period is over. 
Within the past few years the prognosis of mid- 
troposphere flow patterns for 48 or 72 hr in advance 
has become increasingly popular and these charts offer 
a starting point for a more extended prognosis. The 
methods for constructing these prognoses are for the 
most part those briefly discussed under physical meth- 
ods, and involve discriminating use of vorticity prin- 
ciples, thermal fields, warm and cold advection, and 
kinematic (generally pressure-change) methods. As in 
the mean-map technique, there appears to be no es- 
pecial reason why the forecast must be made in steps of 
24-hr intervals. It would appear that synoptic meteorol- 
ogists have too long considered the unit of 24 hr as 
mandatory in the preparation of a series of charts. 
The theory underlying modern extended-range forecast 
techniques is that the forecaster strikes out fora general 
state expected to be observed well ahead of 24 hr and 
then proceeds to determine how this general state will 
evolve. Experienced forecasters of all countries who 
have tried the step-by-step or extrapolation method of 
extended-range forecasting have long ago recognized 
that this method breaks down after about 48 hr. Prob- 
ably the reason for this breakdown lies in the inability 
of man’s brain to cope with the increasing complexity 
of atmospheric interactions. If and when electronic 
machines are used to grind out a reasonably good 24-hr 
mid-troposphere prognosis, meteorologists will have a 
good way of determining how far the step-by-step 
method of long-range prognosis can proceed. 
