EXTENDED-RANGE FORECASTING BY WEATHER TYPES 
Although some investigators who are inclined toward 
statistical analysis advocate the elimination of one or 
more of the intermediate steps by going directly from 
the representation scheme to the forecast of the weather 
elements themselves, in the opinion of the author our 
present knowledge concerning weather processes is so 
limited that the use of the intermediate steps is re- 
quired in order that the forecaster can grasp more 
readily the physical significance of the processes in- 
volved. 
A general discussion of the California Institute of 
Technology weather-type technique is presented in the 
following section. 
THE WEATHER-TYPE APPROACH 
Various sorts of weather types had been developed 
for different areas in the past [2, 14, 23]. Often the earlier 
attempts at typing were based upon single typical 
synoptic charts and therefore did not truly take into 
account weather processes occurring during an extended 
period. After some preliminary investigation during the 
late thirties [10, 18] a six-day North American weather- 
type scheme was developed wherein the type itself em- 
braced a sequence of days characterized by certain 
locations and orientations of the semipermanent ele- 
ments of circulation such as the Pacific anticyclone, the 
Aleutian cyclone, the trajectories of polar outbreaks, 
and cyclone paths. Empirical evidence was discovered 
which indicated that the mean period between the 
passages of cyclone families past a given point was 
about three days and that about two-thirds of the time 
the type characteristics associated with the passage of 
a given family occurred also in either the preceding or 
the following cyclone family. Because of this, the types 
were arranged so as to have a lifetime in any given sec- 
tor or region of six days plus or minus one day. The 
instances where the type characteristic was not main- 
tained during the passage of two successive cyclone 
families were handled by the introduction of the so- 
called “complex” types in which the first three days 
had the characteristics of one type and the last three 
those of another. Considerable work was done on these 
types over a period of several years, resulting in the 
development of a weather-type catalogue covering the 
period 1921—42 [22] and a set of “ideal” types [3] show- 
ing the most probable location of frontal systems and 
centers on successive days of each type, along with 
other statistical information on the weather element 
anomalies associated with these “ideal” types. 
Although these types proved of value, continued 
study mdicated a number of shortcomings, the most 
important of which seemed to be that forcing the types 
into a six-day lifetime seemed somewhat arbitrary. 
Therefore, during World War II, in response to a re- 
quest from the Army Air Force, which was interested 
in extended-range forecasting techniques and had con- 
tracted with the California Institute of Technology for 
continued research on the subject, new three-day types 
[4] were developed wherein the life of the type cor- 
responded to the passage of a single cyclone family 
across a region. This averaged three days, with con- 
835 
siderable variation. The three-day weather types in- 
cluded all the essential and well-tested features of the 
older six-day types plus a number of refinements found 
necessary as a result of experience in their application. 
They were developed for each of four synoptic regions 
consisting of sectors of forty-five degrees of longitude 
extending along the belt of westerlies from 135°H 
through 180° to 45°W and a fifth region extending from 
45°W to 45°H. Some twenty to thirty different types 
were determined for each region and a catalogue of 
weather types for forty-six years was prepared [4, 5, 15]. 
It was early recognized that the large-scale and per- 
sistent distortions of the upper-level pattern controlled 
and determined the weather types in each region. With 
an increasing file of adequate upper-level charts it has 
been possible to establish in a reliable fashion the upper- 
air patterns associated with each weather type [19]. 
Because this file does not extend far back historically, 
it has not been possible to employ upper-air type char- 
acteristics in cataloguing, except in recent years. 
It had early been noted that the weather types for 
each region could be divided into two distinctive cate- 
gories and that types in the same category often oc- 
curred simultaneously in neighboring regions. These 
two categories were the meridional and zonal flow 
types. Meridional flow types are characterized in the 
mean by upper-level crests and troughs of large am- 
plitude which tend to steer the surface migratory sys- 
tems along paths far to the north or south of normal, 
depending upon their location. Most of the large polar 
outbreaks which penetrate to southerly latitudes occur 
in this category. Elsewhere, unseasonably warm 
weather is experienced. Meridional flow patterns are 
particularly imteresting because of the conspicuous 
anomalies of temperature associated with them. An 
interesting example may be cited in this connection: 
during the entire month of January, 1949, in North 
America there existed an extremely strong and per- 
sistent trough aloft in western United States with a 
persistent crest in eastern United States. As a result, 
temperatures were far below normal in the West and 
above normal in the East. 
Zonal flow types are characterized by zonal flow 
aloft. Variations in the strength and latitude of the 
belt of westerlies lead to variations within this category. 
It is beyond the scope of this article to present the 
details of weather types as delineated by daily or 
“Gdeal” charts. However, a set of schematic diagrams 
representing the major features of the mean flow pat- 
tern occurring with a number of the more important 
wintertime weather types of the North American area 
is shown in Fig. 1. The types are arranged in two 
columns; in the left-hand column are meridional flow 
types, in the right-hand column zonal flow types. In 
addition, the position of each type in its column is 
determined by the degree to which it displays the 
characteristics of the category. In the case of the 
meridional flow types, the more the principal mean 
trough over North America is shifted westward, the 
stronger the meridional flow character of the type. In 
the extreme of this category the trough is off the west 
