OBJECTIVE WEATHER FORECASTING 
By R. A. ALLEN 
U. S. Weather Bureau, Washington, D. C. 
and E. M. VERNON 
U. S. Weather Bureau, San Bruno, Calif. 
Tt is the purpose of this paper to outline the breadth 
of the field of study which may be called “objective 
weather forecasting,” to describe in a general way some 
of the recent developments in this field, and to indicate 
the deficiencies and unanswered questions which have 
arisen in such work. 
THE FORECASTING PROBLEM 
Definition of Objective Weather Forecasting. In the 
history of weather forecasting, attempts have often 
been made to devise numerical and objective methods 
for producing the forecast. Thus Besson [2] in 1904 and 
Taylor [24] and Rolf [20] in 1917 produced graphical 
devices for representing lag relationships between se- 
lected weather variables. These studies, in common with 
others made in later years [4, 12, 14, 15, 27], have at- 
tempted to provide an equation or a graphical device 
of some form which would be useful in applying a par- 
ticular relationship or combination of relationships to 
the problem of making a forecast. The distinction be- 
tween an objective forecasting procedure and a pro- 
cedure which depends on subjective judgments and 
subjective experience has not been sharply defined, nor 
‘is 1t intended in this paper to advocate a rigid definition. 
The purpose of this review will be served by defining an 
objective forecasting system as any method of deriving 
a forecast which does not depend for its accuracy upon 
the forecasting experience or the subjective judgment 
of the meteorologist using it. Strictly speaking, an ob- 
jective system is one which can produce one and only 
one forecast from a specific set of data. From the prac- 
tical standpoint it appears reasonable to include as 
objective, however, those forecasts which require me- 
teorological training insofar as such training is stand- 
ardized and is itself based upon a study of well-founded 
physical principles and atmospheric models which are 
commonly recognized from the facts of observation. It 
would be throwing away information of demonstrated 
value in forecasting if, for example, an objective fore- 
casting system were not permitted to make use of iso- 
baric patterns on analyzed maps because of the objec- 
tion that they are arrived at subjectively. The test of 
whether a system is objective is whether different 
meteorologists using the system independently arrive 
at the same forecast from a given set of maps and data. 
Goals of Objective Forecasting Investigations. The 
obvious ultimate goal of forecasting investigations is to 
enable the forecaster to increase the accuracy of fore- 
casts made routinely. Contributions toward this end 
may be made in several ways. The forecaster may 
study the physical characteristics of the atmosphere, 
especially the dynamic relationships which have been 
derived on the basis of simplifying assumptions. Such 
study may enable him, in the course of analyzing given 
situations, to recognize processes in the real atmos- 
phere which have been described analytically, and in 
such cases he will know better what to expect of the 
atmosphere in the immediate future. The success .of 
this method of attack depends on the skill of the theo- 
retical meteorologist in describing the real atmosphere 
when he sets up a model and makes simplifying assump- 
tions, and on the skill of the forecaster in diagnosing 
the present sequence of events in the atmosphere, se- 
leeting the theoretical processes which are most nearly 
applicable, and judging what modifications are neces- 
sary in individual instances. 
On the other hand, the forecaster may search for em- 
pirical relationships between observable characteristics 
of the atmosphere, and with little or no reference to the 
physical validity of the relationships, make use of them 
in forecasting. Many forecasters gain a high degree of 
skill after many years of experience because of this 
second factor, but skill obtained in this way is difficult 
to transfer from place to place or from individual to 
individual. It appears certain, furthermore, that some 
forecasters base forecasts in large part on hypothetical 
relationships that have neither a physical nor a statis- 
tical basis and that cannot even be expressed in 
objective or quantitative terms. In such a case, it is 
impossible to discover from data whether or not these 
relationships exist in the atmosphere. 
Ideas for testing and possible incorporation into an 
objective system can come from several sources: by 
testing new theoretical concepts for their possible con- 
tribution to forecasting practice and providing objec- 
tive ways to use the results; and by testing, combining, 
and systematizing the use of rules and principles which 
have been discovered or are already used by experienced 
forecasters. 
The goal of objective forecasting is simply to elimin- 
ate as many as possible of the subjective elements which 
enter into the application to forecasting of the results of 
such studies. Objective forecasting is not so much con- 
cerned with the source of hypothetical relationships as 
it is with the practical value of the ideas and the extent 
to which they contribute to the accuracy of forecasts. 
Objective forecasting studies and research projects 
which aim to develop objective methods or objective aids 
to forecasting are characterized by the use of historical 
data to demonstrate the reliability of forecasting rela- 
796 
