OBJECTIVE WEATHER FORECASTING 
be misinterpreted if the user is unfamiliar with their 
purpose or if he tends to rush into a statistical treat- 
ment without first considering the meteorology of the 
situation [10, 18]. 
A report by Wadsworth [28] illustrates some of the 
recent work in statistical forecasting; but this subject 
is to receive a more complete treatment elsewhere.! 
Combined Deductive-Empirical Methods. The meth- 
od of numerical calculation has the disadvantage that it 
is difficult (7.c., it requires a large number of calculations) 
to permit actual atmospheric data to determine the 
coefficients of the forecasting equations used, and the 
equations which are presently available do not forecast 
the final weather elements. Other theoretical ap- 
proaches to forecasting often are of little immediate 
value to forecasters because the application of the new 
concepts and new principles which result is not a 
straightforward procedure. The forecaster finds that he 
must expend considerable effort in applying the new 
principles to a variety of cases in order to learn what 
modifications to make and the conditions under which 
they apply. This situation appears to result in a gap 
between the development of new dynamic techniques 
and their application to routine forecasting practice. 
On the other hand, statistical methods often depend 
too largely on empirical results and do not make ade- 
quate use of ideas which result from theoretical de- 
ductions. A new forecasting technique is required which 
will neither neglect the indications of theory nor re- 
quire subjective imterpretations and projections of 
weather maps and charts in order to arrive at a fore- 
cast. The methods of approach which have been under 
recent development in the United States will be de- 
seribed briefly in the following section. In this discus- 
sion, a distinction will be made between the terms 
“forecast method” and “forecast aid.” The term ‘fore- 
cast method” will imply that the objective system pro- 
duces a final forecast, and that in actual practice the 
forecaster should seldom if ever modify this forecast, 
no matter what he may feel about its chances for suc- 
cess. The term “forecast aid,” on the other hand, will 
be used in the sense of a tool or an indication which the 
forecaster must consider and weigh subjectively with 
other available indications. 
METHODS OF DERIVING OBJECTIVE 
FORECASTING AIDS 
Specification of the Problem. A distinguishing char- 
acteristic of work which has been done on objective 
methods of forecasting is that the forecasting problem 
is attacked a little at a time. A problem is defined ob- 
jectively by specifying a single weather element to be 
forecast, a single point or a specified set of points for 
which the forecast is to be made, and a single time or a 
time period to be covered by the forecast. The practical 
value of the result of the study depends in large part on 
the extent to which the specification of the problem 
1. Consult “Application of Statistical Methods to Weather 
Forecasting”? by G. P. Wadsworth, pp. 849-855 in this Com- 
pendium. 
799 
meets practical forecasting requirements. The local fore- 
cast for a given city usually requires, for example, that 
the occurrence or nonoccurrence of precipitation be 
forecast by 12-hr periods or perhaps even by shorter 
periods. Thus an objective method which forecasts 
precipitation occurrence for a 24-hr period does not en- 
tirely meet the usual requirements, although it may 
nevertheless be of value im preparing a shorter-term 
forecast, and may be useful as a preliminary study. 
Limitation of the work to a single problem in this way 
is necessary in order that the weather element being 
forecast can be measured or objectively specified in 
some way. It imposes no limitation on the kinds of 
forecast problems that can be imvestigated, for the 
preparation of any type of forecast that can be objec- 
tively verified, such as area or route forecasts, cold- 
wave warnings, or the deepening of lows, can be studied 
by objective procedures. 
Developing Hypotheses for Testing. The crux of the 
problem of increasing the accuracy of forecasts is that 
of discovering new and more exact relationships be- 
tween observed parameters of the atmosphere and sub- 
sequent weather. The efficient application of such rela- 
tionships to practical forecasting is the goal of objective 
foreeasting, but without an underlying model of the 
atmosphere and its processes, based on the best avail- 
able physical facts and theories, no amount of systema- 
tizing and testing will produce the desired results. 
Numerous sources of ideas are available. In maxi- 
mum-temperature forecasting, for example, equations 
are available for computing the effect of msolation on 
the maximum temperature [14] and, in theory, other 
effects can be evaluated. In practice, the most serious 
disturbing factors seem to be those for which quantita- 
tive data are scarce, such as albedo and outgoing long- 
wave radiation, or those which in turn depend on a 
forecast, such as cloud cover and temperature advec- 
tion. 
Recent developments in dynamic meteorology have 
provided new concepts which have not been adequately 
tested. Relationships involving wave length, zonal wind, 
and wave propagation in the upper air, and the various 
theories subsequently developed which have contrib- 
uted to the understanding of large-scale phenomena in 
the atmosphere, need to be stated in ways that are 
amenable to objective evaluation in connection with 
short-range forecasting. This is the purpose of work 
begun recently at the University of Chicago, and a re- 
port on tropical cyclone forecasting [83] illustrates the 
method of attack. Such a report is only the beginning, 
however, and a large amount of work remains before the 
results can be objectively applied in forecasting. 
A number of papers have been published within the 
last few years which describe forecasting studies wherein 
the ideas seem to have been only those suggested by a 
superficial study of the physical processes which are in- 
volved. A more logical approach has been initiated by 
the Honolulu office of the U.S. Weather Bureau, which, 
as a preliminary step in developing objective methods of 
rainfall forecasting, has prepared a lengthy report on 
the physical causes of rainfall on the Hawaiian Islands, 
