132 
ress has not been made in recent years towards the 
improvement of basic weather forecasting lies in the 
failure to assess the essential nature of the problem 
correctly, with the consequent misdirection of much 
money and effort. Practical weather forecasting has been 
until very recently so exclusively a matter of extrapola- 
tion into the near future of current weather and weather 
trend, on the basis of synoptic experience, empirical 
rules, and statistical probabilities, that most thinking 
about weather forecasting is patterned in the same 
mold. Great amounts of money and effort have been 
expended in increasing the number and content of 
synoptic observations, in experimenting with new forms 
of synoptic representation, and in statistical or synoptic 
analysis of the data. But most of this effort has been 
expended without any planned attack on the basic 
problems of meteorology in the vain hope of finding 
short-cut empirical or synoptic forecast rules or statisti- 
cal relationships which might radically improve weather 
forecasting. Forecasting progress has not been realized 
in proportion to this misdirected effort, and possibly 
it has even been hampered to some extent by a surplus 
of information which is poorly planned and inefficiently 
utilized. 
The Scope and Essential Nature of the Forecast 
Problem. The ultimate practical goal of most meteoro- 
logical research is to improve present knowledge and 
understanding of atmospheric processes to the point 
where accurate scientific weather forecasting becomes 
possible. The all-inclusive scope of the problem is evi- 
dent from its essential character, comprising as it does 
primarily three questions, each of which must be an- 
swered in turn. They may be stated, in their simplest 
terms, as follows: 
1. What is the weather? This question requires a 
knowledge of the distribution of the meteorological 
elements which constitute the weather and whose 
changes reflect the weather processes. Such knowledge is 
required both as the starting point or base from which 
to forecast the future weather, and as a means of verify- 
ing or evaluating past weather forecasts and forecasting 
techniques. All research bearing on the measurement 
and recording of the weather elements throughout the 
atmosphere is directed toward answering this question. 
It is in this technical field that meteorological research 
has been advanced out of all proportion to the utiliza- 
tion of the data which are obtained. 
2. Why is the weather? This question requires the 
quantitative physical explanation, in scientific terms, 
of all weather phenomena and their causative or forma- 
tive processes. This is the basic part of the forecast 
problem which weather-forecasting research has tended 
to neglect in the past, in comparison to the money and 
effort which have been expended upon the unsystematic 
accumulation and the routine synoptic or statistical 
analysis of observational data. It is, however, primarily 
on progress in seeking an answer to this question that 
the future improvement of weather forecasting depends. 
3. What will the weather be? The right answer to this 
question constitutes correct weather forecasting. A satis- 
factory answer requires a comprehensive understanding 
WEATHER FORECASTING 
both of what the weather is and why the weather is. 
It has been the attempt to bypass the second question, 
in the hope of finding short-cut empirical synoptic or 
statistical means of determining the future weather 
directly from the present weather, without understand- 
ing either, that has militated against progress in solving 
the basic forecast problem. 
THE PRESENT PRACTICE AND PERFORM- 
ANCE OF FORECASTING TECHNIQUES 
Types of Weather Forecasts—Time Range, Content, 
and Performance. Since most weather-forecasting prac- 
tice consists essentially of the rather crude extrapolation 
of current weather patterns and tendencies into the 
future, it is axiomatic that the accuracy and justifiable 
detail of the forecasts decrease rapidly with increasing 
range. Consequently weather forecasts are prepared in 
a great variety of form and detail, depending upon the 
degree of detail and accuracy which is required by the 
specific purpose of the forecast. The forecasting tech- 
niques which are used vary greatly with the type (time 
range and detail) of the forecast. Therefore to discuss 
the practice and performance of forecasting techniques, 
it is really necessary to consider briefly the require- 
ments and expected accuracy of the different types of 
forecasts. It must be emphasized, however, that the 
complete lack of any uniform objective verification of 
weather forecasts precludes the possibility of any reli- 
able evaluation or comparison of the performance of 
the different forecast types and the respective fore- 
casting techniques. All estimates of forecasting skill 
in the following discussion are expressed on a basis of 
50 per cent verification by chance as representing pure 
guess work or zero skill. For the most part the figures 
represent rough comparative estimates of forecasting 
slall and are not based on actual extended series of 
numerical verification of forecasts. 
Weather forecasts are most conveniently classified 
according to time range and basic character of the 
forecast in the following four categories: 
1. Short-range forecasts, for periods up to eighteen 
hours from the issue of the forecast. Forecasts of this 
type have been developed almost entirely during the 
past twenty-five years, primarily in response to the 
demands of aviation, both civil and military, although 
short-range forecasts of local conditions of frost, snow 
accumulation, and icing of roads have also received 
increasing attention. Short-range airways forecasts re- 
quire a high degree of accuracy, both im timing and in 
local detail of all elements affecting flight operations, 
particularly of terminal conditions. These elements in- 
clude wind direction and speed up to an elevation of 
from ten to fifteen thousand feet, the state of turbulence 
of the atmosphere, ceiling heights, horizontal visibility, 
and the occurrence of condensation forms (notably when 
the danger of icing, fog, or thunderstorms is present). 
Because this is a relatively new type of forecasting, 
frequently of elements which previously were not fore- 
cast at all, because the usual extrapolation techniques 
are particularly suited to short-range forecasting, and 
because great effort has been expended in meeting the 
