762 
winds aloft (40 knots or more), and the trigger action 
associated with an approaching cold front. Tornadoes 
appear to move with the speed and in the same direc- 
tion as the thunderstorms with which they are normally 
associated, that is, with the winds between 6000 and 
12,000 ft. These winds will normally be from the south- 
west. 
Verification. The primary purpose of any forecast 
verification system is to determine whether satisfactory 
standards of forecasting are being maintained, and, 
secondarily, to ascertain the relative skill of forecasters 
and forecasting candidates. The system should be ra- 
tional, fair to the forecaster, and should not require 
more than a few minutes of his time in paper work. If 
too many elements are verified, too much smoothing 
will result and final values will have little significance. 
Except for very short-term forecasts, spot-check veri- 
fication (e.g., verification of specific elements at exactly: 
0200, 0400, 0600, etc.) should be avoided, since, after 
eight hours, values based on verification of this type 
will show no significant differences between good and 
mediocre forecasters, because of the variability of 
weather. The longer short-term forecasts (12-48 
hr) should not be verified by methods mvolving a 
refinement of detail beyond the current ability of the 
forecasting profession. Verification of trivia should be 
avoided or at least handled with intelligence. Onesystem 
of verification currently in use awards the forecaster 
100 per cent for a measurable amount of precipitation 
(0.01 in.) and 0 per cent for a trace when precipitation 
has been forecast. In many countries a measurable 
amount is 0.1 mm, a unit only four-tenths of the one 
used in the United States. Over large areas of the 
United States, precipitation amounts of either a trace 
or 0.01 in. comprise 50 per cent or more of the total 
periods of precipitation, and, in a large proportion of 
these cases, it is fortuitous whether the observer records 
a trace or 0.01 in. Yet in this system of verification, 
the range in skill score is almost entirely dependent 
upon the forecaster’s luck with these very trivial pre- 
cipitation amounts. 
Indeed, no substitute has been developed in the 
United States which is an improvement over the old 
U.S. Weather Bureau verification system. Ifa sufficient 
number of forecasts are included in the verification, 
chance coincidence and normal expectancies need not 
be considered since contributions from these sources 
will become equal for everyone and the differences in 
the final score will represent a satisfactory approxi- 
mation of the differences in skill. 
STATISTICAL AIDS IN FORECASTING 
Objective Forecast Aids. Gringorten [27] defines an 
objective forecast as one that is made without recourse 
to the personal judgment of the forecaster. He further 
states meteorology is much too complex to allow one to 
believe that objectivity in forecasting will, eventually, 
completely replace subjectivity. To date, wholly objec- 
tive forecast techniques are very rare and have shown 
improvement over accepted subjective methods only 
when the latter are exceptionally inadequate and veri- 
WEATHER FORECASTING 
fication is unusually poor, such as for certain types of 
precipitation forecasting, or when the forecaster is rela- 
tively inexperienced. Objective methods for forecasting 
minimum temperatures, in which subjective selection 
of certain parameters if often required, have at times 
equaled subjective methods but have rarely surpassed 
them. Vernon’s study [56] of winter precipitation at 
San Francisco is an excellent example of a useful and 
reasonably objective forecasting technique. The study 
of wintertime precipitation for Washington, D. C., by 
Brier [9], while a commendable attempt to attack a 
difficult forecasting problem, cannot be regarded as 
particularly successful in providing the forecaster with 
an improved forecasting tool. Forecast methods of this 
type tend to meet with passive resistance from fore- 
casters, who feel that the derived formulas and diagrams 
oversimplify the problem and that other contributory 
variables are frequently neglected. 
Climatological-Statistical Forecast Guides. From 
time to time, and particularly in recent years, fore- 
casters have been promised climatic guides which statis- 
tical climatologists claim will assist and improve fore- 
casting. So far, few have been forthcoming, possibly 
because it is now realized that their value may have 
been overemphasized. Local forecasters soon know the 
earliest and latest dates of killing frost, after answering 
a dozen inquiries on that question, and district fore- 
casters learn climatic normals and extremes of their 
forecast district as part of their training. It does not 
appear that any considerable amount of time and 
expense for this purpose is justified on the basis of 
results to date. 
‘Climatological summaries were of considerable value 
during the recent war in forecasting for foreign areas 
where few or no weather observations were obtainable 
and, of course, are of great value for other purposes. 
Tests of Empirical Forecast Rules. A large number 
of empirical rules have been developed and formulated 
since the beginning of forecasting. Some, such as those 
of Chromoy [15], have been given wide distribution, 
but most forecast centers have their own collection in 
one form or another. Too many such rules are only 
confusing to the forecaster who, in the limited time 
available, may be unable to recall those applicable to 
the synoptic situation at hand. Rules may be classified 
according to seasonal, local, or broad-scale applications, 
types, etc. The value of many empirical rules might 
be enhanced by subjecting them to objective tests with 
further stratification. 
ORGANIZATION OF FORECASTING 
It seems doubtful that the accuracy of weather 
forecasting can be improved materially within the near 
future with present or anticipated new techniques. 
Some radically different approach to short-range fore- 
casting will be required if any marked improvement is 
to be effected. Some significant, although slight, mcrease 
in forecast accuracy may be possible with feasible 
changes in forecast organization. Because of economic 
necessity, forecast organization in many weather serv- 
ices leaves much to be desired. 
