EXTENDED-RANGE WEATHER FORECASTING 
the amount of heat received per unit time by a unit 
mass of a moving particle, c is a constant depending 
on external friction, and the other symbols have their 
customary meanings. 
The replacement of the terms on the right-hand side 
of this equation by quantities which can be statistically 
interpreted—for these quantities the reader is referred 
to the original report [10]|—leads to the following vari- 
ables which are used as headings in a table of multiple 
correlations: 
X, = pressure in mb reduced to sea level, on the 
clue day (day of forecast) at 1400 (4 classes), 
X» = 24-hr temperature change in centigrade degrees 
from the previous day at 1400 to the clue day 
at 1400 (4 classes), 
X3 = 24-hr pressure change in mb from the previous 
day at 1400 to the clue day at 1400 (4 classes), 
X, = cloudiness in tenths at 1400 on the clue day 
(4 classes), 
X; = 7-hr pressure change in mb from 0700 to 1400 
on the clue day (5 classes). 
In Table VI, a section from a multiple-correlation 
table is given whose headings include these five vari- 
ables. The numbers in the blocks are the 48-hr pressure 
changes from 1400 of the day of the forecast to 1400 
two days later, as well as the date of the clue day (in 
parentheses). All observations refer to Potsdam (near 
Berlin). The table is based on observations from June 
21 to August 9 of the years 1893-1937. Of the 1280 
blocks of the table, 698 are covered with a total of 
2250 entries. The boldface numbers of the table repre- 
sent the mathematical expectation of the 48-hr pressure 
changes, in mb. Of course, in practice, only those 
squares are to be used that are most likely to lead to 
an unequivocal result, as is the case with the blocks of 
Table VI, marked by a black star. If one block contains 
a pressure change whose sign is different from all the 
other values of that block, this possibility could be 
excluded if the weather situation on the day in ques- 
tion (for instance July:12, 1928, in Table VI) were 
entirely different from that on the day of the forecast. 
For such comparisons, the method of similar (or, in 
this example, opposite) cases can be applied even at the 
present time. 
The same table can be used for several climatically 
coherent places. However, different tables must be 
computed for different climatic regions and different 
seasons. The tables can be supplemented by the merid- 
ional and zonal pressure gradients or other physically 
important quantities. This, however, is possible only 
if observations over a longer period at several clima- 
tically similar stations can be combined in one table, 
such that the number of blocks, which mcreases with 
every new variable, can be filled with sufficient ob- 
servations. 
On days when sufficiently reliable results cannot be 
obtained from a multiple-correlation table, it is better 
not to draw prognostic pressure charts for the second 
and third consecutive days. 
Mean-Circulation Methods of Forecasting. A judicious 
combination of synoptics and statistics is represented 
829 
in the American investigations which can be classed 
under the name of ‘“‘mean-circulation methods of ex- 
tended forecasting” [82, 40, 41, 45, 53, 57]. With re- 
Tasie VI. Exrract rrom A MuntipLe-CorrELATION TABLE 
FOR THE PREDICTION OF 48-HR PRESSURE 
CHANGES AT PoTspAM 
(X; = 1019.5 to 1036.5 mb) 
(X_ = +4.0 to +11.8 cent. deg.) 
(X; = 0.0 to 3.2 mb) 
X5 Xa Xo 
0 to 4 = 
5 to 7 — 
16.8\ to 441.3 
8 to 9 — 
10 wale 
0 to 4 —2.8 (89-01) 
44.8 (7-4-23) 
—0.5 (8-2-26) 
40.5 
mpwoe (27 1.5 (8-9-14) 
8 to 9 40.4 (7-9-14) 
4.0 (7-11-28) 
aie 
10 es 
0 to 4 | —2.8 (8-93) 40.4 (7-12-28) 
—2.7 (7-20-99) —4.8 (7-11-29) 
=0.8 (77-04) —14.8 (7-23-31) 
3.3 (6-27-08) —5.3 (7-9-32) 
=7.3 (6-30-14) —13.5 (7-17-34) 
4.1 (7-15-21) 
= Full MK 
—0.1 to —1.2| 5 to 7 —4.9 (88-05) 
ES (ES 
9.6 
8 to 9 41.7 (7-1-06) 
—6.0 (83-07) 
2.1 
10 = 
0 to 4 | —11.7 (6-23-97) —10.7 (6-26-24) 
1.2 (7-23-08) —10.7 (7-26-27) 
—219 (7-38-17) —12.5 (7-432) 
~8.3 * 
—1.3 to —2.5 1 5 to 7 —5.6 (6-23-20) 
8 to 9 12.9 (6-22-11) 
10 = 
0 to 4 7.2. (7-21-23) 
aye) ff — 
=I ti =O 
8 to 9 = 
10 ae 
spect to the nature and results of these methods, the 
reader is referred to the article by Namias in this Com- 
pendium.’ Only a few critical remarks will be made in 
3. Consult ‘‘General Aspects of Hxtended-Range Forecast- 
ing” by J. Namias, pp. 802-813. 
