792 
This equation shows, first of all, that convergence 
or divergence in space and time of the absolute iso- 
hypses tends to produce positive or negative angles of 
geostrophic deviation, respectively. In other words, in 
regions of confluence (difluence), the air flows across 
the absolute isohypses toward lower (higher) height. 
Moreover, this equation shows that the angle of geo- 
strophie deviation will be amplified or suppressed ac- 
cording as the horizontal geostrophic shear is anticy- 
clonic or cyclonic, respectively. It can also be seen 
that, in the case of anticyclonic geostrophic shear at 
a constant-pressure surface, anticyclonic curvature of 
the path tends to magnify the horizontal shear effect and 
is therefore also conducive to relatively large angles 
of geostrophic deviation. (For anticyclonic curvature, 
the value of a is greater than one.) Relatively large 
deflection of the streamlines across the absolute iso- 
hypses (or prohypses in the case of prognostic maps) 
can therefore be expected in regions of anticyclonic 
horizontal geostrophic shear and where the curvature 
of the path is anticyclonic. Such conditions may be 
found in the right half of the confluent and difluent 
regions. On the other hand, the conditions for a rela- 
tively small angle of geostrophic deviation are (1) 
that the horizontal shear of the geostrophic wind be 
cyclonic or only slightly anticyclonic and (2) that, if 
the horizontal shear is anticyclonic, the curvature of the 
path be cyclonic or only slightly anticyclonic. (For 
cyclonic curvature, the value of a is less than one.) 
These conditions are found in the left half of the con- 
fluent and difluent regions. Analytic examples showing 
large geostrophic angular deviations of wind have been 
found by Gustafson [37, Figs. 1-3] (cf. also Petterssen 
[61, Fig. 2].) 
The prediction of wind for a mandatory surface is 
thus comparatively simple once the pressure field is 
forecast; the prediction for intermediary pressures is 
achieved by interpolation between two mandatory sur- 
faces. To forecast locally the wind speed aloft, say 12 
hr in advance, the regional forecaster can, often with 
good results, extrapolate the quasi-conservative con- 
figuration of the isotachs. The crescent-shaped cen- 
ters of maximum speed usually have their long axis 
of symmetry along the absolute isohypses. They move 
generally in the direction of the geostrophic wind. Their 
future direction of propagation is thus given by the 
prognostic absolute isohypses and their displacement 
by purely formal extrapolation. These efforts have 
revealed certain deficiencies, however, in the current 
techniques of observing the upper winds, especially at 
high levels with very large speeds. In the United States, 
these deficiencies are expected to be remedied with the 
introduction in the near future of an improved radar 
wind-finding instrument known as the AN/CPS-10. 
Finally, we mention that if no system of upper-air 
maps is available for predicting upper winds, we may 
apply indirect aerology, for instance, the method em- 
ployed by Kibel [47], already introduced by Exner [27]. 
Closely connected with the prediction of wind is 
the problem of forecasting atmospheric turbulence. 
Since the amount of horizontal (vertical) shear of the 
WEATHER FORECASTING 
actual wind is far higher (the same) for turbulent layers 
than (as) for normal ones, the forecaster should be 
on the lookout for pronounced clear-air turbulence 
in the regions adjacent to, but not within, the jet 
stream. The maximum speed of the wind gusts at the 
ground associated with thunderstorms can be forecast 
by the formula: C (in knots) = 11 4/A@ + 15, where 
Aé is the difference (centigrade degrees) between the 
lowest wet-bulb potential temperature of the sounding 
and the wet-bulb potential temperature of the rising 
air above the condensation level [11]. 
Excessively strong winds occur in a small-scale model, 
the tornado (horizontal extent below one km), whose 
prediction is connected with the peculiar difficulty that 
it completely escapes detection on the synoptic map. 
For a long time to come, it will be generally impossible 
to forecast the time and position of occurrence of in- 
dividual tornadoes. The forecaster must be content 
to predict only the general conditions favorable to 
tornado formation and then to forecast in a rather 
vague way that a number of tornadoes are likely to 
occur at scattered pomts in a region (say, 200 km 
square) and during a certain time period (of about one 
to two hours). These general conditions, shown in 
aya 
15 GMT (tg) tt 
\ 
LiKe) 100 
— — GROUND-SURFAGE ISOHUMES (°C) ——— 700-MB ISOHUMES (°C) 
600-MB ISOTACHS (KNOTS) weg GROUND SURFACE COLD FRONT 
12-HR FRONT DISPLACEMENT “a SOO-MB COLD FRONT 
AREA OF TORNADO- GENESIS AT 0300 GMT (tot 12h) 
ISOBARS (MB) ON THE SURFACE OF FREE CONVECTION FOR THE MOIST 
LAYER OF AIR JUST ABOVE THE GROUND 
Fie. 11—Characteristic features of the weather maps (at 
t) for the subsequent development of tornadoes in the shaded 
region at about t + 12). The isohumes connect points of equal 
dew point; isotachs, points of equal wind speed. 
Fig. 11 as they appear on the weather maps, are’ 
(28, 74]: ; 
1. A conditionally unstable, deep tongue of unusu- 
ally dry air (namely, a 700-mb dew point of —10C 
11. Consult ‘‘Tornadoes and Related Phenomena” by E. 
M. Brooks, pp. 673-680 in this Compendium. 
