A PROCEDURE OF SHORT-RANGE WEATHER FORECASTING 
nomena, while (4) enables us to predict the type of 
precipitation. The future location of precipitation due 
to upper-tropospheric instability will be found in the 
forward half of the extrapolated islands of cold air 
aloft (see Figs. 7 and 8). Upper-tropospheric instability 
usually occurs above a cold air mass within a nonfrontal 
trough in the rear of a cyclone on the surface map. At 
sea, where the cold air mass already contains con- 
vective clouds, the presence of upper tropospheric in- 
stability will give the clouds there maximum vertical 
growth. The same development may be found over land 
if the imitial humidity supply and the heating from 
below are sufficient. When the low-level instability is 
missing, the upper-tropospheric instability can’ be seen 
to produce castellatus from layers or stripes of medium 
cloud. Showers and thunderstorms from such cloud 
systems are rather independent of the time of day or 
night and can be followed from map to map. On the 
other hand, in the rear half of the upper island of cold 
air, both the warm advection and the descent of the air 
contribute to the suppression of the unstable character 
of whatever cloudiness and precipitation there may be 
in that area. The forecaster should also take into ac- 
count the orographic strengthening of ascending cloud 
masses due to the nature of the terrain and other more 
indirect orographic effects, such as foehn effects. 
There is a tendency towards formation of cloud and 
precipitation in the entrance regions, whereas there is 
a tendency towards dissolution of hydrometeors in the 
exit regions. There is a tendency towards formation of 
hydrometeors in the areas where the isohypses are 
cyclonically curved and a tendency towards dissolution 
of them in regions of anticyclonic curvature. In pole- 
ward-moving air extending through a deep layer over 
the earth, cloudiness and precipitation may be expected 
in areas where the prognostic isohypses of the lower 
mandatory surfaces are either straight or cyclonically 
curved. This effect will be especially marked within 
tropical air, owing to its great relative and specific 
humidity already before its lifting and stretching. In 
an equatorward flow of similar vertical extent, cloudi- 
ness and precipitation are likely to occur only where 
the prognostic isohypses have pronounced cyclonic 
curvature. 
The centers of nonadvective decrease of relative 
height correspond to regions where the air of the cor- 
responding layer either is ascending or is losing entropy 
by radiation or precipitation. In both instances cloudi- 
ness and precipitation must be expected to be asso- 
ciated with these centers (see Fig. 8a). A detailed 
prediction of the air-mass hydrometeors, of great prac- 
tical importance, must, however, be left to the area 
and local forecasters. 
Warm Air-Mass Hydrometeors. Detailed predictions 
of warm air-mass hydrometeors—fog, stratus, and driz- 
zle—cannot be based completely on the prognostic 
map system, since local effects are also important. 
Warm air-mass fogs, excluding the steam and frontal 
fogs, are abetted by the cooling of the air particles near 
the ground sufficiently below the dew point (Petterssen 
(59]) and are hindered by turbulence-producing wind. 
789 
Upslope fogs may be predicted locally by applying 
a prognosticated lifting condensation level to the large- 
scale orographie ascent of the air which occurs with 
certain wind directions. For predicting local sea and 
coastal fogs, a problem more difficult than the regional 
problem of predicting tropical air fogs, the local fore- 
caster must be familiar with the local changes in the 
sea temperature and wind as well as the local aerologi- 
cal observations, which are necessary to clarify the 
difference between the situation in which fog occurs and 
that in which the fog is lifted by turbulence so that 
stratus cloud appears. Valid for clear skies and slight 
winds (below 5.5 mph), Taylor’s empirical diagram 
[77] for forecasting radiation fog at Kew Observatory 
has as its abscissa the evening temperature and as the 
ordinate the dew-point deficit. For the local prediction 
of advection-radiation fogs, George [382] constructed 
similar but more complete empirical diagrams which 
also take into consideration clouds, wind speed, wind 
direction, and air trajectories. Predicting the morning- 
time dissipation of fog and stratus by synoptic-local 
methods should take into consideration its linear rela- 
tion to cloud thickness [83], in addition to other fog- 
dissolving influences such as wind speed and snow cover. 
Cold Air-Mass Hydrometeors. Cumuliform clouds, 
cold-front and orographic showers, and showers super- 
imposed upon a warm-front rain can all be predicted 
In a general way by the regional forecaster on the 
basis of his prognostic upper-air maps showing the con- 
ditional and potential lability of the air, its humidity, 
and the position of its ice nuclei. In fact, such shower 
areas also form a typical part of the cyclone models 
for the surface map. In nighttime and during winter, 
continental air acquires cold air-mass properties over 
sea, while in the afternoon and during summer, at the 
time of maximum ground heating, all air masses and 
especially maritime air acquire cold air-mass properties 
over land. For the diagnosis and prognosis of the early 
cumulus stage, we may mention that Beers [8] has ap- 
plied a practical method of evaluating for each indi- 
vidual layer its circulational acceleration, a value which 
may also be interpreted as the geopotential difference 
of the ascending column from the descending one. 
For very short-range forecasting (6-12 hr) of thund- 
derstorms, Brooks [12] has found a correlation between 
thunderstorm occurrence and the direction of the lower 
tropospheric winds (ground to 1/4 km), but no evident 
relationship between the speed of these winds and 
thunderstorm occurrence. However, strong winds above 
3 km tend to inhibit thunderstorm activity. Brancato 
[11] states that large thunderstorms move with the 
11,000-ft winds, while (small) nonfrontal storms seem 
to be steered by the 5000-ft winds. Humphreys [43] 
formulates the relation differently; he states that the 
thunderstorms move with the strata containing most 
of the cumulonimbi. But these conclusions are rendered 
indecisive by the report of Byers [15] that the thunder- 
storms move to the right of the winds aloft. 
The formation of hail is favored by high humidity 
and strong upward motions which slow the fall of the 
hailstone and increase the equivalent cloud thickness 
