908 
mén [23], for example, notes that the temperature 
difference between air ascending from the surface and 
the surroundings aloft is much less in winter than in 
summer at Swan Island (17°N, 84°W). There are no 
hurricanes over the Caribbean Sea in winter. Storm 
occurrence is also rare over the cool waters of the eastern 
parts of the tropical oceans. It is most plentiful in the 
west where the sea surface is warmest. By analogy, it is 
possible to argue that storm frequency should be less 
than average in a year when the sea-surface tempera- 
ture 1s below normal. 
It is evident that research following up this suggestion 
can prove most fruitful. Caution, however, is necessary 
in interpretation. Temperatures aloft are as much of a 
variable as surface temperatures. Anomalies of tem- 
perature at different altitudes may not be independent 
of each other but produced by some common dynamical 
factor. The same may be true even for sea-surface 
temperatures. 
Frontal Theory. Ever since the formulation of the 
polar-front theory there have been persistent attempts, 
beginning with that of Brooks and Braby [8], to trans- 
plant the Bergen reasoning to the equatorial zone. 
Advocates of this theory visualize the equatorial con- 
vergence zone (HCZ) as a front with inverted tem- 
perature gradient. The peaks of warm sectors and 
developing cyclone centers should therefore be found 
where the ECZ bulges equatorward. Actually, there is 
practically universal agreement today that forming 
centers are initially situated near poleward bulges. 
This observation has led. some analysts to become 
skeptical regarding the wholesale importation of the 
Norwegian ideas into the tropics. A large group of 
meteorologists, however, today adhere to the view 
that hurricanes are generated at boundaries between 
warm and relatively cold air. Even textbooks of high 
renown (Brunt [4] p. 304) carry statements to that 
effect. Brunt, however, is quite conscious that the 
frontal approach solves the basic cyclone problem no 
more in low than in high latitudes. 
It is not possible here to examine in detail the reason- 
ing of the adherents of the frontal theory. Only one 
additional point will be presented that throws doubt on 
the validity of the frontal concept. This is again the 
matter of the properties of the surface air. If a rapid 
occlusion of a portion of ECZ takes place, when there 
is a real density difference across this boundary, the 
whole surface layer of the nascent cyclone will soon 
consist of the denser air only. This must prevent deep- 
ening. It is a principal difference between high- and low- 
latitude cyclones that all air entering the orbit of the 
latter ascends in the core, while outside the tropics the 
polar air sinks in disturbances and only the tropical 
air rises. Occlusion must be an acute hindrance to hurri- 
cane development. Actually, we can observe that even 
mature hurricanes weaken and sometimes disappear 
when a current with relatively polar characteristics 
enters the core while it is still situated in the tropics. 
A forecast of filling will generally succeed in such cases. 
For comparison, the data of Figs. la and 1b are 
presented in tabular form in Table I. Temperatures at 
TROPICAL METEOROLOGY 
certain isobaric surfaces are given for (1) the average 
tropical atmosphere in the Caribbean [37]; (2) the 
ascent of the surface air of average properties (26C, 
18 g ke), (8) the ascent of the same air after isothermal 
motion to 960 mb and addition of 1.5 g ke of mois- 
ture, and (4) the ascent of air with very slight polar 
characteristics (24C, 15.5 g ke). 
TasiLEe I —TEmprraturEs ALOFT roR THE ConprTrions (1)-(4) 
DESCRIBED IN THE TEXT 
Pressure mb (1) (2) (3) (4) 
surface 26.0C 26.0C 26.0C 24.0C 
900 20.0 20.5 22.5 18.0 
700, 8.0 11.0 13h 8.0 
500 — 6.0 = 33.00) 0.5 = 6.5 
300 —33.0 —28.5 —23.0 —34.0 
200 —55.0 —53.0 —47.0 —59.0 
The temperature differences between the four sound- 
ings are quite large at 700 mb and very large at 300 and 
200 mb. If virtual instead of dry temperature had been 
used, they would be still somewhat larger (up to 1C 
near the ground). It is seen, as stated earlier, that the 
temperature difference between the rain area and the 
surroundings doubles if we consider sounding (3) in- 
stead of sounding (2). The slightly polar air is nowhere 
warmer than the average tropical atmosphere, and it is 
cooler at most levels. This shows clearly in what meas- 
ure a slight temperature and humidity drop acts as a 
deterrent for an incipient hurricane circulation. 
These observations throw doubt on the frontal hy- 
pothesis. Some writers, however, have maintained that 
the frontal temperature contrast is very small and 
merely serves as ‘‘trigger action” to start heavy con- 
vection. But we have seen that heavy convection is not 
a sufficient condition for typhoon development. More- 
over, if the air-mass contrast at the ECZ is as minute 
as suggested by these writers, it would seem that they 
are trying to employ a very dubious mechanism. 
Synoptic Conditions During Storm Formation. Al- 
though the physical reasoning of the frontal advocates 
is open to doubt, their geometrical picture is to some 
extent correct. Storms never develop spontaneously in 
the undisturbed tropical currents but always in a pre- 
existing disturbance. Such a disturbance may be of the 
shear-line type described (ECZ) or it may have the 
character of a transverse wave (Riehl! [25]). Intensifica- 
tion of the bad weather zone attending such a disturb- 
ance and formation of a closed depression generally 
takes place when two or more disturbances meet [8, 
26]. Considerable synoptic evidence supports this state- 
ment. The combination or superposition of disturbances 
usually results from motion in different directions or at 
different rates. Among many possibilities, a common 
type is westward travel of a wave trough in the easter- 
lies that intersects an HCZ extending east-west. At 
other times such wave troughs become coupled with the 
southern extensions of eastward-moving troughs in the 
upper westerlies of the polar zone. The superpositions, 
therefore, can be horizontal and/or vertical. 
To date, only the synoptic circumstances of super- 
