SEA-SURFACE TEMPERATURE 33 
surface temperature on the clear days occurs at 15h, 
and on the cloudy days at 17h and 18h. 
Effect of Wind on the Diurnal Variation 
of Sea-Surface Temperature 
In a similar manner, the mean unperiodic amplitude 
has been computed for days with various wind veloci- 
ties. In the abstracts of the ship’s log, the wind force 
was usually given more explicitly than the cloudiness. 
Therefore, it has not been difficult to select fifty-four 
days in tropical regions with an average wind force 
equal to or greater than 4 on the Beaufort scale, and 
forty-six days within the same latitudes with wind force 
less than 4. These selections were made without regard 
to other meteorological conditions. The results give an 
amplitude of 0°65 for days with a wind force equal to or 
greater than 4, and one of 1°03 for days with wind force 
less than this value. 
The mean diurnal courses of sea-surface tempera- 
ture for these same groups were computed, corrected 
for noncyclic change, and the resulting curves are 
shown in figure 25. The periodic amplitude on windy 
days amounts to 0°11, and on relatively calm days to 
0°84. The maximum of the mean sea-surface tempera- 
ture for windy days falls at 15h and the minimum at O7h. 
Qn calm days, the maximum occurs at 13h, and the min- 
imum at 23h. 
From these data we can conclude that wind, rough 
sea, and cloudiness are conducive to small diurnal 
ranges in sea-surface temperature. The reasons ap- 
_ pear obvious. 
Harmonic Analysis 
of Sea-Temperature Data 
A more detailed study of the diurnal variation of 
sea-surface temperature is possible from an examina- 
tion of the results of Fourier analyses of the mean diur- 
nal curves for each of the groups of Carnegie data. 
From the mean hourly departures, the Fourier coeffi- 
cients for the 24-hour, 12-hour, 8-hour, and 6-hour 
terms have been determined and the results given in 
table 39. 
The amplitudes and phase angles, used as polar co- 
ordinates, were plotted on harmonic dials to facilitate 
Study. It was immediately obvious from a preliminary 
examination of these figures that the coefficients for 
Groups II, XIlIa, XIIIb, and XV were extremely irregu- 
lar, falling completely out of phase with the greater 
number of diurnal curves, and exhibiting amplitudes 
much larger than average. The reasons for these ir- 
regularities are not difficult to explain; namely, Group 
Il includes four days in the region of the Gulf Stream 
where the diurnal variability of sea-surface tempera- 
ture is no doubt completely obscured by noncyclic 
changes; Group XIlla and Group XIlb include sixteen 
days of observation in the Kuroshio Current, where 
the mean is affected bv rapid mixing of water masses 
of very different temperatures; and Group XV embraces 
five days of changeable temperatures due to the cross- 
ing of the California Current where, again, the diurnal 
eon are masked by the large unperiodic varia- 
tions. 
For these reasons, the above-mentioned Groups will 
Table 38. Mean unperiodic daily amplitude of sea- 
surface temperature, tropical latitudes, clear days 
and cloudy days, wind force less than 4 Beaufort 
Scale, Carnegie and Gauss and after Schott 
Cloudy days Clear days 
Amplitude Amplitude 



Source 
Cc Cc 
Carnegie 0.66 10 1.24 10 
Gauss 0.88 28 1.02 19 
Schott 0.93 ? 1.59 a 
Mean 0.82 1.28 
not be considered in this discussion. 
According to values derived for $1, the maximum 
sea-surface temperature, cj, occurs between noon and 
17h except for Groups XVI and XVIIc, which show max- 
ima in the morning, Presumably these two Groups 
were also affected to some extent by large regional var- 
iations in temperature. By averaging the Fourier coef- 
ficients, a1 and bj, for the remaining sixteen Groups, 
it is found that the mean amplitude and phase angle are 
0°12 and 228° respectively. In other words a mean max- 
imum amplitude of 0°12 occurs on the average at 14h 
48m. 
The Carnegie amplitudes and phase angles of the 24- 
hour and 12-hour terms have been compared with values 
for these terms derived from Gauss and Challenger [4, 
p. 509, table 95a] observations in corresponding lati- 
tudes and the results are presented in table 40. The 
Carnegie amplitudes average somewhat lower than the 
Gauss values, in all probability because of differences 
in observational methods. The mean amplitude of the 
Carnegie, Gauss, and Challenger 24-hour term is prac- 
tically the same but the first crest of the double diurnal 
oscillation occurs, according to the Carnegie data, at 
02h 06m, or approximately three-quarters of an hour 
later than is indicated by the Gauss and Challenger data 
(01h 16m and 01h 24m, respectively). 
Sea and Air Temperatures 

General Remarks 
It is obvious that the direct thermal influence of the 
air on surface sea waters is smali as a result of the low 
specific heat of air. Probably for this reason, among 
others, the relations between sea and air temperatures 
have not been given adequate attention by oceanographers. 
On the other hand, the direct effect of sea-surface tem- 
peratures on the temperature of the air immediately 
above the surface is powerful and important. For this 
reason, a knowledge of the differences between sea and 
air temperatures is of extreme importance to the mete- 
orologist, and a study of these differences is essential 
in any consideration of the thermodynamical properties 
of maritime air masses. In view of the importance of 
such a study, the relation between sea and air tempera- 
tures will be considered in detail. 
