64 DEACON AND WEBB [CHAP. 3 



drag coefficient increasing linearly with wind speed and starting from the 

 smooth-surface value at very low wind speeds. As far as may be judged from 

 Fig. 6. this prediction accords with observation and rough quantitative agree- 

 ment would be secured with a sheltering coefficient, s. of rather less than 0.01. 

 A value ,5 = 0.01 was estimated by Sverdrup and Munk (1947) from observed 

 rates of wave growth. 



Neumann (1956) suggests that the sheltering effect of the large waves may be 

 rather different from that visualized by Cox and Munk and puts forward some 

 evidence for a drag coefficient decreasing with wind speed (see also Neumann. 

 1951). This mainly arises from giving undue weight to sea-slope data for light 

 winds (Deacon, 1957a). He argues further that wind-profile results are un- 

 reliable because cup anemometers at low heights over the sea would be sub- 

 jected to considerable fluctuations in wind speed with the passing of the waves 

 and, because of their inertia, record too high a mean wind speed : this would 

 give spuriously low wind gradients and correspondingly low drag coefficients, 

 particularly in light winds for which the over-estimation effect is most marked. 

 Much of the data given in Fig. 6 is, however, almost certainly free from ap- 

 preciable error of this sort. In the studies by Johnson (1927) and Deacon, 

 Sheppard and Webb (1956) and also by Brocks (I960), the heights involved 

 were too great and/or the inertia of the anemometers too small for appreciable 

 over-estimation error. Takahashi (1958) used thermocouple anemometers not 

 subject to the inertia effect, and the eddy-correlation values of Mcllroy (1955) 

 and Vinogradova (1959) give support to the wind-profile results by a com- 

 pletely diiferent technique. 



E. Dejjejidence of Drag on Atmospheric Stability 



The state of the sea surface and the drag depend on the wind speed close 

 above the surface. If it were practicable always to measure the wind speed at 

 some small height, such as 1 m, then the drag coefficients would be found very 

 nearly independent of the air-sea temperature difference. With winds measured 

 at greater heights, such as 10 m, the effect of atmospheric stability is no longer 

 negligible. For a given wind speed at 1 m. the 10 m speed wall be greater under 

 stable than under unstable conditions with the consequence that the drag co- 

 efficient, cio- should be smaller for stable conditions and larger for unstable. 



As yet there is only qualitative evidence on the magnitude of the stability 

 effect over the sea. Darbyshire and Darbyshire (1955) in their analysis of Lough 

 Neagh tilt observations found evidence for a quite marked effect of stability but, 

 as their wind speeds were measured over the land and not over the lake, the 

 interpretation of their result is obscured. 



That the stability effect is of significant magnitude is shown by the results of 

 studies of the state of the sea in relation to wind speed and air-sea temperature 

 difference. The basic data for these studies were wave-height observations at 

 measured wind speeds from weather ships. Roll (1952) found that, for a given 

 wind speed, the mean wave height when the sea temperature is 6.7^C above air 



