SECT. 2] 



SMALL-SCALE INTERACTIONS 



61 



course has, therefore, been taken variously to wind estimates on the Beaufort 

 scale, measured winds at neighbouring land stations and to estimates from 

 atmospheric pressure gradients. Information as to air-sea temperature dif- 

 ferences is also generally lacking. In view of the uncertainties so introduced, it 

 seems only possible to conclude that the sea-slope work gives : 



Cio ~ 0.0025; 10 < uio > 25 m/sec, 



and any variation there may be with wind speed in this range is obscured by 

 experimental scatter. 



0.003 



0.002 



0.001 



20 

 Wind speed , m/sec 



30 



Fig. 7. Drag coefficient, ciq, from surface-tilt observation related to wind speed. 

 O Gulf of Bothnia • Ringkobing Fiord A Lake Erie. 



D. The Dependence of the Drag Coefficient on Wind Speed 



That the extent of the variation of the drag coefficient with wind speed is 

 still rather uncertain is evident from Figs. 6 and 7. Earlier suggestions that 

 the drag coefficient increases rather abruptly by 100% or more at about 

 7 m/sec — the wind speed at which white caps start to become numerous — is not 

 supported by the data shown in Fig. 6. A comparatively slow increase of drag 

 coefficient with wind speed is now seen to be much more probable and the 

 simple linear relationship for near-neutral conditions, 



Cio = (1.00 + 0.07i^io)x 10-3, (22) 



in which uiq is in m/sec, is shown in Fig. 6. At 20 m/sec this would give Cio = 

 0.0024, which agrees reasonably well with values from sea-slope data. Sheppard 

 (1958) has also proposed a linear relationship but with a rather greater slope. 

 His proposal for the range 1-20 m/sec is 



cio = (0.80-f 0.114wio)x 10-3. 



At wind speeds from 2-9 m/sec this differs from (22) by no more than 10% at 

 any point. At 20 m/sec Sheppard's formula gives Cio = 0.0031, which appears to 

 be somewhat large. 



