Jin Wu 
only at high wind velocities. Moreover, since the slope of the fitted 
straightline for the oceanic data is related to the spectral coefficients 
in the equilibrium range, the same slope of the fitted lines shown here 
for laboratory and oceanic data indicates that the spectral coefficients 
are universal constants independent of wind fetches. Consequently, 
the present results also verify the concept on developing the equili - 
brium wind-wave spectrum : high-frequency wind waves saturate at 
short fetches. 
The cross-wind slope has not been measured in the present 
tank. However, the fetch for the growth of resonance waves is about 
half the width of the tank, or 0.5 m, because the slope measurement 
was made at the center of the tank. According to Cox (1958), the 
mean-square slope should not rise until the fetch is greater than lm. 
Therefore, the cross-wind slope should be negligible in the present 
tank. Consequently, the upwind-downwind component of the present 
measurement should be nearly the total slope. 
The upwind-downwind components of the laboratory results 
are plotted along with the mean-square slopes of the sea surface in 
figure 20b. Excellent agreement is seen between the oceanic data and 
the laboratory results except at low wind velocities, where the wind 
boundary layer in the laboratory tank is not even turbulent. The boun- 
dary layer in the present tank becomes fully turbulent when the wind- 
shear velocity is greater than 12 cm/sec. The agreement of oceanic 
and laboratory data further substantiates some previous considerations 
the spectral coefficients are indeed universal; the spreading of wave- 
number vectors is nearly isotropic for the sea and is nearly unidirec- 
tionai in the laboratory. 
The comparison shown in figure 20b explains the discrepan- 
cy between the oceanic (Cox and Munk 1956) and the laboratory (Cox 
1958, Wu 1971) results. Furthermore, the agreement on one hand im- 
plies that short waves are directly generated by the wind and the wind- 
shear velocity is therefore the appropriate parameter for correlating 
results obtained at different fetches. On the other hand, the agreement 
indicates the possibility of modeling microstructures at the sea surfa- 
ce in a laboratory tank. 
VIII. CONCLUSION 
In the present study, the microstructure of the wind-distur- 
bed water surface, characterized by surface-slope and surface-curva- 
ture distributions, is measured in a laboratory tank under various wind 
and wave conditions. It is shown that wind waves arise at about the 
1456 
