292 

 channel are shown in Figure 3 • The pressure gradient was found to increase 



with wind speed, and with depth of the water. 



Velocity Distribution in the Air 



Measurements of the mean horizontal air motion in the vertical direction 

 and across the channel were taken at several sections for Uqq from 6 mps to 

 about ik mps . Typical data for vertical profiles along the center section 

 of the channel are showi in Figure ^A. The vertical profiles of U.(z') 

 indicate that the air flow generally develops a behavior characteristic of 

 turbulent flow in a boundary layer over roughened surfaces. In a few cases, 

 a small kink in the distribution of U(z') was observed which usually appeared 

 at 'V'5 cm height above the mean water level. Using pitot tube measurements, 

 Francis (l95l) also observed these kinks. Schooley (1963) was able to find 

 the kinks by tracing mean trajectories of bubbles over the waves. However, 

 their measurements indicated that the kinks appeared somewhat closer to the 

 water surface, z'= 2-3 cm. The existence of the kinks in the profiles of 

 air velocity indicate that a jet of high velocity air may sometimes develop 

 over wavy water in channel flows. To the authors' knowledge, however, with 

 the possible exception of Sheppard (1952), this phenomenon has not been 

 observed with any measurements over water in the atmosphere . 



Typical measurements of the horizontal distribution of velocity are 

 shown in Figure ^B. These data are representative of flow in wind tunnels 

 of rectangular cross-section. It is interesting to note that the boundary 

 layers associated with the side walls can become rather thick. This thick- 

 ening had no apparent effect, however, on the development of significant 

 waves in the channel. The waves still exhibited a nearly linear crest moving 

 approximately normal to the mean wind direction. 



The lines of constant air velocity plotted for a given cross-section 

 reveal an interesting feature of the channel flow as shown in Figure 5- 

 Because of a secondary circulation in the tunnel, the lines of constant 

 velocity are squeezed down in the comers of the cross-section. This has 

 been observed previously for flow in rectangular ducts (e.g., Schlichting 

 (i960)). However, the effect appears to become somewhat more pronounced 

 when fluid flows over a moving boundary in the CSU channel. 



The three dimensional structure of the air flow does not visibly affect 

 the waves generated on the water surface. However, the pressing of the air 

 moving at higher speeds down along the walls seems to be transmitted to the 

 horizontal velocity in the water. Measurements of the horizontal distribution 

 of velocity in moving water show two maxima developing just underneath the 

 "ears" of the constant velocity curves drawn in Figure 5' Hence, strictly 

 speaJting, the velocity in the air and in water should be written as U(y, z'), 

 and u(y,z), instead of U(z'), and u(z). However, for the purposes of this 

 discussion the motion of the air and the water will be treated as two 

 dimensional. 



