- 19 - 



the l/l density differences indicated by Froude scaling were main- 



6/ 



tained. This decision causes the simulated bleed vater of the 



model to stratify and collect in depressions and practically to 

 fill these depressions before traveling fai-ther along, It also 

 exaggerates the concentrations of bleed water by retarding its 

 rate of mixiJTg with the ambient water which would otherwise allow 

 it to be carried away with greater rapidity and at greater dilu- 

 tion than the model indicates, KJhen at length, under one-to-one 

 density scaling conditions, the bleed water has reached the steady 

 state of distribution and concentration and is mixing away at a 

 rate equal to its rate of supply, the filaments of a given concen- 

 tration are somewhat too long, too thin and somewhat too narrow 

 because of the deficient Reynolds number of the flow and dissimilar 

 eddy diffusivity of the gravity controlled kinematic model. It 

 is significant, and a result of this choice, that the principal 

 effect of wind on the bleed water tracer pattern in the model is 

 to drive it upwind on the bottom. Both in nature and in the model 

 there is a surface layer driven downwind which is usually somewhat 

 thinner and more vigorous than the upwind counter current on the 

 bottom. In nature wave action and more intense eddy diffusion 

 would tend to mix vertically the contents of the two layers, that 

 is, the bleed water accumulated in the bottom counter current 



6/ Question concerning the advisability of departing from Froude 

 scaling arises only in connection with models of very shoal 

 estuaries wherein wave action reaches to the bottom and wind 

 circulation is important. 



