Harleman (1971) has shown that, based on inspectional analysis of the 

 one-dimensional mass transfer equation 



T,^^t'i&{^^t) (^-^^' 



the dispersion coefficient scale can be determined for areas of signifi- 

 cant vertical density gradients. The dimensionless coefficient for the 

 left-hand term is 



UoL' 



therefore, the dispersion coefficient ratio is 



or 



In regions of salinity-induced density gradients, verification of model 

 salinity conditions against prototype observations ensures that the mass 

 dispersion process is satisfactorily reproduced in the model. 



However, for uniform density areas (no longitudinal or vertical salin- 

 ity gradients), Harleman (1971) and Fischer and Holly (1971) state that 

 there is a direct conflict between the required dispersion coefficient 

 ratio determined by inspectional analysis (eq. 3-14) and the ratio pre- 

 dicted on the basis of observed vertical and horizontal distributions in 

 both prototype and model. From the Taylor-Elder equation 



El = 5.9d(gRSE)^/2 (^3_15) 



where d is the depth of flow, R the hydraulic radius, and Sg the 

 slope of energy gradient. Harleman (1971) shows that in this case the 

 dispersion coefficient ratio should be 



1/2 



H 



,1/2 



(3-16) 



For a model with a horizontal scale of 1:1,000 and a vertical scale of 

 1:100, equati 

 (El)_ = 1:316 



1:100, equation (3-14) gives E' = 1:10,000; whereas equation (3-16) gives 



However, this analysis does not consider the roughness elements 

 commonly used in distorted-scale tidal models. The vertical roughness 

 strips generate large-scale mixing by eddies which may be on the order 



55 



