86. Termination time . A termination time Tfin, also in units of 

 decimal hours, must be specified. The model halts when the termination time 

 is reached. Tfin must be greater than TO. 



87. Tolerance values for iteration convergence . Tolerance values for 

 elevation and discharge determine the number of iterations required for the 

 solution to converge to a certain degree of accuracy. These values should be 

 chosen based on consideration of the inlet geometry. If the cross sections 

 are highly irregular, the computation may not converge in a reasonable number 

 of steps, or not at all if the tolerance values are too small. In this case, 

 DYNLETl will stop and a message to this effect will be printed on the screen. 

 If the tolerance values are large, the model will finish rapidly, but the 

 accuracy of the solution is lowered. Tolerance values of 0.02 to 0.05 ft* 

 for elevation and 50 to 200 cfs for discharge are reasonable estimates for 

 most inlets. 



88. Temporal weighting coefficient . The weighting coefficient 9 can 

 vary between 0.5 and 1.0 to define an implicit solution scheme. A weighting 

 coefficient of 0.0 will transform the numerical procedure into an explicit 

 method. However, DYNLETl has not been tested for weighting coefficients less 

 than 0.5. A value 6 = 1.0 is recommended to start a simulation series, and 

 this value is generally recommended for simulations of field conditions. 

 Numerical tests for idealized and severe situations have shown that the value 

 of 9 = 1.0 may introduce some damping, whereas 9 = 0.5 may introduce oscilla- 

 tions. The damping effect is negligible in realistic tidal flows. 



89. Presence of wind . Inclusion of wind in the simulation is regulated 

 by the wind index, which can be either or 1. A value of instructs the 

 model to omit wind shear, but for a wind index of 1, the program needs a 

 coefficient of wind drag to be used in the calculation of surface shear. The 

 drag coefficient, wind speed, and wind direction are specified in data input 

 files . 



90. Computation time step . Niamerical tests (e.g.. Price 1974) of 

 numerical models of the shallow-water wave equations have shown that solution 

 accuracy depends on the size of the computation step. In natural channels, 



A table of factors for converting non-SI units of measurement to SI 

 (metric) units is presented on page 4. 



36 



