In all four subreaches, the differences between means for 

 the IFG-4 versus the measured water surface -elevations were 

 considerably less than those for the WSP versus the measured 

 elevations. The differences ranged from .003 to .007 ft for 

 the IFG-4 model and .026 to .215 ft for the WSP model. The 

 IFG-4 model is clearly the better predictor of water surface 

 elevations. 



In three of four subreaches, the differences for the IFG-4 

 model were less than those for the WSP m.odel for both velocity 

 and depth. Velocity differences ranged from .046 to .253 ft/sec 

 for the IFG-4 model and .100 to .6 70 ft/sec for the V7SP model. 

 EJepth differences ranged from .066 to .135 ft for the IFG-4 

 model and .018 to ,192 ft for the WSP model. 



A better measure of the bias of the predicted values is 

 provided by the standard error for the differences between 

 means (Table 30) . Again, the model having the smaller standard 

 error is considered the better predictor. In all four sub- 

 reaches, the IFG-4 model produced smaller standard errors for 

 both water surface elevation and velocity. In two of four 

 subreaches, the standard errors for the differences between 

 means for depth were smaller for the IFG-4 model and were 

 equal for both models for the remaining two subreaches. 



Based on the above evaluation of the differences between 

 means and their standard errors, the IFG-4 model was undoubtedly 

 the better hydraulic simulation model in this study. 



The lower reliability of the WSP predictions of mean seg- 

 ment velocity and depth may in fact have little practical sig- 

 nificance. A comparison of the differences between means in 

 Table 30 suggests that, except for the velocity predictions 

 for the Beaverhead subreach, the bias of the WSP predictions 

 of velocity and depth are not of a magnitude to be of major 

 concern. Results of the study suggest that in single channels 

 where mean segment velocities and depths are desired, the WSP 

 model should provide reasonably accurate predictions. 



The application of any hydraulic simulation model to sub- 

 reaches containing island complexes should proceed with cau- 

 tion. If islands or multiple channels are unavoidable, the 

 IFG-4 model is preferred. In these situations, it may be 

 unwise to extrapolate the IFG-4 data beyond the highest and 

 lowest calibration flows. 



Wetted Perimeter Predictions 



The reliability of the predictions of wetted perimeter 

 for the two models could not be determined since the actual 

 values were not available for comparison. The IFG-4 model 



85 



