Table 28- Differences between the means for the IFG-4 pre- 

 dicted versus the measured velocities in five s\abreaches 

 of the Madison, Beaverhead, Gallatin, and Big Hole Rivers, 

 The standard error is in parenthesis. 



Sub reach 

 Madison (#1) 

 Madison (#3) 

 Beaverhead (#2) 

 Gallatin (#2) 

 Big Hole (#1) 



■■■r 



IFG-4 Predicted Vs. Measured Velocities 



Difference Between 



Means 

 in ft/sec 



.144(.009) 



-.109(.009) 



■,.,... ., .115 (.019) -'- 



-.161(.011) 



* ^'''- -.OSK.OIO) 



A better measure of the bias of the velocity predictions is 

 the standard error for the differences between the predicted versus 

 the measured means (Table 28) . The subreach having the smallest 

 standard error would have the most reliable velocity predictions. 

 For the Madison #1, Madison #3, Big Hole, and Gallatin subreaches, 

 the standard errors were similar, ranging from .009 to .011 ft/ 

 sec, and highest (.019 ft/sec) for the Beaverhead subreach. The 

 standard errors indicate that the predicted velocities were the 

 least reliable for the Beaverhead s\±>reach while the reliability 

 of the predictions for the remaining four subreaches was about 

 equal. 



The Beaverhead is the only subreach in which the reliability 

 of the IFG-4 velocity predictions is questionable. The author 

 believes that the bias, however, is not large enough to invalidate ' 

 the predictions within the range of the calibration flows. 



Comparison of IFG-4 and WSP Models 



In this section the predictions of water surface elevation, 

 velocity and depth generated by the two models are compared to 

 the measured values. A comparison could not be made without 

 first adjusting the IFG-4 and measured velocities and depths. 

 The WSP model only predicts a mean depth and velocity by segment 



82 



