9625 - Laserfiche WebLink

Home Browse Search

26 Probable Effects of the Proposed Sulphur Gulch Reservoir on Colorado River Quantity and Quality near Grand Junction, Colorado In this study, the lower and upper confidence intervals for all Monte Carlo trials agreed within 1 percent. The standard error of average simulated daily streamflow values for Colo- rado River near Cameo, however, continued to decrease until the total number of trials was equal to or greater than 1,500 (see fig. 17). For this reason, the use of 1,500 Monte Carlo trials is deemed sufficient to ensure convergence and stability of fore- cast distributions in this study. Because the daily time step is used to compute water-budget components over a calendar year, the 1,500 Monte Carlo trials is representative of 1,500 alternate calendar years and therefore hydrologic condi- tions. Comparison of Simulated and Measured Forecasts In the second phase of model validation, the objective is to evaluate the stochastic mixing model accuracy. To evaluate the accuracy of the stochastic mixing model, selected forecasts are summarized statistically and compared to selected hydrologic and water-quality data that were collected at various streamflow gaging stations along the Colorado River study reach. This val- idation phase uses the same decision variables previously described in the Stability and Convergence Section. The sto- chastic hydrology forecasts of interest include annual and daily simulations of the Colorado River streamflow near Cameo and Palisade and at Plateau Creek, a tributary to the Colorado River. A comparison of simulated and measured annual streamflow statistics is provided in table 7. Inspection of the annual streamflow statistics indicates excellent correspondence between simulated and measured streamflow at the Colorado River near Cameo gage station. This finding underscores the validity of using correlated daily ran- dom streamflow variables in this study. Whereas the simulated and measured statistics associated with streamflow at Plateau 1,500 Creek near Cameo and the Colorado River near Palisade are of the same order of magnitude, these statistical streamflow values are not as accurate as those at Cameo. One plausible reason is that the residual analysis used to convert the deterministic equa- tions (fit using nonlinear regression) to stochastic equations is not as accurate as using correlated random variables (see fig. 10). Individual stochastic daily streamflow simulations for USGS Colorado River gage sites near Cameo, Palisade, and Plateau Creek near Cameo, however (shown in figs. 18-20), demonstrate good visual correspondence to the measured daily median values. In addition to streamflow, the mean-monthly simulated evaporation values were computed and found compa- rable to basin estimates published by Farnsworth and Thomp- son (1982). A comparison of simulated and measured daily streamflow statistics is provided in tables 8-10. In general, good correspondence exists between measured and simulated daily streamflow statistics. Whereas the median and average statistics appear to give the best correspondence between actual and simulated values, the extreme hydrologic conditions represented by the minimum and maximum stream- flow values are characterized by more uncertainty. The fact that measured and simulated streamflow statistics appear similar at the Colorado River near Cameo gage site (close to the begin- ning of study reach) and then at Colorado River near Palisade gage (end of study reach) indicates that the water-budget process of adding and subtracting stochastic diversions, return flow, streamflow tributary water, and evaporation works. Because the stochastic hydrologic model appears to provide adequate simulations of daily and annual streamflow through- out the study reach, the next section is used to provide informa- tion to better understand background hydrologic and water-quality conditions for selecting appropriate reservoir parameters that satisfy the PBO water-delivery requirements. U j 1,250 U ~ Z Z ~ 1,000 ~ W O cn ~ W 750 W 0_ O ~ Q ~ 500 Z 250 0` 0 500 1,000 1,500 2,000 2,500 3,000 NUMBER OF MONTE CARLO SAMPLES Figure 17. Mean standard error of median simulated daily streamflow values for Colorado River near Cameo as function of Monte Carlo sampling.