Laserfiche WebLink
<br /> <br />4.4.3 Monthly Properties <br /> <br />N <br />I-" <br />CJ:l <br />c.;l <br /> <br />In addition to mllintsining the .correct total flow in a ballin, the <br />data generation procedure must also preserve the annual periodicity <br />reflecting the seasonal runoff pattern. Since it would require an <br />enormous volume to compare the periodic patterns of all stations, two <br />major ones will be illustrated here. The present modified values were <br />taken from Appendix 1 and the synthetic monthly parameters from an <br />analysis of the generated traces. Figures.9 and 10 of the July 1974 <br />report show monthly means and standard deviations, respectively, for <br />the flow of the Gunnison River at Blue Hesa Reservoir. This station <br />was chosen to illustrate the behavior of synthetic traces at an <br />upstream or "rim" station. The population values, indicated by solid <br />lines connecting them, are generally in the center of the synthetic <br />trace values. This indicates the streamflow generation me~hanism <br />creates values which might be expected if the flow were sampled for <br />the given periods. The good agreement in the standard deviation plot <br />further indicates that the monthly variation is well preserved .in the <br />synthetic traces. <br /> <br />Figures 11 and 12 of the July 1974 report show the same properties <br />for the Lees Ferry station. This station is especially significant <br />since it represents the SUDI of many flow increments and "rim stations" <br />from above. The good comparison between properties of the synthetic <br />traces and those of the present modified flows clearly illustratesthe <br />efficacy of the generating procedure as well as the adequacy of the <br />coefficients used in the model. <br /> <br />4.4.4 Probability Distributions <br /> <br />The final comparison was made in the July 1974 report to illustrate <br />the validity of synthetic flows is a series of plots containing the <br />probability distributions of flows. To construct long sequences of <br />data, 30-year traces were used together to provide five 60-year traces. <br />These data were then averaged for three periods (3, 5, and 9 years) <br />by a moving mean process.11 Plotting positions were calculated by <br />the equation <br /> <br />m <br />P(x) . N+f . 100 <br /> <br />(15) <br /> <br />In the equation P(x) is the probability of the random variable being less than <br />or equal to x, m is the ordered sequence of a particular x value and <br />N is the sample size. P(x) is in percentage units. These probabili- <br />ties were calculated for the annual as well as 3-, 5-, and 9-year <br /> <br />11 Calculations and results produced by the computer program "SUTl" <br />described in the unpublished user I s guide, "Simulation Util ity <br />Program I," by R. W. Ribbens, Division of Planning Coordination, USBR. <br /> <br />32 <br />