Laserfiche WebLink
<br />r::) <br />Cl <br />f\:) <br />t-' <br />W <br />N <br /> <br />201 <br /> <br />uncertain. This uncertainty is important in that its presence in- <br /> <br />creases the risks inherent in making management decisions. <br /> <br />The following questions may be posed: (1) how much of the ob- <br /> <br />served variation in response is the result of properties of the hydro- <br /> <br />logic processes modeled, and how much is the result of a failure to <br /> <br />model all relevant streamflow characteristics; (2) can the uncertainty <br /> <br />in response be lessened through refinement of the model; and (3) given <br /> <br />the uncertainty in response, how should the output of the model be <br /> <br />interpreted and used in making management decisions? <br /> <br />The first and second questions concern the validity of the model <br /> <br />and are discussed in this section. The third question involves model <br /> <br />application and is addressed in Chapter 7. <br /> <br />The limited data base used for calibration introduces an inherent <br /> <br />uncertainty into the values of all quantities examined with the model. <br /> <br />This uncertainty can be decreased by reca1ibrating the existing model <br /> <br />as more data become available. However, there are streamflow charac- <br /> <br />teristics that are not necessarily reproduced by the, synthetic stream- <br /> <br />flow generator used in the model. One suggestion for improving the <br /> <br />streamflow generator is presented here. <br /> <br />An examination of the ten simulations performed with the model <br /> <br />indicated that the variation in frequency of reservoir failure was <br /> <br />related to the differences in patterns of years of high and low runoff. <br /> <br />This observation suggests that information about naturally occurring <br /> <br />patterns of high and low runoff might indicate modifications to the <br /> <br />model that would provide more accurate determination of system response. <br /> <br />For example, analyses of run-length (duration) and run-sum (severity) <br />