Laserfiche WebLink
<br />C;) <br />o l~ <br />~ <br />l~ <br />~ periods of time are required for reservoir storage to ,ecover from <br />~ <br />severe drawdown, barring curtailment of either upstream or downstream <br /> <br />depletions. <br /> <br />6.2 Determination of Simulation Time and Stability of System Response <br /> <br />The simulation time is the number of years for which the system is <br /> <br />observed. It is desired that there be a sufficient number of observa- <br /> <br />tions to provide stable estimates of system response. <br /> <br />A common practice for obtaining stable measurements of system <br /> <br />response from simulation studies is to make several simulations and <br /> <br />treat them as replications of a single experiment. Analyzing the <br /> <br />results from several runs may indicate convergence and provide estimates <br /> <br />of average response and variance of response (Emshoff and Sisson, 1970; <br /> <br />Mize and Cox, 1968). The length of each simulation and the number of <br /> <br />simulations required for a given degree of convergence determine the <br /> <br />efficiency and feasibility of this approach. No standard technique <br /> <br />for determining either the length or number of simulations presently <br /> <br />exists, and their determination has been labeled an art (Mize and Cox, <br /> <br />1968). <br /> <br />Factors affecting the total length of simulation required to obtain <br /> <br />convergence of system response are (1) the variability of the stochas- <br /> <br />tic inputs to the model (Me1entijevich, 1965); (2) the degree of <br /> <br />dependency or autocorrelation between observations within a simulation <br /> <br />run (Emshoff and Sisson, 1970); and (3) the degree of convergence <br /> <br /> <br />desired. <br />