Laserfiche WebLink
<br />~ <br /> <br />sensitive. A -30 percent to +30 percent change in KSAT produced a +48 percent <br />to -53 percent change in computed runoff. The next most sensitive parameter <br />is surface-retention capacity (SURF). A -30 percent to +30 percent change in <br />SURF resulted in a +6 percent to -6 percent change in runoff. Changes of -30 <br />percent to +30 percent in the other parameters resulted in less than 5 percent <br />changes in runoff. <br /> <br />The precipitation section of figure 7 il(ustrates the sensitivity of <br />model output to errors in precipitation input. A -20 percent to +20 percent <br />change in precipitation results in a -59 percent to +59 percent change in <br />runoff. The precipitation input to model calibration runs is a constant, <br />measured value. The sensitivity analysis is included here to demonstrate the <br />effect of possible errors in precipitation measurements on larger watersheds. <br />On all simulator runs on plots 1-4, water applied was measured in from 10 to <br />15 rain gages. The average standard error of the mean precipitation for all <br />simulator runs was 2.3 percent. <br /> <br />Examples of the effect of changes in all parameters used in developing <br />best-fit curves are shown in figures 8 through 12. Changes in KSAT (fig. 8) <br />affected not only the shape of the rising hydrograph; the change in infiltra- <br />tion rate throughout the run created large differences in volume of runoff. <br />Changes in P and FRIC affect the shape of the rising hydrograph, but create <br />very small changes in runoff volume (figs. 9, 12). The surface-retention <br />parameter is a direct subtraction from applied water before runoff can begin <br />(fig. 11). Therefore, it affects the timing of initial runoff. Affect of <br />errors in determination of initial moisture content is shown in figure 10. <br />The rising hydro graph is changed slightly, but runoff volumes are changed <br />very little. Although the fitting of calculated runoff to observed runoff <br />through the adjustment of parameters in the model is somewhat subjective, <br />it appears that a fairly unique set of parameters is obtained when calculated <br />runoff matches observed runoff. <br /> <br />A large reduction in the hydraulic conductivity term is required to <br />reproduce observed runoff for the October runs (figs. 5 and 6). Values of <br />about 1 to 1.2 in/h fit the summer runs, and values of 0.5 and 0.75 are <br />required for the October runs. Antecedent soil-moisture content for the <br />October runs was similar to that for the dry runs of summer: on the order <br />of 0.05 (5 percent of volume). However, temperature of surface soil was <br />considerably less .in October (about 50.F) than during the summer runs (70 <br />to 80.F). Hydraulic conductivity is a function of many factors, including <br />viscosity of water. The effect of lower soil temperatures on viscosity <br />of water is in approximate agreement with the reductions in conductivity <br />required to fit the October runs. Viscosity of water at 75.F is 1.92, and <br />at 50 of is 2.74, or an increase of 42 percent. KSAT needed to fit summer <br />runs was about 1.1; KSAT needed for fall runs was about 0.6, or a decrease <br />of 45 percent. <br /> <br />The moisture content of surface-soil samples (0 to 2 in) taken before <br />and after simulation runs are the basis for assigning values to the initial <br />moisture content, WINT, and transmission zone moisture content, WWET. The <br /> <br />16 <br />