Laserfiche WebLink
<br />The storm of August 10 produced 0.74 in of rain; results for this storm <br />are shown in figures 30 and 31. Both the observed volume and peak were under- <br />estimated by about 70 percent using simulator parameter specifications. Using <br />the natural parameters, volume was underestimated by 28 percent and peak by <br />37 percent. A summary of these modeling results is shown in table 10. <br /> <br />Sources of Error <br /> <br />As shown in figures 26 through 31, the general shape and timing of modeled <br />hydrographs is in fair agreement with the shape and timing of observed hydro- <br />graphs. The major source of error in reproducing the observed hydro graph is <br />the prediction error in runoff volume. A good prediction of the runoff volume <br />is contingent on many factors; the two most significant factors are: (1) An <br />accurate representation of storm rainfall, and (2) an accurate computation <br />of infiltration losses (a function of KSAT). The effect on model results <br />of changing the specifications of KSAT values and of adjusting storm rainfall <br />are shown in figures 32, 33, and 34. For the June 25 event shown in figure <br />32, KSAT values were lowered from the natural specification, and storm rain- <br />fall was increased by 40 percent. The predicted volume is 12 percent more, <br />and the predicted peak is 9 percent less than the observed values. Results <br />of modeling the July 26 event using simulator parameters, and aID-percent <br />reduction in storm rainfall are shown in figure 33. The predicted and <br />observed runoff volumes are the same; the predicted peak is 18 percent less <br />than the observed. Observed hydrographs for both the June 25 and July 26 <br />storm events show the observed rise lagging the predicted rise by about <br />6 minutes. On June 14, a tornado crossed the watershed of the North Fork <br />and deposited many trees in the drainageways (fig. 35). Flood waters of <br />the storm of June 25 carried one of these trees into the approach section <br />of the measuring flume where it became lodged (fig. 36). These trees may <br />have partially dammed the channels and slowed the runoff. Model results <br />for the August 10 storm are shown in figure 34; KSAT values are lowered, <br />(as was done for the June 25 storm), and storm rainfall is increased by <br />10 percent. Predicted runoff volume is 7 percent larger, and predicted <br />peak is 6 percent smaller, than observed values. <br /> <br />Results of the foregoing adjustments were included to demonstrate <br />sensitivity of the model to possible variations in rainfall input and KSAT <br />parameter. Rainfall data from one gage located on the edge of the watershed <br />could vary considerably from total rainfall on the watershed. <br /> <br />RESULTS AND CONCLUSIONS <br /> <br />Runoff data from both simulator runs and observed rainstorms on plots <br />were used to develop best-fit parameters of the Green-Ampt infiltration <br />equation. In all fitting attempts, the hydraulic conductivity term, KSAT, <br />grossly controlled the goodness of fit. High variability in soil-water <br /> <br />49 <br />