Laserfiche WebLink
<br />computatIons of thiS model. The volume of excess precipitation is routed to the oUllet using the <br />SCS triangular hydro graph approximation. Basins are modeled using a lumped approach in both <br />SCS CN and HEC-HMS. <br /> <br />A method of evaluating the various models must be established to aid in calibrating and <br />validating the models. The concern with this analysis is the ability for a model to accurately <br />estimate peak discharge values for a given basin. Total event volume is also of importance, but <br />assumes a secondary role to the peak discharge in this analysis. <br /> <br />The assumption has been made that a rainstorm of a given return period will produce a runoff <br />event with an equal return period. This assumption, although standard practice, may be a poor <br />one. The amount of antecedent moisture in the soil can have a dramatic impact on the peak <br />discharge (Urbonas, 1979). This would tend to have a greater impact on the less intense, more <br />frequent storms where rainfall depth is relatively small. The affect this has on the frequency <br />distribution is inconclusive as antecedent moisture appears to be a random phenomena (Urbonas, <br />1979). <br /> <br />Given the above assumption, a model should be able to match the frequency curve within <br />acceptable limits. General hydrologic data can be obtained from topographic maps and soil <br />surveys to develop initial model parameters for simulation. Often times recommended parameter <br />space is available in the model documentation or other references. When these recommended <br />values exist, they are often used in place of more detailed estimation. This is especially true in <br />areas where systematic records do not exist. Calibrating the model against various measured <br />rainfall-runoff events will help to refine the model parameters. These data generally reflect the <br />more frequent floods on the frequency curve. <br /> <br />Recorded storms and associated runoff are used to develop the model calibrations, when <br />available. Calibration attempts to match the peak and volume of the events within adopted <br />limits. The peak flow was deemed to be niost critical for this study. For calibration, the <br />observed peak should be replicated within 15 percent. Parameters are adjusted such that the <br />volume residual is minimized without set limits. (V\v~Y< iV";"". <br />~A)IY""' <br />The generous bounds given to the volume car ration can be justified due to unreasonable <br />discharge readings at the beginning of the hydrograph. While the majority of the <br />hydrographs appear reasonable within the. brunt of the runoff, several storms at each site <br />contained readings at the onset of runoff that had no hydrologic explanation. Such oddities <br />included constant, non-zero discharge readings several time steps into the event and initial <br />discharge readings greater than half the magnitude of the peak. Such data problems impart <br />significant uncertainty in volume calculations. <br /> <br />In several cases, the storm data may not exist or is of questionable quality. The flood frequency <br />curves for the sites in this study, as developed in Section 3.3, are the best estimate of the <br />frequency relationship under the uncertainty of the available data. This statement should be <br />qualified noting that the frequency curve is established under Bulletin 17B guidelines and the <br />"best" estimate is subject to all assumptions and limitations ofthese guidelines. <br /> <br />10 <br /> <br />v/ <br />