Laserfiche WebLink
<br />A monthly flow balance at the mainstem gages was performed which indicated <br />significant errors between flow volumes calculated using the tributary and <br />diversion data and the observed volumes at the mainstem gages. These errors <br />were apparently due to a combination of the following: <br /> <br />1. Irrigation return flow and sand sluicing flows that are not gaged. <br /> <br />2. Routing effects, attenuation and travel times <br /> <br />3. Interaction of river flows with ground water: infiltration and/or <br />exfiltration <br /> <br />4. Use of constant diversion rates for one month periods whereas daily <br />variation probably occurs <br /> <br />5. Random gage errors. <br /> <br />It was necessary to remove the error in the water balance; <br />long-term sediment accounting requires accurate water accounting. <br />decided that the ungaged irrigation return flows were the most likely <br />the water balance errors. <br /> <br />reliable <br />It was <br />cause of <br /> <br />To obtain the correct (observed) flow volumes at each of the three <br />intermediate gages, an additional tributary was inserted immediately <br />downstream of each of those gages. This tributary accounted for the ungaged <br />irrigation return flows in the next reach. The tributary flows were <br />calculated as daily flows, constant for a month, such that a volumetric <br />balance was maintained at the intermediate gages. No sediment discharge was <br />associated with these flows; consistent with the theory that these flows are <br />irrigation returns. <br /> <br />The final adjusted daily flow data <br />pre-processor to generate a sequence of <br />computational efficiency and accuracy. <br />ranged from one day to one month. <br /> <br />were then processed by the HEC-6 data <br />flows of varying time step to optimize <br />The final computational time steps <br /> <br />4.4 Kodel Performance and Calibration <br /> <br />Long period, movable boundary, simulations are subject to many data <br />uncertainties. In this study. uncertainties involved the geometric data (a <br />complete set of cross sections for the entire study reach for one point in <br />time was not available), the sediment data (no main stem inflowing load <br />measurements), and hydrologic data (ungaged areas and ungaged agricultural <br />return flows). These various data gaps were filled in appropriate fashion as <br />described in previous sections. Once assembled, however, test simulations <br />must be performed to ascertain whether the model is performing adquately and <br />if necessary adjust (calibrate) various model parameters to improve the <br />correspondence between simulatIon results and the observed stream behavior. <br />Only after the model has been calibrated (and verified if data are available) <br />can it be used for credible predictions of future channel behavior. <br /> <br />Selection of appropriate measures of model performance for a movable <br />boundary model such as HEC-6 is not straightforward. Ideally, one would have <br />complete sets of sut'Veyed cross sections and measured sediment transport rates <br />periodically throughout the calibration period. These data sets are extremely <br />rare. Consequently, different calibration measures may be used for different <br />studies depending on study objective, data availability, etc. Some of these <br />calibration techniques are described in reference (7). <br /> <br />26 <br />