DWR_3066022 - Laserfiche WebLink

Western Dam Engineering Technical Note May 2017 15 range of events and conditions and can be used as part of study watershed pseudo-calibrations. Due to the regional nature of some pseudo-calibration data, calibrations may not be able to adequately replicate predicted runoff rates; however, general discharge-frequency tendencies should be retained, if possible, which provides some degree of reliability and confidence. Figure 4 shows an example in which the modeled discharge frequencies do not have the same tendency as the observed data and do not lie within the 90 percent confidence band. Ideally, a verified model (Figure 5) should produce results that retain the same tendency as observed or predicted data and lie close to the line of equal values. Figure 4. Unacceptable Simulation for an Uncalibrated Model. Figure 5. Acceptable Simulation from a Calibrated Model. Flood modeling runoff result verifications are often focused on replicating peak runoff rates; however, in some cases, the runoff volume can prove to be more critical than the peak runoff rate. Runoff volume verifications require measured or observed event data; therefore, these types of verifications are calibration based. Conversely, peak runoff rate verifications can be based on watershed specific and/or regional data and associated calibrations and/or pseudo- calibrations. Calibration Techniques Hydrologic model runoff result calibrations involve varying estimated watershed input parameters such that the model results match well with observed and/or estimated data under similar conditions. It is important to note that model input parameters are generally adopted based on methodologies and empirical relationships whose own parameters are subject to interpretation and engineering judgement. As such, a range of potentially reasonable values/magnitudes for a given watershed parameter is expected and appropriate. These ranges provide a basis (i.e., upper and lower limit) for which model input parameters can be varied and justified. Figure 6 presents a flow chart of a typical hydrologic model runoff result calibration process. Common watershed input parameters varied as part of this process include:  Excess precipitation – Infiltration, evaporation, and transpiration rates, as well as initial abstraction (i.e., interception). Initial abstraction and the infiltration rate are the most pertinent to dam safety studies. o Runoff volume modeling results are most sensitive to watershed loss and excess precipitation parameters.  Excess precipitation transformation – Physical watershed characteristics like shape, topography, surface roughness, etc. o Runoff rates and timing are most sensitive to excess precipitation transformation parameters.  Watercourse routing – Physical watercourse characteristics like hydraulic roughness, slope, cross-sectional geometry, etc.