DWR_3066022 - Laserfiche WebLink

Western Dam Engineering Technical Note May 2017 12 Calibration, Validation, and Verification: Bringing More Certainty to the Uncertainty of Hydrologic Modeling Results Introduction Engineers and regulators alike are frequently confronted with the task of attempting to accurately simulate and estimate processes that are both highly complex and variable. The dam safety arena is no different in this respect, particularly as it applies to the “black magic” better known as hydrologic watershed modeling. The hydrologic responses of a watershed are dependent on sundry variables—all of which are difficult to confidently and accurately estimate—and when combined, do not usually give one a warm and fuzzy feeling of confidence. For this reason, hydrologic modeling for dam safety evaluations is often performed using conservative methodologies. While a conservative approach helps us all to sleep a little easier at night, it can significantly increase hydrologic model runoff results and associated dam flood passage infrastructure requirements. This of course can lead to costly dam modifications, which may or may not be entirely necessary based on the level of conservatism adopted as part of the modeling. Wouldn’t it be nice to be able to justify and verify that the parameters and approaches adopted as part of hydrologic watershed modeling are accurate and appropriate? Of course it would—and you can (well, sort of). A model verification process can be employed to provide degrees of confidence and reliability in modeling results ranging from very high to very low, but hey, very low is better than extremely low or zero! More often than not, though, it is likely that model result confidence and reliability will lie somewhere between these extremes; but, like most applications to engineering evaluations, available data quantity and quality is extremely influential and important. The model verification process [1, 2] is summarized on Figure 1. The goal of a verified hydrologic model is to be able to adequately replicate observed, measured, and predicted watershed data, like runoff rates and volumes, for a suite of events, conditions and scenarios. This is accomplished through a process of:  Using a hydrologic model to estimate runoff;  Comparing modeled runoff estimates to available watershed data to assess the adequacy of the results; and  Adjusting and refining model watershed input parameters to improve agreement between modeled runoff and available watershed data for a suite of events, conditions and scenarios. Figure 1. Watershed Hydrologic Model Verification Process. A verified hydrologic model can be used to confidently predict the watershed responses resulting from a range of precipitation and flood events; however, reliable return intervals often fall within a range more frequent than a 500-year return period due to available data limitations. For this reason, it can be more difficult to verify hydrologic models for dams with higher hazard potentials as the inflow design floods (IDF) are often significantly less frequent than a 500-year event. Runoff Estimation – A Quick Review This article builds on previous Western Dam Engineering Technical Note (WDETN) articles—one of which focuses entirely on the process of estimating runoff though flood modeling [3]. Rather than referring the reader to that article alone, let’s quickly review the steps required to transform precipitation to runoff. Step 1: Define the precipitation and corresponding IDF event  For the purposes of dam safety, the IDF is the flood event required to be safely routed through the reservoir. The IDF is typically based on a dam’s hazard