DWR_3066022 - Laserfiche WebLink

Western Dam Engineering Technical Note May 2017 14 Figure 2. Unacceptable Simulation from an Unverified Model. Figure 3 presents an example in which the modeled runoff (from Figure 2) has been calibrated and has a high degree of accuracy and reliability. Figure 3. Acceptable Simulation from a Verified Model. The Extreme Storm Working Group Summary Report (MT ESWG) prepared by DOWL in December 2016 for the Montana Dam Safety Program (Department of Natural Resources and Conservation) provided an excellent overview of model verification and calibration techniques and methodologies [4]. As described by the MT ESWG, model simulation calibrations are based on two general methodologies: 1) Calibration – The process in which the parameters of a hydrologic model are adjusted to replicate a measured or observed event (gaged watersheds). Gaged data include those from precipitation, snowpack, and stream discharge gages. This process lends itself to a high degree of confidence and reliability. 2) Pseudo-Calibration – The process in which the parameters of a hydrologic model are adjusted to reasonably approximate a range of flood frequency values obtained independently from the hydrologic model, such as the 100- and 500-year events (gaged and ungaged watersheds). Ungaged data include those from local and regional regression equations (e.g., USGS regional regression for peak runoff rates) and correlations with similar, neighboring gaged watersheds. This process lends itself to more of a “sanity check”. Ideally, a calibration to a single or series of measured or observed events can be performed for a given watershed. Unfortunately, most watersheds lack sufficient data to perform a calibration due to:  No measured or observed data;  Generally short or incomplete data records and associated statistical limitations (i.e., attempting to reliably estimate infrequent event data based on a small sample size); and  An absence of observed infrequent precipitation, snowpack/melt, and stream discharge event data. For these reasons, pseudo-calibrations are often necessary to verify flood modeling runoff results; however, the use of measured or observed data for pseudo-calibrations is often met with the same limitations described for calibrations. As such, pseudo- calibrations are commonly performed using data that are generally regional and not specific to the watershed. This can be accomplished by using regression equations developed by organizations like the USGS (https://water.usgs.gov/osw/streamstats/) or USACE, as well as those developed based on Log Pearson III or other probabilistic comparisons for neighboring and similar watersheds, in which, more robust observed data are available. Neighboring and similar watersheds could include those that are adjacent to the study watershed or lie within the same overall hydrologic basin and have similar watershed characteristics like area, elevation, shape, topography, vegetative cover and general precipitation loss parameters. By “normalizing” these characteristics and parameters, a comparison of runoff rates per area (e.g., cfs per acre) can be estimated for a