Description |
-
Abstract: Practicing hydrologic engineers have limited resources for documenting and justifying the use of flood frequency analysis (FFA) methods when, due to changing watershed conditions, commonly used methods are inapplicable. A practical parsimonious method is proposed to improve the accuracy of traditional approaches to FFA under nonstationary conditions. Using 75- and 77- years of annual max... read moreima discharge measurements from two highly urbanized watersheds, we demonstrate that traditional estimates (of the 100-year flood) are significantly downward biased when an overt trend in the mean is omitted. Providing corrections to remove this bias, we capture the nonstationarity in a flood series by fitting a single regression using method of moments from which we get estimates of three conditional moments for two- and three-parameter log-normal, log-Pearson and Generalized Extreme Value probability distributions, yielding similar and more reasonable estimates of the design discharge than a traditional stationary analysis would yield. Standardized probability plots are constructed for comparison between hypothesized distributions and their stationary and nonstationary fits. Issues regarding the application of this procedure and the needs for future study are proposed for the practicing engineer.
Thesis (M.S.)--Tufts University, 2016.
Submitted to the Dept. of Civil Engineering.
Advisor: Richard Vogel.
Committee: Ellen Douglas, and James Limbrunner.
Keywords: Hydrologic sciences, Water resources management, and Civil engineering.read less
|
This object is in collection