|
Copyright
|
(Copyright © 1981, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing) |
|
Abstract
|
A case study of the Colorado River at Glenwood Springs, Colorado, is presented to demonstrate several statistical tests for identifying watersheds in which conditions are changing with time.
RESULTS of the tests indicate that annual peak flows of the Colorado River are influenced significantly by reservoir regulation. Consequently, conventional methods of frequency analysis are not suitable for obtaining flood estimates from the data series. Time-series analysis is a versatile approach to flood-frequency determinations when conventional statistical methods are not appropriate. The basic strategy of time-series analysis is to treat each value of the regulated annual peak-flow series as a combination of two elements-a deterministic component and a stochastic component. The deterministic component is quantified and removed from the flood series. The residual stochastic components, found to be stationary and independent, are then fitted to a probability distribution from which annual floods are estimated.
RESULTS of the time-series analysis show that the 2 percent and 1 percent chance floods, both required for Interstate highway design, are substantially less than corresponding log-Pearson type III estimates. Because the time-series analysis is able to detect and to treat the impact of reservoir regulation on the peak-flow series, the resulting flood frequency estimates are more representative of the watershed. |