ESTIMATING THE T-YEAR FLOOD BY THE EXTENSION OF RECORDS OF PARTIAL DURATION SERIES

Open Access
Publisher Website
Google Scholar
Abstract
The discharge likely to be exceeded once in T years on average is the ‘T-year flood’. This may be estimated in at least three ways: by fitting a probability distribution to the record of annual maximum discharges, by Monte Carlo simulation using a time-series model of the instantaneous peak discharges in short time-intervals (commonly days), or by a ‘partial duration series’ approach, in which a stochastic model is formulated for discharges exceeding a threshold. The latter method commonly assumes a Poisson distribution for the occurence of exceedances and an exponential distribution for the magnitudes of exceedances. The paper considers an extension of the partial duration series method to the commonly-encountered situation in which (a) the T-year flood must be estimated for a gauging site with a short record, and (b) a longer record is available for a nearby gauging site. The conclusion is that such an extension is possible if the following assumptions are valid: (a) the incidence of exceedances follows a bivariate binomial distribution for the period of common records; (b) the magnitudes of exceedances follow a bivariate exponential distribution.

This publication has 0 references indexed in Scilit: