TR-97 : The effect of measurement errors on design flood estimation by flood frequency analysis

Abstract

Flood frequency analysis is very commonly used for estimation of design floods of different return periods. For correct estimation of the design flood the observed data are required to be accurate, reliable and of sufficiently long duration.

Hydrological data contain errors which are introduced during measurement, recording and processing. The measurement errors in the data result in erroneous assessment of the design flood. The under-estimation or over estimation of design flood due to measurement errors in the data lead to under design or over design of the water resources structures. If any structure is under designed, then the risk of its failure increases. On the other hand, over design of any structure may lead to an uneconomic proposal.

The effects of errors in the observed data are highly pronounced on the design flood estimation. In case of design flood estimation by flood frequency analysis, the observed values of the peak discharges are used which are more prone to measurement and other types of errors than the normal flow data. Such errors may lead to wrong estimation of design flood during extrapolation; as the design floods are usually estimated for a very large recurrence interval on the basis of short record length.

In this study, an attempt has been made to examine the effects of various types of measurement errors in the data on the estimated values of design floods of different recurrence intervals using the flood frequency analysis approach. The effect of systematic measurement errors has been studied using the observed data as well as the generated data of the three sites by introducing systematic errors to the annual peak flow sequences pertaining to i) high floods, ii) medium floods, iii) low floods ; iv) all the peak floods; and v) all events except high floods. The effects of random measurement errors on design flood estimates have also been examined.