Flood Peak Estimates from Limited At‐Site Historic Data

Abstract

Estimates of peak flood discharge on ungauged watersheds are often determined from regional relationships. This approach, however, ignores sitespecific information that might be available from sources such as historic high water marks, etc. A systematic procedure to estimate a given quantile flood, employing the limited at‐site historic data, is developed for the ungauged watershed. In this procedure, the estimates of discharge based upon historic sitespecific evidence are combined with independent regional estimates by inverse variance weighting, to obtain weighted estimates at the ungauged site. The variance of flow determined by the regional method is shown to be adequately represented by the square of the standard error of regression. The variance of flow determined on the basis of historic evidence is represented by the square of the regional standard error determined for gauged sites within the study region. The methodology based on inverse variance weighting appears to give better results than either independent estimate by itself, when compared to the station values at gauged sites within the study region. The simple inverse variance weighting procedure has also been shown to offer computational advantages over the optimal variance weighting scheme, with no sacrifice in accuracy.