Estimation of an index flood using data transfer in the UK

Abstract

An important part of the statistical procedure for flood frequency analysis in the UK outlined in the Flood Estimation Handbook (FEH) is concerned with estimation of an index flood at an ungauged site. This is carried out through application of a multivariate regression model linking the index flood, defined as the median annual maximum flood, to a set of catchment descriptors. The FEH then emphasises the importance of data transfer from nearby gauged (donor) sites, or from catchments considered to be hydrologically similar but located anywhere in the UK (analogue sites). This paper considers the suggested methods for estimating the index flood at ungauged sites and develops a new and improved data transfer scheme. A study of 728 gauged catchments located in the UK found that the new data transfer method performs better than both using the FEH regression model only and the FEH data transfer method.