Improving the FEH statistical procedures for flood frequency estimation: Science Report: SC050050

T. R. Kjeldsen, D. A. Jones, A. C. Bayliss

Research output: Book/ReportCommissioned report


This report recommends changes to the procedures contained in the Flood Estimation Handbook (FEH), which have been adopted as standard practice by the principal bodies engaged in flood frequency estimation in the UK and, in particular, by the Environment Agency. These procedures provide estimates of the flows that will occur in rivers on moderately rare occasions: flow values that have an exceedance probability in any given year of 50 per cent (a 2-year return period) to 1 per cent (a 100-year return period), or even more rare. In the majority of cases where such estimates are required, the locations affected will be ungauged and too far from established river gauging stations to provide data records that can be immediately transferred. The changes recommended arise, in part, because the HiFlows-UK project has led to the creation of a much-improved database of systematically recorded flood data. Not only are the data records now much longer than those used previously but the HiFlows-UK project put substantial effort into the quality control and assessment of the whole data-set. This means that the data available for analysis have been dramatically improved. Another influence on the renewed procedures has been feedback from users of the FEH, both informal and formal. Without substantially changing the overall framework of the methodology, most technical details of the method have been updated to improve the performance of the procedure. The updates include significant improvements to the theoretical statistical framework underlying the method. In addition, it has been possible to consider some new descriptors of catchment topography and local climate that have been proposed since the FEH study. In particular, a new descriptor that measures floodplain extent has been devised and is now included in the improved procedures. This report is largely a technical description of the studies that have led to the new recommendations. The folllowing are the key improvements. • A new regression model for estimating the median annual maximum flood (QMED) at ungauged catchments (Chapter 4). • An improved procedure for the use of donor catchments for estimation of QMED at ungauged catchments (Chapter 5). • An improved procedure for formation of pooling groups and estimation of pooled growth curves (Chapter 6). Flood estimates produced by the new procedures can be substantially different from those produced using the original FEH procedures. On taking the catchments whose data have been analysed as typical examples, and treating them as if they were ungauged, the ratios of the new estimates to the FEH estimates indicate the following changes. • The changes in QMED range from 0.55 to 2.01, with half being greater than 1.15 (25 per cent of the ratios are less than 1.00, and 25 per cent are greater than 1.24). • For floods with an annual probability of exceedance of 1 per cent (the 1 per cent flood), the changes range from 0.48 to 2.24, with half being greater than 1.14 (25 per cent of the ratios are less than 0.97 and 25 per cent are greater than 1.32). For both QMED and the 1 per cent flood, the new procedure produced lower estimates than the FEH in the East of England, whereas increases in both quantities were generally observed in West England, Wales and Scotland.
Original languageEnglish
Place of PublicationBristol, U. K.
PublisherEuropean Environment Agency (EEA)
Commissioning bodyThe Environment Agency
Number of pages137
ISBN (Print)9781844329205
Publication statusPublished - 2008


Dive into the research topics of 'Improving the FEH statistical procedures for flood frequency estimation: Science Report: SC050050'. Together they form a unique fingerprint.

Cite this