Localism: New endorsement of an old paradigm for flood frequency estimation

Design flood estimates in ungauged UK catchments are routinely obtained using the improved Flood Estimation Handbook (FEH) statistical procedure, and are based on a combination of regional methods and the national dataset of AMAX event available in the HiFlows-UK database. In this paper a relatively simple assessment of the uncertainty associated with FEH design events in ungauged sites will be presented. The assessment is based on a comparison, across a large number of gauged sites, the differences between design flood estimates obtained as-if a site is ungauged and the corresponding at-site estimates obtained by fitting a distribution directly to the available AMAX data at each site.
The results are provided for a range of return periods, and it was found that the level of uncertainty of a design flood in an ungauged catchment with a return period of 100 years correspond to a factorial standard error value of 1.54. Using data transfer method for one or more nearby gauged donor catchments, it is shown that the factorial standard error associated with the generalised method can be reduced considerably by considering local data. Finally, results are presented to show that the information content in local data can be utilised to compensate for the lack of hydrologically relevant information contained in the
available catchment descriptors. This result is potentially important when flood frequency estimates are required in regions outside the UK where only a limited number of catchment descriptors are available.