Use of Flood Seasonality in Pooling‐Group Formation and Quantile Estimation: An Application in Great Britain. Issue 2 (22nd February 2018)

• Record Type: Journal Article
• Authors: Formetta, Giuseppe; Bell, Victoria; Stewart, Elizabeth
• Is Part Of: Water resources research. Volume 54:Issue 2(2018)
• Page Start: 1127
• Page End: 1145
• Publication Date: 2018-02-22
• Subjects: flood seasonality -- flood frequency analysis -- regionalization -- hydrological modeling; Hydrology -- Periodicals; 333.91
• DOI: 10.1002/2017WR021623
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Regional flood frequency analysis is one of the most commonly applied methods for estimating extreme flood events at ungauged sites or locations with short measurement records. It is based on: (i) the definition of a homogeneous group (pooling‐group) of catchments, and on (ii) the use of the pooling‐group data to estimate flood quantiles. Although many methods to define a pooling‐group (pooling schemes, PS) are based on catchment physiographic similarity measures, in the last decade methods based on flood seasonality similarity have been contemplated. In this paper, two seasonality‐based PS are proposed and tested both in terms of the homogeneity of the pooling‐groups they generate and in terms of the accuracy in estimating extreme flood events. The method has been applied in 420 catchments in Great Britain (considered as both gauged and ungauged) and compared against the current Flood Estimation Handbook (FEH) PS. Results for gauged sites show that, compared to the current PS, the seasonality‐based PS performs better both in terms of homogeneity of the pooling‐group and in terms of the accuracy of flood quantile estimates. For ungauged locations, a national‐scale hydrological model has been used for the first time to quantify flood seasonality. Results show that in 75% of the tested locations the seasonality‐based PS provides an improvement in the accuracy of the flood quantile estimates. The remaining 25% were located in highly urbanized, groundwater‐dependent catchments. The promising results support the aspiration that large‐scale hydrological models complement traditional methods for estimating design floods. Key Points: We quantified the effect of flood seasonality on the estimation of floods with a given exceedance probability The new methodology has been applied and verified in Great Britain on 420 catchments considered in turn as gauged and ungauged The proposed method outperforms the current method for flood frequency estimation in UK in terms of accuracy in gauged and in ungauged sites