Assessment of Peak Flow Scaling and Its Effect on Flood Quantile Estimation in the United Kingdom. Issue 4 (6th April 2021)

• Record Type: Journal Article
• Authors: Formetta, Giuseppe; Over, Thomas; Stewart, Elizabeth
• Is Part Of: Water resources research. Volume 57:Issue 4(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-04-06
• Subjects: peak flow quantile estimation -- prediction in ungauged basin -- spatial scaling of streamflow; Hydrology -- Periodicals; 333.91
• DOI: 10.1029/2020WR028076
• Languages: English
• ISSNs: 0043-1397

Abstract

Abstract: Regional flood frequency analysis (RFFA) methods are essential tools to assess flood hazard and plan interventions for its mitigation. They are used to estimate flood quantiles when the at‐site record of streamflow data is not available or limited. One commonly used RFFA method is the index flood method (IFM), which assumes that peak floods satisfy the simple scaling hypothesis. In this work we present an integrated approach to assess the spatial scaling behavior of floods in the United Kingdom (UK) for 540 catchments, where the IFM is currently used operationally. This assessment employs product moments, probability weighted moments, and quantile analysis, and is applied to two different types of "hydrologically homogeneous" UK regions: geographical regions as defined in the Flood Studies Report (NERC, 1975) and pooling‐groups as defined in the updated Flood Estimation Handbook (FEH; Institute of Hydrology, 1999). To understand which variables play a significant role in the flood‐peak generating mechanism, the assessment approach considers scaling not only of drainage area alone but also of other hydro‐geomorphological variables. Results provided by the different methodologies consistently showed that only part (ranging from 30% to 70%) of the peak flow variability is explained by drainage area alone; this fraction increases (up to 80%–95%) when multiple regression is used. Supported by the peak flow spatial scaling assessment, we compared the proposed approach for peak flow quantile estimation with the current FEH method in ungauged catchments. The quantile regression method based on the pooling‐group outperforms the current FEH‐ungauged method, providing a 14% relative improvement in root mean square error over the entire country. Key Points: Multivariate regressions (MVR) better represent the scaling behavior of the UK peak flow compared to univariate regression Only part (from 30% to 70%) of the peak flow variability is explained by drainage area alone; it increases (up to 95%) when the MVR is used The presented methodology provides a 14% relative improvement in root mean square error over the entire UK country compared to the current Flood Estimation Handbook ungauged method