Estimating the index flood with continuous hydrological models: an application in Great Britain

Abstract

Estimating peak river discharge, a critical issue in engineering hydrology, is essential for designing and managing hydraulic infrastructure such as dams and bridges. In the UK, practitioners typically apply the Flood Estimation Handbook (FEH) statistical method which estimates the design flood as the product of a relatively frequent flow estimate (the index flood, IF) and a regional growth factor. For gauged catchments the IF is estimated from observations. For ungauged catchments it is computed through a multiple regression model. While the FEH IF method provides peak flow estimates that are statistically robust, it does not readily take into account catchment heterogeneity or effect of environmental change on river flows. This study presents a new methodology to estimate the IF at national scale using continuous simulation from a physically based hydrological model (Grid-to-Grid). The methodology is tested across Great Britain and compares well with IF estimates at 550 gauging stations (R2 = 0.91). The promising results for Great Britain support the aspiration that continuous simulation from large-scale hydrological models coupled with increasing availability of global weather and climate products, could be used to estimate design floods in regions with limited gauge data or affected by environmental change.

Benchmarking the predictive capability of hydrological models for river flow and flood peak predictions across over 1000 catchments in Great Britain

“…The GLUE framework is based on the equifinality concept that there are many different model structures and parameter sets for a given model structure which result in acceptable model simulations of observed river flow (Beven and Freer, 2001). This methodology has been widely applied to explore parameter uncertainty within hydrological modelling (Freer et al, 1996;Gao et al, 2015;Jin et al, 2010;Shen et al, 2012) and includes approaches to directly deal with observational uncertainties in the quantification of model performance (Coxon et al, 2014;Freer et al, 2004;Krueger et al, 2010;Liu et al, 2009). For every catchment and model structure, an efficiency score was calculated for each of the 10 000 Monte Carlo (MC) sampled parameter sets.…”

Benchmarking the predictive capability of hydrological models for river flow and flood peak predictions across a large-sample of catchments in Great Britain

“…Lumped and semi-distributed hydrological models, applied singularly or within nested sub-catchment networks, are used for a wide range of applications. These include water resources planning, flood/drought impact assessment, comparative analyses of catchment and model behaviour, regionalisation studies, simulations at ungauged locations, process based analyses, and climate or land-use change impact studies (see for example Coxon et al, 2014;Formetta et al, 2017;Melsen et al, 2018;Parajka et al, 2007;Perrin et al, 2008;Poncelet et al, 2017;Rojas-Serna et al, 2016;Salavati et al, 2015;van Werkhoven et al, 2008). However, model skill varies between catchments due to differing catchment characteristics such as climate, land use and topography.…”

Refining flood estimation in urbanized catchments using landscape metrics

“…For national methods of flood estimation at ungauged sites, there remains in many countries a reliance on the simplicity of empirical formulae relating the index flood to catchment characteristics (Bocchiloa et al, 2003) that include land class data to inform upon levels of imperviousness for more urbanized locations (Formetta et al, 2017). National agencies across Europe continue to employ such methods (Castellarin et al, 2012), based on regressions of index flood data to catchment characteristics in gauged basins.…”