Comparison of Flood Frequency Analysis Methods for Ungauged Catchments in France

Abstract

Abstract:The objective of flood frequency analysis (FFA) is to associate flood intensity with a probability of exceedance. Many methods are currently employed for this, ranging from statistical distribution fitting to simulation approaches. In many cases the site of interest is actually ungauged, and a regionalisation scheme has to be associated with the FFA method, leading to a multiplication of the number of possible methods available. This paper presents the results of a wide-range comparison of FFA methods from statistical and simulation families associated with different regionalisation schemes based on regression, or spatial or physical proximity. The methods are applied to a set of 1535 French catchments, and a k-fold cross-validation procedure is used to consider the ungauged configuration. The results suggest that FFA from the statistical family largely relies on the regionalisation step, whereas the simulation-based method is more stable regarding regionalisation. This conclusion emphasises the difficulty of the regionalisation process. The results are also contrasted depending on the type of climate: the Mediterranean catchments tend to aggravate the differences between the methods.

Delineation of flood-prone areas using modified topographic index for a river basin

“…All extreme value distribution functions are considered. Among them Gumbel is found to fit well (Odry & Arnaud 2017;Onen & Bagatur 2017;Kobierska et al 2018). Accordingly, the Gumbel method is used for the comparison of flood magnitude and area under flood inundation.…”

Review: Continuous simulation modelling for design flood estimation – a South African perspective and recommendations

“…For this reason stochastic rainfall generators are commonly used with CS models, as well as many other rainfall-runoff simulation-based approaches, to generate or extended rainfall records (e.g. Beven, 1987;Smithers et al, 2000;Clothier and Pegram, 2002;Frezghi, 2005;Sivapalan et al, 2005;Rogger et al, 2012;Sharma et al, 2016;Arnaud et al, 2017;Odry and Arnaud, 2017). Similarly, rainfall disaggregation models, or simple disaggregation techniques, are also commonly applied to generate short duration data from longer time-steps, e.g., daily to hourly (Calver et al, 2005;Knoesen, 2005;Knoesen and Smithers, 2009;Grimaldi et al, 2012;Haberlandt and Radtke, 2014;Nathan and Ling, 2016).…”

“…This includes the GR4H model, as implemented in ARR Revision Project 8, which is an hourly version of the GR4J daily rainfall-runoff model developed by Perrin et al (2003) with 2 storages and 4 parameters (Van Esse et al, 2013;Bennett et al, 2014). Although not a CSM approach, another noteworthy event-based simulation approach -SHYREG, developed over several years, also by IRSTEA, has recently been established as a national DFE method in France (Arnaud et al, 2016;Arnaud et al, 2017;Odry and Arnaud, 2017). Arnaud et al (2016);Arnaud et al (2017) and Odry and Arnaud (2017) have compared the SHYREG approach to several other FFA methods applied in the country and highlight several advantages of the approach, including greater stability of the regionalised rainfall-runoff model parameter for different regionalisation methods, in comparison to a regional FFA for example.…”

“…The SHYREG approach utilises a stochastic hourly rainfall generator to simulate extended rainfall time-series, on an event basis, at any point in France at a 1 km resolution. The rainfall-runoff model converts this rainfall into a flood quantile at the point, and these point estimates are scaled to the catchment using reduction factors (Odry and Arnaud, 2017 (Brocca et al, 2011). The model is composed of two components.…”

“…A regionalisation approach is required to estimate the model parameters at ungauged locations. Several regionalisation techniques are available such as spatial proximity and similarity pooling, a region-of-influence type approach, regression-based methods or cluster analysis (Smithers, 2012;Blöschl et al, 2013;COST, 2013;Odry and Arnaud, 2017). In the case of the ACRU model in South Africa direct calibration of model parameters is not performed (Schulze, 1995).…”