Bayesian MCMC flood frequency analysis with reconstructed paleofloods
Résumé
The choice of an acceptable and cost-effective solution for the design of hydraulic structures depends upon the estimation of quantiles for different characteristics of floods, usually peak flows. However, series of observed floods have a limited length, and quantile estimates associated to high return periods are subject to large uncertainties. In this study, we propose a novel and complementary approach which aims at combining reconstructed peak flows of the Rhône river with the series of observations. These reconstructions cover the last 350 years and are obtained using measurements of sediment volumes in the Bourget Lake1 (at Aix-les-Bains, France).
A Bayesian approach is adopted in order to properly treat the non-systematic nature of the reconstructed flow data, as well as the uncertainties related to the reconstruction method. While this methodology has already been applied to historical floods, similar applications to paleofloods are absent and promising. We first estimate extreme quantiles using direct measurements of peak flows (1853-2004). Direct observations are then combined to the sedimentary information (1650-2013). The comparison of the resulting estimates demonstrates the added value of the sedimentary information, and its impact on the associated uncertainties. In particular, 4 major floods which have occurred during the 18th century are very unlikely in comparison to the floods observed during the last 150 years.