Climate Variability and Change—Hydrological Impacts (Proceedings of the Fifth FRIEND World Conference held at Havana, Cuba, November 2006), IAHS Publ. 308, 2006, 226–231.

---

Bayesian priors based on regional information: application to regional flood frequency analysis

MATHIEU RIBATET^1,2, ERIC SAUQUET², JEAN-MICHEL GRESILLON² & TAHA B. M. J. OUARDA¹

1 INRS-ETE, University of Quebec, 490 de la Couronne, Quebec G1K 9A9, Canada

mathieu.ribatet@ete.inrs.ca

2 Cemagref, 3 bis quai Chauveau CP 220, F-69336 Lyon Cedex 9, France

Abstract Flood frequency analysis is usually based on the fitting of an extreme value distribution to the series of local streamflow. However, when the at-site time series is short, frequency analysis results become unreliable. In this work, a regional Bayesian model to estimate flood quantile from a few years of stream flow data is proposed. This model is less restrictive than the Index Flood model while preserving the formalism of “homogeneous regions”. Performance of the proposed model is assessed on a set of French gauging stations. The accuracy of quantile estimates as a function of homogeneity level of the pooling group is also analysed. Results indicate that the regional Bayesian model outperforms the Index Flood model and local estimators. Furthermore, it seems that working with relatively large and homogeneous regions may lead to more accurate results than working with smaller and highly homogeneous regions.

Key words regional frequency analysis; Bayesian inference; Index Flood; MCMC

---