Estimating extreme sea levels combining systematic observed skew surges and historic record sea levels

The estimation of the sea levels with a high return period is crucial for coastal planning and the assessment of coastal flooding risk. Coastal facilities are designed to very low probabilities of failure and hence the design values are affected by significant uncertainties. Some recent coastal floods due to exceptional surges suggest that the design performed with the current statistical approaches may sometimes be significantly underestimated. This presentation is a contribution to the use of historical observations to improve the estimation of extreme sea levels. Historic records consist in observed major sea level values. The corresponding skew surges may be estimated but the exhaustiveness of historical skew surges, which is an essential criterion for an unbiased statistical inference cannot be guaranteed. . Indeed, Extreme skew surges can easily remain unnoticed if they occur at low or moderate high tide and do not generate extreme sea levels. This study proposes to combine, in a single Bayesian inference procedure, series of measured skew surges for the recent period and extreme sea levels for the historic period. The method is tested on four sites (tide gauges) located on the French Atlantic and Channel coasts. The proposed method appears to provide unbiased quantile estimates and to be more reliable than previously proposed approaches to include historic records in coastal sea level or surge statistical analyses.