B1973
Title: An improved method for extreme sea level estimation
Authors: Eleanor DArcy - Lancaster University (United Kingdom) [presenting]
Jonathan Tawn - Lancaster University (United Kingdom)
Abstract: Storm surges, combined with high tide, pose an increasing risk to coastline communities. To reduce their impact, accurate return level estimates are required to provide information for coastal defence engineering. Early methods modelled sea levels directly, but this ignores the known tidal component and results were biased due to stationary assumptions being violated. Instead, we filter out waves and remove the mean sea level trend, to consider peak tide and skew surge as the only components of sea levels. Skew surges are stochastic and define the difference between the peak tide and maximum observed sea level within a tidal cycle. They are driven meteorologically, so they are more extreme in winter. Methods currently used in practice make several restrictive and unrealistic assumptions; our approach corrects these. We model extreme skew surges using the GPD. We capture seasonality, longer-term trends and skew surge-peak tide dependence through daily, yearly and tidal covariates in the scale and rate parameters. We also account for skew surge temporal dependence using a Gaussian copula, assuming the series follows a Markov process. Since peak tides are predictable, we choose tidal samples to reflect monthly and interannual variations. To derive a distribution for the annual maximum sea levels, we combine the distributions of the skew surge and peak tide. Our return level estimates are more accurate than those currently used in the UK for coastal flood defence design.
