Trends in UK extreme sea-levels: a spatial approach

Abstract

S U M M A R YThe majority of studies of long-term sea-level change have concentrated on the trend in mean sea-level which is just one constituent of the trend in extreme sea-level. By fitting a spatial model to sea-level annual maxima from 62 UK sites, extreme sea-level trend estimates a r e obtained for the entire British coastline. These estimates exhibit smooth, but significant, spatial variation which arises from a combination of eustatic extreme sea-level change a n d local vertical land movements. Once the latter effect is removed, eustatic extreme sea-level trends are found to have no significant spatial variation and t o be similar in value to trends in UK eustatic mean sea-level.

Spatial and temporal analysis of extreme sea level and storm surge events around the coastline of the UK

“…Most past studies of extreme sea levels have analyzed records from different sites individually. Studies that have assessed the spatial nature of events have typically done this by considering secondary parameters 49,50 (e.g., the shape parameter of an extreme value distribution), rather than the actual levels among sites. Studies that have compared levels across sites have tended to focus on a limited (<5) number of events 10,21,51 .…”

“…We remove the influence of MSL, to directly compare the magnitude of events throughout the period analysed. Many past studies of storm surges in the UK 49,63–66 have focused on analysing the non-tidal residual—the component that remains once the astronomical tidal component has been removed from the measured sea level record (Fig. 2).…”

Bayesian Inference for Extremes: Accounting for the Three Extremal Types

“…We believe that there is considerable additional flexibility obtained by constructing approximately independent summary statisticsq q 1 ;q q 2 ;q q 3 ð Þwhich can then be modelled nonparametrically, as opposed to jointly modelling a dependent trivariate distribution with a fully parametric model. The above discussion glosses over the clear presence of an increasing trend in the Sheerness maxima (see Figure 1) and that similar trends have been identified at neighbouring sites (Dixon and Tawn, 1992). To overcome this in the construction of p c (q) the maximum likelihood estimates, taken from Coles and Tawn (1990), are calculated with respect to the year 1990, using a linear trend for each location parameter.…”

Multivariate declustering techniques

“…Indeed, observations of a number of different physical processes observed at one site (Joe et al, 1992;Joe, 1994), or a number of summarizing features of behaviour of a single process at a particular location (Walshaw, 1991;Anderson and Turkman, 1992;Coles and Walshaw, 1994), or consecutive observations during extreme events of one process (Barnett and Lewis, 1967;Smith, 1990;Smith et al, 1997) or most indicatively a spatial process observed at a ®nite number of sites (Gumbel and Goldstein, 1964;Gumbel and Musta®, 1967;Raynal-Villasenor and Salas, 1987;Smith et al, 1990;Tawn, 1990, 1991;Dixon and Tawn, 1992) are all examples of observations leading naturally to multivariate extreme value distributions.…”