Some important considerations in the analysis of annual maximum sea-levels

Abstract

Recommendations are made for an improved method of analysing annual maximum sea-levels involving confidence intervals which should provide the coastal engineer with a clearer indication of the realibility of sea-level estimates. It is suggested that assessing the effect of trends in the observed annual maxima and estimating the variability of the estimates should be treated separately. The reason for this is that the variability depends on the parameters of the underlying extreme-value distribution which can vary considerably from port to port. This is illustrated for several ports using confidence intervals for estimated sea-levels for specific return periods and series lengths. It is also shown that the interpretation of an alternative procedure involving 10-year cumulative subsets is far from straight-forward.