Estimation of return levels with long return periods for extreme sea levels in a time-varying framework

Abstract

AbstractAt nuclear power plants, risk analysis concerning environmental extremes is crucial. Based on historical data, estimation of return levels is usually performed. For long return periods, a problem is that the related uncertainties of the return levels often get large. Moreover, models need to take into account possible effects of climate change. In this paper, extreme sea levels close to Swedish nuclear power plants are considered. Non-stationary statistical models and the related results of conditional prediction during a typical time horizon of an infrastructure are studied. The influences of parameters in extreme-value distributions and the lengths of observation records are discussed. The effect of land uplift in parts of the Baltic Sea is seen.