Effects of parameter estimation method and sample size in metocean design conditions

The accurate estimation of extreme values for metocean parameters (wind speed, wave height, etc.) plays a crucial role in marine renewable energy industry and in coastal and offshore engineering applications. The most fundamental approach for extreme value analysis is the annual maxima approach that is directly related to the Generalized Extreme Value (GEV) distribution. Since the performance of the GEV parameter estimation methods is dependent on both the method and the available sample size, the exploration of these issues is analytically performed. Firstly, a simulation study is implemented based on the Maximum Likelihood (ML), the L-moments (LMOM), the Elemental Percentile and the Maximum Product of Spacings methods for different sample sizes. It is concluded that the ML should not be taken for granted since LMOM method performs better in many respects. Afterwards, both methods are applied for the estimation of the GEV parameters of wind speed annual maxima series. LMOM method provided the best fits for the overwhelming majority of cases considered. Finally, the 50- and 100-year wind speed return levels are estimated. With respect to the relative confidence intervals of the return level estimates, no solid conclusions can be drawn since there is lack of a systematic behaviour.