Statistics of extreme climatic actions based on the Gumbel probability distributions with an upper limit

• A reliable forecast of extreme storms is needed to assess the storm risk of structures.
• We have analysed data from 168 weather stations using the methodology PROGUMBEL.
• We tested the fit of the Gumbel distribution type III with an upper limit to the data.
• An upper limit could be derived for the annual extremes of the storm data.
• Applying the type III distribution, the partial factor of the wind load decreases.

The “Comité Européen de Normalisation” developed unified technical standards for application within the EU. The Eurocode EN 1990 is the basic document regarding the requirements for structural reliability. The code is based on the partial factor method. The Gumbel Type I probability distribution is applied to predict extreme values of climatic actions. The authors extend this approach to the generalised Gumbel distribution allowing including an upper limit of the climatic variable. For estimating the distribution parameters, a statistical methodology called ProGumbel has been developed. The implications for statistical analyses of wind speeds and determination of design wind speeds are discussed.