Roberto Mínguez

12-04-2021

In this talk an overview of extreme value statistics, with particular attention to applications in engineering and environmental variables, will be presented. We will review extreme value distributions and the more modern “threshold” approach based on generalized Pareto distributions. In addition, we will briefly describe recent developments for the analysis of environmental variables, which allows:
·        Combining different sources of information, such as reanalysis databases and instrumental records.
·        Automatic parameter selection for non-stationary generalized extreme value models.
·        Sensitivity analysis.
Finally, we will discuss new challenges on the topic, and how the new sources of information available might contribute to enhance the quality of extreme value analysis predictions.