Author(s): Panagiota Galiatsatou; Panayotis Prinos; Yannis Krestenitis
Linked Author(s): Panagiotis Prinos
Keywords: Pressure Data; Extreme Events; Dependence Measures χ and χ; Coefficient of Tail Dependence η; Point Process Representation; Components R and W; Joint Probabilities; Gumbel-McF adden and Husler-Reiss Models
Abstract: In the present paper, an estimation of dependence between extreme atmospheric pressure data at five locations of the Aegean sea is performed. Dependence measures χ and χ, as well as the coefficient of tail dependence η, are defined to give a summary of the correlations among data at different stations. A classical point process representation seems to be appropriate for the modelling of pressure data. The process requires an examination of the radial and angular components R and W of the data, focusing on the distribution of W for large values of R. Finally, joint probabilities of extreme events are evaluated using the Gumbel-McFadden and the Husler-Reiss bivariate threshold models.
Year: 2005