A Multi-point Model for Annual Maximum Wind Speed via Max-stable Process

Abstract

A multi-point stochastic model for annual maximum wind speed is presented, and its consistency is confirmed through the application of the model to some areas in Japan. To estimate the damage level of the group of buildings against strong winds, it is necessary to evaluate the simultaneous wind hazard at each site. So far this is mainly done by the Monte Carlo simulation methods, usually called Typhoon simulation. These are rather simple methods and do not require the explicit stochastic modeling for annual maximum wind speed but some stochastic models for the characteristics of the typhoon itself. In this paper, instead, an explicit model for annual wind speeds at multi-point is proposed via the multivariate extreme value distribution and, its natural extension to a process, i.e. the max-stable process. The Explicit probabilistic wind hazard model at multi-point is especially useful for optimal design for a group of facilities or a system spatially spread in a region as well as estimations of insurance premium of a portfolio. The spatial effects of typhoons are directly considered in the presented model so that the correspondence to the physical phenomenon is clearly observed.