Extreme value analysis for gridded data

The Risk and Impact Analysis Group of Geoscience Australia has been developing models to assess the hazard and risk produced by a number of natural phenomena. This paper describes a model to assess severe wind hazard over a region rather than at a recording station. The model integrates three sub-models: a statistical model that calculates return periods for the event using extreme value distributions; a model to extract and process wind speeds from a high-resolution (regional) climate model; and a Monte Carlo simulation model to generate wind gust speeds from mean wind speeds. Large scale high resolution gridded data requires a fast, efficient way to calculate wind hazard. A computer-based algorithm to achieve this aim is presented in this paper. To illustrate the methodology, wind hazard calculation over the Australian island state of Tasmania will be presented.