Application of stochastic and deterministic modeling to hurricane wind risk assessment

Assessments of hurricane wind risk must necessarily focus on the most intense events, which cause a disproportionate share of hurricane-related wind damage. Yet there are too few such events in the historical record to infer robust probabilities from the data alone. One partial solution to this problem is to generate a very large number of synthetic hurricane tracks, whose most important statistical properties conform to those of historical tracks. Although this is an important step forward, it remains to estimate the evolution of the storm wind field along the tracks. Our approach begins by drawing randomly fro the historical space-time genesis PDF, and then generating tracks using two independent methods: A Markov Chain technique, using statistics from historical tracks, and a beta-and-advection model (BAMS) driven by general circulation statistics derived from NCEP reanalysis data. Once the tracks are generated, a coupled ocean-atmosphere numerical model (CHIPS) is then used to estimate the evolution of the storm's wind field along each track. For large samples, such as all the satellite-era North Atlantic storms, the intensity statistics generated by our method can be checked against historical storm data. We will show applications of your technique for wind risk assessment at particular locations and for the present and also for hypothetical future climates.