253 – Contributed Oral Poster Presentations: Section on Risk Analysis
Modeling Extreme Hurricane Damage in the United States
Sponsor: Section on Risk Analysis
Keywords: Extreme value theory, Generalized Pareto distribution, Mean Excess Plot, Return period, Threshold, Bootstrap Sampling
Asim Kumer Dey
Lamar University
Kumer Pial Das
Lamar University
Hurricanes are one of the costliest natural disasters in the United States. After more than two decades of relatively little hurricane activity, the past decade saw a heightened hurricane activity and more than $200 billion in damage in 2004 and 2005. This study analyzes US hurricane damages from 1900 to 2012. Based on the analysis we propose an extreme value model for predicting extreme hurricane damage. Finally, a simulated hurricane series are generated by bootstrap sampling to quantify the uncertainty in the inference of extreme return levels of hurricane losses.