Empirical Simulation of Future Hurricane Storm Histories as a Tool in Engineering and Economic Analysis

Borgman, Leon; Miller, Martin; Butler, Lee; Reinhard, Robin; ,

Empirical Simulation of Future Hurricane Storm Histories as a Tool in Engineering and Economic Analysis

by Leon Borgman, Univ of Wyoming, Laramie, United States,
Martin Miller, Univ of Wyoming, Laramie, United States,
Lee Butler, Univ of Wyoming, Laramie, United States,
Robin Reinhard, Univ of Wyoming, Laramie, United States,

Document Type: Proceeding Paper

Part of: Civil Engineering in the Oceans V

Abstract:

Although the database of past storm occurrences and their detailed histories and geometries at or near a site is essential to planning safe and cost-effective future engineering operations and to estimating economic consequences of various alternative actions, the storm history at the site has many inadequacies if it is used directly as the only environmental conditions imposed in the planning and estimation. The data at most sites is extremely limited and sparse in terms of number of storms, paths of approach, and other features. This is often particularly true for the most severe of the storms which typically are the most critical to high loss aspects. Even more importantly, the future will resemble the past only in a statistical way. The actual occurrences of number of storms, their geometries and path time histories, their severities, and so forth will certainly be different from what has occurred in the past. If engineering planning and economic analysis is only based on past storm experience, future events may produce substantially different combinations of conditions and lead to costly failure of the operation and gross errors in the economic estimation. A new procedure, labeled here empirical simulation, has been developed in a joint study between the U.S. Army Engineer Waterways Experiment Station Coastal Engineering Reswarch Center (CERC) and the USAE Mobile District which generates numerous multiyear intervals of possible future hurricane storm histories at a selected site. The ensemble of simulations are consistent with the statistics and correlations of past storm activity at the site, but allow random deviations in behavior such as is likely to develop in the future. The computer software is also designed to minimize the cost and effort required in associated finite modeling of storm surge and currents by using a carefully selected set of `training storms' which extend over the range of likely occurring storms. The predictions from the modeling of these are mathematically extended to the storms produced in the empirical simulations. The empirical simulation method is nonparametric in the sense that it does not require estimation of parameters in a selected formula for joint probabilities, but instead proceeds directly from past data to simulation of future events using techniques which grow out of recent studies of `bootstrap' methods in statistics.

Return to search