Extreme Winds Simulated From Short‐Period Records

Abstract

In order to utilize a limited length of wind records in providing extreme wind speeds for structural design, a stochastic model for generating long‐term annual extreme winds in a well‐behaved climate, on the basis of short period records, is developed and presented herein. Basically, this method uses historical wind data to establish Markov transition probabilities at an intended project site. These probabilities will be the guide for producing synthesized hourly wind speeds of a desired period, i.e., 100 annual extreme wind speeds would be obtained if hourly wind data for 100 yr were generated. Furthermore, if correlation is observed between the historical annual extreme winds and annual fastest mile wind speeds at a given site, then long‐term annual fastest mile wind speeds can be obtained. Applications of this model are demonstrated. The simulated 25‐yr, 50‐yr and 100‐yr wind speeds compared very favorably with those obtained by fitting Type I distribution of the largest values to the 33 historical annual fastest mile wind speeds for the Honolulu International Airport. However, the results from the proposed simulation model are less favorable for generating extreme winds for typhoon‐prone regions on the basis of short period historical data.