Characteristics of measured extreme thunderstorm near-surface wind gusts in the United States

Franklin T. Lombardo, Alexander S. Zickar

Research output: Contribution to journalArticlepeer-review


Winds generated from thunderstorms have been shown to be important in wind engineering. However, the collection of data for these events has been limited. In this study, thunderstorm wind speeds from airport surface stations (i.e., ASOS) in the United States (U.S.) were analyzed. Overall the analysis showed significant directional and regional preferences. The results revealed that a large majority of thunderstorm wind speeds >75 mph (33.4 m/s) had a westerly component and had a preference toward the central United States. Annual maximum wind speeds >29 mph (12.4 m/s) from over 400 ASOS stations were employed in a cluster analysis using k-means. For both non-directional and directional analyses, the statistical character of the wind speeds is different for different wind speed magnitudes. Probability distributions fit to the data showed best fit to a generalized extreme value distribution with an average shape parameter of approximately −0.1, suggesting a Gumbel fit would be conservative. Statistical analysis also revealed the significance of small sample sizes and the presence of ‘outliers’, that when analyzed, had a significant effect on the parameters of the fitted probability distribution. A simple data-driven methodology is proposed to utilize both site-specific and regional data in extreme wind analysis.

Original languageEnglish (US)
Article number103961
JournalJournal of Wind Engineering and Industrial Aerodynamics
StatePublished - Oct 2019


  • Clustering
  • Extreme value analysis
  • Thunderstorm
  • Wind directionality

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Renewable Energy, Sustainability and the Environment
  • Mechanical Engineering


Dive into the research topics of 'Characteristics of measured extreme thunderstorm near-surface wind gusts in the United States'. Together they form a unique fingerprint.

Cite this