Comparison of various statistical methods for estimating extreme wind speed at the pedestrian level in idealized and actual urban areas

Pedestrian-level wind environment is a key aspect in sustainable urban planning. Especially, evaluating extreme wind speed in urban areas is important, since strong winds increase the risk of accidents and injuries of pedestrians. Although several statistical methods were developed to estimate extreme wind speed using statistics in previous studies, their robustness and accuracy remain unclear for the applications on the entire pedestrian level of urban areas. This study validated the effectiveness of the Weibull distribution (KB, 2W, and 3W) method and the Gram–Charlier series (GCS) method in four scenarios: three idealized urban cases and one actual urban case, all at the pedestrian level. The validation was performed using time-series data from the large-eddy simulations (LESs). It was found that for the extreme wind speed with exceedance probability q = 10%, 1%, and 0.1%, the adaptive GCS method (GCS-A) and 3W method (the third-order Weibull distribution method) were found to be robust and accurate. The sixth-order GCS method (GCS-6th) shows high estimation accuracy at most regions, except several regions with large high-order statistics. The outcomes of this study are anticipated to make a valuable contribution to urban planning and design, particularly in the context of wind environment considerations.