Statistical analysis for predicting low-occurrence strong wind speeds at the pedestrian level in an actual urban case

Abstract

This study validated the applicability of two statistical methods (3W and 2W methods) for estimating low-occurrence strong wind speeds (LOSWSs) of an actual urban case (Case-TPU) based on the three-parameter and two-parameter Weibull distributions, respectively. The 3W method necessitates the mean, standard deviation, and skewness of wind speed, while the 2W method only requires the mean and standard deviation. The large-eddy simulation (LES) results of Case-TPU were validated using the wind tunnel experimental results and applied to the statistical analysis of the 3W and 2W methods. An analysis was conducted to assess the suitability of the 3W and 2W for fitting wind speed data obtained from the LES. The results showed that even though the 3W had a better fit than the 2W owing to the location parameter of 3W enhancing its adaptability, both methods exhibited a satisfactory fit to the wind speed data at the majority of points. Furthermore, the 3W and 2W methods showed high estimation accuracies of LOSWSs. For LOSWSs with exceedance probabilities of q = 10%, 1%, and 0.1%, the 3W method produced relative errors within 10%, while the 2W method yielded relative errors within 20%. The high estimation accuracy of LOSWSs for Case-TPU proved the robustness of the 3W and 2W methods for the urban area case with complicated building layout.