Using a set of skewed distributions, we fit and analyzed wind speed data including the most common models used in previous studies. Finding suitable distributions for modeling is crucial since wind speed data generates valuable information about wind energy and power. The goal of this study is to demonstrate a framework for modeling wind speed data using different distributions. Since wind speed data exhibits skewed and bimodal characteristics, we concentrated on modeling with flexible-skewed and mixture distributions. Suitability of each model is assessed by fitting two real data sets from the National Buoy Center (NBC) and data from the Minnesota Department of Natural Resources (Minnesota DNR) to the distributions. All model parameters were derived using maximum likelihood estimation (MLE) method. Accuracy of each distribution is tested using Kolmogorov-Smirnov (K-S) and R2 goodness-of-fit tests. Our findings showed that the Weibull, Gamma, and skew-normal distributions can model wind speed data accurately. In addition, mixture models such as Gamma-Weibull and GEV-Lognormal showed an exceptional fit due to bimodal features of one of the NBC data. Furthermore, we explored some popular estimation methods for the three-parameter Weibull and Skew Normal (SN) distributions. All estimators were compared based on their MSE obtained from Monte Carlo. Our MSE results showed that the method of weighted least squares performed the best in both distribution, followed by method of maximum likelihood and method of least squares.
