Comparative analysis of Weibull parameter estimation methods for wind resource assessment: A case study in Mexico

Wind energy, a key renewable source, requires an accurate assessment of its resources, where the Weibull distribution stands out for its ability to model wind variability. This paper examines eleven methods for estimating Weibull parameters: Graphical method (GM), Maximum likelihood method (MLM), Empirical method of Justus (EMJ), Energy pattern factor method (EPFM), Alternative maximum likelihood method (AMLM), Modified maximum likelihood method (MMLM), Generalized method of moments (GMM), Empirical method of Lysen (EML), method of moment –1 (MOM–1), method of moment –2 (MOM–2) and method of moment –3 (MOM–3). To compare the performance of these methods, three goodness-of-fit tests and two probability error metrics were employed: Anderson–Darling, Kolmogorov–Smirnov, Chi-square, Root Mean Square Error, and determination coefficient. The analysis was applied to wind speed data recorded from seven different locations in Mexico. Despite the absence of a consistently superior Weibull distribution parameter estimation method across all scenarios, the analysis reveals that MOM–3, MOM–1, and GM generally performed better in meeting the various evaluation criteria across the datasets from the seven locations.