Mixture bivariate distribution of wind speed and air density for wind energy assessment

Abstract

The probabilistic structures of wind speed and air density are indispensable for the wind energy assessment. Note that the statistical patterns of these parameters might involve various modes due to the complex climate systems. The present study focuses on the bivariate distribution of wind speed and air density with a mixture copula model, each component of which was constructed using a Weibull distribution for the wind speed, a lognormal distribution for the air density, and a Gaussian copula function for the description of the dependency structure. The optimum component number and maximum likelihood estimate of the mixture model were determined using the Bayesian information criterion and expectation-maximization algorithm, respectively. Reanalysis data in six locations planned for wind resource projections along the coast of China were utilized to verify the mixture bivariate distribution model. The results suggested that the proposed model is capable of capturing the multimodal characteristics of the joint distribution for both the total and seasonal data adequately, allowing for a thorough assessment of offshore wind energy potential. It turned out that the assumption of constant air density cannot be justified in predicting the energy production of a wind farm and an accurate description of its probabilistic structure is required.