On the Evaluation of Wind Power from Short Wind Records

Abstract

A method to estimate the meteorological mean wind power of a given site and year based upon data of only three months is presented. It is based on the minimization of the rms errors between observations and a representation by means of empirical orthogonal functions. These functions are calculated from a matrix formed by annual series of wind speed data from stations belonging to the same region. A numerical experiment has been performed using wind series of 20 stations in the U.S. Midwest and East. The method allowed the recovery of both the mean wind speed value and its variance with a small error. The meteorological mean wind power of the year was calculated using the Weibull-II and Rayleigh probability distributions. Using the Weibull-II distribution, the method outlined above, and only 13-week observations, estimations of the true meteorological wind power were obtained with a 6% average relative error when only 11 central stations were computed. The experiment showed that the method is also potentially useful to calculate monthly wind power.