Use of multi-parameter distributions for offshore wind speed modeling: The Johnson SB distribution

Abstract

Weibull distribution is widely suggested for modeling the behavior of offshore wind speeds. However, it has been often proved to be inadequate, while its indiscriminate use is not justified. Thus, in order to minimize estimation errors in offshore wind energy, it is necessary to select the most appropriate distribution for the wind climate description of a specific area. In this context, the performance of several probability distributions for offshore wind speed modeling, using long-term time series from 11 buoys in Eastern Mediterranean Sea (Greek waters) and 8 buoys in Western Mediterranean Sea (Spanish coastal waters) will be assessed for the first time here. We focus on the efficiency of three bounded multi-parameter distributions: Wakeby and Kappa, that have performed very well in completely different areas of the US coasts, Atlantic and Pacific Oceans, and Johnson SB distribution, which is introduced here for the first time. It is shown that Johnson SB, Kappa and Wakeby distributions accurately describe the empirical distribution of offshore wind speed; they have better adaptability than the 3-parameter Weibull distribution and qualify as reliable and prominent candidates for the assessment of offshore wind speed in any sea area. Moreover, Johnson SB is the only distribution that suits very well for all examined cases, providing consistently fair fits with respect to all goodness-of-fit tests applied. Alternative criteria, such as the performance of the examined probability models in terms of wind power density and average wind turbine power, have also been used for evaluating the fitted wind speed distributions. In this case the results proved to be different.