Abstract
The probability distribution of wind speed data over the world's oceans is studied using a two-parameter Weibull distribution. The parameters are estimated following a linearized least-squares approach. The seasonal and latitudinal variation are described. A bootstrap statistical stability criterion is developed to select the appropriate method to estimate the Weibull parameters A and C. The method with the most stable estimate of parameters gives acceptable goodness-of-fit values. The Kolmogorov–Smirnov test also shows that the distribution adequately fits the data.
The seasonal and latitudinal variations are presented using Hovmöller diagrams of the Weibull parameters. In general, these diagrams showed a seasonal change in fair agreement with other independent estimates of wind speed statistics. The results are more reliable in the Northern Hemisphere because more adequate data are available. The uneven geographical distribution and the scarcity of data at high latitudes and the Southern Hemisphere do not permit precise determination of Weibull statistics and remain unsolved problems.
