Near-surface wind data of the global ocean derived from ship observations and models of the atmosphere are used for weather and climate predictions. Recently, wind data from the scatterometer of the first European Remote Sensing (ERS-1) satellite became available. These global observations are prepared for operational assimilation into the model of the European Centre for Medium Range Weather Forecasts (ECMWF). The remotely sensed winds have been validated in dedicated campaigns. However, data from such measurement campaigns incompletely cover the possible range of global winds. Here, characteristics of the general population distribution of global ocean wind speeds are investigated.
The dynamic properties of the atmospheric circulation suggests the separation of the global winds into extratropical and tropical subsets. In situ wind speeds from the Ocean Weather Station M (OWS M) in the North Atlantic and the Tropical Ocean Global Atmosphere (TOGA) buoys in the tropical Pacific are representative for the high and low latitudes, respectively. In situ and modelled wind speeds from extratropical latitudes are reasonably simulated by the Weibull distribution, and from the tropical Pacific by the generalized extreme value distribution. The remotely sensed wind speeds agree poorly with the corresponding empirical distributions. The comparison of frequencies of wind speeds with empirical distributions is seen to contribute to the mutual quality assessment of global wind speeds.