Copyright 2005 Royal Meteorological Society. International Journal of Climatology 25: 735–752 (2005). Published online 26 April 2005 in Wiley InterScience ( DOI: 10.1002/joc.1151


Wind speeds over the Baltic significantly increased over the second half of the 20th century (C20th), with the majority of the increase being focused on the upper quartile of the wind speed distribution and in the southwest of the region. These changes have potentially profound implications for the wind energy resource. For example, based on the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data it is shown that, owing to this non-stationarity, using the normalization period of 1987–98 to determine the wind resource (as in the Danish wind index) leads to overestimation of the wind energy index (and hence the wind energy resource) in western Denmark relative to 1958–2001 by approximately 10%. To address whether the increased prevalence of high wind speeds at the end of the C20th will be maintained in the future, we provide a first prognosis of annual wind indices from the HadCM3 coupled atmosphere–ocean general circulation model. The results suggest the 21st century (C21st) will be similar to the 1958–2001 period with respect to the wind energy density, but that the northeastern Baltic will exhibit slightly higher wind energy indices over the course of the C21st relative to the latter half of the C20th, whereas the southwest of the Baltic exhibits some evidence of declining wind indices towards the end of the C21st. These changes may indicate a tendency in HadCM3 towards more northerly tracking of mid-latitude cyclones in the future, possibly due to evolution of the North Atlantic oscillation. As a caveat to this finding, it should be noted that the NCEP–NCAR and European Centre for Medium-Range Weather Forecasts reanalysis data sets and HadCM3 simulations, although exhibiting commonalities during the period of overlap, differ quantitatively in terms of the spatial fields and empirical cumulative probability distributions at individual grid cells.