Climate driven variability of wind energy over Europe

Sammanfattning: In the following report, two methods of wind downscalling are applied to reanalysed 10-m wind speed data from the ERA5 reanalysis of ECMWF, in order to derive wind speed data from 01/01/1979 till 31/12/2019 at the location of onshore wind farms from the ”World Wind Farm Database”.  Being unable to use complex downscalling methods such as Generalisation (WAsP method- ology) nor Kriging interpolation, the study has been restricted to simple downscalling methods. The physical model for wind speed was derived from the 1D-Reynold Averaged Navier-Stokes (RANS) equations. The interpolation scheme was chosen to be an IDW interpolation whose weighting function was derived from a correlation versus distance plot for geostrophic wind speed. This plot was derived from Thiessen interpolated wind speed at wind farm’s locations.  The results from an holdout cross-validation of the interpolation scheme (Thiessen, IDW) showed that the calibrated IDW performed better than Thiessen, yet with a good margin of improvement. The physical downscalling model was also assessed to provide better result than a sole interpolation, especially for capturing local effects influencing wind speed.  The resulting wind speed timeseries at the power plant locations where finally transformed into power timeseries, using the power curve of the General Electric 2.5 MW turbine. The hourly timeseries where averaged into daily timeseries using a simple averaging pattern. The final product is a database of daily mean Power timeseries for the 22010 wind farms from the ”World Wind Farm Database”, ranging from the 01/01/1979 till 31/12/2019.