Statistical post-processing of COSMO-REA6's wind speeds at hub heights using a diagnostic wind model and neural networks

The correct representation of wind speeds at hub height (100m above ground) is becoming more and more important with respect to the expansion of renewable energies. Since there are only a few long-term measurements at hub heights available in Europe, we rely on wind speed estimates from reanalyses. Reanalyses provide a physically consistent state of the atmospheric dynamics over long periods, but are not able to represent local effects due to their limited horizontal resolution. We perform a post-processing of the wind speeds from the regional reanalysis COSMO-REA6 in Central Europe based on a combined physical and statistical approach. The physical basis is provided by downscaling wind speeds with the help of a diagnostic wind model, which reduces the horizontal grid spacing by a factor of eight compared to COSMO-REA6 (to approx. 800m) and considers different vertical atmospheric stabilities. While the downscaled wind fields might be better in line with the orography, the data still has inherent uncertainties (e.g., fit of the COSMO-REA6 input to the orography, errors in COSMO-REA6, assumptions in the wind model) and thus may still deviate considerably from the observations.

Therefore, in a second step, a statistical correction based on various reanalysis parameters as predictors. These corrections are performed using a neural network approach as well as a generalized linear model as reference. Although only few measurements by masts or lidars are available at hub heights, a reduction of wind speed RMSE of up to 30% can be achieved depending on location. A comparison with radiosonde observations also confirms the added value of combining the physical and statistical approach in wind speed post-processing.