An evaluation and comparison of wind speeds from different reanalysis models in the context of wind energy – the influence of topography

Wind speed data from reanalysis models is used for a wide range of industrial and research applications in the wind energy sector. These range from resource estimation in the development phase to planning of the future (wind) energy system and the integration of wind energy into electrical grids. Validation of reanalysis models for offshore conditions far away from the coast is relatively straightforward, for which very good agreement between reanalysis models and wind speed measurements has been demonstrated. In contrast, validation in heterogeneous and complex terrain is much more difficult.

This study uses an extensive measurement dataset with high-quality wind speed measurements at heights relevant for modern wind energy applications (100 – 200 m) to evaluate and compare a set of reanalysis models, i.e. ERA5, COSMO-REA6, CERRA, and NEWA. The evaluation dataset in this study comprises approximately 100 lidar and mast measurements, mainly carried out for wind park developments, distributed over Germany and covering a wide range of topographic conditions.

The analysis focuses on identifying local topographic effects and indicators that influence the quality of the agreement between the reanalysis models and the measurements (i.e. correlation) as well as systematic deviations (biases). The investigated topographic effects include orographic exposure, land /forest cover, and sub-grid scale orography. In a preliminary analysis based on 44 measurement stations, a clear correlation between the bias in the reanalysis models (compared to the measurements) and the orographic exposure of the measurement location could be demonstrated. At measurement locations with ground heights exceeding the height of the grid cell of the reanalysis model, the model data underestimated the observed wind speeds. The opposite could be observed at measurement location below grid height.

These findings are especially important for wind energy applications, as wind farm developments tend to concentrate on areas with high exposure and specific land use types. Moreover, the observed relationships with local geography potentially provide the possibility to empirically correct and downscale wind speeds from reanalysis models. This allows to better represent the wind resource available for wind parks and to estimate the uncertainties based on local geographical conditions.