Evaluating renewable-energy-relevant parameters of COSMO-REA6 by comparison with satellite data, station observations and other reanalyses

Energy-meteorology applications have recently started to make intensive use of reanalysis data. Due to their higher spatial resolution, regional reanalyses (RRA) are especially suitable for a wide range of applications concerning wind- and solar-energy generation. This study evaluates the wind speed from various European reanalyses by comparing reanalysis wind fiel s against near-surface wind speed from more than 200 German station observations. For the evaluation of solar irradiance, reanalysis fiel s are compared with two satellite datasets, SARAH-2 (Surface Solar Radiation Data Set - Heliosat - Edition 2) and HelioMont. We show that, for the 10-m wind speed, the maximum peak correlation between observations and regional reanalyses is reached at a weekly timescale. The investigated European RRA have significantly higher correlations than the global reanalysis ERA-Interim. The biases of the reanalyses strongly depend on the location of interest, the Numerical Weather Prediction (NWP) model used, and the wind speed. The lowest mean absolute error of 0.69 m/s, averaged for stations below heights of 500 m, is achieved by COSMO-REA6 (Consortium of small scale modelling - Reanalysis - 6 km). Using a decomposition scheme, the mean square error (MSE) with respect to tower measurements shows, on the one hand, that the error increases with higher temporal scales due to decreasing correlation. On the other hand, on monthly and longer time scales, the MSE is dominated by the bias between the observations and reanalysis. The comparison of the solar irradiance of the reanalyses with satellite datasets reveals a negative bias for the reanalysis systems based on the NWP model COSMO, while other European regional reanalysis products show an overestimation of solar irradiance. Over the Alpine region, all RRA show high positive anomalies against the SARAH-2 dataset, which vanish when using a second high-quality satellite reference dataset HelioMont, which is especially useful for snow-covered regions.