Towards global validation of wind power simulations: A multi-country assessment of wind power simulation from MERRA-2 and ERA-5 reanalyses bias-corrected with the global wind atlas

Reanalysis data are widely used for simulating renewable energy and in particular wind power generation. While MERRA-2 has been a de-facto standard in many studies for a long time, the newer ERA5reanalysis recently gained importance. Here, both datasets were used to simulate wind power generation and evaluate their quality in terms of correlations and error measures compared to historical data of wind power generation. Due to their coarse resolution, reanalyses are known to fail to represent local climatic conditions adequately. Hence, mean bias correction was applied with two versions of the Global Wind Atlas (GWA) to the reanalysis data and the quality of the resulting simulations was assessed. Potential users of these datasets can also benefit from our analysis of the impact of spatial and temporal aggregation on indicators of simulation quality. We also assessed regions which differ significantly in terms of the prevailing climate, some of which are underrepresented in similar studies: the US, Brazil, South-Africa, and New Zealand. Our principal findings are threefold. (i) ERA5 outperforms MERRA-2 in terms of the assessed error measures. (ii) Bias-correction with GWA2 does not improve simulation quality substantially, while bias-correction with GWA3 is detrimental. (iii) Temporal aggregation increases correlations and reduces errors, while spatial aggregation does so consistently only when comparing very fine and very coarse granularities. (c) 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Automated summary

Reanalysis data, including MERRA-2 and ERA5, were used to simulate wind power generation. ERA5 outperforms MERRA-2 in terms of error measures, while bias-correction with GWA2 did not substantially improve simulation quality. Temporal aggregation increases correlations and reduces errors, while spatial aggregation only consistently improves simulation quality with very fine and very coarse granularities.