Wind Speed Evaluation of MERRA-2, ERA-Interim and ERA-5 Reanalysis Data at a Wind Farm Located in Brazil

Climatological normal is a powerful tool to understand the behavior of meteorological variables like wind speed, and it is typically employed in wind resource assessments. This paper compares three different sources of long-term wind speed data at a wind farm, located in the Northeast of Brazil, with meteorological mast observations at the site from 2009 to 2018. The main goal is to determine which one better describes the wind climate at the wind farm. All long-term sources, MERRA-2, ERA-Interim and ERA-5, showed satisfactory correlations with meteorological mast data for monthly means, which were 0.91, 0.94 and 0.95, respectively. Even though the wind speed observations were systematically underestimated, all reanalysis databases represented well the seasonal variability, but were not able to capture extreme winds. Although negative bias was reduced using a linear regression, it remained in the months of austral winter in all datasets. The correlation coefficient was not improved by the vertical correction of ERA-Interim wind speed, therefore the dataset from 1000 hPa represents satisfactorily the mean wind speed behavior at the target site.