Validation of the representativeness of wind speed time series obtained from reanalysis data for Brazilian territory

In recent years, consideration of reanalysis data has gained space and importance globally as a promising alternative for climate studies that suffer from an absence or scarcity of data. Wind speed time series can be obtained from these bases for various purposes, such as inferring the potential of sites for wind power generation. These projections can be useful to analyze the feasibility of building new wind farms and the formation of historical series of wind power generation to enable better planning for existing facilities. Therefore, reliable wind speed time series is essential to obtain accurate projections. The reanalysis databases are characterized for having extended historical series. On the other hand, one of their drawbacks is the arrangement of data in a grid with low spatial resolution, so not cover all points on the Earth's surface. This study aims to verify whether the wind speed time series of the MERRA-2 dataset can represent the values at points in Brazilian territory. For this purpose, we examine the use of strategies for interpolation, extrapolation, and bias correction to overcome these limits and obtain time series that better approximate the most probable values, as suggested in the specialized literature. The results are compared with historic series recorded in Brazil to evaluate the method's applicability and indicate whether the data extracted from MERRA-2, after treatment, provide a relevant representation. This study contributes to the literature by (i) measuring the quality of MERRA-2 data to represent high spatial resolution locations in Brazil, (ii) evaluating the impacts of the natural variability of these wind speed series on the results, (iii) describing new bias correction approaches, (iv) verifying the impact of the temporal and spatial scales utilized on the results, and (v) assessing the results by comparing wind speeds. (c) 2022 Elsevier Ltd. All rights reserved.