Evaluation of the Wind Speed Time Series from Climate Reanalysis for Brazil

Abstract

A major limitation of regional climate studies in Brazil is related to the availability of meteorological datasets in terms of data quality, quantity, and its spatial density distribution. Wind speed data from climate reanalysis has been used as input in several studies for climate and hydrological modeling however the quality of these datasets over Brazilian territory remains unknown for daily datasets at spatial resolutions < 1°. In this study, the performances of 3 sets of reanalysis data (ERA5, GLDAS 2.1, and MERRA-2) are assessed in reproducing surface wind speed on a daily scale for the period of 2000-2018 using data from 521 meteorological stations as a baseline. The reanalysis's accuracy is evaluated by comparing simulated trends, mean and seasonal variability, and spatial heterogeneity of observation data. All 3 sets show several uncertainties in simulated wind speed concerning its trends and dispersion in daily intensity compared to observed series. Our results show the wind speed derived from reanalysis presents a significant increase in trends on daily wind speed throughout the Brazilian climate zones. From the results of the comparative analysis, the ERA5 dataset simulated more accurately wind speed daily trends and seasonality over the climate zones in Brazil.

Keywords
wind speed; meteorological variables; climate reanalysis; accuracy