Monteiro, Ana Flávia Martins, Martins, Fabrina Bolzan, Torres, Roger Rodrigues, de Almeida, Vitor Hugo Marrafon, Abreu, Marcel Carvalho, and Mattos, Enrique Vieira
All methods for estimating evapotranspiration (ETo) require accurate and complete meteorological datasets. However, the common lack of such datasets in Brazil, as well as the definition of the method that best represents the spatiotemporal pattern of ETo, are the main challenges to assess and mitigate the effects of climate variability (natural or due to anthropogenic climate change) in the Brazilian agricultural production systems. In this sense, this work aims to assess the spatiotemporal pattern of ETo, identify, and select among twenty-nine methods the one that presents the best performance in estimating ETo for different regions of Brazil using a high-resolution gridded weather dataset (GWD). In this study, performance is evaluated by comparing the ETo results obtained through the different methods to that estimated by the Penman–Monteith method. The weather variables used were near surface air temperature (maximum and minimum), relative humidity, wind speed at 2 m, global solar radiation, ETo, and sea level pressure in a daily basis from 1980 to 2017. Through principal component analysis (PCA), the behavior of ETo was mainly influenced by the global solar radiation, maximum air temperature, and relative humidity. For this reason, the performance of the methods varied across the Brazilian regions and seasons. The Turc and Abtew methods showed the best performance in estimating daily ETo, with lower RMSE (~ 0.5 mm day−1) and MAPE (~ 12%) and higher c-index values (~ 0.75), with slight advantage of Turc method, for all Brazilian regions and seasons. Also, the ETo estimation by Turc and Abtew using the GWD dataset showed a good agreement with Penman–Monteith method. Finally, the Hargreaves, Penman Original, and Stephens Stewart methods stood out for the Brazilian Northeast region (mean RMSE of 0.7 mm day−1, mean MAPE of 14%, mean c-index of 0.7), in areas that presents predominantly arid and semiarid climate conditions. [ABSTRACT FROM AUTHOR]