Assessment of empirical methods for estimation of reference evapotranspiration in the Brazilian Savannah

Abstract

Evapotranspiration can be sufficiently estimated when meteorological data are available to implement robust models such as Penman-Monteith (PM). However, due to data scarcity, alternative approaches are necessary. In this context, this study aims to compare the reference evapotranspiration (ETo) obtained from the PM standard method with eight empirical equations to identify the simplest method that can be alternative to the reference method (Penman Monteith method) for ten places in state of Goiás (located in west-central Brazil, Brazilian Savanna). To estimate the ETo, air temperature and relative humidity air, wind speed, sunshine and solar radiation data, which were obtained from the data platform National Institute of Meteorology and the Meteorological and Hydrological System of the State of Goiás, were used. For comparison of empirical methods with PM standard method, we used the following statistical indicators: slope and intercept coefficients (β0 and β1) of regressions equations, the coefficient of determination (r²), Pearson's correlation (r), mean bias error (MBE), root mean square error (RMSE) concordance index refined (dr) and performance index (Pi). Our results indicated that the Turc method is the best option for the state of Goiás when meteorological data are not suffeciently available to use the standard PM method. On the other hand, the method of Romanenko did not present acceptable performance in nine of the ten studied localities. Therefore, its use is advised only in the municipality of the Itumbiara. Among evaluated methods the Hargreaves-Samani method is the best alternative, when there is only air temperature data.