Estimating reference evapotranspiration with limited meteorological data

Abstract

The objective of this work was to evaluate the estimate of reference evapotranspiration (ETo) for the Southeast Region of Brazil using limited meteorological data. The FAO 56 Penman‑Monteith method (PM_p) was used as reference. Three scenarios of limited meteorological data, obtained from automatic network stations, were used for estimating ETo: standard method (PM_p), with solar radiation estimated from the balance between short and wide wavelengths (PM_Krs); standard method using vapor pressure estimated from minimum and maximum temperatures, and from air relative humidity (PM_ea); and standard method using constant wind speed (2 m s^‑1; PM_u2). ETo was also estimated by the Hargreaves‑Samani (HS) and Turc methods. The models were analyzed by statistical indicators of mean absolute deviation (MBE), relative error (RE), root mean square error (RMSE), and Willmott index (d). The PM_ea method is the best alternative to estimate ETo, followed by PM_u2, for the states of Espírito Santo and Rio de Janeiro, and PM_Krs, for the states of São Paulo and Minas Gerais. The largest errors are obtained with the Hargreaves‑Samani method, which overestimated ETo compared with PM_p, for most of the evaluated stations.