Evaluation of reference evapotranspiration estimation methods for the municipalities of Paulistana and Piripiri in the State of Piauí, northeastern Brazil

Abstract

Evapotranspiration is a major component for agricultural water management. Direct measurements of crop evapotranspiration (ETc) are very difficult and requires a lot of time and cost. ETc can be obtained by multiplying the specific crop coefficient (Kc) by reference evapotranspiration (ETo). According to the literature, there are a large number of empirical methods to estimate ETo from different climatic variables. Thus, the objective of this study was to evaluate the performance of Makkink, Hamon, Jensen-Haise, Benavides-Lopez, Camargo, Priestley-Taylor, Linacre, Radiação FAO-24, and Hargreaves-Samani methods for the estimation of monthly reference evapotranspiration (ETo) compared with the FAO-56 Penman-Monteith method, in the municipalities of Paulistana and Piripiri in the State of Piauí, northeastern Brazil. The work was developed based on maximum and minimum air temperatures, relative air humidity, wind speed, and sunshine duration of conventional weather stations over the period from 01 January 2005 to 31 December 2014, obtained from the National Institute of Meteorology. To compare the values of monthly ETo by FAO-56 PM and the other methods the following goodness-of-fit criteria were used: mean error (ME, in mm day^-1), standard error of estimation (SEE, in mm day^-1), mean percentage error (MPE, in %), ratio between the mean (RM, in %), coefficient of determination (R²), Willmott’s index of agreement (id), and Camargo and Sentelhas index of performance (c > 0.85 is great, 0.76 ≤ c ≤ 0.85 is very good, 0.66 ≤ c ≤ 0.75 is good, 0.61 ≤ c ≤ 0.65 is median, 0.51 ≤ c ≤ 0.60 is not good, 0.41 ≤ c ≤ 0.50 is bad, and c ≤ 0.40 is terrible). For the municipality of Piripiri, the Hargreaves-Samani and Camargo methods showed the best performance in the monthly ETo estimation according to the statistical indices (ME of 0.41 and 0.13 mm day^-1, SEE of 0.45 and 0.33 mm day^-1, MPE of 8.29 and 2.56%, RM of 108.29 and 102.56%, R² of 0.96 and 0.89, id of 0.89 and 0.91, and c of 0.86 – great and 0.81 – very good, respectively). In the municipality of Paulistana, the Hargreaves-Samani and Jensen-Haise methods showed the best performance (ME of -0.14 and 0.80 mm day^-1, SEE of 0.45 and 1.03 mm day^-1, MPE of -2.76 and 15.53%, RM of 97.24 and 115.53%, R² of 0.80 and 0.96, id of 0.88 and 0.74, and c of 0.70 – good and 0.71 – good, respectively).