HARGREAVES-SAMANI EQUATION CALIBRATED IN DIFFERENT TIME BASES FOR SETE LAGOAS, MG, BRAZIL

Abstract

The FAO Penman-Monteith method (FAO-PM) for determination of the reference evapotranspiration (ETo) may have its application restricted by lack of one or more wheather data. One of the alternative methods with lowest data requirement is the Hargreaves-Samani (HS). This study aimed to evaluate the performance of the HS method on its original form and after calibration on annual, semiannual, quarterly and monthly basis, compared to the FAO-PM method. For this, it was used a time series of 86 years of daily weather data obtained from a conventional meteorological station of INMET in Sete Lagoas, MG, Brazil. It was calibrated the coefficients and exponent of the HS equation by minimizing the mean absolute error. Afterwards the analysis of descriptive statistics was performed. To test the feasibility of comparative tests and to evaluate the fit of the probability distributions of the daily values of ETo, it was applied the Kolmogorov-Smirnov and Mann-Whitney test, respectively, at 5% significance level. Calibration of HS method removed the trend of systematic overestimation of daily ETo values as compared to FAO-PM. There was no significant difference between the probability distributions of daily ETo values calculated by the HS method calibrated at different time basis.