Calibrated models to estimate Referensce evapotranspiration, for the city of Botucatu/Sp, in relation to the weighing lysimeter

Abstract

This work aims to adjust the Thornthwaite, Blaney and Criddle, Makkin, Camargo, Turc, Jensen and Haise, Priestley-Taylor, Linacre, Tomar and O'Toole, Hargreaves and Samani, Abtew, Thornthwaite modified by Camargo and Bruin and Stricker equations to estimate reference evapotranspiration (ETo) for the city of Botucatu-SP. For this adjustment, the model coefficients were calibrated using the General Reduced Gradient optimization technique. Model efficiency was measured in comparison to real values of evapotranspiration (lysimeter). For model validation, simple sampling technique was used. The Penman–Monteith model was also evaluated in this study. Results show that the original Thornthwaite e Penman–Monteith models, in a monthly scale, overestimate values over reference ET0 in 15.52% and 10,81%, respectively. In a daily scale, the original model from Tomar and O’Toole underestimated ETo in 52.79%. Meanwhile, the Penman–Monteith model had the highest correlation with the daily reference value (r= 0.81). The adjusted models of Turc, Camargo, Bruin and Stricker e Priestley-Taylor show difference among averages when compared to the reference value. For these models ETo estimation efficiency was verified for periods of 3, 5 and 7 days. For all periods, every model showed values near reference values, with high correlation and efficiency. For model validation, the Bruin and Stricker and Priestley-Taylor equations underestimated ETo for the daily period. For the monthly period, the model from Thornthwaite overestimated ETo in 18%. In all periods, all models presented values close to the reference value, with high correlation and efficiency.