A comparison of approaches for evapotranspiration estimation

Abstract

Flux (sensible heat, ground heat and net radiation) and meteorological data measured from the Lucky Hills and Kendall subwatersheds during the winter period (41 days) were used to obtain the daily actual evapotranspiration and to compare the daily evapotranspiration estimated from energy blance approach (Penman-model), radiation approach (Priestley-Taylor model), mass transfer approach (adjusted Dalton model), single source approach (adjusted Penman-Monteins model), and temperature approach (Thornthwaite model). The numericals and graphical tests were performed to compare the measured and estimated evapotranspiration. For the numerical tests, root mean square error (RMSE) and mean absolute error (MAE) were applied. For graphical tests, time series plots were used. The study results indicate that Penman-Monteith single source approach does not provide the satisfactory estimation of actual evapotranspiration. For the Lucky Hills subwatershed, the RMSE of adjusted Penman-Monteith model is 2.031, and the MAE is 1.645. For the Kendall subwatershed, the root mean square error of adjusted Penman-Monteith model is 2.132, and the mean absolute error is 1.851. Adjusted Dalton model gives the better estimation of evapotranspiration compared with adjusted Penman-Monteith model for the Kendall subwatershed (RMSE=1.984; MAE=1.744). The evapotranspiraton rate estimated from Penman potential ET model (energy balance approach) is much higher than those estimated from other potential ET models such as Priestley-Taylor model and Thornthwaite model. The evapotranspiraton rate estimated from Thornthwaite potential evapotranspiration model is even much lower than actual evapotranspiration.