Comparison of seven models for estimation of evapotranspiration and groundwater recharge using lysimeter measurement data in Germany

Abstract This study evaluates seven evapotranspiration models and their performance in water balance studies by using lysimeter measurement data at the Mönchengladbach hydrological and meteorological station in Germany. Of the seven evapotranspiration models evaluated, three models calculate actual evapotranspiration directly using the complementary relationship approach, i.e. the CRAE model of Morton, the advection–aridity (AA) model of Brutsaert and Stricker, and the GG model of Granger and Gray, and four models calculate first potential evapotranspiration and then actual evapotranspiration by considering the soil moisture condition. Two of the four potential evapotranspiration models belong to the temperature‐based category, i.e. the Thornthwaite model and the Hargreaves model, and the other two belong to the radiation‐based category, i.e. the Makkink model and the Priestley–Taylor model. The evapotranspiration calculated by the above seven models, together with precipitation, is used in the water balance model to calculate other water balance components. The results show that, for the calculation of actual evapotranspiration, the GG model and the Makkink model performed better than the other models; for the calculation of groundwater recharge using the water balance approach, the GG model and the AA models performed better; for the simulation of soil moisture content using the water balance approach, four models (GG, Thornthwaite, Makkink and Priestley–Taylor) out of the seven give equally good results. It can be concluded that the lysimeter‐measured water balance components, i.e. actual evapotranspiration, groundwater recharge, soil moisture, etc., can be predicted by the GG model and the Makkink model with good accuracy. Copyright © 2005 John Wiley & Sons, Ltd.