Errors in output of hydrologic models due to errors in input potential evapotranspiration

A study was conducted to give some perspective to the importance of accurate evapotranspiration (ET) input data to hydrologic models. For the analysis, computed output from three hydrologic models was considered as the true watershed response. Variations in the form of random fluctuations and fixed biases were introduced into the potential evapotranspiration (Pet) input data of the test models. By leaving the other inputs and parameters unchanged, such an analysis shows the effect of evapotranspiration on streamflow under the regulation of the other components in the model. It is shown that a constant bias of 20% in the Pet input data has a cumulative effect and results in considerable error in the computed hydrograph peaks and recession characteristics, whereas the influence of the random error on estimated streamflow was generally not measurable for the watersheds and models studied.