Evaluation of macroscopic root water uptake models using lysimeter data

Three macroscopic root water uptake models (Molz-Remson, Feddes, and Selim-Iskandar) have been widely used in modeling water flow in soil-plant systems. In this article, we use evapotranspiration data and soil water content data obtained from lysimeter measurements and root distribution data obtained in a nearby wheat field to evaluate the accuracy of these root water uptake models in predicting the soil water content profiles in the winter wheat field. To account for the role of root distribution, we modified the Molz-Remson model and the Selim-Iskandar model with the Feddes reduction function, and the Feddes model with the root length density. These models, with and without modifications, were then individually incorporated into the soil water flow equations as a sink term. The flow equations were solved numerically with the measured evapotranspiration data as input, and the predicted soil water content profiles were compared with the measured profiles to evaluate the validity of the root water uptake models. The comparison showed that: (1) the average deviation of the Molz-Remson (linear version) model, the Feddes model, and the Selim-Iskandar model could reach as high as 25,70, and the non-linear version of the Molz-Remson model performed slightly better, with an average deviation of 13%; (2) modifications made to the Molz-Remson models and the Selim-Iskandar model did not achieve any improvement in the model predictions, but the modified Feddes model significantly improved the prediction accuracy, reducing the average deviation to 5.6%.