Hydraulic and transport properties of the plant-soil system estimated by inverse modelling

Abstract

Modelling soil water flow and solute transport under field conditions requires the knowledge of many parameters that are difficult to determine directly. Values determined on small isolated samples in the laboratory are often not representative of field situations. We investigated the applicability of inverse modelling to a soil-plant system in lysimeter experiments. We also tested whether parameters obtained from one experiment could be applied to another with the same soil. In a lysimeter planted with young trees, we first did a multistep drainage experiment and then a long-term bromide tracer experiment with atmospheric boundary conditions at the soil surface. To estimate the unsaturated hydraulic properties, we linked the inverse program SUFI (Sequential Uncertainty FItting) to the flow and transport model HYDRUS5. A comparison of several scenarios showed that the resulting values of parameters depended strongly on the data used for calibration and the formulation of the objective function. The results suggested that inverse modelling could be used to identify important processes. Inversely obtained parameters gave better predictions for a second experiment when more variables were considered in the objective function and when the range of hydraulic conditions was wider. Furthermore, with retention curves directly fitted to measured water retention data we achieved acceptable results. Despite some limitations, the inverse approach was found to be a sound and useful procedure for estimating parameters of a complex system involving water uptake by roots, solute transport and unsaturated flow.