No abstract

The coupling of soil and root water fluxes at the plant scale is a particularly challenging task. Numerical three‐dimensional plant‐scale models exist that consider these soil–root interactions. The influence of the hydraulic conductivity drop at the microscopic scale and especially the effect on root water uptake is not yet assessed in such models. In this study, an analytical approach describing the hydraulic conductivity drop from the bulk soil to the soil–root interface for a three‐dimensional plant‐scale model was derived and validated by numerical means. With these tools, quantification of the local hydraulic conductivity drop with time was possible. Furthermore, the effect of the hydraulic conductivity drop on the time occurrence of plant stress was evaluated. Root water uptake was assessed, with and without considering the hydraulic conductivity drop around single roots in a three‐dimensional plant‐scale model in terms of total water uptake at the root collar under different soil and root properties. It was shown that the total root water uptake was strongly affected, especially under conditions where the radial root hydraulic conductivity, which regulates root water uptake, was larger than the soil hydraulic conductivity, which regulates water flow in the soil. These findings were backed up by numerical validation of the model using mesh refinement. Incorporation of the hydraulic conductivity drop around individual roots in a three‐dimensional plant‐scale model can solve problems with greater accuracy for larger grid resolutions, and with smaller computational times, than not considering the hydraulic conductivity drop.

To understand how water uptake locally affects and is affected by the soil water distribution, three‐dimensional soil–root models need to be developed. Nowadays, fully coupled three‐dimensional soil–root flow models at the plant scale are available that simulate water flow along water potential gradients in the soil–root continuum, but the problems that arise by the coupling of soil and root have not been investigated thoroughly. In a previous work, we introduced and numerically validated a microscopic model to be used on a coarse numerical soil grid, describing the hydraulic conductivity drop between the bulk soil and the soil–root interface within a voxel of a three‐dimensional soil–root model. In this study, the impact of the local hydraulic conductivity drop on denser root architectures and in drier soil regions was assessed. When a coarse discretization of the soil grid is used, the local hydraulic conductivity drop has a significant effect on the water potential distribution at the soil–root interface and in the xylem, especially under conditions near plant stress where the local soil conductivity is lower than the radial root conductivity regulating root water uptake. As a consequence, plant stress conditions will be reached earlier than if the local conductivity drop within a soil voxel is neglected. In comparison with a fine soil discretization, the soil water potential gradient calculated by including the local conductivity drop at a coarser discretization does not fit the soil water potential gradient resulting from the fine soil discretization. Estimation of accurate water potential gradients throughout the soil requires a fine soil discretization.

No abstract

[1] Three dimensional soil-root water transfer models require a fine soil and root discretization in order to obtain accurate results. This goes along with a considerable computational effort. One way of reducing the computational effort is the usage of grid refinement techniques. With such techniques irregular grids are obtained that combine the accuracy of a fine grid resolution with a considerable reduction in computational costs. As a consequence of plant transpiration roots take up water and large soil water potential gradients around roots are created. Especially in these regions a fine soil discretization is needed. The root spatial distribution can therefore be used for refinement of the soil grid, a priori. Simulations show that the accuracy is indeed maintained for a priori refined grids but with reduced computational costs as compared to regular fine grids. Comparison with a well recognized a posteriori error estimate strengthen these results.