Estimation of root water uptake parameters by inverse modeling with soil water content data

In this paper we have tested the feasibility of the inverse modeling approach to derive root water uptake parameters (RWUP) from soil water content data using numerical experiments for three differently textured soils and for an optimal drying period. The RWUP of interest are the rooting depth and the bottom root length density. In a first step, a thorough sensitivity analysis was performed. This showed that soil water content dynamics is relatively insensitive to RWUP and that the sensitivity depends on the texture of the considered soil. For medium‐fine textured soil, the sensitivity is particularly low due to relatively high unsaturated hydraulic conductivity values. These ones allow a “compensating effect” to occur, i.e., vertical unsaturated water fluxes overshadowing in some way the root water uptake. In a second step, we analyzed the well‐posedness of the solution (stability and nonuniqueness) when only RWUP are optimized. For this case, the inverse problem is clearly ill‐posed except for the estimation of the rooting depth parameter for coarse and the very fine textured soils. In a third step, we addressed the case where RWUP are estimated simultaneously with additional parameters of the system (i.e., with soil hydraulic parameters). For this case, our study showed that the inverse problem is well‐posed for the coarse and very fine textured soils, allowing for the estimation of both RWUP of interest provided that a powerful global optimization algorithm is used. On the contrary, the estimation of RWUP is unfeasible for medium‐fine textured soil due to the “compensating effect” of the vertical unsaturated water flows. In conclusion, we can state that the inverse modeling approach can be applied to derive RWUP for some soils (coarse and very fine textured) and that the feasibility is strongly improved if the RWUP are simultaneously optimized with additional parameters. Nevertheless, more detailed research is needed to apply the inverse modeling approach to real cases for which additional issues are likely to be encountered such as soil heterogeneity and root dynamics.