Determination of Parameters for Bimodal Hydraulic Functions by Inverse Modeling

Abstract

AbstractTo determine the hydraulic properties of soils, transient flow experiments on soil samples and their evaluation by parameter identification methods have become common. In the past, the feasibility of this method has been investigated for soils exclusively with unimodal poresize distributions. In this work we investigated, both from a theoretical and from an experimental point of view, whether hydraulic properties of structured soils with bimodal pore‐size characteristics can be identified by inverse simulation. Multistep outflow experiments were simulated for three hypothetical soils with different degrees of bimodality. The outflow data thus created were then used in the inverse procedure, both in the original form and with an added random error. An analysis of sensitivity coefficients for the model parameters showed that if the bimodality of the pore system is well developed, the parameters of the bimodal hydraulic functions are uncorrelated and can be identified. This was confirmed by inverse optimization runs with simultaneous optimization of up to six parameters, which consistently converged from different initial values to the true parameter values, indicating uniqueness of the inverse problem. Attempts to fit the bimodal outflow data with simulations using unimodal hydraulic functions were not successful, as expressed by systematic disagreement between fitted and “observed” data and nonuniqueness of the inverse solution. We validated these theoretical results with experimental data from an undisturbed sandy forest soil, which showed that it is possible to determine the retention parameters of bimodal hydraulic properties by inverse modeling of multistep outflow experiments.