Bayesian inverse modelling of in situ soil water dynamics: using prior information about the soil hydraulic properties
Abstract. In situ observations of soil water state variables under natural boundary conditions are often used to estimate field-scale soil hydraulic properties. However, many contributions to the soil hydrological literature have demonstrated that the information content of such data is insufficient to reliably estimate all the soil hydraulic parameters. In this case study, we tested whether prior information about the soil hydraulic properties could help improve the identifiability of the van Genuchten-Mualem (VGM) parameters. Three different prior distributions with increasing complexity were formulated using the ROSETTA pedotransfer function (PTF) with input data that constitutes basic soil information and is readily available in most vadose zone studies. The inverse problem was posed in a formal Bayesian framework and solved using Markov chain Monte Carlo (MCMC) simulation with the DiffeRential Evolution Adaptive Metropolis (DREAM) algorithm. Synthetic and real-world soil water content data were used to illustrate our approach. The results of this study corroborate and explicate findings previously reported in the literature. Indeed, soil water content data alone contained insufficient information to reasonably constrain all VGM parameters. The identifiability of these soil hydraulic parameters was substantially improved when an informative prior distribution was used with detailed knowledge of the correlation structure among the respective VGM parameters. A biased prior did not distort the results, which inspires confidence in the robustness and effectiveness of the presented method. The Bayesian framework presented in this study can be applied to a wide range of vadose zone studies and provides a blueprint for the use of prior information in inverse modelling of soil hydraulic properties at various spatial scales.