PROBABILISTIC IDENTIFICATION OF UNSATURATED SOIL HYDRAULIC PROPERTIES FROM SINGLE RING INFILTRATION EXPERIMENT

Abstract

: Knowledge of soil hydraulic properties is essential for modelling hydrodynamic phenomenons in soils under variably saturated conditions. This contribution focuses specifically on the properties of the top layer of soil which fundamentally influences the rainfall-runoff process, but the laboratory experimental proce-dures for the top-soil are very limited due to e.g. vegetation disturbing the experimental samples. The Richards equation describing the soil water flow process is here parameterised by the Mualem-van Genuchten model. The identification of the soil hydraulic parameters of this model is based on an in situ measurement of the single ring infiltration experiment. The experimental data and a priori information about parameters’ values are combined according to Bayes’ rule in order to obtain an updated probabilistic description of unknown parameters. The Bayesian inference is realized by the Markov chain Carlo method based on a less time-consuming polynomial approximation of the Mualem-van Genuchten model.