Bayesian Data-Worth Analysis for Estimating Unsaturated Soil Hydraulic Parameters.

To make reasonable predictions of water movement in unsaturated soil, it is essential to accurately estimate the soil hydraulic conductivity and water retention parameters. While it is a common practice to estimate soil hydraulic parameters from observed hydraulic state variables (negative pressure head and water content), heat transfer in unsaturated soil is also closely coupled with water movement. Thus, it is of interest to incorporate soil temperature data in estimating soil hydraulic parameters. In this study, a data-worth analysis of different types of measurements (i.e., pressure head, water content and temperature) was employed to identify unknown soil hydraulic parameters in a rigorous Bayesian framework. The relative entropy was used to quantify the information content of different types of measurements. Our numerical simulations using the estimated soil hydraulic conductivity and water retention parameters showed that the most informative measurement was pressure head, followed by temperature data, with water content providing the least information. Our simulations further revealed that the estimation of soil hydraulic parameters can be improved by jointly assimilate the temperature data with the pressure head or water content measurements, which implies that the temperature data carried non-redundant information in estimating the hydraulic parameters.