Coupled hydrogeophysical inversion for soil hydraulic property estimation from time-lapse geophysical data

Abstract

Good knowledge of the hydraulic properties of the vadose zone is important for understanding water flow and solute transport processes therein. This can help to promote sustainable use and mitigate anthropogenic threats to soil and water resources. The use of time-lapse geophysical data to constrain our understanding of the flow and transport properties of the vadose zone is now well recognised. Conventional use of geophysical data to estimate the hydraulic properties of the vadose zone is based on an uncoupled inversion approach including an ill-posed tomograhic inversion step which can lead to error propagation to the estimated hydraulic properties. One way of improving the accuracy of estimating soil hydraulic properties is to use a so-called coupled hydrogeophysical inversion approach. In this inversion approach, the tomograhic inversion step is avoided as geophysical measurements are directly used in the hydrological inverse problem by coupling a forward model of the geophysical measurements with a hydrological model describing the hydrologic processes under investigation. Although the potential benefits of the coupled inversion approach have been illustrated with synthetic data, there are very few applications of the approach to actual field or laboratory data. Moreover, most studies using this approach focused on electrical resistivity tomography (ERT) and ground penetration radar (GPR), and the usefulness of this inversion approach remains to be explored for a range of other geophysical methods. Although coupled hydrogeophysical inversion frameworks are flexible enough for the integration of multiple hydrologic and geophysical data types, this data fusion aspect has also received less attention. Therefore, the aim of this thesis was to develop inversion frameworks for the estimation of effective subsurface hydraulic parameters from: i) the fusion of ERT and inflow data obtained under constant head infiltration in a field sandy loam, ii) SP data acquired during primary drainage of a sandy soil column, and iii) TDR data obtained under falling head infiltration into an initially dry sandy loam.<br /> Based on synthetic and actual data we showed that it is feasible to estimate three key Mualem-van Genuchten parameters (<em>α, n </em>and<em> K_s</em>), using the developed coupled hydrogeophysical inversion frameworks for ERT, SP and TDR. In all cases, the inversion results compared well with independently obtained values. With respect to the fusion of ERT and inflow data, it was observed that the success of the procedure depends on the choice of an appropriate objective function. The best results were obtained when an objective function defined as the sum of the root mean square error of both data types normalized by the standard deviation of the respective measurements was used. On the other hand, successful inversion of the SP data depended on efficient pretreatment of the measured signals prior to inversion and the availability of an adequate model for the voltage coupling coefficient at partial saturation. By comparing different models for the voltage coupling coefficient at partial saturation to the experimental data, it was observed that models that relate the voltage coupling coefficient to the relative permeability of the porous medium in addition to the saturation in water were most appropriate. In the case of inversion of TDR data, a comparison of the coupled and uncoupled inversion approaches revealed that the coupled inversion approach is more practical and less uncertain. Particularly it was observed that coupled hydrogeophysical inversion enables simultaneous monitoring of ponding depth and water infiltration, which avoids the laborious task of manually measuring the ponding depths and can thus enable rapid estimation of the soil hydraulic parameters for multiple locations through automatic measurements of ponded infiltration for multiple rings through TDR multiplexing.<br /> Future studies should focus on using the coupled hydrogeophysical inversion approach to estimate spatially varying hydraulic properties which are more characteristic of the vadose zone. At the expense of a higher computational cost, better estimates of parameter uncertainties can be obtained with the use of MCMC algorithms that provide posterior probability distributions of the inverted parameters.