An integrated data fusion approach to use geophysical measurements in hydrological models
Final Report Abstract
Geophysical measurements are a valuable source of information for the parameterization of hydrological models. Traditionally, relevant information on hydrological properties and/or state variables is obtained in a sequential approach from geophysical measurements: the geophysical survey data are inverted first, and the information thus obtained is used within the hydrological model. The aim of this project is to further develop an alternative so-called coupled hydrogeophysical inversion approach to use geophysical data in hydrological models that overcomes some of the limitations of the sequential approach. In this coupled inversion approach, geophysical measurements are directly included in the hydrological inverse problem by coupling a forward model of the geophysical measurements with a hydrological model and minimizing the difference between modelled and observed data by perturbing the relevant hydrological flow parameters. Whereas previous studies relied on synthetic modelling data to evaluate coupled inversion, this study focussed on the evaluation of coupled inversion using actual geophysical data. In particular, coupled inversion frameworks were developed for the estimation of effective soil hydraulic properties from timelapse electrical resistivity tomography, time domain reflectometry, and streaming potential measurements. The soil hydraulic property estimates obtained with coupled inversion compared favourably with independent reference values in all our case studies, thus confirming the power of coupled hydrogeophysical inversion. A prerequisite for coupled inversion is a good description of the hydrological processes that determine the time-lapse changes in the geophysical monitoring data. In case of considerable model structural uncertainty, the advantage of the coupled inversion is reduced and alternative interpretation methods might be more appropriate. A Bayesian formulation of coupled inversion was also developed, and it was found to be useful to assess the posterior uncertainty of the estimated model parameters and the model predictions. This kind of information is of great value and can be used to analyze the worth of different experimental data sources and for experimental design questions (e.g. optimization of amount and timing of measurements). Finally, coupled inversion was established as a natural framework for the fusion of multiple types of geophysical and hydrological data. However, the trade-offs between fitting different data types need to be carefully considered for hydrological interpretation. The remaining challenges are mainly associated with the computational burden associated with coupled hydrogeophysical inversion frameworks, in particular when using a Bayesian approach in combination with three-dimensional hydrological and geophysical models. This is also the reason that the envisioned coupled inversion for spatially variable model parameters was not achieved. Nevertheless, the work in this project has helped to firmly establish coupled inversion as a viable approach to interpret time-lapse geophysical data in a hydrological context.
Publications
-
2009. Coupled hydrogeophysical parameter estimation using a sequential Bayesian approach. Hydrology and Earth System Sciences, 14: 545–556
Rings, J., Huisman, J.A., and Vereecken, H.
-
2009. Improved extraction of hydrologic information from geophysical data through coupled hydrogeophysical inversion. Water Resources Research, 46: W00D40
Hinnell, A.C., Ferré, T.P.A., Vrugt, J.A., Huisman, J.A., Moysey, S., Rings, J. and Kowalsky, M.B.
-
2010. Hydraulic properties of a model dike from coupled Bayesian and multi-criteria hydrogeophysical inversion. Journal of Hydrology, 380: 62-73
Huisman, J.A., Rings, J., Vrugt, J.A., Sorg, J. and Vereecken, H.
-
2011. Feasibility of sequential and coupled inversion of TDR data to infer soil hydraulic parameters under falling head infiltration. Soil Science Society of America Journal, 75(3): 775-786
Mboh, C.M., Huisman, J.A., and Vereecken, H.,
