Electromagnetic Inversion of GPR Signals and Subsequent Hydrodynamic Inversion to Estimate Effective Vadose Zone Hydraulic Properties

methods, intensive efforts have been undertaken to supplement the scarcity of hydrogeological data with densely We combine electromagnetic inversion of ground penetrating radar sampled geophysical data (Beres and Haeni, 1991; Ru(GPR) signals with hydrodynamic inverse modeling to identify the effective soil hydraulic properties of a sand in laboratory conditions. bin et al., 1992; Hubbard et al., 1997; Hubbard and RuGround penetrating radar provides soil moisture time series that are bin, 2000; Gloaguen et al., 2001). subsequently used as input in the hydrodynamic inverse procedure. In particular, GPR has been used increasingly to reThe technique relies on an ultrawide band (UWB) stepped frequency motely identify soil stratigraphy (Davis and Annan, 1989; continuous wave (SFCW) radar combined with an off-ground monoKung and Lu, 1993; Boll et al., 1996), to locate the water static transverse electromagnetic (TEM) horn antenna. Ground penetable (Nakashima et al., 2001), to follow wetting front trating radar signal forward modeling is based on the exact solution movement (Vellidis et al., 1990), to measure soil water of the three-dimensional Maxwell equations for describing free wave content (Greaves et al., 1996; van Overmeeren et al., propagation and on linear systems in series and parallel for describing 1997; Weiler et al., 1998; Huisman et al., 2001; Rucker wave propagation in the antenna. Water flow in the sand is described by and Ferré, 2003; Grote et al., 2003), to assist in subsurthe one-dimensional Richards equation using the Mualem–van Genuchten parameterization. Both model inversions are formulated by the face hydraulic parameter identification (Hubbard et al., classical least-squares problem and are performed iteratively using 1997; Gloaguen et al., 2001), to assess soil salinity (al advanced global optimization techniques. Compared with time doHagrey and Müller, 2000), and to support the monitormain reflectometry (TDR), results demonstrated the appropriateness ing of contaminants (Brewster and Annan, 1994; Daraof the GPR integrated approach to measure soil moisture remotely. yan et al., 1998; Yoder et al., 2001). An excellent review In particular, the approach was found to be less sensitive to the inof GPR principles and history was given by Annan herent small-scale heterogeneities. Hydrodynamic inversion of soil (2002), and reviews of its application for measuring soil moisture data led to hydraulic parameters agreeing reasonably well water content in particular were given by Davis and with direct measurements. The observed discrepancies were attributed Annan (2002) and Huisman et al. (2003). to the different characterization scales and samples. The overall inteGround penetrating radar data can be utilized as ingrated approach offers great promise to map the effective hydraulic properties of the shallow subsurface at a high spatial resolution. formation on the state of the flow system to infer the soil hydraulic properties using hydrodynamic inverse modeling methods (Šimunek and van Genuchten, 1996; Durner et al., 1999; Romano and Santini, 1999; Hughson C of vadose zone hydraulic propand Yeh, 2000; Hopmans et al., 2002). With these metherties remains a major challenge in the soil and hyods, state variables of the dynamic flow system are comdrologic sciences. New appropriate methods are needed bined with a validated hydrodynamic model and an for many agricultural and environmental engineering appropriate optimization algorithm to estimate the soil applications involving water flow and solute transport hydraulic parameters. The inversion is based on the opmodeling. It has become evident that modeling detailed timization of an objective function that represents the spatial distributions of water and solutes in the heterogeerror between the measured response of the soil system neous subsurface requires extensive site characterizaand the simulated response with the given model subject tion for determining the spatial distribution of the soil to a trial parameter set. Among existing hydrodynamic hydraulic properties, including the water retention curve inverse modeling procedures, the method proposed by and the hydraulic conductivity function (Yeh, 1998). CharLambot et al. (2002, 2004a) appears to be very promising acterizing this variability with conventional laboratory to perform nondestructive hydraulic characterizations or in situ downhole methods is invasive, and thus, timeof the subsurface. Indeed, it requires only soil moisture consuming, costly, and subject to a high degree of untime series gathered during a natural infiltration process certainty due to a lack of densely sampled in situ hydroas input. Soil moisture, in contrast to the conventionally logical measurements. Given the limitations of these used pressure head or flow data, is potentially obtainable from near-surface remote sensing using GPR. S. Lambot, M. Antoine, and M. Vanclooster, Department of EnvironThe objective of this study is to combine the hydrodymental Sciences and Land Use Planning, Catholic University of Lounamic inversion method of Lambot et al. (2002) with advain, Croix du Sud 2, Box 2, B-1348 Louvain-la-Neuve, Belgium; I. van den Bosch, Microwave Laboratory, Catholic University of Louvanced GPR characterization techniques to identify the vain, Place du Levant 3, B-1348 Louvain-la-Neuve, Belgium; E.C. effective hydraulic properties of a sandy soil in laboraSlob, Department of Geotechnology, Delft University of Technology, tory conditions. The method provides the basis of a promMijnbouwstraat 120, 2628 RX Delft, The Netherlands. Received 14 Jan. 2004. Special Section: Hydrogeophysics. *Corresponding author Abbreviations: GMCS, global multilevel coordinate search; GPR, ground ([email protected]). penetrating radar; MVG, Mualem–van Genuchten model; NMS, Nelder–Mead simplex; SFCW, stepped frequency continuous wave; Published in Vadose Zone Journal 3:1072–1081 (2004). © Soil Science Society of America TDR, time domain reflectometry; TEM, transverse electromagnetic; UWB, ultrawide band; VNA, vector network analyzer. 677 S. Segoe Rd., Madison, WI 53711 USA