Material And Methods 2521 GPR Technique

The propagation velocity of GPR electromagnetic waves is determined by the dielectric constant or permittivity of the medium to be investigated, which on its part is a function of the water content. The dielectric constant e for water is 80, for various soil and geological materials 5 to 15, and 1 for air. Measurements of the wave propagation velocity can be taken to determine soil water contents using various functional relationships between the two quantities (e.g., Topp et al., 1980). The GPR technique is a nondestructive measurement, which is an advantage in comparison to other electromagnetic wave-based methodologies, such as the TDR technique.

In principle, all kinds of GPR measurements require one transmitting and one receiving antenna. In this study, the fixed offset (FO) method in which the distance between the two antennas is held constant was used. Although a 1 GHz antenna (MALA Geoscience, Mala, Sweden) was found to be suitable for the investigated lysimeters, we preferred a 500 MHz antenna (SIR-10A from Geophysical Survey Systems Inc., Salem, NH) for the large sand tank. The measurements were recorded at twelve times during the vegetation period from March to September in the lysimeter study in order to identify absolute soil water content changes. In the sand tank study, where we were aiming at depicting the proceeding water front following single irrigation events, GPR data were acquired at 15 to 20 min intervals.

25.2.1.1 Migration of Radargrams

The processing steps applied to the acquired GPR data were first counting back the applied field gain curve, then normalizing in space to 1 cm trace interval, and afterward reapplying a realistic gain curve taking the attenuation of electromagnetic waves in a midelectroconductive environment into account. No filtering was applied to the data beneath a low-cut filter during acquisition for signal stability.

In order to obtain depth profiles, the GPR signal originally acquired in the time domain has to be converted to a depth section. This procedure called "migration" is a common process in seismic data processing. During the migration, the data are not simply rescaled, but, for example, effects of the acquisition in the time domain such as diffraction hyperbolas at edges or isolated bodies are taken into analysis. In this case, a classic Stolt migration algorithm was used to convert the time section into a depth section (Stolt, 1978). The procedure is based on a basic velocity model that could not be computed directly from the acquired database. However, from the independent TDR measurements in eight depths, it was possible to set up a bedded velocity model neglecting strong variations in the horizontal plane. Some artifacts may be introduced using such a simplified model, but these are not critical compared with analysis on those arising from time-domain data.

25.2.1.2 Normalized Maximum Reflection Amplitude Analysis (NMRA)

The energy transmitted into the soil will be, in part, reflected when contrasts in soil permittivity (e.g., soil water contents) are encountered. Figure 25.1 shows a theoretical derived GPR radargram with a fixed antenna separation (single offset) for a layered soil profile with soil water content differences. The first reflection from the top is the direct wave, and the second indicates the different reflection amplitudes with changing water content.

The maximum reflection amplitude is computed by (a) selecting an interval of interest and (b) calculating the absolute value of the maximal amplitude. The values can be presented in normalized form by dividing the selected value with the maximum value of the considered interval (NMRA). In this study, NMRA values derived from one soil depth (from migrated radargrams) were statistically analyzed and compared to soil water content distributions.

25.2.2 Experimental Setup

Two setups were used to generate the experimental database. In both cases, a 2 m thick sand body was investigated. Although a lysimeter stand with a natural vegetation cover subjected to atmospheric conditions served for absolute water content investigation on a seasonal scale, we used a sheltered large physical sand tank (Schmalz et al., 2003) for temporal and spatial high-resolution measurements.

2nd Layer Properties Identifier 0 1 2 3 4 5 6 7

2nd Layer Properties Identifier 0 1 2 3 4 5 6 7

FIGURE 25.1 Idealized ground-penetrating radar (GPR) transect measured with a fixed separation as obtained from a layered soil profile with contrasting soil water contents. The soil water content in the second soil layer is increasing from the left to the right. The first reflection from the top is the direct or air wave, and the second indicates the different reflection amplitudes with changing water content. The normalized maximum reflection amplitude (NMRA) is computed by (i) selection of an interval of interest, (ii) picking the absolute value of the maximal amplitude, and (iii) normalizing the value with the maximum value of the considered transect and soil depth.

FIGURE 25.1 Idealized ground-penetrating radar (GPR) transect measured with a fixed separation as obtained from a layered soil profile with contrasting soil water contents. The soil water content in the second soil layer is increasing from the left to the right. The first reflection from the top is the direct or air wave, and the second indicates the different reflection amplitudes with changing water content. The normalized maximum reflection amplitude (NMRA) is computed by (i) selection of an interval of interest, (ii) picking the absolute value of the maximal amplitude, and (iii) normalizing the value with the maximum value of the considered transect and soil depth.

25.2.2.1 Lysimeter Stand

The lysimeter station consists of twelve lysimeter cylinders, each with an area of 1 m2 (0 = 113 cm). The two cylinders considered in this study stand on a platform scale that is connected to an electronic scale. The changes in water content can be measured with an accuracy of 100 g, which corresponds to 0.1 mm of the area or 0.000067 m3/m3 of the volume of the lysimeter. The temporal solution of the measurement is 15 min. An artificial groundwater table was established at 2.10 m with the help of a suction system.

25.2.2.2 large physical Sand Tank

Infiltration experiments were carried out using a large physical sand model having a base of 5 m x 3 m and a surface of 6 m x 5.6 m and containing three sloped side walls. The chosen construction with three sloped side walls resulted from statical constraints. All soil hydraulic measuring devices were installed from the vertical wall. The tank was filled with a 2 m layer of homogeneous sand (Hagrey et al., 1999). Measurements of the pressure head (two vertical tensiometer profiles) and the water content (one vertical TDR profile) were conducted at eight depths allowing the derivation of 0(h) relationships at 20, 40, 60, 100, 120, 140, 160, and 180 cm soil depth. An automatic irrigation device ensured the homogeneous distribution of the irrigation water over the soil surface. In this study, we consider an irrigation event of 287 mm applied within 14 h, which produced a total discharge of 197 mm over a 14-day period.

25.2.3 Soil Water Content Profiles

As a comparison bases for the GPR signal analysis, we numerically generated two-dimensional soil water content profiles assuming a heterogeneous soil water distribution according to results from earlier studies on homogeneous sand packages (Schmalz et al., 2003). Numerical experiments were conducted using the Hydrus-2D software package of Simunek et al. (1996) to obtain two-dimensional views of the infiltration and redistribution process. The Windows-based Hydrus-2D package solves the Richards equation for variably saturated flow numerically, assuming applicability of the van Genuchten-Mualem soil hydraulic functions.

Because of the geometric features of the sand tank, only flow in one particular cross section was considered. To run Hydrus-2D, we implemented a relatively fine numerical mesh involving 4488 nodes depicting the geometry of the sand tank. An atmospheric boundary condition accounting for infiltration and evaporation was imposed at the soil surface, whereas a seepage face was used at the bottom boundary between the sand and a gravel drainage layer.

A heterogeneous soil water distribution could be depicted by assuming a spatially noncorrelated distribution of the 8(h) and K(h) relations as measured in situ in eight depths with soil hydraulic probes. The saturated hydraulic conductivity was assessed from the grain distribution using Hazen's empirical formula (Hölting, 1996) and was found to vary from 0.3 to 0.47 mh-1 over the entire profile.

Was this article helpful?

0 0
Taming Taxes

Taming Taxes

Get All The Support And Guidance You Need To Permanently Get A Handle On Your Taxes. This Book Is One Of The Most Valuable Resources In The World When It Comes To A Guide To Home Business Taxes.

Get My Free Ebook


Post a comment