The fundamentals of GPR are described in Chapter 7 and practical considerations are emphasized in Case History 12.7. In the studies presented here, tree roots were the target of interest, and GPR was applied to detect electrical discontinuities between roots and the surrounding soil. A variety of frequencies (400 to 1500 MHz) have been used to locate coarse roots; 400 MHz antenna can resolve discontinuities to a depth of 2 m or more, however the smallest detectable root is 4 to 5 cm in diameter (Butnor et al., 2001; Hruska et al., 1999); 900 MHz antennas consistently resolve roots >2 to 3 cm; and under optimal conditions, while 1500 MHz center frequencies can detect roots as small as 0.5 cm, penetration is usually limited to ~50 cm (Butnor et al., 2001). Using the aforementioned frequencies, it is not possible to resolve fine roots that are commonly classified as <0.2 cm diameter. Because most tree roots are located at relatively shallow soil depths <0.5 m (Hruska et al., 1999; Jim, 2003) the high-frequency antenna is a good choice to maximize root detection. A trade-off exists where low-frequency antennas penetrate deeper in the soil, but provide lower data resolution, and higher frequencies provide greater resolution with decreased depth penetration. Ground-coupled antennas need to be in close contact with the soil surface; this is especially important with higher-frequency antennas (e.g., 1500 MHz) where air-gaps greater than a couple of centimeters will cause a deleterious loss in resolution. This presents problems in forested terrains where the presence of herbaceous vegetation, fallen trees limbs, and irregular soil surfaces impedes the travel of an antenna, requiring additional site preparation before data acquisition (Butnor et al., 2001).
The goals of this study were to test the feasibility of using GPR to quantify root mass and root distribution at two sites that share similar soils but markedly different vegetation types. One site was a scrub-oak ecosystem located on a subtropical barrier island in the northern part of the Kennedy Space Center (KSC), Merritt Island, Brevard County, Florida. Several oak species (Quercus sp.) dominate the system because of a prescribed fire ten years earlier, and they represent most of the aboveground biomass (Stover et al., 2007). The other was a 5-year-old loblolly pine (Pinus taeda) plantation in northern Florida near Sanderson. Both sites exhibited sandy soils in the near surface, the KSC being excessively well drained and the Sanderson site being moderately well drained.
In 2005, a SIR-2000 GPR system, manufactured by Geophysical Survey Systems, Inc. (GSSI; Salem, NH) was fitted with a custom-designed sampling rig that steadied the high-frequency antenna (model 5100, 1500 MHz antenna) and incorporated a survey wheel to meter electromagnetic pulses (Figure 30.1) to make measures at Sanderson. Measures at KSC were made with a SIR-3000, using a model 5100, 1500 MHz antenna fitted with a much smaller survey wheel than depicted in Figure 30.1. Before spatial distribution can be assessed, it is necessary to sample test transects to be able to correlate radar data with destructively sampled soil cores. With both systems, measures
were made by slowly drawing the survey rig along a measurement transect (200 scans per m) while ensuring that the antenna remained in contact with the soil surface. The locations of the verification cores were electronically marked on the data file during collection. The resulting scan was a two-dimensional profile (transect length by depth) where electromagnetic anomalies are located.
Postcollection processing was accomplished using RADAN 6.5 software (GSSI) with the following steps (Butnor et al., 2003; Cox et al., 2005; Stover et al., 2007):
1. Crop. Each radargram was cropped to the diameter of a 15 cm area soil core.
2. Background Removal. Root structures appear as hyperbolic reflectors, and parallel bands represent plane reflectors such as ground surface and soil layers. Parallel bands were removed with a horizontal finite impulse response filter (FIR) method called background removal.
3. Migration. Kirchoff migration was applied to correct the position of objects and collapse hyperbolic diffractions based on signal geometry (Daniels, 2004).
4. Hilbert Transformation. Hilbert transformation was applied to express the relationship between magnitude and the phase of the signal, allowing the phase of the signal to be reconstructed from its amplitude, allowing subtle discontinuities to be detected (Oppenheim and Schafer, 1975).
Root mass was quantified by converting radargrams to bitmap images and using Sigma Scan Pro Image Analysis software (Systat) with the following steps (Butnor et al., 2003; Cox et al., 2005; Stover et al., 2007):
1. Gray Scale. Each image was converted to an 8-bit grayscale image.
2. Intensity Threshold. Root mass was quantified by using pixel intensity (proxy for amplitude derived from Hilbert transform). Intensity is a relative measure ranging from 0 (black) to 255 (white). An intensity threshold range of 60 to 255 pixels, which was able to delineate roots as small as 0.5 cm was applied.
3. Statistical Analysis. Linear regression was performed to quantify the relationship between total root biomass from the soil cores and the GPR-derived index (pixels within the threshold range).
Once transects were scanned, extreme care was used to extract the cores in the correct location to correspond to the radargram. The cores were screened, and the biomass was separated into live roots and dead organic debris, dried and weighed.
Was this article helpful?