Ground penetrating radar (GPR) operates by sending out an electromagnetic wave pulse into the ground that reflects off materials with contrasting electrical properties (Figure 1) (Weymouth 1986:371; Conyers and Goodman 1997:23). This is related primarily to the electrical conductivity and magnetic permeability of the materials (Conyers and Goodman 1997:32). Relative dielectric permittivity (RDP), the ability of a material to store and pass a magnetic field, is the accepted property used to describe the materials. RDP (K) ranges from 1 for air to 81 for water and is expressed by K = c2 / V 2, where c is the velocity of light and V is the velocity of the wave (Conyers and Goodman 1997:33; Reynolds 1997:689). For soils, the RDP ranges from 3 from the driest sand to 40 for saturated clay. The strength of the reflection is proportional to the difference in RDP of the two materials and relies on an abrupt change between the materials (Conyers and Goodman 1997:34; Geophysical Survey Systems Inc. 1999:36). A contrast in RDP as small as 1 can cause a reflection in some cases (Geophysical Survey Systems Inc. 1999:31).
Furthermore, the travel time of the interaction is recorded as a matter of course in GPR surveys and this can be related to the depth of the target. When a radar wave is bounced off a subsurface reflector, the total travel time is recorded in nanoseconds (ns). This time is directly proportional to the depth of that target. Therefore, if the RDP is known for the medium, the depth can be found. RDP is difficult to determine accurately in the field, but can be estimated by several methods (Conyers and Goodman 1997:32; Geophysical Survey Systems Inc. 1999:79). One commonly used technique is geometric scaling in which a curve is fit to the properties of hyperbolic reflections in the data generated by strong reflectors. Because of the geometry of reflectance as the antenna passes over a target, the reflection will be expressed as a hyperbola and the width of that hyperbola is determined by the dielectric permittivity of the soil (Geophysical Survey Systems Inc. 199:83).
An interface is visible if the electrical properties of two substances contrast enough to produce a reflection. The magnitude of the reflection depends on the amount of contrast in the dielectric properties of the materials at an interface. This characteristic of GPR can contribute substantially to the study of stratigraphy. For example, a sand layer overlying a packed clay floor, a buried stone wall, or an air filled cavity will be likely produce a measurable reflection.
GPR antennas are available in various center frequencies, usually between 100 MHz and 1500 Mhz, which are related to the optimum depth of propagation and the resolution of the signal (Geophysical Survey Systems Inc. 1999:51). In general, lower frequency antennas propagate energy to greater depths. However, the vertical resolution also decreases (Geophysical Survey Systems Inc. 1999:56). For example, low frequency antennas can penetrate as far as 50 meters in ideal circumstances. In contrast, a 1000 Mhz antenna may only penetrate to about .5 meters, but can resolve features to a thickness of a centimeter (Geophysical Survey Systems Inc. 1999:52). A 400 MHz antenna is often used in archaeological applications because of the intermediate depth abilities. For all frequencies of antenna, a cone of energy is sent out that is roughly 90 degrees from front to back and 60 degrees from side to side (Geophysical Survey Systems Inc. 1999:45).
Limitations in GPR are related to the mechanics of sending electromagnetic energy through materials with high dielectric values, such as clayey soil (Reynolds 1997:688). Such soils cause the electromagnetic energy to attenuate at shallower depths as a result of the dispersion of the energy (Conyers and Goodman 1997:55). Attenuation causes the resultant data to be blurry when viewed and returns from even strong reflectors can be obscured. Wetter soils, often including clays, and high salinity materials are not ideal conditions for GPR survey. Dry sand, however, can often produce exceptional results.
Limitations in GPR are related to the mechanics of sending electromagnetic energy through materials with high dielectric values, such as clayey soil (Reynolds 1997:688). Such soils cause the electromagnetic energy to attenuate at shallower depths as a result of the dispersion of the energy (Conyers and Goodman 1997:55). Attenuation causes the resultant data to be blurry when viewed and returns from even strong reflectors can be obscured. Wetter soils, often including clays, and high salinity materials are not ideal conditions for GPR survey. Dry sand, however, can often produce exceptional results.
The data processing that is necessary in order that GPR data be used to its maximum potential by archaeologists is more involved than any of the other geophysical methods. Analysis begins by locating targets in the radar profiles, estimating the average RDP, and estimating the depth to targets. In the radar profiles, the amplitude of a reflection is positive if a high dielectric medium is encountered below a lower dielectric medium and negative when the reverse occurs. A strong narrow reflector will often produce an anomaly alternating between signs in a hyperbolic shape. Further processing is somewhat complex and includes creating plan view amplitude slice maps and three dimensional data cubes. Usually, the amplitudes are squared so that strong positive or negative anomalies appear the same.