In practice, GPR measurements can be made by towing the antennas continuously over the ground, or at discreet points along the surface. These two modes of operation are illustrated in Figure 7.5. The fixed-mode antenna arrangement consists of moving antennas independently to different points and making discrete measurements. The moving-mode arrangement keeps the transmit and receive antennas at a fixed distance with the antenna pair moved along the surface by pulling them by hand or with a vehicle. Transmit and receive antennas are moved independently in the fixed mode of operation. This allows more flexibility of field operation than when transmit and receive antennas are contained in a single box. For example, different polarization components can be recorded easily when transmit and receive antennas are separate. In the fixed mode of operation, a trace is recorded at each discrete position of transmit and receive antennas through the following sequence of events in the GPR system: (1) a wave is transmitted, (2) the receiver is turned on to receive and record the received signals, and (3) after a certain period of time the receiver is turned off. The resulting measurements recorded during the period of time that the receiver is turned on are called a trace, and the spacing between measurement points is called the trace spacing. The chosen trace spacing should be a function of the target size and the objectives of the survey. Traces displayed side by side form a GPR time-distance record, or GPR cross section, which shows how the reflections vary in the subsurface. If the contrasts in electrical properties (e.g., changes in permittivity) are relatively simple, then the GPR time-distance record can be viewed as a two-dimensional pseudoimage of the earth, with the horizontal axis the distance along the surface, and the vertical axis the two-way travel time of the radar wave. The two-way travel time on the vertical axis can be converted to depth, if the permittivity (which can be converted to velocity) is known. The GPR time-distance record is the simplest display of GPR data that can be interpreted in terms of subsurface features. A GPR time-distance record can also be produced by making a series of fixed-mode measurements at a constant interval between traces on the surface.
Field procedures for GPR measurements follow the basic design that must be followed for most other geophysical methods. The field procedures for a specific objective are determined by answering a number of questions, including the following:
• What are the objectives of the survey?
• What is the nature of the subsurface environment?
• What are the electrical properties of the materials at the site?
• What is the nature of the site access?
• How will cultural features affect the measurements?
The fundamental questions that must be asked prior to any survey are as follows: (1) What is the maximum depth of penetration?, (2) What are the line and trace spacing (horizontal resolution)?, and (3) What is the vertical resolution needed to achieve the goals of a survey?
The survey objectives dictate the depth of investigation, the lateral resolution that must be achieved, and the orientation of the antennas. However, the depth of penetration and resolution are also determined by the electrical properties of the material that contains the objects that are the targets of the survey. These combined factors help to establish the operating frequencies of the antenna that must be used, and the optimum spacing between measurements on the surface. A good rule of thumb for establishing adaquate trace measurement spacing is that the spacing should be less than one quarter of the size of the smallest object that is to be detected by the survey. This value is twice the Nyquist sampling frequency and should be adequate for most situations. Contrary to popular statements, it is impossible to "oversample": oversampling is a myth propagated by lazy people, which is the spatial analog to the popular myths of "overstudying" for exams, or being "overeducated." Also, the rule that "less is best" when it comes to determining the trace measurement spacing is erroneous. Economics and other practical considerations (e.g., spatial survey accuracy) are the only limitations to using a very small measurement spacing between traces and lines.
Surveys should be designed so they record a two-way travel time that is twice the amount of time required to see the deepest object anticipated in the survey area. There is nothing more frustrating than to run a survey and discover that the objects of interest were below (beyond, in time) the data that were recorded. The depth of penetration of a radar wave depends upon the electrical properties and the center-band frequency of the antenna. The theoretical depth of penetration for a given antenna frequency can be computed from the equation for the skin depth, if the values of conductivity and permittivity are known. However, from a practical point of view, the only way to know the depth of penetration for a particular survey site is to make an initial guess based on past experience and to determine the actual depth of penetration by testing at the site. Dry homogenous rocks and soils (e.g., beach sand, nonvuggy limestone, granite, etc.), permafrost regions, lignite, and peat bogs, can yield penetration depths up to 20 m, and sometimes even greater depths. In contrast, the normal penetration depth for most soils is on the order of 1 to 3 m, with a low range of a few centimeters in a soil that is predominantly montmorillonite clay, up to several tens of meters for a clean sand.
There are no fixed rules for determining the optimum antenna frequency range that should be used for a given survey, but the choice of antenna frequency should be based on the survey objectives and the electrical properties at the site. Because the depth of penetration decreases as the centerband frequency increases, the choice of antenna frequency is often determined by the depth of penetration that is needed. Lower frequencies generally improve the depth of penetration for most soil and rock types. However, a lower frequency has an adverse effect on the resolution of the radar wave for shallow investigations. A high-frequency antenna (500 MHz) would provide a detailed image of very shallow features, but it would not penetrate very deeply. A low-frequency antenna (50 MHz) would significantly improve the depth of penetration of the radar wave, but the resolution of very shallow features would be less distinct. The center-band frequency of the antenna should be computed so that the wavelength is smaller than one half the size of the smallest target. However, this does not mean that you will necessarily miss objects that are smaller than twice the size of the wavelength, because scattering depends upon a number of factors in addition to target size.
The size of the site, the access to the survey area, and the nature of surface features at the site influence the location and spacing of measurements. A large site area may necessitate the use of a vehicle to tow the antennas and preclude the possibility of making fixed-mode measurements. Conversely, site access problems (e.g., confined space) may make it necessary to make fixed-mode measurements. Surface metallic objects (e.g., fences), buildings, overhead utilities, and underground objects may influence both the frequency of the antenna that is selected and the location and spacing of measurements. For example, a small confined area containing a lot of cultural features above the surface may require the use of a high-frequency antenna that can be shielded from scattered energy above the surface.
Finally, the desired detail of the output display should strongly influence the survey design. If a three-dimensional pseudo-image is the desired output, then it is necessary to make measurements at a very close spacing. However, if the objective is to simply detect a large object, then measurements at a wide spacing may be adequate.
Filtering GPR raw field data is a necessary step to obtaining good data, and all field recording systems involve some type of analog and digital filtering. This is clearly illustrated by the three traces shown in Figure 7.7. The first trace is the raw field data, the second trace is the trace with a low-cut filter, the third trace has had a band-pass filter applied, and the fourth trace is that with band-pass filtering and amplification. The raw field data (Figure 7.7) do not even look like they contain any useful information. The low-cut frequency filter, which removes the low-frequency components, improves the appearance of the trace so that it looks like a GPR trace with positive and negative polarities situated properly above and below the zero line. The process of applying a low-cut filter in the early stages of processing is call "de-wowing" by GPR processing practitioners. The final trace filtering step removes the externally generated high-frequency noise that may be present on the data, and removes more of the low-frequency components beyond the basic de-wow filtering. Energy that occurs on a trace at later times can be enhanced by applying amplification to the trace that increases with time. This amplitude increase as a function of time is called gain. The output trace after applying these two filtering stages and gain is data that can be used for display and interpretation.
Static corrections consist of applying a time correction to data measured at a different distance from the object, when the difference in distance is caused by changes in elevation on the surface. The static correction simply consists of subtracting the travel time for each trace, using the elevation and velocity of the surface material, as follows: ts = 2d/v, where ts is the two-way static time correction and d is the elevation between the baseline and the elevation where the trace is measured, as defined in Figure 7.8. Figure 7.8 also illustrates applying static corrections to field data where the elevation is changing.
7.4.4 Velocity Analysis and Two-Dimensional Filtering
One of the primary features of GPR is the fact that data can be displayed, processed, and interpreted in two and three spatial dimensions. Data are commonly measured along a line to create a cross section that is analogous to a seismic cross section. The vertical axis is the two-way travel time of the pulse that was transmitted and received at each position along the line. If the velocity of the host
Log power Spectrum
Log power Spectrum
Was this article helpful?