GPR
|
GPR
|
|
Civil and Environmental Engineering 540 - Term Paper Brigham Young University Fall Semester 2000 Table of Contents1.1
The
History of Ground Penetrating Radar (GPR)
1.2
An
Overview of GPR Technology
1.2.4
A Note on GPR Penetration Depth
1.3
Applications
of GPR in Characterizing Hydrogeologic Conditions
2.2.1
Procedure for Determining Shallow Subsurface Geology
2.2.2
Procedure for Mapping Contaminant Plumes
2.2.3
Procedure for Determining Hydrogeologic Conditions
3.1
Determining
Shallow Subsurface Geology
3.2
Mapping
Contaminant Plumes
3.3
Determining
Hydrogeologic Conditions
AbstractGround-penetrating radar (GPR) has been used for several years as a non-destructive method of locating subsurface anomalies. In this paper, the history of GPR and methods used in GPR surveys are first discussed. Next, the applications of GPR for assessing shallow hydrogeologic conditions are presented. Procedures similar to those used in seismic reflection surveys can be used to obtain, process, visualize, and interpret GPR data. Case studies have been conducted that show how GPR data can be used for determining shallow subsurface geology, mapping contaminant plumes, determining the soil water content, and locating the water table. 1 IntroductionWhen people think of radar (radio detection and ranging), they probably think of a military naval vessel scanning the skies and the seas for other military aircraft or vessels. Alternatively, they think of an airport control tower scanning the skies for incoming aircraft. On the other hand, people may think of the local Doppler radar weather forecast on the evening news. Nevertheless, people seldom think of scanning the subsurface for structures, cavities, and other anomalies using radar. But just as anomalies on the surface can be scanned using radar, anomalies in the subsurface can also be detected. A radar transmitter sends a radio signal from an antenna into the surrounding space. A receiver picks up reflections received from this radio signal, and the strength and direction of the reflected signal gives the size and distance to the reflecting object. For example, if a battleship wants to spot enemy aircraft in the area, the ship could switch on its radar system. The antenna for the radar would rotate, sending out short, intense pulses of energy. After each pulse of energy, a receiver antenna would pick up all the reflections off surrounding aircraft, exciting electrons and showing a blip on a monitor for each vessel (from Complete Computer Solutions, web site). Ground Penetrating Radar, or GPR, works much the same way as conventional radar. Some basic theory about how GPR works is described in Benson (1995). GPR sends out pulses of energy between 10 and 1000 MHz into the ground from an antenna, and then listens for reflections for a small amount of time. GPR is often compared with seismic reflection surveys. Just as seismic reflections are generated when a seismic wave hits a layer in the subsurface with different material properties, GPR reflections are generated when a pulse hits an object or layer with different electromagnetic characteristics. Objects with different electromagnetic characteristics may be buried tanks, sedimentary layers, the water table, or the boundary of a contaminant plume. Essentially, a reflection occurs when there is an increase in the dielectric constant of materials in the subsurface. The dielectric constant is defined as the capacity of a material to store a charge when an electric field is applied relative to the same capacity in a vacuum, and can be computed using Equation 1: Where: er = The relative dielectric constant, c = The speed of light (30 cm/nanosecond), and v = The velocity of electromagnetic energy passing through the material. So if a GPR pulse hits a layer or object with a different dielectric constant, the pulse is reflected back, picked up by the receiving antenna, and the time and magnitude of that pulse is recorded, as in Figure 1. In many cases, the transmitting and receiving antennas are the same.
Figure 1: GPR schematic, from EPA (from Environmental Protection Agency, web site). As a GPR transmitter is moved along the ground at a constant velocity, it can take a “picture” of the shallow subsurface. The GPR continuously records reflection profiles as the transmitter is moved along the surface, as shown in Figure 2.
Figure 2: A 2D GPR profile, from Benson (1995). This paper discusses the history and applications of using GPR for assessing in situ hydrogeologic conditions. Methods for processing and interpreting GPR data are discussed in detail, and case studies where these methods were used are presented. To begin with, the history and basic theory behind using GPR data are discussed before describing some applications of GPR. 1.1
The
History of Ground Penetrating Radar (GPR)
James Clerk Maxwell in 1864 and Heinrich Hertz in 1886 developed the basic theory behind electromagnetic waves and their reflections. But it was not until 1924 that the British physicist Sir Edward Victor Appleton estimated the height of the ionosphere (a layer in the upper atmosphere that reflects long radio waves) using basic electromagnetic reflection principles. Then, in 1935, the British physicist Sir Robert Watson-Watt developed the first practical radar system. By the time the British entered World War II, they had constructed a network of radar systems along their south and east coasts to detect enemy vessels. (Calligeros, Hehir, and Jacobs, web site) According to Gary R. Olhoeft (see Olhoeft, web site), a GPR survey was first performed by the German geophysicist W. Stern in 1929. But GPR was largely forgotten until the late 1950’s when the radar systems in US Air Force planes saw through ice in Greenland, causing them to misread their altitude and crash into the ice. In 1960, John C. Cook made the first proposal for using radar to detect subsurface reflections in his article “Proposed monocycle-pulse, VHF radar for airborne ice and snow measurements” (Cook, 1960). Cook and others continued to develop radar systems to detect reflections beneath the ground surface (Moffatt and Puskar, 1976). One of the original and most promising ground penetrating radars was presented by Moffatt and Puskar (1976). Their system used an improved antenna that gave a better target-to-clutter ratio and was able to more accurately detect important subsurface reflections. Moffatt and Puskar used their system for several applications. With their GPR unit, they estimated the location of an underground tunnel, a fault, and mines. They also attempted to detect the variation of moisture content in subsurface soils. Their conclusion was that GPR is a useful tool for detecting anomalies and variations in subsurface rocks and soils. Moffatt and Puskar (1976) also presented some basic GPR theory and computations for computing subsurface wave velocities. Ulriksen (1982) and other scientists described better methods of processing and analyzing subsurface GPR data. Then, Wyatt, Waddell, and Sexton (1996) published a brief list of articles describing methods of obtaining, processing, and analyzing GPR data. Today, varied references exist that cover topics ranging from building GPR units, obtaining GPR data, processing GPR data, and analyzing GPR data. Some technologies have emerged in the past ten years that give GPR users better methods of processing and analyzing the GPR data than were available before. One of these technologies is the ability to visualize GPR data in three dimensions, with the ability to add time as a fourth dimension. Adding a time component to the visualization allows scientists to visualize the movement of contaminant plumes in the subsurface. Brewster and Annen (1994) were among the first to visualize GPR results in three dimensions, and Birken and Versteeg (2000) continued research in this area. 1.2
An
Overview of GPR Technology
A handful of companies exist that specialize in GPR data acquisition and processing. One such company is Geophysical Survey Systems, Inc. (see web site). A person interested in performing a GPR survey can first purchase a GPR system from one of these companies. Then, procedures described in this paper can be used to obtain, process, and interpret the GPR data. Other geophysical methods, such as seismic methods, borehole geophysical methods, cone penetrometer, and direct sampling methods can be used in conjunction with GPR results to fully characterize the subsurface (Atekwanna, Sauck, and Werkema, 2000). This section will describe the procedures used to obtain, process, visualize, and interpret GPR data. 1.2.1 Obtaining GPR DataThere are several different methods for obtaining GPR data. One of the most common is to drag a GPR unit along the ground or tow a GPR unit behind a vehicle, as in Figure 3 .
Figure 3: A GPR antenna towed behind an ATV (from Environmental Protection Agency, web site). As the GPR unit moves along the survey line, pulses of energy are transmitted from the transmission antenna and reflected to the receiving antenna (the transmission and reception antennas can be the same). The receiving antenna sends the received signal to a recorder. The data are recorded on a visual readout, a paper chart, a computer, or any combination of the three. The two most important components to consider when obtaining GPR data are the type of transmission and receiving antennas used and the available frequency range for the electromagnetic pulse from the GPR unit. Moffatt and Puskar (1976) described a new 4-foot antenna configuration that gave a better signal-to-noise ratio that the 6-foot antennas used up to that time. They used two different GPR configurations, one with a 250-picosecond pulse repeated 250 times per second, the other with a 45-nanosecond pulse repeated up to 1 million times per second. Their pioneering work in GPR data acquisition and analysis provided useful tools. Guy and others (1999) improved the original co-pole (parallel transmitting and receiving antennas) antenna configuration. They developed a cross-pole antenna configuration with orthogonal transmitting and receiving antennas. Their configuration reduced clutter and more effectively focused on the subsurface targets of interest. Benson (1995) describes the importance of choosing a GPR unit with the correct frequency. Energy is lost as the electromagnetic radar signal passes from the transmitting antenna through the subsurface. Eventually, all energy is lost. After a certain depth, the reflecting signals do not return to the surface. Higher signal frequencies provide high subsurface resolutions, but only penetrate to shallow depths. Low signal frequencies provide low resolution, but can penetrate to depths of 10 meters or more (Beres and Haeni, 1991). Typically, signal losses are high in soils having high water content and in lake or river surveys. But Ulriksen (1982) discovered that in some conditions, a 0.5-meter depth of water below the GPR unit (when taking a GPR survey in a lake) could enhance the GPR signal and act as a reflecting lens to illuminate subsurface features. Given the frequency of the GPR unit (many have variable frequencies), the velocity of the electromagnetic signal can be determined from equation 2, and the signal’s wavelength can be determined from equation 3:
Where: v = The velocity of the wave through the subsurface material. c = The speed of light (30 cm/nanosecond). er = The relative dielectric constant.
Where: l = Wavelength. v = The velocity of the wave through the subsurface material. f = Frequency. The relative dielectric constant (also called relative dielectric permittivity) can be determined from testing or from the chart below .
Figure 4: Relative dielectric permittivity of selected materials (from Beres and Haeni, 1991). The resolution of a GPR is approximately equal to 1/3 to 1/2 of a wavelength of the GPR signal. If an engineer is looking for a small target, he should use the equations above to determine the required GPR frequency. If looking for a larger target, such as the location of the water table, he would use lower GPR unit frequencies to determine the location and depth of the target. In Benson’s (1995) description of the data acquisition process, he stated that GPR units capable of transmitting electromagnetic waves with frequencies between 10 and 1000 MHz existed, but units capable of frequencies up to 1.5 GHz were found at the web site of Geophysical Survey Systems, Incorporated (see web site). These high-frequency GPR units can be used for profiling bituminous and concrete pavements and for inspecting concrete bridges and other structures. Lower frequency GPR units can be used for detecting general trends in the subsurface, such as the location of the water table, the depth to bedrock, and the general structure of subsurface sediments. After determining the type of GPR system to use, how do you acquire survey data? Brewster and Annan (1994) discussed how they obtained data for their project. In their experiment, they released 770 liters of PCE into the center of a 9-meter by 9-meter aquifer surrounded by sheet pile walls. They surveyed the area using a GPR unit at several times during the course of the experiment. The layout of their survey was in a grid format, as shown in Figure 5 .
Figure 5: Layout of the 9 x 9 experimental cell, from Brewster and Annan (1994). Most likely, typical field surveys would not be as accurate as Brewster and Annan’s survey, but their grid-based survey pattern may be a good way to cover an area of interest. 1.2.2 Processing GPR DataAfter obtaining GPR data, this data must be processed. Processing GPR data involves modifying it so it is more easily visualized and interpreted. Since data obtained from GPR surveys is similar to data obtained from seismic reflection surveys, many of the same techniques used to process seismic data can be used to process GPR data. In many cases, such as that presented by Benson (1995), very little processing of the data is required to locate the target of interest. The processing techniques and some equations for these techniques, as well as a definition of each of the techniques, will be discussed later in this paper. Normally, processing the data involves the following steps: 1. Converting the data to a usable digital format. 2. Removal or minimization of direct and air waves from the data. 3. Amplitude adjustments to the data. 4. Gain adjustments to the data. 5. Static adjustments to the data—this involves removing the effects of changes in elevation and effects from leveling the GPR. 6. Filtering the data. 7. Velocity analysis. 8. Migration. After processing GPR data, the data is ready to be visualized. Methods for visualizing GPR data are discussed in the next section. 1.2.3 Visualizing GPR DataLuca Baradello describes three different methods of visualizing GPR data. An A-scan is a 1D representation of a single GPR profile (trace), a B-scan is a 2D representation of a series of GPR traces, and a C-scan is a 3D representation of a series of 2D traces, as shown in Figure 6 .
Figure 6: A-scan (1D), B-scan (2D), and C-scan (3D) images of GPR data (from Baradello, web site). Many times, the travel time of a GPR trace is displayed on one side of the axis, while this travel time is converted to depth and displayed on the other side of the axis, as in Figure 2 . Birken and Versteeg (2000) used the dataset created by Brewster and Annan (1994) to visualize GPR data in four dimensions (3D + time, similar to a motion picture). In their experiment, they took the data provided by Brewster and Annan and viewed the differences between each data time step to analyze the movement of a PCE contaminant plume. Such complex 3D-time plots are not necessary, however, since Brewster and Annan (1994) were able to determine the downward movement of the PCE plume using a 1D plot of a single radar trace over time, shown in Figure 7 .
Figure 7: The change of a single radar trace along line 4N in Brewster and Annan's experiment (1994, see Figure 5 above ) over time, from Brewster and Annan (1994). Any of the above-mentioned methods of visualizing data can be used. Different methods may be better for different applications. For example, for locating the groundwater table, a simple 2D representation of the data may be adequate. But for visualizing the movement of a contaminant plume, a 4D representation may be required. 1.2.4 A Note on GPR Penetration DepthNo matter how the data is processed and visualized, GPR units can only penetrate to a limited depth. As the electromagnetic pulses dissipate with depth, they eventually fade to nothingness. According to Benson (1995), the penetration depth of a GPR signal depends on: 1. The frequency of the GPR source signal. 2. The GPR antenna radiation efficiency. 3. The electrical properties of the subsurface materials. In a study by Beres and Haeni (1991) to determine applications of GPR for hydrogeologic studies, they obtained depths of GPR penetration between 20 and 70 feet with an 80-MHz (relatively low-frequency) antenna. Greater attenuation of the radar data occurred in areas with saturated and clay-based soils. 90-foot or greater penetration depths were obtained in areas of low electric conductivity (unsaturated sands and gravels). In conditions where deep ice or salt deposits exist, GPR can penetrate to depths of up to 5000 meters (Environmental Protection Agency, web site). Because the frequency and antenna radiation efficiency of GPR units can be controlled, it is necessary to consider these parameters when considering a GPR unit for characterizing hydrogeologic conditions. 1.3
Applications
of GPR in Characterizing Hydrogeologic Conditions
GPR data can be used to interpret and describe hydrogeologic data in three ways. First, GPR data can be used for determining geologic conditions in the shallow subsurface. Second, GPR data can be used for mapping the extent of contaminant plumes. Third, GPR data can be used for determining hydrogeologic conditions, such as depth to the water table. GPR surveys can be useful for determining geologic conditions in the shallow subsurface. Beres and Haeni (1991) conducted an interesting study in which they used extensive amounts of GPR data to characterize shallow subsurface geologic conditions at five Connecticut test sites. In their application of GPR data, they attempted to evaluate the bedding of sediments, depth to the water table, grain size variation, and geologic structures in the subsurface. They interpreted the GPR data, characterized the subsurface, and then compared their interpretations to borehole test logs along the survey lines. The borehole and GPR methods of sampling the subsurface gave about the same results. In this way, they used borehole data to verify their GPR interpretation. Van Overmeeren (1998) characterized unconsolidated sediments in much of The Netherlands in much the same way. Besides being able to determine geologic conditions in the subsurface using GPR surveys, these surveys can also be used for mapping contaminant plumes. There have been several studies where GPR data has been used to determine the location and extent of contaminant plumes. Benson (1995) mapped the extent of hydrocarbon contaminants from a leaky underground storage tank. Brewster and Annan (1994) injected PCE into the subsurface and monitored the movement of the PCE plume over time with high-resolution GPR surveys. Atekwanna, Sauck, and Werkema (2000) used GPR surveys in conjunction with other geophysical methods to determine the extent of a contaminant plume. Others (Birken and Versteeg, 2000 and Peters et al., 1994) have also used GPR data to characterize the extent of contaminant plumes. Finally, GPR data can be used for determining hydrogeologic conditions, such as depth to the water table. In many GPR surveys, it is easy to detect the location of the water table because of the strong contrast in relative dielectric constants between unsaturated and saturated materials (See Figure 4 for the magnitude of these contrasts). The location of the water table was easily detected in studies conducted by Benson (1995), Beres and Haeni (1991), and Mellett (1995). Greaves et al. (1996) were able to approximate the soil water content in soils beneath the surface from GPR survey data. The methods used for processing and interpreting GPR data will now be discussed. 2 MethodsAfter you have obtained GPR data for a particular purpose, how is this data processed? And what is this data good for? One useful way of viewing GPR data is to view a survey line of 1D traces, converting the two-way travel time to reflections to depth from the ground surface, as discussed in section 1.2.3 and as shown in Figure 8 . The hyperbolas in the figure are created from scattered reflections as the GPR unit approaches and moves away from reflection surfaces.
Figure 8: A GPR profile with the approximate depth of subsurface reflectors (from Daniel, 1993). While some GPR data can be left unprocessed, processing the data can be a vital key to visualizing what is beneath the surface. A few of the methods for processing and interpreting GPR data will be presented in this section. After a little bit of practice, these processing and interpretation techniques can be used in conjunction with geophysical, borehole, and other techniques to accurately characterize the subsurface. 2.1
Processing
GPR Data
In many cases, it is possible to use the results from a GPR survey with very little processing. In these cases, the only adjustments that need to be made are to convert the data to a usable digital format, to make gain adjustments to the data, and to determine the depth to each reflector (such as the water table) in the subsurface (this involves converting time to depth). In other cases, several techniques that are also used in seismic data processing (Smemoe, 1999) are used. Brewster and Annan (1994) released PCE into the subsurface and used extensive GPR surveys to monitor the migration of the PCE plume. In their experiment, they needed their data to be as accurate and correct as possible to completely monitor the migration of the plume through the subsurface. These data processing techniques have been developed through years of research. Though some of the simple data processing equations are presented here, many of them involve complex mathematical equations that are beyond the scope of this paper. Some of the equations are described in Smemoe (1999). References that go into a little more detail about how to process data include Olhoeft (2000), Roberts and Daniels (1997), Leuschen and Plumb (2000), and several other books and journal articles. Here are the steps required for processing GPR survey data: 1. Convert the data to a usable digital format. In most recently manufactured GPR units, the data is automatically recorded in digital format. In a survey by Benson (1995), the data was both plotted and recorded on a magnetic tape device for use on a computer. Data from GPR units purchased from Geophysical Survey Systems, Incorporated (see web site) can be uploaded to computers and processed with software available from the same company. Some of the GPR units are sophisticated enough to allow processing and printing the data while still in the field. There are lots of different file formats to work with, and there are also lots of utility programs available for converting between different file formats. 2. Removal or minimization of direct and air waves from the data. Many times, there are large reflection amplitudes at the interface between the air and soil surface immediately below the GPR antennas. The high contrast between the air and soil conductivities can create direct and air waves that may mask reflections from important objects just below the subsurface.
Figure 9: Direct and air waves. These direct and air waves can be removed by computing the time of arrival and wavelength of these waves, then by subtracting the theoretical wave along the width of the wavelength from the actual wave in each GPR trace. In many cases, such as Brewster and Annan (1994), these waves were not removed since they did not effect the interpretation of the GPR data. 3. Amplitude adjustments to the data. In many cases the battery on the GPR unit may wear down as each survey progresses. This results in GPR traces with lower and lower reflection amplitudes during the survey. Determining the battery power loss over time, then multiplying each trace by a constant value to recover these losses can correct this problem. 4. Gain adjustments to the data. As the transmitted signal from a GPR unit penetrates into the ground, attenuation of the GPR trace occurs. This attenuation can be corrected by applying gain adjustments to each trace. Several models exist for computing gain adjustments (Smemoe, 1999). In one model, each data value in the entire trace is multiplied by a factor related the depth of the signal:
Where: ti’ = The gain-corrected trace amplitude. a = The constant gain adjustment factor. zi = The depth of the trace. ti = The original trace amplitude. Since the above model tends to give less than realistic results, an exponential gain adjustment factor can be used to correct for attenuation losses:
Where: ti’ = The gain-corrected trace amplitude. ti = The original trace amplitude. a = The exponential gain adjustment factor. zi = The depth of the trace. An example of how gain corrections can correct for attenuation losses is shown by data from an experiment performed by Brewster and Annan (1994) in Figure 10 and Figure 11 .
Figure 10: GPR profile data before gain correction was applied to the data, from Brewster and Annan (1994).
Figure 11: GPR profile data after gain correction was applied to the data, from Brewster and Annan (1994). 5. Static adjustments to the data—these adjustments remove the effects of changes in elevation and effects from leveling the GPR antennas. The theory and equations behind these adjustments are beyond the scope of this paper and will not be presented here. 6. Filtering the data. The purpose of filtering is to remove unwanted background noise. For example, if a cellular phone antenna or power transmission line is located in the area, it may create unwanted noise along the trace at a certain frequency. To remove this unwanted noise, the time-domain trace data is converted to the frequency domain using the Fourier transform. Desired frequencies are zeroed out, then the trace is converted back to the time domain using the inverse Fourier transform. Olhoeft (2000) discusses filtering the results from GPR surveys. Filtering strategies can include band pass (removing frequencies in a certain range), low pass (removing low-frequency signals), and high pass filtering (removing high-frequency signals). 7. Velocity analysis. Velocity analysis involves determining the velocity of subsurface materials, then converting the reflection travel times to depths. This is most easily done using methods described by Benson (1995). In Benson’s approach, he determined the velocity of soils near the subsurface by testing for the relative dielectric constant and using the equation below : Where: v = The velocity of the wave through the subsurface material. c = The speed of light (30 cm/nanosecond). er = The relative dielectric constant. Then, the depth of each reflection in the subsurface (such as the depth to the water table) was determined from: Where: dr = The depth to the reflector. v = The velocity of the wave through the subsurface material. tr = The two-way travel time to the reflector (taken from the GPR trace). Greaves et al. (1996) used common midpoint (CMP) tests and more complex methods of determining variations in subsurface velocities. Their methods gave more accurate results for depths of subsurface reflectors, though the methods used by Benson (1995) are adequate for most analyses. 8. Migration—Migration is a procedure to transform surface-recorded GPR data into data with the subsurface heterogeneities located at the correct position (Smemoe, 1999, Brewster and Annan, 1994, Conyers and Goodman, 1997, and Olhoeft, 2000). For example, pre- and post-migrated data from an experiment by Brewster and Annan (1994) is shown in Figure 12 and Figure 13 . In Figure 12 , reflections from sheet pile walls along the boundary of the experiment show up as reflections in the lower corners of the figure. These reflections have been removed by migration in Figure 13 .
Figure 12: Pre-migrated data from an experiment by Brewster and Annan (1994).
Figure 13: Post-migrated data from an experiment by Brewster and Annan (1994). 9. Other processing techniques. Birken and Versteeg (2000) determined the differences between 3D datasets to process four-dimensional (3D over time) data for visualization. Stickley et al. (2000) used a unique approach where a receiver “gate” was used to pass only reflections from certain desirable depths and “blank” the unwanted reflections. 2.2
Interpreting
GPR Data
After GPR data has been processed, how can this data be interpreted? This section will present GPR data interpretation procedures for determining shallow subsurface geology, for mapping contaminant plumes, and for determining hydrogeologic conditions. These interpretations can then be combined with interpretations from borehole logs and other geophysical methods to accurately characterize the subsurface. 2.2.1 Procedure for Determining Shallow Subsurface GeologyFor the most part, determining the shallow subsurface geology is an art more than a science. In other words, it takes lots of practice before you get it right. In a GPR survey performed by Beres and Haeni (1991), they outline the procedure for determining shallow subsurface geology from GPR data. In their paper, they used the chart in Figure 14 to interpret the GPR surveys and use equations presented in Benson (1995, mentioned above ) for determining the depth to reflectors in the subsurface.
Figure 14: Chart for determining types of subsurface sediments (from Beres and Haeni, 1991). After interpreting the surveys, they used borehole and geophysical data in the area of the survey to verify and calibrate the results from the subsurface interpretation. In all cases of interpreting the data, the interpretations were calibrated with known subsurface data, compared to known subsurface data, or compared with data obtained using other geophysical methods. The following approach is recommended for determining shallow subsurface geology from GPR data: 1. Perform the GPR survey. 2. Process the data from the GPR survey. 3. Use Figure 14 for determining the subsurface geology. 4. Calibrate the subsurface geology obtained from the GPR data using borehole data, known subsurface geologic records, and other geophysical methods. 2.2.2 Procedure for Mapping Contaminant PlumesBesides being able to determine shallow subsurface geology from GPR data, there has been lots of research done on mapping the extent of contaminant plumes using GPR surveys. For example, Brewster and Annan (1994) performed an extensive GPR survey of a controlled PCE release over long period of time. The presence of PCE provides a reflecting surface similar to that of the water table in the area of contamination, though the reflection is less intense. The following approach, used by Brewster and Annan, is recommended to define the location of contaminant plumes in the subsurface: 1. Collect GPR data over a period of time. 2. Perform extensive processing on the data to correct for signal loss, migration, and other problems to make sure the data is as correct as possible. Processed GPR data toward the end of Brewster and Annan’s experiment is shown in Figure 15 .
Figure 15: A GPR profile collected 984 days after the start of a PCE release (from Brewster and Annan, 1994). 3. Interpret the data for the location of the contaminant pools and calibrate the location of these pools with other geophysical and/or borehole log data. One method of visualizing the movement of contaminants in the subsurface is to plot differences in the GPR surveys over time. This should reveal the movement of contaminant pools. Another method is to plot the changes in a single GPR trace over time, as shown in Figure 7 . 2.2.3 Procedure for Determining Hydrogeologic ConditionsGPR data can be useful in characterizing the hydrogeologic conditions at a site. Benson (1995) discussed a method for determining the depth to the water table. In his method, he first estimates the velocity of the electromagnetic GPR pulse through the unsaturated subsurface sediments according to the equation below (this equation was mentioned in the section describing the processing techniques for GPR data, I will repeat it again here):
Where: v = The velocity of the wave through the subsurface material. c = The speed of light (30 cm/nanosecond). er = The relative dielectric constant. In the velocity equation, the only unknown is the relative dielectric constant. The relative dielectric constant of a soil can be measured in the laboratory by comparing the capacitance of capacitors encased with air and capacitors encased with the soil (Hendricks, web site). Next, Benson interpreted the location of the water table reflection on a processed GPR profile. The depth to the water table was determined from:
Where: dw = The depth to the water table. v = The velocity of the wave through the subsurface material. tw = The two-way travel time to the water table (from GPR profile). Finally, the depth to the water table was compared with known depths to the water table from wells close to the GPR survey site. The procedure to determine the depth to the water table is about the same as the procedure to characterize the subsurface and to characterize contaminant plumes, with a few twists for each method. Greaves et al. (1996) were able to successfully estimate the water content in soils by making measurements and assumptions for the velocity of GPR pulses through soils along a GPR survey. 3 Case StudiesThe methods for processing GPR survey data, determining shallow subsurface geology, mapping contaminant plumes, and characterizing the hydrogeologic conditions at a site have been applied to real-world conditions. These applications have shown that GPR surveys can be a viable method for characterizing the subsurface. Most of the studies were performed in conjunction with other geophysical and borehole studies in the area surveyed. All of the studies recommend combining GPR data with geophysical and borehole data to obtain an accurate characterization of the subsurface. 3.1
Determining
Shallow Subsurface Geology
An overview of the work of Beres and Haeni (1991) and Van Overmeeren (1998) has been covered in section 1.3 . In their work, Beres and Haeni studied subsurface geologic conditions at five test sites in Connecticut. Using an 80-MHz antenna, they obtained a near-surface resolution (resolution = wavelength/2) of 1-2 feet with an antenna attached to a 75-foot rope behind a four wheel-drive vehicle. Their GPR data required very little data processing. After performing the GPR surveys, they used the chart in Figure 14 to interpret their data. Their interpretation was then compared with interpretations from borehole logs of the subsurface at each of the 5 survey locations. The work of Beres and Haeni produced the following conclusions: 1. High-quality profiles of the subsurface can be obtained with little or no data processing. 2. Penetration depth of a GPR survey can range between 20 and 70 feet, depending on the type of subsurface sediments. 3. Errors in analysis of borehole logs, incorrect radar velocity estimations, poor GPR resolution, interference between GPR reflectors, and other factors can give poor correlation between GPR characterization and borehole logs. 4. For the most part, GPR is a fast, economical method of characterizing the lithology (types of sediments and structures) of the subsurface. GPR may be less accurate in highly conductive deposits, such as deposits of saturated clays. Van Overmeeren (1998) conducted GPR surveys in several parts of the Netherlands over a period of several years. He consolidated the radar patterns into an atlas similar to that of Beres and Haeni, shown in Figure 16 . Using this atlas, subsurface structures in the Netherlands can be interpreted from GPR survey results.
Figure 16: Radar patterns of unconsolidated sediments in the Netherlands (from Van Overmeeren, 1998). 3.2
Mapping
Contaminant Plumes
Several researchers have successfully mapped contaminant plumes using GPR survey data. Benson (1995) conducted a GPR survey in an area of contaminated groundwater in north-central Arizona near Tuba City. In this area, at least one of four underground storage tanks was known to have leaked hydrocarbons, causing contamination of groundwater in an area surrounding the small gas station. Benson’s study of the area used a three-phase approach. First, a GPR study of the area was performed. Second, monitoring wells were installed in and surrounding the area of the interpreted plume. Water samples taken from the interpreted plume revealed concentrations of hydrocarbons with concentrations of 5 PPM (Benzene compounds) and 80 PPM (total petroleum hydrocarbons) higher than surrounding wells. The GPR survey successfully mapped the extent of the contaminant plume without the need for extensive use of monitoring wells. Brewster and Annan (1994) conducted a controlled release of 770 liters PCE into a 9-meter by 9-meter area of the subsurface. This area was surrounded by sheet pile walls and underlain by a low-permeability aquitard. They then performed a detailed GPR survey of the area using a 200 MHz GPR unit at certain time intervals over a period of several days. Since they had GPR data for the entire area at several times, they were able to identify the change in the GPR survey data over time and to detect the motion of the contaminant plume. After using extensive data processing techniques, they were able to fully characterize the motion of the DNAPL plume in the subsurface. Their conclusion was that it is fairly difficult to determine the location of DNAPL plumes in the subsurface. However, detailed surveys in conjunction with monitoring wells can determine the plume location. They also concluded that slow-moving or static DNAPL plumes may be difficult to detect from GPR surveys alone. Birken and Versteeg (2000) used ingenious methods to manipulate Brewster and Annan’s GPR data and characterize the extent of contaminant plumes in the subsurface. Atekwanna, Sauck, and Werkema (2000) used GPR measurements in conjunction with downhole resistivity measurements. Their study determined that the conductivity in the area of a liquid contaminant might actually be higher than the conductivity through uncontaminated water, depending on the hydrogeochemical conditions at the site. Atekwanna, Sauck, and Werkema stressed the importance of using other geophysical and borehole measurements at a site in addition to the GPR measurements. 3.3
Determining
Hydrogeologic Conditions
Benson’s (1995) work in Tuba City, Arizona was discussed previously. In this study, he was able to interpret the depth to the water table from his GPR measurements. In the Connecticut study by Beres and Haeni (1991), they successfully determined the depth to the water table. In studies where the water table was located at a reasonable depth, there was a clear reflection at the location of the water table. In these cases, the depth of the water table was easily determined. Greaves et al. (1996) estimated soil moisture content from GPR data using a detailed analysis of GPR wave velocities at their site of interest. 4 ConclusionsGPR methods were first used when radar methods were originally developed, but were largely forgotten. Cook (1960) renewed interest in using GPR surveys to monitor the environment. Since his research, extensive research has been performed and methods have been developed for processing and interpreting GPR data. GPR is now an established, viable technology for probing the shallow subsurface. Several companies exist which manufacture GPR units and antennas of various frequencies. These companies also provide software for processing and visualizing GPR data. GPR surveys have been used for remote sensing of the shallow subsurface for several projects. If GPR data are correctly processed and interpreted, then GPR can be used effectively to determine shallow subsurface geology, map contaminant plumes, and determine subsurface hydrogeologic conditions. ReferencesAtekwanna, E. A., Sauck, W. A., and Werkema, D. D. (2000). “Investigations of geoelectrical signatures at a hydrocarbon contaminated site.” J. of Applied Geophysics, 44(2), 167-180.
Baradello, L.
Acquisition and Processing of GPR Data. (http://web.interpuntonet.it/baradello/georadar_processing.html)
Benson, A. K. (1995). “Applications of ground penetrating radar in assessing some geological hazards: examples of groundwater contamination, faults, cavities.” J. of Applied Geophysics, 33(1-3), 177-193.
Beres, M., and Haeni, F. P. (1991). “Application of ground-penetrating-radar methods in hydrogeologic studies.” Ground Water, 29(3), 375-386.
Birken, R., and Versteeg, R. (2000). “Use of four-dimensional ground penetrating radar and advanced visualization methods to determine subsurface fluid migration.” J. of Applied Geophysics, 43(2-4), 215-226.
Calligeros, E., Hehir, D., and Jacobs, R. Invention of Radar. (http://murray.newcastle.edu.au/users/staff/eemf/ELEC351/SProjects/Calligeros/invent_radar.htm)
Complete Computer Solutions. Radar. (http://www.iao.com/howthing/radarhtm.htm)
Conyers, L. B., and Goodman, D. (1997). Ground-Penetrating Radar: An Introduction for Archaeologists. AltaMira Press.
Cook, John C. (1960). “Proposed monocycle-pulse, VHF radar for airborne ice and snow measurements.” AIEE Trans. Commun. And Electron., 79(2), 588-594.
Daniel, D. E., Editor. (1993). Geotechnical Practice for Waste Disposal. Chapman & Hall.
Geophysical Survey Systems, Inc. Geophysical Survey Systems, Inc. web site. (http://www.geophysical.com/)
Greaves, R. J., Lesmes, D. P., Lee, J. M., and Toksöz, M. N. (1996). “Velocity variations and water content estimated from multi-offset, ground-penetrating radar.” Geophysics, 61(3), 683-695.
Guy, E. D., Daniels, J. J., Radzevicius, S. J., and Vendl, M. A. (1999). “Demonstrating of using crossed dipole GPR antennae for site characterization.” Geophysical Res |
